diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,43681 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999541893810986,
+  "global_step": 7276,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.132420091324201e-08,
+      "loss": 0.874,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8264840182648401e-07,
+      "loss": 1.5194,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.73972602739726e-07,
+      "loss": 0.2726,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6529680365296803e-07,
+      "loss": 1.3258,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5662100456621004e-07,
+      "loss": 0.2823,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.47945205479452e-07,
+      "loss": 1.3791,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.39269406392694e-07,
+      "loss": 1.4914,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.305936073059361e-07,
+      "loss": 1.3354,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.219178082191781e-07,
+      "loss": 1.3604,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.132420091324201e-07,
+      "loss": 1.3179,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0045662100456622e-06,
+      "loss": 1.3774,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.095890410958904e-06,
+      "loss": 1.3958,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1872146118721462e-06,
+      "loss": 1.3657,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.278538812785388e-06,
+      "loss": 1.3133,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3698630136986302e-06,
+      "loss": 1.4062,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4611872146118721e-06,
+      "loss": 1.3197,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5525114155251142e-06,
+      "loss": 1.2872,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6438356164383561e-06,
+      "loss": 1.259,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7351598173515982e-06,
+      "loss": 1.2737,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8264840182648401e-06,
+      "loss": 0.2769,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9178082191780823e-06,
+      "loss": 1.2027,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0091324200913244e-06,
+      "loss": 1.1641,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.100456621004566e-06,
+      "loss": 1.1656,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.191780821917808e-06,
+      "loss": 1.1618,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2831050228310503e-06,
+      "loss": 1.1232,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3744292237442924e-06,
+      "loss": 1.1749,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4657534246575345e-06,
+      "loss": 1.0707,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.557077625570776e-06,
+      "loss": 1.2762,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6484018264840183e-06,
+      "loss": 1.1743,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7397260273972604e-06,
+      "loss": 1.0485,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8310502283105025e-06,
+      "loss": 1.1501,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9223744292237442e-06,
+      "loss": 1.0513,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0136986301369864e-06,
+      "loss": 1.1587,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1050228310502285e-06,
+      "loss": 1.1085,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1963470319634706e-06,
+      "loss": 1.1051,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2876712328767123e-06,
+      "loss": 1.0042,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3789954337899544e-06,
+      "loss": 1.0514,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4703196347031965e-06,
+      "loss": 1.1125,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5616438356164386e-06,
+      "loss": 1.0099,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6529680365296803e-06,
+      "loss": 1.0518,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7442922374429224e-06,
+      "loss": 1.0304,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8356164383561645e-06,
+      "loss": 0.9735,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.926940639269407e-06,
+      "loss": 1.0506,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.018264840182649e-06,
+      "loss": 1.0269,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.109589041095891e-06,
+      "loss": 0.9497,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.200913242009132e-06,
+      "loss": 1.0081,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.292237442922374e-06,
+      "loss": 1.1073,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.383561643835616e-06,
+      "loss": 1.0413,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4748858447488585e-06,
+      "loss": 1.0439,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.566210045662101e-06,
+      "loss": 1.0242,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.657534246575343e-06,
+      "loss": 1.0998,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.748858447488585e-06,
+      "loss": 1.0191,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.840182648401827e-06,
+      "loss": 1.0871,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.931506849315069e-06,
+      "loss": 1.0599,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.02283105022831e-06,
+      "loss": 1.0051,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.114155251141552e-06,
+      "loss": 1.0491,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2054794520547945e-06,
+      "loss": 0.9369,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.296803652968037e-06,
+      "loss": 1.0072,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.388127853881279e-06,
+      "loss": 1.0835,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.479452054794521e-06,
+      "loss": 0.9565,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.570776255707763e-06,
+      "loss": 0.9364,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.662100456621005e-06,
+      "loss": 0.9851,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.753424657534246e-06,
+      "loss": 1.0406,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8447488584474885e-06,
+      "loss": 1.0275,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.936073059360731e-06,
+      "loss": 0.9688,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.027397260273973e-06,
+      "loss": 0.9552,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.118721461187215e-06,
+      "loss": 1.0275,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.210045662100457e-06,
+      "loss": 1.0059,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.301369863013699e-06,
+      "loss": 0.9555,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.392694063926941e-06,
+      "loss": 0.9981,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.484018264840182e-06,
+      "loss": 0.9552,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5753424657534245e-06,
+      "loss": 0.9351,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.9392,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.757990867579909e-06,
+      "loss": 0.8819,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.849315068493151e-06,
+      "loss": 0.8873,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.940639269406393e-06,
+      "loss": 0.3196,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.031963470319635e-06,
+      "loss": 1.0012,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.123287671232877e-06,
+      "loss": 1.0022,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.214611872146119e-06,
+      "loss": 0.9389,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.305936073059361e-06,
+      "loss": 0.9975,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.397260273972603e-06,
+      "loss": 0.9306,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.488584474885845e-06,
+      "loss": 0.9856,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.579908675799087e-06,
+      "loss": 1.0381,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.671232876712329e-06,
+      "loss": 0.3162,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.76255707762557e-06,
+      "loss": 1.0076,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.853881278538813e-06,
+      "loss": 0.9634,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.945205479452055e-06,
+      "loss": 1.0081,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.036529680365297e-06,
+      "loss": 0.9938,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.127853881278539e-06,
+      "loss": 0.9217,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.219178082191782e-06,
+      "loss": 0.9448,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.310502283105023e-06,
+      "loss": 1.0198,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.401826484018264e-06,
+      "loss": 0.2204,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.493150684931507e-06,
+      "loss": 0.9776,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.584474885844748e-06,
+      "loss": 0.9605,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.675799086757991e-06,
+      "loss": 1.0017,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.767123287671233e-06,
+      "loss": 0.9403,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.858447488584476e-06,
+      "loss": 0.9094,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.949771689497717e-06,
+      "loss": 0.9006,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.04109589041096e-06,
+      "loss": 0.9836,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.132420091324201e-06,
+      "loss": 0.2847,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.223744292237442e-06,
+      "loss": 0.955,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.315068493150685e-06,
+      "loss": 0.986,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.406392694063927e-06,
+      "loss": 0.9592,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.49771689497717e-06,
+      "loss": 1.0039,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.589041095890411e-06,
+      "loss": 0.9751,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.680365296803654e-06,
+      "loss": 1.0287,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.771689497716895e-06,
+      "loss": 0.9436,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.863013698630138e-06,
+      "loss": 0.9834,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.95433789954338e-06,
+      "loss": 0.3442,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.004566210045662e-05,
+      "loss": 0.9136,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0136986301369864e-05,
+      "loss": 0.8825,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0228310502283105e-05,
+      "loss": 0.9029,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0319634703196348e-05,
+      "loss": 0.9687,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0410958904109589e-05,
+      "loss": 0.8392,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0502283105022832e-05,
+      "loss": 0.8895,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0593607305936073e-05,
+      "loss": 0.981,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0684931506849316e-05,
+      "loss": 0.9634,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0776255707762558e-05,
+      "loss": 0.9592,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.08675799086758e-05,
+      "loss": 0.9956,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0958904109589042e-05,
+      "loss": 1.0058,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1050228310502285e-05,
+      "loss": 0.9535,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1141552511415526e-05,
+      "loss": 1.0642,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1232876712328769e-05,
+      "loss": 1.0373,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.132420091324201e-05,
+      "loss": 0.876,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1415525114155253e-05,
+      "loss": 1.0255,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1506849315068493e-05,
+      "loss": 0.9915,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1598173515981737e-05,
+      "loss": 0.9634,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1689497716894977e-05,
+      "loss": 0.9478,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.178082191780822e-05,
+      "loss": 0.9929,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1872146118721461e-05,
+      "loss": 0.9173,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1963470319634704e-05,
+      "loss": 1.049,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2054794520547945e-05,
+      "loss": 0.9159,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2146118721461188e-05,
+      "loss": 0.9117,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.223744292237443e-05,
+      "loss": 0.9871,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2328767123287673e-05,
+      "loss": 0.9677,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2420091324200914e-05,
+      "loss": 0.8851,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2511415525114157e-05,
+      "loss": 0.9687,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2602739726027398e-05,
+      "loss": 0.3222,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2694063926940641e-05,
+      "loss": 1.0427,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2785388127853882e-05,
+      "loss": 0.9197,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2876712328767125e-05,
+      "loss": 0.9551,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2968036529680365e-05,
+      "loss": 0.9347,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.305936073059361e-05,
+      "loss": 0.9994,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3150684931506849e-05,
+      "loss": 0.9873,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3242009132420092e-05,
+      "loss": 0.9247,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.8699,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3424657534246576e-05,
+      "loss": 1.0492,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3515981735159818e-05,
+      "loss": 0.923,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.360730593607306e-05,
+      "loss": 0.9798,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3698630136986302e-05,
+      "loss": 0.9914,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3789954337899545e-05,
+      "loss": 0.9667,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3881278538812786e-05,
+      "loss": 0.9991,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3972602739726029e-05,
+      "loss": 0.8783,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.406392694063927e-05,
+      "loss": 0.9542,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4155251141552513e-05,
+      "loss": 0.9302,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4246575342465754e-05,
+      "loss": 0.8925,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4337899543378997e-05,
+      "loss": 0.9894,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4429223744292239e-05,
+      "loss": 0.8562,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4520547945205482e-05,
+      "loss": 0.9165,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4611872146118721e-05,
+      "loss": 0.8943,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4703196347031964e-05,
+      "loss": 0.8826,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4794520547945205e-05,
+      "loss": 0.944,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4885844748858448e-05,
+      "loss": 0.9847,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.497716894977169e-05,
+      "loss": 0.9199,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5068493150684933e-05,
+      "loss": 0.9666,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5159817351598174e-05,
+      "loss": 0.9177,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5251141552511417e-05,
+      "loss": 0.9285,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5342465753424658e-05,
+      "loss": 0.3016,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.54337899543379e-05,
+      "loss": 0.3286,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.552511415525114e-05,
+      "loss": 1.0207,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5616438356164384e-05,
+      "loss": 0.8958,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5707762557077627e-05,
+      "loss": 0.9057,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.579908675799087e-05,
+      "loss": 0.9494,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.589041095890411e-05,
+      "loss": 0.9307,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5981735159817352e-05,
+      "loss": 0.3156,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6073059360730595e-05,
+      "loss": 0.9161,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6164383561643838e-05,
+      "loss": 0.9794,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6255707762557078e-05,
+      "loss": 0.9652,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.634703196347032e-05,
+      "loss": 0.9199,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6438356164383563e-05,
+      "loss": 0.301,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6529680365296806e-05,
+      "loss": 0.9788,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6621004566210046e-05,
+      "loss": 1.0257,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.671232876712329e-05,
+      "loss": 0.9169,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.680365296803653e-05,
+      "loss": 0.8666,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6894977168949775e-05,
+      "loss": 0.8678,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6986301369863014e-05,
+      "loss": 0.9237,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7077625570776257e-05,
+      "loss": 0.9244,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7168949771689497e-05,
+      "loss": 0.9748,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.726027397260274e-05,
+      "loss": 0.9131,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7351598173515983e-05,
+      "loss": 0.9306,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7442922374429226e-05,
+      "loss": 0.2928,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7534246575342465e-05,
+      "loss": 0.9205,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.762557077625571e-05,
+      "loss": 0.976,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.771689497716895e-05,
+      "loss": 1.008,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7808219178082194e-05,
+      "loss": 0.9653,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7899543378995434e-05,
+      "loss": 0.9814,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7990867579908677e-05,
+      "loss": 0.9534,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.808219178082192e-05,
+      "loss": 0.8821,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8173515981735163e-05,
+      "loss": 0.873,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8264840182648402e-05,
+      "loss": 0.8825,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8356164383561645e-05,
+      "loss": 0.9867,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8447488584474885e-05,
+      "loss": 0.9161,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8538812785388128e-05,
+      "loss": 0.3177,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.863013698630137e-05,
+      "loss": 0.9515,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8721461187214614e-05,
+      "loss": 0.9806,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8812785388127853e-05,
+      "loss": 0.9901,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8904109589041096e-05,
+      "loss": 0.9653,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.899543378995434e-05,
+      "loss": 0.9719,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9086757990867582e-05,
+      "loss": 0.9334,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9178082191780822e-05,
+      "loss": 0.9504,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9269406392694065e-05,
+      "loss": 0.972,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9360730593607308e-05,
+      "loss": 0.8793,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.945205479452055e-05,
+      "loss": 0.9694,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.954337899543379e-05,
+      "loss": 0.9798,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9634703196347033e-05,
+      "loss": 0.9246,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9726027397260276e-05,
+      "loss": 0.8494,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.981735159817352e-05,
+      "loss": 0.9009,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.990867579908676e-05,
+      "loss": 0.9897,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.2962,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999900910073e-05,
+      "loss": 0.9757,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999996036403125e-05,
+      "loss": 0.9454,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999991081907764e-05,
+      "loss": 0.8544,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999984145615637e-05,
+      "loss": 0.9718,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999975227528113e-05,
+      "loss": 0.9491,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999964327646967e-05,
+      "loss": 0.9328,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999951445974347e-05,
+      "loss": 0.9123,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999936582512818e-05,
+      "loss": 0.9722,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999919737265317e-05,
+      "loss": 0.9141,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999990091023519e-05,
+      "loss": 0.9348,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999880101426165e-05,
+      "loss": 0.8962,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999857310842362e-05,
+      "loss": 0.9895,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99998325384883e-05,
+      "loss": 0.2807,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999980578436889e-05,
+      "loss": 0.9034,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999777048489433e-05,
+      "loss": 0.9505,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999746330855623e-05,
+      "loss": 0.877,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999713631473548e-05,
+      "loss": 0.2995,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999678950349694e-05,
+      "loss": 0.9641,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999642287490926e-05,
+      "loss": 0.9819,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999960364290451e-05,
+      "loss": 0.9709,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999956301659811e-05,
+      "loss": 0.9745,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999520408579778e-05,
+      "loss": 0.9509,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999947581885795e-05,
+      "loss": 0.9388,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999942924744147e-05,
+      "loss": 0.9099,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999380694339567e-05,
+      "loss": 0.91,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999933015956186e-05,
+      "loss": 0.9041,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999277643118368e-05,
+      "loss": 0.9858,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999922314501949e-05,
+      "loss": 0.9925,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999166665276038e-05,
+      "loss": 0.8717,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999108203899196e-05,
+      "loss": 0.9019,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999047760900557e-05,
+      "loss": 0.8604,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998985336292092e-05,
+      "loss": 0.9731,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998920930086182e-05,
+      "loss": 0.9295,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998854542295582e-05,
+      "loss": 0.9113,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998786172933452e-05,
+      "loss": 0.9526,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998715822013342e-05,
+      "loss": 0.9904,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998643489549193e-05,
+      "loss": 0.9173,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999856917555534e-05,
+      "loss": 0.9444,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998492880046517e-05,
+      "loss": 0.8603,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998414603037832e-05,
+      "loss": 0.948,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998334344544808e-05,
+      "loss": 0.8961,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998252104583344e-05,
+      "loss": 0.9599,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998167883169742e-05,
+      "loss": 0.9358,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998081680320694e-05,
+      "loss": 0.9865,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999799349605328e-05,
+      "loss": 0.9806,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999790333038498e-05,
+      "loss": 0.8576,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999781118333366e-05,
+      "loss": 0.9342,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999771705491758e-05,
+      "loss": 0.9196,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99976209451554e-05,
+      "loss": 0.9943,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997522854066166e-05,
+      "loss": 0.9036,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997422781669315e-05,
+      "loss": 0.9135,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997320727984683e-05,
+      "loss": 0.8653,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997216693032487e-05,
+      "loss": 0.8459,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999711067683335e-05,
+      "loss": 0.9421,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997002679408285e-05,
+      "loss": 0.9456,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996892700778692e-05,
+      "loss": 0.9913,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996780740966367e-05,
+      "loss": 0.9814,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996666799993496e-05,
+      "loss": 0.3059,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996550877882664e-05,
+      "loss": 0.8407,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999643297465684e-05,
+      "loss": 0.9844,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996313090339394e-05,
+      "loss": 0.9195,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999619122495408e-05,
+      "loss": 0.9424,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996067378525054e-05,
+      "loss": 0.9538,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995941551076855e-05,
+      "loss": 0.9386,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995813742634427e-05,
+      "loss": 0.9783,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995683953223092e-05,
+      "loss": 0.8405,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995552182868576e-05,
+      "loss": 1.0102,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999541843159699e-05,
+      "loss": 0.9289,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999528269943484e-05,
+      "loss": 0.8772,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999514498640903e-05,
+      "loss": 0.9309,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995005292546848e-05,
+      "loss": 0.319,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994863617875982e-05,
+      "loss": 0.9769,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994719962424505e-05,
+      "loss": 0.8322,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999457432622089e-05,
+      "loss": 0.8635,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994426709293996e-05,
+      "loss": 0.9505,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999427711167308e-05,
+      "loss": 0.9265,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999412553338779e-05,
+      "loss": 0.8598,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993971974468166e-05,
+      "loss": 0.9144,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993816434944637e-05,
+      "loss": 1.008,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993658914848028e-05,
+      "loss": 0.8793,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993499414209562e-05,
+      "loss": 1.0131,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999333793306084e-05,
+      "loss": 0.9269,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999317447143387e-05,
+      "loss": 0.9897,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993009029361046e-05,
+      "loss": 0.9223,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992841606875158e-05,
+      "loss": 0.9919,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999267220400938e-05,
+      "loss": 0.8748,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992500820797286e-05,
+      "loss": 0.8538,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992327457272846e-05,
+      "loss": 0.9053,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992152113470407e-05,
+      "loss": 0.8914,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991974789424732e-05,
+      "loss": 0.8515,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991795485170952e-05,
+      "loss": 0.3413,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99916142007446e-05,
+      "loss": 0.8804,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991430936181615e-05,
+      "loss": 1.0664,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991245691518307e-05,
+      "loss": 1.0076,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999105846679139e-05,
+      "loss": 0.9231,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999086926203797e-05,
+      "loss": 0.9506,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990678077295542e-05,
+      "loss": 1.0122,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990484912601992e-05,
+      "loss": 0.9052,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990289767995606e-05,
+      "loss": 0.9234,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990092643515055e-05,
+      "loss": 0.9814,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989893539199406e-05,
+      "loss": 0.8663,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998969245508812e-05,
+      "loss": 0.9152,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998948939122104e-05,
+      "loss": 0.8689,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989284347638423e-05,
+      "loss": 1.0374,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998907732438089e-05,
+      "loss": 0.9548,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988868321489475e-05,
+      "loss": 0.9078,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99886573390056e-05,
+      "loss": 0.9119,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998844437697107e-05,
+      "loss": 0.9811,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99882294354281e-05,
+      "loss": 0.9553,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988012514419282e-05,
+      "loss": 1.0081,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987793613987606e-05,
+      "loss": 0.9868,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987572734176454e-05,
+      "loss": 0.8646,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987349875029597e-05,
+      "loss": 0.8763,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987125036591205e-05,
+      "loss": 0.7894,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986898218905834e-05,
+      "loss": 0.9503,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986669422018433e-05,
+      "loss": 0.9438,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986438645974352e-05,
+      "loss": 0.9778,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998620589081932e-05,
+      "loss": 0.9812,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985971156599467e-05,
+      "loss": 0.9486,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998573444336131e-05,
+      "loss": 0.9839,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985495751151765e-05,
+      "loss": 0.9107,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985255080018132e-05,
+      "loss": 0.9376,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998501243000811e-05,
+      "loss": 0.9436,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984767801169784e-05,
+      "loss": 0.8314,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984521193551635e-05,
+      "loss": 0.9566,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984272607202542e-05,
+      "loss": 0.3352,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984022042171762e-05,
+      "loss": 0.9683,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983769498508954e-05,
+      "loss": 0.9548,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983514976264172e-05,
+      "loss": 0.9159,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998325847548785e-05,
+      "loss": 0.9008,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982999996230824e-05,
+      "loss": 1.0542,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998273953854432e-05,
+      "loss": 0.9514,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998247710247996e-05,
+      "loss": 0.9228,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998221268808974e-05,
+      "loss": 0.9052,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998194629542608e-05,
+      "loss": 0.9457,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981677924541757e-05,
+      "loss": 0.9446,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981407575489964e-05,
+      "loss": 0.8687,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981135248324283e-05,
+      "loss": 0.953,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980860943098675e-05,
+      "loss": 0.9366,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998058465986751e-05,
+      "loss": 0.9741,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980306398685532e-05,
+      "loss": 0.943,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99800261596079e-05,
+      "loss": 0.9243,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997974394269014e-05,
+      "loss": 1.0024,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979459747988187e-05,
+      "loss": 0.9426,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997917357555836e-05,
+      "loss": 0.9453,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978885425457373e-05,
+      "loss": 0.8511,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997859529774234e-05,
+      "loss": 0.9205,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978303192470744e-05,
+      "loss": 0.9379,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978009109700486e-05,
+      "loss": 0.9744,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997771304948984e-05,
+      "loss": 0.9311,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977415011897486e-05,
+      "loss": 0.97,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977114996982482e-05,
+      "loss": 0.9228,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997681300480429e-05,
+      "loss": 0.8857,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976509035422754e-05,
+      "loss": 0.9227,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976203088898122e-05,
+      "loss": 0.9129,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975895165291018e-05,
+      "loss": 0.8381,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975585264662473e-05,
+      "loss": 0.8845,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975273387073895e-05,
+      "loss": 0.9201,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974959532587103e-05,
+      "loss": 0.9186,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997464370126429e-05,
+      "loss": 0.917,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974325893168045e-05,
+      "loss": 0.877,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974006108361357e-05,
+      "loss": 0.8724,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99736843469076e-05,
+      "loss": 0.9485,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997336060887053e-05,
+      "loss": 0.8278,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973034894314323e-05,
+      "loss": 0.8623,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997270720330352e-05,
+      "loss": 0.9505,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972377535903057e-05,
+      "loss": 0.9683,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972045892178276e-05,
+      "loss": 0.9356,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99717122721949e-05,
+      "loss": 1.0472,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997137667601905e-05,
+      "loss": 0.9071,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971039103717222e-05,
+      "loss": 0.8722,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997069955535633e-05,
+      "loss": 0.9393,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970358031003655e-05,
+      "loss": 0.8429,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997001453072689e-05,
+      "loss": 1.0006,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969669054594097e-05,
+      "loss": 0.9779,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969321602673753e-05,
+      "loss": 0.9692,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968972175034715e-05,
+      "loss": 1.0002,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996862077174623e-05,
+      "loss": 0.9688,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996826739287794e-05,
+      "loss": 0.9232,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967912038499872e-05,
+      "loss": 0.9515,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996755470868246e-05,
+      "loss": 0.86,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996719540349651e-05,
+      "loss": 0.9742,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996683412301324e-05,
+      "loss": 0.8691,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996647086730424e-05,
+      "loss": 0.9084,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99661056364415e-05,
+      "loss": 0.8854,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965738430497406e-05,
+      "loss": 1.0083,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965369249544724e-05,
+      "loss": 0.9454,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964998093656627e-05,
+      "loss": 0.8715,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964624962906666e-05,
+      "loss": 0.8859,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964249857368787e-05,
+      "loss": 0.8135,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996387277711733e-05,
+      "loss": 0.9607,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963493722227024e-05,
+      "loss": 0.9551,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963112692772992e-05,
+      "loss": 0.864,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996272968883074e-05,
+      "loss": 0.8911,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962344710476183e-05,
+      "loss": 0.8874,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961957757785604e-05,
+      "loss": 0.8524,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961568830835697e-05,
+      "loss": 0.9411,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961177929703537e-05,
+      "loss": 0.9492,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960785054466593e-05,
+      "loss": 0.3791,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960390205202726e-05,
+      "loss": 0.9357,
+      "step": 419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995999338199018e-05,
+      "loss": 0.9636,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959594584907605e-05,
+      "loss": 0.8875,
+      "step": 421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995919381403404e-05,
+      "loss": 0.8984,
+      "step": 422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958791069448893e-05,
+      "loss": 0.8306,
+      "step": 423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958386351231997e-05,
+      "loss": 0.9222,
+      "step": 424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957979659463546e-05,
+      "loss": 0.9162,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957570994224148e-05,
+      "loss": 0.955,
+      "step": 426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957160355594785e-05,
+      "loss": 0.9406,
+      "step": 427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956747743656842e-05,
+      "loss": 0.9478,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956333158492086e-05,
+      "loss": 0.8545,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955916600182685e-05,
+      "loss": 0.8247,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955498068811185e-05,
+      "loss": 0.8207,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955077564460535e-05,
+      "loss": 0.9612,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995465508721407e-05,
+      "loss": 0.8918,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995423063715552e-05,
+      "loss": 0.9218,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953804214369002e-05,
+      "loss": 0.931,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953375818939015e-05,
+      "loss": 0.9339,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952945450950468e-05,
+      "loss": 0.9602,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952513110488646e-05,
+      "loss": 0.8977,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952078797639235e-05,
+      "loss": 0.9152,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951642512488305e-05,
+      "loss": 0.8913,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995120425512232e-05,
+      "loss": 1.02,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950764025628127e-05,
+      "loss": 0.8597,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950321824092976e-05,
+      "loss": 0.9115,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949877650604506e-05,
+      "loss": 0.8954,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949431505250736e-05,
+      "loss": 0.8985,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994898338812009e-05,
+      "loss": 0.9695,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994853329930137e-05,
+      "loss": 0.9074,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948081238883773e-05,
+      "loss": 0.8623,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947627206956897e-05,
+      "loss": 0.9281,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947171203610717e-05,
+      "loss": 0.8844,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946713228935603e-05,
+      "loss": 0.9314,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946253283022318e-05,
+      "loss": 0.2939,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994579136596201e-05,
+      "loss": 0.8973,
+      "step": 453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945327477846228e-05,
+      "loss": 0.8875,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944861618766898e-05,
+      "loss": 0.9478,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994439378881635e-05,
+      "loss": 0.925,
+      "step": 456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943923988087298e-05,
+      "loss": 0.9035,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943452216672846e-05,
+      "loss": 0.8904,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942978474666484e-05,
+      "loss": 0.8881,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994250276216211e-05,
+      "loss": 0.8739,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942025079253988e-05,
+      "loss": 0.8294,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941545426036795e-05,
+      "loss": 0.9188,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994106380260558e-05,
+      "loss": 0.8623,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99405802090558e-05,
+      "loss": 0.9745,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940094645483286e-05,
+      "loss": 0.9314,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939607111984268e-05,
+      "loss": 0.9662,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939117608655372e-05,
+      "loss": 0.9606,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99386261355936e-05,
+      "loss": 0.8926,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993813269289636e-05,
+      "loss": 0.927,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993763728066143e-05,
+      "loss": 0.3072,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937139898987003e-05,
+      "loss": 0.9009,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936640547971646e-05,
+      "loss": 0.3448,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993613922771432e-05,
+      "loss": 0.8753,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993563593831437e-05,
+      "loss": 0.8649,
+      "step": 474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935130679871548e-05,
+      "loss": 0.9152,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993462345248598e-05,
+      "loss": 0.8648,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993411425625819e-05,
+      "loss": 0.8165,
+      "step": 477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933603091289094e-05,
+      "loss": 0.9185,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993308995767999e-05,
+      "loss": 0.9578,
+      "step": 479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932574855532572e-05,
+      "loss": 0.873,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993205778494892e-05,
+      "loss": 0.8418,
+      "step": 481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931538746031513e-05,
+      "loss": 0.9444,
+      "step": 482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993101773888321e-05,
+      "loss": 0.8934,
+      "step": 483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930494763607265e-05,
+      "loss": 0.9488,
+      "step": 484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992996982030732e-05,
+      "loss": 0.9451,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992944290908741e-05,
+      "loss": 0.957,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928914030051956e-05,
+      "loss": 0.8908,
+      "step": 487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928383183305774e-05,
+      "loss": 0.8783,
+      "step": 488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992785036895407e-05,
+      "loss": 0.8416,
+      "step": 489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927315587102428e-05,
+      "loss": 0.883,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926778837856836e-05,
+      "loss": 0.8549,
+      "step": 491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992624012132367e-05,
+      "loss": 0.8555,
+      "step": 492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925699437609688e-05,
+      "loss": 0.8407,
+      "step": 493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992515678682204e-05,
+      "loss": 0.9061,
+      "step": 494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992461216906828e-05,
+      "loss": 0.9122,
+      "step": 495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924065584456324e-05,
+      "loss": 0.948,
+      "step": 496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923517033094512e-05,
+      "loss": 0.9583,
+      "step": 497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992296651509154e-05,
+      "loss": 0.9341,
+      "step": 498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992241403055652e-05,
+      "loss": 0.9625,
+      "step": 499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921859579598938e-05,
+      "loss": 0.9129,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921303162328677e-05,
+      "loss": 0.9279,
+      "step": 501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920744778856008e-05,
+      "loss": 0.9862,
+      "step": 502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992018442929159e-05,
+      "loss": 0.9315,
+      "step": 503
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919622113746474e-05,
+      "loss": 0.8619,
+      "step": 504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919057832332097e-05,
+      "loss": 0.88,
+      "step": 505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918491585160295e-05,
+      "loss": 0.9277,
+      "step": 506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991792337234328e-05,
+      "loss": 0.8722,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991735319399366e-05,
+      "loss": 0.8788,
+      "step": 508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991678105022444e-05,
+      "loss": 0.8387,
+      "step": 509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916206941149e-05,
+      "loss": 0.8448,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915630866881123e-05,
+      "loss": 0.8623,
+      "step": 511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915052827534968e-05,
+      "loss": 0.8683,
+      "step": 512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914472823225095e-05,
+      "loss": 0.9233,
+      "step": 513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991389085406645e-05,
+      "loss": 0.9089,
+      "step": 514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991330692017437e-05,
+      "loss": 0.8574,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912721021664573e-05,
+      "loss": 0.8685,
+      "step": 516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991213315865317e-05,
+      "loss": 0.9525,
+      "step": 517
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911543331256674e-05,
+      "loss": 0.9195,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991095153959197e-05,
+      "loss": 0.961,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991035778377634e-05,
+      "loss": 0.9034,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909762063927453e-05,
+      "loss": 0.9506,
+      "step": 521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909164380163373e-05,
+      "loss": 0.8675,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908564732602546e-05,
+      "loss": 0.9025,
+      "step": 523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907963121363808e-05,
+      "loss": 0.8314,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907359546566393e-05,
+      "loss": 0.8331,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990675400832991e-05,
+      "loss": 0.8922,
+      "step": 526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990614650677437e-05,
+      "loss": 0.3596,
+      "step": 527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990553704202016e-05,
+      "loss": 0.9484,
+      "step": 528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904925614188074e-05,
+      "loss": 0.8823,
+      "step": 529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990431222339928e-05,
+      "loss": 0.955,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903696869775334e-05,
+      "loss": 0.9501,
+      "step": 531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903079553438197e-05,
+      "loss": 0.9172,
+      "step": 532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902460274510204e-05,
+      "loss": 0.8275,
+      "step": 533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901839033114075e-05,
+      "loss": 0.9564,
+      "step": 534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901215829372943e-05,
+      "loss": 0.8963,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900590663410306e-05,
+      "loss": 0.9209,
+      "step": 536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989996353535006e-05,
+      "loss": 0.3324,
+      "step": 537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899334445316487e-05,
+      "loss": 0.8391,
+      "step": 538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898703393434266e-05,
+      "loss": 0.9414,
+      "step": 539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898070379828452e-05,
+      "loss": 0.8591,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897435404624497e-05,
+      "loss": 0.9374,
+      "step": 541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896798467948248e-05,
+      "loss": 0.9812,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896159569925924e-05,
+      "loss": 0.943,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895518710684145e-05,
+      "loss": 0.9361,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894875890349917e-05,
+      "loss": 0.8294,
+      "step": 545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894231109050635e-05,
+      "loss": 0.8976,
+      "step": 546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893584366914075e-05,
+      "loss": 0.9292,
+      "step": 547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892935664068416e-05,
+      "loss": 0.9051,
+      "step": 548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892285000642215e-05,
+      "loss": 0.8929,
+      "step": 549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989163237676442e-05,
+      "loss": 0.9803,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989097779256437e-05,
+      "loss": 0.8656,
+      "step": 551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890321248171786e-05,
+      "loss": 0.8218,
+      "step": 552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889662743716784e-05,
+      "loss": 0.869,
+      "step": 553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988900227932987e-05,
+      "loss": 0.8519,
+      "step": 554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988833985514193e-05,
+      "loss": 0.8451,
+      "step": 555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887675471284244e-05,
+      "loss": 0.9181,
+      "step": 556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988700912788848e-05,
+      "loss": 0.9438,
+      "step": 557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886340825086695e-05,
+      "loss": 0.375,
+      "step": 558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988567056301133e-05,
+      "loss": 0.8882,
+      "step": 559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884998341795224e-05,
+      "loss": 0.8549,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988432416157159e-05,
+      "loss": 0.8344,
+      "step": 561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883648022474046e-05,
+      "loss": 0.9587,
+      "step": 562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882969924636576e-05,
+      "loss": 0.8801,
+      "step": 563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882289868193576e-05,
+      "loss": 0.9044,
+      "step": 564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881607853279817e-05,
+      "loss": 0.9394,
+      "step": 565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988092388003046e-05,
+      "loss": 0.8883,
+      "step": 566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880237948581052e-05,
+      "loss": 0.9671,
+      "step": 567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879550059067533e-05,
+      "loss": 0.842,
+      "step": 568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878860211626233e-05,
+      "loss": 0.8918,
+      "step": 569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987816840639386e-05,
+      "loss": 0.8162,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877474643507518e-05,
+      "loss": 0.9843,
+      "step": 571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876778923104697e-05,
+      "loss": 0.9286,
+      "step": 572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987608124532327e-05,
+      "loss": 0.9128,
+      "step": 573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875381610301514e-05,
+      "loss": 0.8301,
+      "step": 574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874680018178076e-05,
+      "loss": 0.8807,
+      "step": 575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873976469091994e-05,
+      "loss": 0.9363,
+      "step": 576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98732709631827e-05,
+      "loss": 0.3461,
+      "step": 577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872563500590013e-05,
+      "loss": 0.8958,
+      "step": 578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871854081454137e-05,
+      "loss": 0.8038,
+      "step": 579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987114270591566e-05,
+      "loss": 0.9415,
+      "step": 580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870429374115572e-05,
+      "loss": 0.8816,
+      "step": 581
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869714086195234e-05,
+      "loss": 0.2895,
+      "step": 582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868996842296398e-05,
+      "loss": 0.7858,
+      "step": 583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868277642561214e-05,
+      "loss": 0.9013,
+      "step": 584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867556487132215e-05,
+      "loss": 0.8542,
+      "step": 585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866833376152314e-05,
+      "loss": 0.9311,
+      "step": 586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986610830976482e-05,
+      "loss": 0.8416,
+      "step": 587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986538128811342e-05,
+      "loss": 0.9018,
+      "step": 588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98646523113422e-05,
+      "loss": 0.9417,
+      "step": 589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863921379595636e-05,
+      "loss": 0.963,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863188493018573e-05,
+      "loss": 0.8778,
+      "step": 591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862453651756257e-05,
+      "loss": 0.8215,
+      "step": 592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986171685595432e-05,
+      "loss": 0.8678,
+      "step": 593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986097810575878e-05,
+      "loss": 0.8985,
+      "step": 594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986023740131604e-05,
+      "loss": 0.3469,
+      "step": 595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859494742772897e-05,
+      "loss": 0.862,
+      "step": 596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985875013027653e-05,
+      "loss": 0.9035,
+      "step": 597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858003563974503e-05,
+      "loss": 0.8747,
+      "step": 598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857255044014777e-05,
+      "loss": 0.8808,
+      "step": 599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856504570545682e-05,
+      "loss": 0.9414,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855752143715957e-05,
+      "loss": 0.9985,
+      "step": 601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854997763674718e-05,
+      "loss": 0.8595,
+      "step": 602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854241430571458e-05,
+      "loss": 0.9602,
+      "step": 603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853483144556082e-05,
+      "loss": 0.9119,
+      "step": 604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985272290577885e-05,
+      "loss": 0.9339,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985196071439044e-05,
+      "loss": 0.9456,
+      "step": 606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851196570541894e-05,
+      "loss": 0.9416,
+      "step": 607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850430474384655e-05,
+      "loss": 0.8868,
+      "step": 608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849662426070546e-05,
+      "loss": 0.9209,
+      "step": 609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848892425751777e-05,
+      "loss": 0.9065,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984812047358095e-05,
+      "loss": 0.9579,
+      "step": 611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984734656971105e-05,
+      "loss": 0.948,
+      "step": 612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984657071429545e-05,
+      "loss": 0.8127,
+      "step": 613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845792907487905e-05,
+      "loss": 0.9231,
+      "step": 614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984501314944256e-05,
+      "loss": 0.8668,
+      "step": 615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844231440313954e-05,
+      "loss": 0.8858,
+      "step": 616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843447780257002e-05,
+      "loss": 0.936,
+      "step": 617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842662169427008e-05,
+      "loss": 0.8954,
+      "step": 618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841874607979663e-05,
+      "loss": 0.8889,
+      "step": 619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984108509607105e-05,
+      "loss": 0.868,
+      "step": 620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840293633857636e-05,
+      "loss": 0.9097,
+      "step": 621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983950022149627e-05,
+      "loss": 0.3716,
+      "step": 622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838704859144188e-05,
+      "loss": 0.9082,
+      "step": 623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837907546959014e-05,
+      "loss": 0.9223,
+      "step": 624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837108285098766e-05,
+      "loss": 0.9255,
+      "step": 625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836307073721835e-05,
+      "loss": 0.9522,
+      "step": 626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983550391298701e-05,
+      "loss": 0.9102,
+      "step": 627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983469880305346e-05,
+      "loss": 0.8702,
+      "step": 628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833891744080736e-05,
+      "loss": 0.8701,
+      "step": 629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833082736228792e-05,
+      "loss": 0.8821,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832271779657945e-05,
+      "loss": 0.8059,
+      "step": 631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983145887452892e-05,
+      "loss": 0.8313,
+      "step": 632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830644021002812e-05,
+      "loss": 0.953,
+      "step": 633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982982721924111e-05,
+      "loss": 0.8957,
+      "step": 634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829008469405688e-05,
+      "loss": 0.924,
+      "step": 635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828187771658808e-05,
+      "loss": 0.8778,
+      "step": 636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827365126163115e-05,
+      "loss": 0.839,
+      "step": 637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826540533081636e-05,
+      "loss": 0.9032,
+      "step": 638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825713992577796e-05,
+      "loss": 0.8931,
+      "step": 639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824885504815393e-05,
+      "loss": 0.959,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824055069958615e-05,
+      "loss": 0.9221,
+      "step": 641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823222688172047e-05,
+      "loss": 0.9084,
+      "step": 642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982238835962064e-05,
+      "loss": 0.8675,
+      "step": 643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821552084469746e-05,
+      "loss": 0.9575,
+      "step": 644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98207138628851e-05,
+      "loss": 0.8768,
+      "step": 645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819873695032816e-05,
+      "loss": 0.8917,
+      "step": 646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98190315810794e-05,
+      "loss": 0.869,
+      "step": 647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818187521191746e-05,
+      "loss": 0.9742,
+      "step": 648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817341515537123e-05,
+      "loss": 0.9469,
+      "step": 649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816493564283195e-05,
+      "loss": 0.9497,
+      "step": 650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981564366759801e-05,
+      "loss": 0.9183,
+      "step": 651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814791825649998e-05,
+      "loss": 0.8642,
+      "step": 652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981393803860798e-05,
+      "loss": 0.886,
+      "step": 653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981308230664116e-05,
+      "loss": 0.8503,
+      "step": 654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981222462991912e-05,
+      "loss": 0.8493,
+      "step": 655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981136500861184e-05,
+      "loss": 0.9349,
+      "step": 656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981050344288968e-05,
+      "loss": 0.8067,
+      "step": 657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809639932923383e-05,
+      "loss": 0.8204,
+      "step": 658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808774478884085e-05,
+      "loss": 0.8989,
+      "step": 659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807907080943288e-05,
+      "loss": 0.8609,
+      "step": 660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807037739272904e-05,
+      "loss": 0.8713,
+      "step": 661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806166454045215e-05,
+      "loss": 0.9329,
+      "step": 662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98052932254329e-05,
+      "loss": 0.9461,
+      "step": 663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804418053609e-05,
+      "loss": 0.9168,
+      "step": 664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803540938746968e-05,
+      "loss": 0.884,
+      "step": 665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802661881020628e-05,
+      "loss": 0.902,
+      "step": 666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801780880604193e-05,
+      "loss": 1.059,
+      "step": 667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800897937672256e-05,
+      "loss": 0.939,
+      "step": 668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800013052399802e-05,
+      "loss": 0.7955,
+      "step": 669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799126224962192e-05,
+      "loss": 0.8614,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798237455535184e-05,
+      "loss": 0.844,
+      "step": 671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979734674429491e-05,
+      "loss": 0.9039,
+      "step": 672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979645409141789e-05,
+      "loss": 0.8116,
+      "step": 673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795559497081033e-05,
+      "loss": 0.8942,
+      "step": 674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979466296146163e-05,
+      "loss": 0.3237,
+      "step": 675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793764484737354e-05,
+      "loss": 0.962,
+      "step": 676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792864067086266e-05,
+      "loss": 0.8884,
+      "step": 677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979196170868681e-05,
+      "loss": 0.9102,
+      "step": 678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791057409717816e-05,
+      "loss": 0.8656,
+      "step": 679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790151170358497e-05,
+      "loss": 0.9276,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789242990788453e-05,
+      "loss": 0.9017,
+      "step": 681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788332871187666e-05,
+      "loss": 0.805,
+      "step": 682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787420811736503e-05,
+      "loss": 0.9148,
+      "step": 683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786506812615712e-05,
+      "loss": 0.9056,
+      "step": 684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978559087400644e-05,
+      "loss": 0.9253,
+      "step": 685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784672996090198e-05,
+      "loss": 0.8863,
+      "step": 686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978375317904889e-05,
+      "loss": 0.8116,
+      "step": 687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782831423064813e-05,
+      "loss": 0.8591,
+      "step": 688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781907728320636e-05,
+      "loss": 0.8448,
+      "step": 689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780982094999417e-05,
+      "loss": 0.8735,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780054523284596e-05,
+      "loss": 0.8758,
+      "step": 691
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779125013360003e-05,
+      "loss": 0.3533,
+      "step": 692
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778193565409845e-05,
+      "loss": 0.9292,
+      "step": 693
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977726017961872e-05,
+      "loss": 0.8861,
+      "step": 694
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776324856171597e-05,
+      "loss": 0.9474,
+      "step": 695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977538759525385e-05,
+      "loss": 0.8903,
+      "step": 696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977444839705122e-05,
+      "loss": 0.8746,
+      "step": 697
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773507261749836e-05,
+      "loss": 0.9251,
+      "step": 698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772564189536215e-05,
+      "loss": 0.8874,
+      "step": 699
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977161918059725e-05,
+      "loss": 0.8903,
+      "step": 700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770672235120226e-05,
+      "loss": 0.8616,
+      "step": 701
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769723353292812e-05,
+      "loss": 0.9194,
+      "step": 702
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768772535303054e-05,
+      "loss": 0.9843,
+      "step": 703
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976781978133938e-05,
+      "loss": 0.9414,
+      "step": 704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766865091590614e-05,
+      "loss": 0.8561,
+      "step": 705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765908466245954e-05,
+      "loss": 0.8583,
+      "step": 706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764949905494984e-05,
+      "loss": 0.7951,
+      "step": 707
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976398940952767e-05,
+      "loss": 0.8822,
+      "step": 708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976302697853436e-05,
+      "loss": 0.8934,
+      "step": 709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97620626127058e-05,
+      "loss": 0.8265,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761096312233094e-05,
+      "loss": 0.8744,
+      "step": 711
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760128077307752e-05,
+      "loss": 0.9084,
+      "step": 712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759157908121652e-05,
+      "loss": 0.9003,
+      "step": 713
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975818580486707e-05,
+      "loss": 0.9352,
+      "step": 714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757211767736658e-05,
+      "loss": 0.8858,
+      "step": 715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975623579692344e-05,
+      "loss": 0.8925,
+      "step": 716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975525789262084e-05,
+      "loss": 0.884,
+      "step": 717
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975427805502266e-05,
+      "loss": 0.8803,
+      "step": 718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975329628432308e-05,
+      "loss": 0.9567,
+      "step": 719
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975231258071667e-05,
+      "loss": 0.9775,
+      "step": 720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751326944398385e-05,
+      "loss": 0.8644,
+      "step": 721
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750339375563546e-05,
+      "loss": 0.9186,
+      "step": 722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974934987440788e-05,
+      "loss": 0.8732,
+      "step": 723
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748358441127485e-05,
+      "loss": 0.84,
+      "step": 724
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747365075918842e-05,
+      "loss": 0.9315,
+      "step": 725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974636977897881e-05,
+      "loss": 0.9178,
+      "step": 726
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745372550504645e-05,
+      "loss": 0.9175,
+      "step": 727
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974437339069397e-05,
+      "loss": 0.9449,
+      "step": 728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974337229974481e-05,
+      "loss": 0.8049,
+      "step": 729
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742369277855546e-05,
+      "loss": 0.8803,
+      "step": 730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741364325224967e-05,
+      "loss": 0.9823,
+      "step": 731
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740357442052235e-05,
+      "loss": 0.8372,
+      "step": 732
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739348628536886e-05,
+      "loss": 0.863,
+      "step": 733
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738337884878852e-05,
+      "loss": 0.3384,
+      "step": 734
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737325211278447e-05,
+      "loss": 0.8547,
+      "step": 735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973631060793635e-05,
+      "loss": 0.878,
+      "step": 736
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735294075053642e-05,
+      "loss": 0.8808,
+      "step": 737
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734275612831783e-05,
+      "loss": 0.3357,
+      "step": 738
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733255221472607e-05,
+      "loss": 0.9269,
+      "step": 739
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732232901178334e-05,
+      "loss": 0.8784,
+      "step": 740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731208652151568e-05,
+      "loss": 0.9809,
+      "step": 741
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730182474595295e-05,
+      "loss": 0.911,
+      "step": 742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729154368712885e-05,
+      "loss": 0.9241,
+      "step": 743
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972812433470809e-05,
+      "loss": 0.8643,
+      "step": 744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727092372785032e-05,
+      "loss": 0.885,
+      "step": 745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726058483148236e-05,
+      "loss": 0.7782,
+      "step": 746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725022666002588e-05,
+      "loss": 0.9252,
+      "step": 747
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723984921553377e-05,
+      "loss": 0.3582,
+      "step": 748
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722945250006254e-05,
+      "loss": 0.8498,
+      "step": 749
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721903651567264e-05,
+      "loss": 0.8423,
+      "step": 750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720860126442832e-05,
+      "loss": 0.8934,
+      "step": 751
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719814674839767e-05,
+      "loss": 0.9111,
+      "step": 752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718767296965247e-05,
+      "loss": 0.8824,
+      "step": 753
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717717993026847e-05,
+      "loss": 0.8111,
+      "step": 754
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716666763232516e-05,
+      "loss": 0.8941,
+      "step": 755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715613607790592e-05,
+      "loss": 0.8658,
+      "step": 756
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714558526909782e-05,
+      "loss": 0.7854,
+      "step": 757
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713501520799186e-05,
+      "loss": 0.3131,
+      "step": 758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971244258966828e-05,
+      "loss": 0.9073,
+      "step": 759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711381733726926e-05,
+      "loss": 0.8262,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971031895318536e-05,
+      "loss": 0.9454,
+      "step": 761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709254248254202e-05,
+      "loss": 0.8466,
+      "step": 762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708187619144462e-05,
+      "loss": 0.3563,
+      "step": 763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707119066067518e-05,
+      "loss": 0.9139,
+      "step": 764
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706048589235138e-05,
+      "loss": 0.8898,
+      "step": 765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970497618885947e-05,
+      "loss": 0.3247,
+      "step": 766
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970390186515304e-05,
+      "loss": 0.9013,
+      "step": 767
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970282561832876e-05,
+      "loss": 0.8638,
+      "step": 768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701747448599918e-05,
+      "loss": 0.9148,
+      "step": 769
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970066735618019e-05,
+      "loss": 0.9325,
+      "step": 770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969958534128362e-05,
+      "loss": 0.9158,
+      "step": 771
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969850140412465e-05,
+      "loss": 0.9017,
+      "step": 772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969741554491809e-05,
+      "loss": 0.9001,
+      "step": 773
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696327763879137e-05,
+      "loss": 0.8861,
+      "step": 774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695238061223367e-05,
+      "loss": 0.8847,
+      "step": 775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694146437166736e-05,
+      "loss": 0.893,
+      "step": 776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693052891925583e-05,
+      "loss": 0.8699,
+      "step": 777
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691957425716628e-05,
+      "loss": 0.847,
+      "step": 778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690860038756967e-05,
+      "loss": 0.8016,
+      "step": 779
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689760731264082e-05,
+      "loss": 0.8971,
+      "step": 780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688659503455836e-05,
+      "loss": 0.8848,
+      "step": 781
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687556355550466e-05,
+      "loss": 0.8754,
+      "step": 782
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686451287766594e-05,
+      "loss": 0.9012,
+      "step": 783
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685344300323226e-05,
+      "loss": 0.3397,
+      "step": 784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684235393439743e-05,
+      "loss": 0.8982,
+      "step": 785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683124567335902e-05,
+      "loss": 0.7976,
+      "step": 786
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682011822231855e-05,
+      "loss": 0.8792,
+      "step": 787
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968089715834812e-05,
+      "loss": 0.8954,
+      "step": 788
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679780575905605e-05,
+      "loss": 0.8153,
+      "step": 789
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678662075125593e-05,
+      "loss": 0.8919,
+      "step": 790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677541656229745e-05,
+      "loss": 0.807,
+      "step": 791
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676419319440108e-05,
+      "loss": 0.8972,
+      "step": 792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675295064979107e-05,
+      "loss": 0.9521,
+      "step": 793
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674168893069545e-05,
+      "loss": 0.9208,
+      "step": 794
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673040803934612e-05,
+      "loss": 0.8717,
+      "step": 795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671910797797864e-05,
+      "loss": 0.8464,
+      "step": 796
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967077887488325e-05,
+      "loss": 0.8906,
+      "step": 797
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669645035415093e-05,
+      "loss": 0.8363,
+      "step": 798
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668509279618103e-05,
+      "loss": 0.9375,
+      "step": 799
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667371607717356e-05,
+      "loss": 0.9299,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666232019938318e-05,
+      "loss": 0.8108,
+      "step": 801
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665090516506832e-05,
+      "loss": 0.8429,
+      "step": 802
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663947097649124e-05,
+      "loss": 0.9186,
+      "step": 803
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662801763591796e-05,
+      "loss": 0.9101,
+      "step": 804
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661654514561825e-05,
+      "loss": 0.8867,
+      "step": 805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660505350786582e-05,
+      "loss": 0.8102,
+      "step": 806
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659354272493797e-05,
+      "loss": 0.8605,
+      "step": 807
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658201279911596e-05,
+      "loss": 0.8184,
+      "step": 808
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965704637326848e-05,
+      "loss": 0.9295,
+      "step": 809
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965588955279333e-05,
+      "loss": 0.828,
+      "step": 810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654730818715398e-05,
+      "loss": 0.8689,
+      "step": 811
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653570171264328e-05,
+      "loss": 0.8655,
+      "step": 812
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652407610670132e-05,
+      "loss": 0.8974,
+      "step": 813
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965124313716321e-05,
+      "loss": 0.8835,
+      "step": 814
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650076750974333e-05,
+      "loss": 0.9951,
+      "step": 815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964890845233466e-05,
+      "loss": 0.8732,
+      "step": 816
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647738241475723e-05,
+      "loss": 0.3976,
+      "step": 817
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646566118629434e-05,
+      "loss": 0.9195,
+      "step": 818
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964539208402808e-05,
+      "loss": 0.916,
+      "step": 819
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644216137904334e-05,
+      "loss": 0.8353,
+      "step": 820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964303828049125e-05,
+      "loss": 0.901,
+      "step": 821
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641858512022246e-05,
+      "loss": 0.3324,
+      "step": 822
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964067683273114e-05,
+      "loss": 0.8538,
+      "step": 823
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639493242852103e-05,
+      "loss": 0.8558,
+      "step": 824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638307742619708e-05,
+      "loss": 0.9429,
+      "step": 825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637120332268894e-05,
+      "loss": 0.872,
+      "step": 826
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635931012034987e-05,
+      "loss": 0.8753,
+      "step": 827
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963473978215368e-05,
+      "loss": 0.9164,
+      "step": 828
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633546642861053e-05,
+      "loss": 0.974,
+      "step": 829
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632351594393564e-05,
+      "loss": 0.8369,
+      "step": 830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631154636988045e-05,
+      "loss": 0.9029,
+      "step": 831
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629955770881707e-05,
+      "loss": 0.8872,
+      "step": 832
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962875499631215e-05,
+      "loss": 0.8909,
+      "step": 833
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962755231351733e-05,
+      "loss": 0.9634,
+      "step": 834
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626347722735603e-05,
+      "loss": 1.0028,
+      "step": 835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625141224205696e-05,
+      "loss": 0.8675,
+      "step": 836
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623932818166706e-05,
+      "loss": 0.8825,
+      "step": 837
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622722504858123e-05,
+      "loss": 0.9076,
+      "step": 838
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96215102845198e-05,
+      "loss": 0.9191,
+      "step": 839
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620296157391978e-05,
+      "loss": 0.8821,
+      "step": 840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961908012371527e-05,
+      "loss": 0.9103,
+      "step": 841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617862183730672e-05,
+      "loss": 0.3424,
+      "step": 842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616642337679553e-05,
+      "loss": 0.3709,
+      "step": 843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615420585803663e-05,
+      "loss": 0.801,
+      "step": 844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961419692834513e-05,
+      "loss": 0.8467,
+      "step": 845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612971365546454e-05,
+      "loss": 0.8386,
+      "step": 846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611743897650526e-05,
+      "loss": 0.8889,
+      "step": 847
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961051452490059e-05,
+      "loss": 0.8943,
+      "step": 848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96092832475403e-05,
+      "loss": 0.3656,
+      "step": 849
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960805006581366e-05,
+      "loss": 0.3257,
+      "step": 850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606814979965063e-05,
+      "loss": 0.8848,
+      "step": 851
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960557799023928e-05,
+      "loss": 0.8978,
+      "step": 852
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604339096881458e-05,
+      "loss": 0.8971,
+      "step": 853
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960309830013712e-05,
+      "loss": 0.8394,
+      "step": 854
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960185560025216e-05,
+      "loss": 0.9114,
+      "step": 855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600610997472867e-05,
+      "loss": 0.8624,
+      "step": 856
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959936449204589e-05,
+      "loss": 0.8804,
+      "step": 857
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598116084218265e-05,
+      "loss": 0.9066,
+      "step": 858
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95968657742374e-05,
+      "loss": 0.8999,
+      "step": 859
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959561356235108e-05,
+      "loss": 0.8746,
+      "step": 860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594359448807466e-05,
+      "loss": 0.9097,
+      "step": 861
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593103433855102e-05,
+      "loss": 0.8672,
+      "step": 862
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591845517742903e-05,
+      "loss": 0.8894,
+      "step": 863
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590585700720166e-05,
+      "loss": 0.836,
+      "step": 864
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589323983036554e-05,
+      "loss": 0.9806,
+      "step": 865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588060364942123e-05,
+      "loss": 0.9304,
+      "step": 866
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958679484668729e-05,
+      "loss": 0.7906,
+      "step": 867
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585527428522856e-05,
+      "loss": 0.9335,
+      "step": 868
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95842581107e-05,
+      "loss": 0.8963,
+      "step": 869
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582986893470275e-05,
+      "loss": 0.9052,
+      "step": 870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581713777085612e-05,
+      "loss": 0.8787,
+      "step": 871
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958043876179831e-05,
+      "loss": 0.9302,
+      "step": 872
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579161847861058e-05,
+      "loss": 0.7611,
+      "step": 873
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577883035526914e-05,
+      "loss": 0.7812,
+      "step": 874
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957660232504931e-05,
+      "loss": 0.8778,
+      "step": 875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575319716682065e-05,
+      "loss": 0.8998,
+      "step": 876
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574035210679356e-05,
+      "loss": 0.8331,
+      "step": 877
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572748807295746e-05,
+      "loss": 0.8939,
+      "step": 878
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571460506786183e-05,
+      "loss": 0.9545,
+      "step": 879
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957017030940598e-05,
+      "loss": 0.8392,
+      "step": 880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956887821541082e-05,
+      "loss": 0.8783,
+      "step": 881
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567584225056778e-05,
+      "loss": 0.8139,
+      "step": 882
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566288338600296e-05,
+      "loss": 0.9364,
+      "step": 883
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564990556298188e-05,
+      "loss": 0.8641,
+      "step": 884
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563690878407654e-05,
+      "loss": 0.9236,
+      "step": 885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562389305186263e-05,
+      "loss": 0.8939,
+      "step": 886
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561085836891954e-05,
+      "loss": 0.9215,
+      "step": 887
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955978047378306e-05,
+      "loss": 0.8864,
+      "step": 888
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558473216118266e-05,
+      "loss": 0.8262,
+      "step": 889
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557164064156645e-05,
+      "loss": 0.8936,
+      "step": 890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555853018157657e-05,
+      "loss": 0.9551,
+      "step": 891
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554540078381112e-05,
+      "loss": 0.9465,
+      "step": 892
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553225245087214e-05,
+      "loss": 0.8536,
+      "step": 893
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955190851853653e-05,
+      "loss": 0.9312,
+      "step": 894
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955058989899002e-05,
+      "loss": 0.8663,
+      "step": 895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549269386709004e-05,
+      "loss": 0.909,
+      "step": 896
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547946981955174e-05,
+      "loss": 0.899,
+      "step": 897
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546622684990608e-05,
+      "loss": 0.9483,
+      "step": 898
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545296496077757e-05,
+      "loss": 0.8397,
+      "step": 899
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954396841547944e-05,
+      "loss": 0.8615,
+      "step": 900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542638443458864e-05,
+      "loss": 0.9465,
+      "step": 901
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541306580279594e-05,
+      "loss": 0.8682,
+      "step": 902
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539972826205585e-05,
+      "loss": 0.8339,
+      "step": 903
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538637181501157e-05,
+      "loss": 0.9368,
+      "step": 904
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537299646431006e-05,
+      "loss": 0.9049,
+      "step": 905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535960221260213e-05,
+      "loss": 0.8454,
+      "step": 906
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534618906254215e-05,
+      "loss": 0.8439,
+      "step": 907
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533275701678834e-05,
+      "loss": 0.8943,
+      "step": 908
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531930607800276e-05,
+      "loss": 0.7805,
+      "step": 909
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530583624885105e-05,
+      "loss": 0.9123,
+      "step": 910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529234753200263e-05,
+      "loss": 0.8394,
+      "step": 911
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527883993013072e-05,
+      "loss": 0.878,
+      "step": 912
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526531344591226e-05,
+      "loss": 0.8774,
+      "step": 913
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525176808202794e-05,
+      "loss": 0.7932,
+      "step": 914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523820384116213e-05,
+      "loss": 0.8425,
+      "step": 915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522462072600306e-05,
+      "loss": 0.8807,
+      "step": 916
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952110187392426e-05,
+      "loss": 0.9264,
+      "step": 917
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519739788357636e-05,
+      "loss": 0.8717,
+      "step": 918
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951837581617037e-05,
+      "loss": 0.8361,
+      "step": 919
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517009957632785e-05,
+      "loss": 0.904,
+      "step": 920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515642213015554e-05,
+      "loss": 0.8818,
+      "step": 921
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514272582589743e-05,
+      "loss": 0.8426,
+      "step": 922
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512901066626785e-05,
+      "loss": 0.8679,
+      "step": 923
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511527665398487e-05,
+      "loss": 0.8524,
+      "step": 924
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510152379177027e-05,
+      "loss": 0.9348,
+      "step": 925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508775208234963e-05,
+      "loss": 0.8319,
+      "step": 926
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507396152845216e-05,
+      "loss": 0.9126,
+      "step": 927
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950601521328109e-05,
+      "loss": 0.8538,
+      "step": 928
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504632389816265e-05,
+      "loss": 0.8577,
+      "step": 929
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503247682724783e-05,
+      "loss": 0.8671,
+      "step": 930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501861092281062e-05,
+      "loss": 0.9712,
+      "step": 931
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500472618759903e-05,
+      "loss": 0.9388,
+      "step": 932
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499082262436474e-05,
+      "loss": 0.8817,
+      "step": 933
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497690023586307e-05,
+      "loss": 0.8603,
+      "step": 934
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496295902485326e-05,
+      "loss": 0.892,
+      "step": 935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494899899409808e-05,
+      "loss": 0.8908,
+      "step": 936
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493502014636422e-05,
+      "loss": 0.899,
+      "step": 937
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949210224844219e-05,
+      "loss": 0.8532,
+      "step": 938
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949070060110453e-05,
+      "loss": 0.8567,
+      "step": 939
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489297072901214e-05,
+      "loss": 0.8839,
+      "step": 940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948789166411039e-05,
+      "loss": 0.9022,
+      "step": 941
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486484375010587e-05,
+      "loss": 0.8153,
+      "step": 942
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485075205880698e-05,
+      "loss": 0.8285,
+      "step": 943
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483664156999992e-05,
+      "loss": 0.8504,
+      "step": 944
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482251228648112e-05,
+      "loss": 0.9238,
+      "step": 945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948083642110507e-05,
+      "loss": 0.8914,
+      "step": 946
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479419734651256e-05,
+      "loss": 0.9395,
+      "step": 947
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478001169567422e-05,
+      "loss": 0.9832,
+      "step": 948
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476580726134706e-05,
+      "loss": 0.8282,
+      "step": 949
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947515840463461e-05,
+      "loss": 0.8547,
+      "step": 950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473734205349005e-05,
+      "loss": 0.9018,
+      "step": 951
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472308128560145e-05,
+      "loss": 0.9496,
+      "step": 952
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9470880174550646e-05,
+      "loss": 0.936,
+      "step": 953
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469450343603503e-05,
+      "loss": 0.903,
+      "step": 954
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946801863600207e-05,
+      "loss": 0.9012,
+      "step": 955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466585052030093e-05,
+      "loss": 0.384,
+      "step": 956
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465149591971677e-05,
+      "loss": 0.9521,
+      "step": 957
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94637122561113e-05,
+      "loss": 0.8254,
+      "step": 958
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462273044733807e-05,
+      "loss": 0.8888,
+      "step": 959
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460831958124435e-05,
+      "loss": 0.853,
+      "step": 960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459388996568766e-05,
+      "loss": 0.9438,
+      "step": 961
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457944160352772e-05,
+      "loss": 0.7835,
+      "step": 962
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456497449762787e-05,
+      "loss": 0.8268,
+      "step": 963
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945504886508552e-05,
+      "loss": 0.9404,
+      "step": 964
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453598406608057e-05,
+      "loss": 0.8773,
+      "step": 965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945214607461784e-05,
+      "loss": 0.9213,
+      "step": 966
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450691869402703e-05,
+      "loss": 0.8491,
+      "step": 967
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449235791250833e-05,
+      "loss": 0.9186,
+      "step": 968
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447777840450794e-05,
+      "loss": 0.8889,
+      "step": 969
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446318017291527e-05,
+      "loss": 0.8941,
+      "step": 970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444856322062337e-05,
+      "loss": 0.9225,
+      "step": 971
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9443392755052907e-05,
+      "loss": 0.87,
+      "step": 972
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441927316553277e-05,
+      "loss": 0.923,
+      "step": 973
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440460006853877e-05,
+      "loss": 0.9166,
+      "step": 974
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438990826245496e-05,
+      "loss": 0.856,
+      "step": 975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943751977501929e-05,
+      "loss": 0.8315,
+      "step": 976
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94360468534668e-05,
+      "loss": 0.9753,
+      "step": 977
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434572061879923e-05,
+      "loss": 0.8934,
+      "step": 978
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943309540055094e-05,
+      "loss": 0.7903,
+      "step": 979
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431616869772484e-05,
+      "loss": 0.8329,
+      "step": 980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430136469837582e-05,
+      "loss": 0.8517,
+      "step": 981
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428654201039614e-05,
+      "loss": 0.9206,
+      "step": 982
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9427170063672335e-05,
+      "loss": 0.9192,
+      "step": 983
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425684058029874e-05,
+      "loss": 0.8695,
+      "step": 984
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424196184406728e-05,
+      "loss": 0.853,
+      "step": 985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422706443097757e-05,
+      "loss": 0.8673,
+      "step": 986
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421214834398205e-05,
+      "loss": 0.8756,
+      "step": 987
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419721358603677e-05,
+      "loss": 0.8271,
+      "step": 988
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418226016010147e-05,
+      "loss": 0.8467,
+      "step": 989
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416728806913963e-05,
+      "loss": 0.949,
+      "step": 990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415229731611843e-05,
+      "loss": 0.8979,
+      "step": 991
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941372879040087e-05,
+      "loss": 0.8734,
+      "step": 992
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941222598357851e-05,
+      "loss": 0.8842,
+      "step": 993
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410721311442574e-05,
+      "loss": 0.8602,
+      "step": 994
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940921477429127e-05,
+      "loss": 0.7901,
+      "step": 995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940770637242316e-05,
+      "loss": 0.9458,
+      "step": 996
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406196106137175e-05,
+      "loss": 0.9542,
+      "step": 997
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404683975732625e-05,
+      "loss": 0.8552,
+      "step": 998
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403169981509178e-05,
+      "loss": 0.8645,
+      "step": 999
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401654123766884e-05,
+      "loss": 0.9103,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400136402806148e-05,
+      "loss": 0.9414,
+      "step": 1001
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939861681892776e-05,
+      "loss": 0.8843,
+      "step": 1002
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397095372432863e-05,
+      "loss": 0.8371,
+      "step": 1003
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395572063622984e-05,
+      "loss": 0.3848,
+      "step": 1004
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394046892800004e-05,
+      "loss": 0.9292,
+      "step": 1005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392519860266187e-05,
+      "loss": 0.9218,
+      "step": 1006
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939099096632416e-05,
+      "loss": 0.8273,
+      "step": 1007
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389460211276917e-05,
+      "loss": 0.8964,
+      "step": 1008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387927595427823e-05,
+      "loss": 0.3562,
+      "step": 1009
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938639311908061e-05,
+      "loss": 0.8459,
+      "step": 1010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938485678253939e-05,
+      "loss": 0.8625,
+      "step": 1011
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938331858610862e-05,
+      "loss": 0.8706,
+      "step": 1012
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381778530093145e-05,
+      "loss": 0.8957,
+      "step": 1013
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380236614798177e-05,
+      "loss": 0.8221,
+      "step": 1014
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937869284052929e-05,
+      "loss": 0.9508,
+      "step": 1015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377147207592427e-05,
+      "loss": 0.9282,
+      "step": 1016
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93755997162939e-05,
+      "loss": 0.8058,
+      "step": 1017
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93740503669404e-05,
+      "loss": 0.8149,
+      "step": 1018
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372499159838968e-05,
+      "loss": 0.8263,
+      "step": 1019
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937094609529702e-05,
+      "loss": 0.8412,
+      "step": 1020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936939117362235e-05,
+      "loss": 0.7333,
+      "step": 1021
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936783439512311e-05,
+      "loss": 0.8345,
+      "step": 1022
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366275760107817e-05,
+      "loss": 0.8938,
+      "step": 1023
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364715268885366e-05,
+      "loss": 0.8949,
+      "step": 1024
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363152921765014e-05,
+      "loss": 0.88,
+      "step": 1025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361588719056384e-05,
+      "loss": 0.9405,
+      "step": 1026
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360022661069474e-05,
+      "loss": 0.889,
+      "step": 1027
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358454748114646e-05,
+      "loss": 0.9294,
+      "step": 1028
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356884980502623e-05,
+      "loss": 0.9858,
+      "step": 1029
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355313358544503e-05,
+      "loss": 0.8313,
+      "step": 1030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935373988255175e-05,
+      "loss": 0.9124,
+      "step": 1031
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352164552836198e-05,
+      "loss": 0.8877,
+      "step": 1032
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9350587369710042e-05,
+      "loss": 0.8349,
+      "step": 1033
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934900833348585e-05,
+      "loss": 0.8565,
+      "step": 1034
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347427444476556e-05,
+      "loss": 0.8624,
+      "step": 1035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345844702995456e-05,
+      "loss": 0.8616,
+      "step": 1036
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344260109356227e-05,
+      "loss": 0.8599,
+      "step": 1037
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342673663872894e-05,
+      "loss": 0.9047,
+      "step": 1038
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934108536685986e-05,
+      "loss": 0.8763,
+      "step": 1039
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339495218631895e-05,
+      "loss": 0.8993,
+      "step": 1040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337903219504134e-05,
+      "loss": 0.8933,
+      "step": 1041
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336309369792082e-05,
+      "loss": 0.9096,
+      "step": 1042
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334713669811603e-05,
+      "loss": 0.9472,
+      "step": 1043
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933311611987894e-05,
+      "loss": 0.9562,
+      "step": 1044
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331516720310682e-05,
+      "loss": 0.88,
+      "step": 1045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329915471423814e-05,
+      "loss": 0.826,
+      "step": 1046
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932831237353566e-05,
+      "loss": 0.8813,
+      "step": 1047
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326707426963926e-05,
+      "loss": 0.9071,
+      "step": 1048
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932510063202668e-05,
+      "loss": 0.9202,
+      "step": 1049
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323491989042352e-05,
+      "loss": 0.8803,
+      "step": 1050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321881498329747e-05,
+      "loss": 0.8822,
+      "step": 1051
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320269160208032e-05,
+      "loss": 0.8985,
+      "step": 1052
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931865497499674e-05,
+      "loss": 0.9983,
+      "step": 1053
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317038943015772e-05,
+      "loss": 0.9076,
+      "step": 1054
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315421064585384e-05,
+      "loss": 0.9626,
+      "step": 1055
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313801340026216e-05,
+      "loss": 0.8523,
+      "step": 1056
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931217976965926e-05,
+      "loss": 0.811,
+      "step": 1057
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931055635380588e-05,
+      "loss": 0.5557,
+      "step": 1058
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308931092787806e-05,
+      "loss": 0.9193,
+      "step": 1059
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930730398692713e-05,
+      "loss": 0.8693,
+      "step": 1060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930567503654631e-05,
+      "loss": 0.8794,
+      "step": 1061
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304044241968176e-05,
+      "loss": 0.4816,
+      "step": 1062
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302411603515915e-05,
+      "loss": 0.8626,
+      "step": 1063
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930077712151308e-05,
+      "loss": 0.8504,
+      "step": 1064
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92991407962836e-05,
+      "loss": 0.9034,
+      "step": 1065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297502628151758e-05,
+      "loss": 0.8267,
+      "step": 1066
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295862617442206e-05,
+      "loss": 0.9017,
+      "step": 1067
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929422076447996e-05,
+      "loss": 0.8993,
+      "step": 1068
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292577069590407e-05,
+      "loss": 0.9388,
+      "step": 1069
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290931533099284e-05,
+      "loss": 0.8382,
+      "step": 1070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928928415533271e-05,
+      "loss": 0.9122,
+      "step": 1071
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928763493661717e-05,
+      "loss": 0.905,
+      "step": 1072
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928598387727949e-05,
+      "loss": 0.8835,
+      "step": 1073
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284330977646884e-05,
+      "loss": 0.4309,
+      "step": 1074
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282676238046926e-05,
+      "loss": 0.9253,
+      "step": 1075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281019658807548e-05,
+      "loss": 0.8896,
+      "step": 1076
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279361240257057e-05,
+      "loss": 0.8383,
+      "step": 1077
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927770098272411e-05,
+      "loss": 0.8576,
+      "step": 1078
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927603888653774e-05,
+      "loss": 0.9052,
+      "step": 1079
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927437495202734e-05,
+      "loss": 0.945,
+      "step": 1080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272709179522674e-05,
+      "loss": 0.8879,
+      "step": 1081
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271041569353857e-05,
+      "loss": 0.8496,
+      "step": 1082
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269372121851377e-05,
+      "loss": 0.8745,
+      "step": 1083
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267700837346087e-05,
+      "loss": 0.8871,
+      "step": 1084
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92660277161692e-05,
+      "loss": 0.9019,
+      "step": 1085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264352758652303e-05,
+      "loss": 0.9082,
+      "step": 1086
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926267596512733e-05,
+      "loss": 0.8623,
+      "step": 1087
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260997335926585e-05,
+      "loss": 0.9089,
+      "step": 1088
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259316871382747e-05,
+      "loss": 0.8802,
+      "step": 1089
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257634571828846e-05,
+      "loss": 0.9596,
+      "step": 1090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255950437598278e-05,
+      "loss": 0.7959,
+      "step": 1091
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254264469024814e-05,
+      "loss": 0.8907,
+      "step": 1092
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252576666442565e-05,
+      "loss": 0.8566,
+      "step": 1093
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250887030186028e-05,
+      "loss": 0.9123,
+      "step": 1094
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249195560590052e-05,
+      "loss": 0.9545,
+      "step": 1095
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247502257989858e-05,
+      "loss": 0.8217,
+      "step": 1096
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245807122721017e-05,
+      "loss": 0.9147,
+      "step": 1097
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244110155119473e-05,
+      "loss": 0.9224,
+      "step": 1098
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242411355521535e-05,
+      "loss": 0.7649,
+      "step": 1099
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240710724263864e-05,
+      "loss": 0.9016,
+      "step": 1100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239008261683492e-05,
+      "loss": 0.8138,
+      "step": 1101
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237303968117816e-05,
+      "loss": 0.8601,
+      "step": 1102
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235597843904595e-05,
+      "loss": 0.9012,
+      "step": 1103
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233889889381944e-05,
+      "loss": 0.9127,
+      "step": 1104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232180104888345e-05,
+      "loss": 0.8751,
+      "step": 1105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230468490762644e-05,
+      "loss": 0.7706,
+      "step": 1106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922875504734405e-05,
+      "loss": 0.9033,
+      "step": 1107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227039774972128e-05,
+      "loss": 0.9286,
+      "step": 1108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225322673986814e-05,
+      "loss": 0.8652,
+      "step": 1109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223603744728408e-05,
+      "loss": 0.8516,
+      "step": 1110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922188298753756e-05,
+      "loss": 0.9331,
+      "step": 1111
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220160402755286e-05,
+      "loss": 0.8293,
+      "step": 1112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218435990722977e-05,
+      "loss": 0.8902,
+      "step": 1113
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921670975178237e-05,
+      "loss": 0.7557,
+      "step": 1114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214981686275577e-05,
+      "loss": 0.8798,
+      "step": 1115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921325179454506e-05,
+      "loss": 0.9638,
+      "step": 1116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211520076933653e-05,
+      "loss": 0.8583,
+      "step": 1117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209786533784543e-05,
+      "loss": 0.8592,
+      "step": 1118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9208051165441285e-05,
+      "loss": 0.8676,
+      "step": 1119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9206313972247797e-05,
+      "loss": 0.8763,
+      "step": 1120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204574954548348e-05,
+      "loss": 0.8966,
+      "step": 1121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202834112687588e-05,
+      "loss": 0.8486,
+      "step": 1122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201091447010508e-05,
+      "loss": 0.8435,
+      "step": 1123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199346957862475e-05,
+      "loss": 0.8921,
+      "step": 1124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197600645589204e-05,
+      "loss": 0.8665,
+      "step": 1125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9195852510536782e-05,
+      "loss": 0.9014,
+      "step": 1126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919410255305166e-05,
+      "loss": 0.8525,
+      "step": 1127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9192350773480637e-05,
+      "loss": 0.8316,
+      "step": 1128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190597172170886e-05,
+      "loss": 0.9467,
+      "step": 1129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9188841749469928e-05,
+      "loss": 0.8878,
+      "step": 1130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187084505725657e-05,
+      "loss": 0.8568,
+      "step": 1131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9185325441286323e-05,
+      "loss": 0.9109,
+      "step": 1132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183564556500543e-05,
+      "loss": 0.9306,
+      "step": 1133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918180185171728e-05,
+      "loss": 0.8705,
+      "step": 1134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918003732728587e-05,
+      "loss": 0.8569,
+      "step": 1135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178270983556004e-05,
+      "loss": 0.7956,
+      "step": 1136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917650282087774e-05,
+      "loss": 0.8775,
+      "step": 1137
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917473283960149e-05,
+      "loss": 0.8873,
+      "step": 1138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9172961040078025e-05,
+      "loss": 0.8819,
+      "step": 1139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917118742265849e-05,
+      "loss": 0.8309,
+      "step": 1140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916941198769437e-05,
+      "loss": 0.8921,
+      "step": 1141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167634735537525e-05,
+      "loss": 0.9312,
+      "step": 1142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165855666540166e-05,
+      "loss": 0.8388,
+      "step": 1143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916407478105488e-05,
+      "loss": 0.9359,
+      "step": 1144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916229207943459e-05,
+      "loss": 0.8752,
+      "step": 1145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160507562032598e-05,
+      "loss": 0.8693,
+      "step": 1146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158721229202556e-05,
+      "loss": 0.8478,
+      "step": 1147
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156933081298486e-05,
+      "loss": 0.8107,
+      "step": 1148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155143118674754e-05,
+      "loss": 0.8216,
+      "step": 1149
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91533513416861e-05,
+      "loss": 0.901,
+      "step": 1150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915155775068762e-05,
+      "loss": 0.9455,
+      "step": 1151
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914976234603476e-05,
+      "loss": 0.8068,
+      "step": 1152
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147965128083336e-05,
+      "loss": 0.8723,
+      "step": 1153
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146166097189526e-05,
+      "loss": 0.9185,
+      "step": 1154
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144365253709854e-05,
+      "loss": 0.852,
+      "step": 1155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142562598001217e-05,
+      "loss": 0.8146,
+      "step": 1156
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914075813042086e-05,
+      "loss": 0.8986,
+      "step": 1157
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138951851326398e-05,
+      "loss": 0.913,
+      "step": 1158
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913714376107579e-05,
+      "loss": 0.8577,
+      "step": 1159
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9135333860027374e-05,
+      "loss": 0.8163,
+      "step": 1160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913352214853983e-05,
+      "loss": 0.8831,
+      "step": 1161
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9131708626972202e-05,
+      "loss": 0.8407,
+      "step": 1162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129893295683893e-05,
+      "loss": 0.8577,
+      "step": 1163
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128076155034668e-05,
+      "loss": 0.9533,
+      "step": 1164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912625720538465e-05,
+      "loss": 0.8144,
+      "step": 1165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912443644709431e-05,
+      "loss": 0.9152,
+      "step": 1166
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122613880524496e-05,
+      "loss": 0.8543,
+      "step": 1167
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120789506036393e-05,
+      "loss": 0.8407,
+      "step": 1168
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118963323991562e-05,
+      "loss": 0.8761,
+      "step": 1169
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117135334751916e-05,
+      "loss": 0.3966,
+      "step": 1170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911530553867972e-05,
+      "loss": 0.8526,
+      "step": 1171
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911347393613761e-05,
+      "loss": 0.9153,
+      "step": 1172
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911164052748857e-05,
+      "loss": 0.7652,
+      "step": 1173
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109805313095935e-05,
+      "loss": 0.9504,
+      "step": 1174
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107968293323427e-05,
+      "loss": 0.8127,
+      "step": 1175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910612946853509e-05,
+      "loss": 0.8683,
+      "step": 1176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910428883909535e-05,
+      "loss": 0.8232,
+      "step": 1177
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910244640536898e-05,
+      "loss": 0.8815,
+      "step": 1178
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100602167721113e-05,
+      "loss": 0.9477,
+      "step": 1179
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098756126517243e-05,
+      "loss": 0.8044,
+      "step": 1180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909690828212321e-05,
+      "loss": 0.8737,
+      "step": 1181
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909505863490523e-05,
+      "loss": 0.8157,
+      "step": 1182
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093207185229857e-05,
+      "loss": 0.8004,
+      "step": 1183
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909135393346402e-05,
+      "loss": 0.9044,
+      "step": 1184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089498879974986e-05,
+      "loss": 0.8011,
+      "step": 1185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087642025130395e-05,
+      "loss": 0.9088,
+      "step": 1186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908578336929824e-05,
+      "loss": 0.8757,
+      "step": 1187
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9083922912846864e-05,
+      "loss": 0.4305,
+      "step": 1188
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082060656144978e-05,
+      "loss": 0.8524,
+      "step": 1189
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080196599561636e-05,
+      "loss": 0.8496,
+      "step": 1190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907833074346626e-05,
+      "loss": 0.9151,
+      "step": 1191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907646308822863e-05,
+      "loss": 0.8248,
+      "step": 1192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074593634218873e-05,
+      "loss": 0.8354,
+      "step": 1193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907272238180748e-05,
+      "loss": 0.9043,
+      "step": 1194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070849331365287e-05,
+      "loss": 0.9093,
+      "step": 1195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068974483263507e-05,
+      "loss": 0.8567,
+      "step": 1196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067097837873686e-05,
+      "loss": 0.9355,
+      "step": 1197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9065219395567748e-05,
+      "loss": 0.8928,
+      "step": 1198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9063339156717956e-05,
+      "loss": 0.9781,
+      "step": 1199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9061457121696933e-05,
+      "loss": 0.8656,
+      "step": 1200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9059573290877666e-05,
+      "loss": 0.9103,
+      "step": 1201
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905768766463349e-05,
+      "loss": 0.8365,
+      "step": 1202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055800243338096e-05,
+      "loss": 0.9225,
+      "step": 1203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053911027365538e-05,
+      "loss": 0.9035,
+      "step": 1204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052020017090216e-05,
+      "loss": 0.7602,
+      "step": 1205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905012721288689e-05,
+      "loss": 0.8619,
+      "step": 1206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904823261513068e-05,
+      "loss": 0.8967,
+      "step": 1207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046336224197057e-05,
+      "loss": 0.8138,
+      "step": 1208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904443804046184e-05,
+      "loss": 0.8499,
+      "step": 1209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042538064301222e-05,
+      "loss": 0.8504,
+      "step": 1210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904063629609173e-05,
+      "loss": 0.8468,
+      "step": 1211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903873273621026e-05,
+      "loss": 0.8785,
+      "step": 1212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903682738503406e-05,
+      "loss": 0.7758,
+      "step": 1213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034920242940733e-05,
+      "loss": 0.8489,
+      "step": 1214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033011310308232e-05,
+      "loss": 0.8101,
+      "step": 1215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903110058751487e-05,
+      "loss": 0.8709,
+      "step": 1216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029188074939317e-05,
+      "loss": 0.8973,
+      "step": 1217
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027273772960596e-05,
+      "loss": 0.8363,
+      "step": 1218
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025357681958076e-05,
+      "loss": 0.907,
+      "step": 1219
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023439802311493e-05,
+      "loss": 0.8275,
+      "step": 1220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902152013440093e-05,
+      "loss": 0.8581,
+      "step": 1221
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019598678606827e-05,
+      "loss": 0.8521,
+      "step": 1222
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901767543530998e-05,
+      "loss": 0.8365,
+      "step": 1223
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015750404891532e-05,
+      "loss": 0.925,
+      "step": 1224
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013823587732988e-05,
+      "loss": 0.8212,
+      "step": 1225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9011894984216205e-05,
+      "loss": 0.8976,
+      "step": 1226
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9009964594723395e-05,
+      "loss": 0.905,
+      "step": 1227
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008032419637123e-05,
+      "loss": 0.8467,
+      "step": 1228
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006098459340298e-05,
+      "loss": 0.8884,
+      "step": 1229
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9004162714216202e-05,
+      "loss": 0.8389,
+      "step": 1230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9002225184648457e-05,
+      "loss": 0.8431,
+      "step": 1231
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9000285871021043e-05,
+      "loss": 0.8495,
+      "step": 1232
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998344773718293e-05,
+      "loss": 0.8868,
+      "step": 1233
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996401893124888e-05,
+      "loss": 0.8813,
+      "step": 1234
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994457229625877e-05,
+      "loss": 0.8322,
+      "step": 1235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992510783606645e-05,
+      "loss": 0.8615,
+      "step": 1236
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8990562555452945e-05,
+      "loss": 0.8757,
+      "step": 1237
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898861254555087e-05,
+      "loss": 0.8427,
+      "step": 1238
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898666075428688e-05,
+      "loss": 0.8672,
+      "step": 1239
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984707182047777e-05,
+      "loss": 0.8599,
+      "step": 1240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898275182922072e-05,
+      "loss": 0.873,
+      "step": 1241
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980794696193218e-05,
+      "loss": 0.8461,
+      "step": 1242
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978835783353137e-05,
+      "loss": 0.7724,
+      "step": 1243
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89768750910887e-05,
+      "loss": 0.8393,
+      "step": 1244
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974912619788464e-05,
+      "loss": 0.8355,
+      "step": 1245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897294836984136e-05,
+      "loss": 0.8076,
+      "step": 1246
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970982341636664e-05,
+      "loss": 0.9,
+      "step": 1247
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969014535564e-05,
+      "loss": 0.9376,
+      "step": 1248
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967044952013347e-05,
+      "loss": 0.8207,
+      "step": 1249
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965073591375038e-05,
+      "loss": 0.8706,
+      "step": 1250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8963100454039758e-05,
+      "loss": 0.9026,
+      "step": 1251
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896112554039854e-05,
+      "loss": 0.8421,
+      "step": 1252
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959148850842773e-05,
+      "loss": 0.7744,
+      "step": 1253
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89571703857642e-05,
+      "loss": 0.8947,
+      "step": 1254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955190145554915e-05,
+      "loss": 0.9162,
+      "step": 1255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895320813060735e-05,
+      "loss": 0.8861,
+      "step": 1256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951224341314313e-05,
+      "loss": 0.9037,
+      "step": 1257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8949238778068943e-05,
+      "loss": 0.8999,
+      "step": 1258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894725144126474e-05,
+      "loss": 0.9001,
+      "step": 1259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894526233129556e-05,
+      "loss": 0.9396,
+      "step": 1260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943271448555604e-05,
+      "loss": 0.8661,
+      "step": 1261
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8941278793439415e-05,
+      "loss": 0.9163,
+      "step": 1262
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8939284366341902e-05,
+      "loss": 0.8484,
+      "step": 1263
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937288167658327e-05,
+      "loss": 0.7352,
+      "step": 1264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935290197784288e-05,
+      "loss": 0.8685,
+      "step": 1265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933290457115746e-05,
+      "loss": 0.8349,
+      "step": 1266
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893128894604901e-05,
+      "loss": 0.8199,
+      "step": 1267
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8929285664980733e-05,
+      "loss": 0.8948,
+      "step": 1268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927280614307934e-05,
+      "loss": 0.8381,
+      "step": 1269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892527379442797e-05,
+      "loss": 0.8939,
+      "step": 1270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892326520573855e-05,
+      "loss": 0.8794,
+      "step": 1271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8921254848637735e-05,
+      "loss": 0.4035,
+      "step": 1272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8919242723523944e-05,
+      "loss": 0.8852,
+      "step": 1273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8917228830795935e-05,
+      "loss": 0.8947,
+      "step": 1274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8915213170852817e-05,
+      "loss": 0.8524,
+      "step": 1275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913195744094063e-05,
+      "loss": 0.8851,
+      "step": 1276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891117655091948e-05,
+      "loss": 0.7737,
+      "step": 1277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890915559172923e-05,
+      "loss": 0.39,
+      "step": 1278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890713286692383e-05,
+      "loss": 0.9287,
+      "step": 1279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890510837690414e-05,
+      "loss": 0.8626,
+      "step": 1280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890308212207138e-05,
+      "loss": 0.8769,
+      "step": 1281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8901054102827105e-05,
+      "loss": 0.9165,
+      "step": 1282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889902431957323e-05,
+      "loss": 0.8049,
+      "step": 1283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889699277271202e-05,
+      "loss": 0.9166,
+      "step": 1284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889495946264608e-05,
+      "loss": 0.854,
+      "step": 1285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889292438977838e-05,
+      "loss": 0.8021,
+      "step": 1286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890887554512223e-05,
+      "loss": 0.8896,
+      "step": 1287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8888848957251276e-05,
+      "loss": 0.8636,
+      "step": 1288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888680859839954e-05,
+      "loss": 0.8938,
+      "step": 1289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884766478361375e-05,
+      "loss": 0.8675,
+      "step": 1290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8882722597541495e-05,
+      "loss": 0.8677,
+      "step": 1291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880676956344946e-05,
+      "loss": 0.8555,
+      "step": 1292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887862955517714e-05,
+      "loss": 0.8384,
+      "step": 1293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8876580394443826e-05,
+      "loss": 0.8748,
+      "step": 1294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887452947455111e-05,
+      "loss": 0.9479,
+      "step": 1295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872476795905443e-05,
+      "loss": 0.8219,
+      "step": 1296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887042235891362e-05,
+      "loss": 0.385,
+      "step": 1297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868366163982794e-05,
+      "loss": 0.8514,
+      "step": 1298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8866308211520462e-05,
+      "loss": 0.8565,
+      "step": 1299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864248501934465e-05,
+      "loss": 0.8824,
+      "step": 1300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862187035632994e-05,
+      "loss": 0.9017,
+      "step": 1301
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8860123813024597e-05,
+      "loss": 0.8959,
+      "step": 1302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858058834518157e-05,
+      "loss": 0.8832,
+      "step": 1303
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855992100522914e-05,
+      "loss": 0.8998,
+      "step": 1304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885392361144845e-05,
+      "loss": 0.8282,
+      "step": 1305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851853367704707e-05,
+      "loss": 0.7466,
+      "step": 1306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8849781369701955e-05,
+      "loss": 0.7948,
+      "step": 1307
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847707617850824e-05,
+      "loss": 0.8677,
+      "step": 1308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845632112562295e-05,
+      "loss": 0.8385,
+      "step": 1309
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884355485424769e-05,
+      "loss": 0.885,
+      "step": 1310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8841475843318678e-05,
+      "loss": 0.8473,
+      "step": 1311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8839395080187275e-05,
+      "loss": 0.9115,
+      "step": 1312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883731256526585e-05,
+      "loss": 0.8783,
+      "step": 1313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8835228298967118e-05,
+      "loss": 0.8824,
+      "step": 1314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883314228170413e-05,
+      "loss": 0.9429,
+      "step": 1315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88310545138903e-05,
+      "loss": 0.8011,
+      "step": 1316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828964995939375e-05,
+      "loss": 0.845,
+      "step": 1317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882687372826546e-05,
+      "loss": 0.8112,
+      "step": 1318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824780711283007e-05,
+      "loss": 0.8793,
+      "step": 1319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88226859454068e-05,
+      "loss": 0.8997,
+      "step": 1320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820589431051984e-05,
+      "loss": 0.8416,
+      "step": 1321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8818491168634043e-05,
+      "loss": 0.8907,
+      "step": 1322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8816391158568817e-05,
+      "loss": 0.9265,
+      "step": 1323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814289401272477e-05,
+      "loss": 0.905,
+      "step": 1324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812185897161558e-05,
+      "loss": 0.9407,
+      "step": 1325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810080646652928e-05,
+      "loss": 0.9153,
+      "step": 1326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88079736501638e-05,
+      "loss": 0.8809,
+      "step": 1327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8805864908111745e-05,
+      "loss": 0.8868,
+      "step": 1328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8803754420914675e-05,
+      "loss": 0.8736,
+      "step": 1329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880164218899084e-05,
+      "loss": 0.8734,
+      "step": 1330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799528212758842e-05,
+      "loss": 0.8554,
+      "step": 1331
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797412492637632e-05,
+      "loss": 0.8657,
+      "step": 1332
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8795295029046504e-05,
+      "loss": 0.8327,
+      "step": 1333
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879317582240509e-05,
+      "loss": 0.9007,
+      "step": 1334
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791054873133384e-05,
+      "loss": 0.3817,
+      "step": 1335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788932181651704e-05,
+      "loss": 0.9165,
+      "step": 1336
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786807748380735e-05,
+      "loss": 0.9109,
+      "step": 1337
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878468157374149e-05,
+      "loss": 0.8924,
+      "step": 1338
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8782553658155337e-05,
+      "loss": 0.8781,
+      "step": 1339
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8780424002043985e-05,
+      "loss": 0.8005,
+      "step": 1340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877829260582949e-05,
+      "loss": 0.8271,
+      "step": 1341
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877615946993425e-05,
+      "loss": 0.8826,
+      "step": 1342
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8774024594781015e-05,
+      "loss": 0.8601,
+      "step": 1343
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8771887980792867e-05,
+      "loss": 0.8181,
+      "step": 1344
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876974962839324e-05,
+      "loss": 0.9044,
+      "step": 1345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876760953800592e-05,
+      "loss": 0.8474,
+      "step": 1346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8765467710055023e-05,
+      "loss": 0.9392,
+      "step": 1347
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763324144965016e-05,
+      "loss": 0.8478,
+      "step": 1348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8761178843160715e-05,
+      "loss": 0.8942,
+      "step": 1349
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8759031805067272e-05,
+      "loss": 0.8578,
+      "step": 1350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875688303111019e-05,
+      "loss": 0.8152,
+      "step": 1351
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875473252171531e-05,
+      "loss": 0.387,
+      "step": 1352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752580277308815e-05,
+      "loss": 0.9106,
+      "step": 1353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750426298317246e-05,
+      "loss": 0.853,
+      "step": 1354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748270585167473e-05,
+      "loss": 0.8628,
+      "step": 1355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746113138286716e-05,
+      "loss": 0.8087,
+      "step": 1356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8743953958102538e-05,
+      "loss": 0.8778,
+      "step": 1357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741793045042845e-05,
+      "loss": 0.811,
+      "step": 1358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739630399535885e-05,
+      "loss": 0.8618,
+      "step": 1359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737466022010248e-05,
+      "loss": 0.9032,
+      "step": 1360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873529991289488e-05,
+      "loss": 0.9701,
+      "step": 1361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8733132072619048e-05,
+      "loss": 0.8853,
+      "step": 1362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730962501612383e-05,
+      "loss": 0.8914,
+      "step": 1363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728791200304845e-05,
+      "loss": 0.8658,
+      "step": 1364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872661816912675e-05,
+      "loss": 0.8816,
+      "step": 1365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724443408508738e-05,
+      "loss": 0.9424,
+      "step": 1366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872226691888181e-05,
+      "loss": 0.8666,
+      "step": 1367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87200887006773e-05,
+      "loss": 0.9255,
+      "step": 1368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717908754326886e-05,
+      "loss": 0.8274,
+      "step": 1369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715727080262594e-05,
+      "loss": 0.9012,
+      "step": 1370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713543678916783e-05,
+      "loss": 0.8964,
+      "step": 1371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871135855072216e-05,
+      "loss": 0.8748,
+      "step": 1372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870917169611178e-05,
+      "loss": 0.8016,
+      "step": 1373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706983115519023e-05,
+      "loss": 0.7975,
+      "step": 1374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704792809377626e-05,
+      "loss": 0.894,
+      "step": 1375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702600778121666e-05,
+      "loss": 0.8122,
+      "step": 1376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870040702218556e-05,
+      "loss": 0.8217,
+      "step": 1377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698211542004057e-05,
+      "loss": 0.8894,
+      "step": 1378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869601433801227e-05,
+      "loss": 0.8675,
+      "step": 1379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8693815410645627e-05,
+      "loss": 0.9435,
+      "step": 1380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8691614760339925e-05,
+      "loss": 0.7922,
+      "step": 1381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868941238753128e-05,
+      "loss": 0.8835,
+      "step": 1382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868720829265616e-05,
+      "loss": 0.927,
+      "step": 1383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868500247615137e-05,
+      "loss": 0.4087,
+      "step": 1384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682794938454064e-05,
+      "loss": 0.8151,
+      "step": 1385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680585680001724e-05,
+      "loss": 0.9029,
+      "step": 1386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678374701232186e-05,
+      "loss": 0.7858,
+      "step": 1387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867616200258362e-05,
+      "loss": 0.8266,
+      "step": 1388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8673947584494536e-05,
+      "loss": 0.9354,
+      "step": 1389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671731447403793e-05,
+      "loss": 0.8446,
+      "step": 1390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866951359175058e-05,
+      "loss": 0.8668,
+      "step": 1391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866729401797443e-05,
+      "loss": 0.8588,
+      "step": 1392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8665072726515224e-05,
+      "loss": 0.8332,
+      "step": 1393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662849717813166e-05,
+      "loss": 0.8903,
+      "step": 1394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8660624992308823e-05,
+      "loss": 0.8204,
+      "step": 1395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8658398550443092e-05,
+      "loss": 0.9382,
+      "step": 1396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656170392657197e-05,
+      "loss": 0.8482,
+      "step": 1397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653940519392723e-05,
+      "loss": 0.8419,
+      "step": 1398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865170893109158e-05,
+      "loss": 0.8564,
+      "step": 1399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864947562819603e-05,
+      "loss": 0.8518,
+      "step": 1400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8647240611148664e-05,
+      "loss": 0.8992,
+      "step": 1401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8645003880392422e-05,
+      "loss": 0.8324,
+      "step": 1402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8642765436370576e-05,
+      "loss": 0.3998,
+      "step": 1403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864052527952674e-05,
+      "loss": 0.9071,
+      "step": 1404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8638283410304868e-05,
+      "loss": 0.3966,
+      "step": 1405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636039829149255e-05,
+      "loss": 0.8268,
+      "step": 1406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863379453650453e-05,
+      "loss": 0.9171,
+      "step": 1407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631547532815672e-05,
+      "loss": 0.8397,
+      "step": 1408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629298818527984e-05,
+      "loss": 0.8503,
+      "step": 1409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8627048394087118e-05,
+      "loss": 0.9105,
+      "step": 1410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8624796259939067e-05,
+      "loss": 0.7989,
+      "step": 1411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622542416530154e-05,
+      "loss": 0.8306,
+      "step": 1412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862028686430705e-05,
+      "loss": 0.8732,
+      "step": 1413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8618029603716753e-05,
+      "loss": 0.8411,
+      "step": 1414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861577063520661e-05,
+      "loss": 0.8619,
+      "step": 1415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8613509959224306e-05,
+      "loss": 0.8412,
+      "step": 1416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8611247576217856e-05,
+      "loss": 0.9082,
+      "step": 1417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8608983486635625e-05,
+      "loss": 0.8454,
+      "step": 1418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8606717690926302e-05,
+      "loss": 0.8827,
+      "step": 1419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860445018953893e-05,
+      "loss": 0.9406,
+      "step": 1420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602180982922878e-05,
+      "loss": 0.4038,
+      "step": 1421
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599910071527862e-05,
+      "loss": 0.3751,
+      "step": 1422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859763745580392e-05,
+      "loss": 0.8791,
+      "step": 1423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8595363136201448e-05,
+      "loss": 0.7489,
+      "step": 1424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859308711317117e-05,
+      "loss": 0.8349,
+      "step": 1425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859080938716414e-05,
+      "loss": 0.9073,
+      "step": 1426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858852995863177e-05,
+      "loss": 0.3829,
+      "step": 1427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8586248828025785e-05,
+      "loss": 0.8102,
+      "step": 1428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583965995798267e-05,
+      "loss": 0.8605,
+      "step": 1429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858168146240162e-05,
+      "loss": 0.7948,
+      "step": 1430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85793952282886e-05,
+      "loss": 0.8878,
+      "step": 1431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857710729391229e-05,
+      "loss": 0.9073,
+      "step": 1432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857481765972611e-05,
+      "loss": 0.8405,
+      "step": 1433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857252632618382e-05,
+      "loss": 0.8624,
+      "step": 1434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570233293739517e-05,
+      "loss": 0.8569,
+      "step": 1435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856793856284764e-05,
+      "loss": 0.8971,
+      "step": 1436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8565642133962945e-05,
+      "loss": 0.8443,
+      "step": 1437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8563344007540547e-05,
+      "loss": 0.4253,
+      "step": 1438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856104418403589e-05,
+      "loss": 0.8416,
+      "step": 1439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855874266390475e-05,
+      "loss": 0.9191,
+      "step": 1440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8556439447603242e-05,
+      "loss": 0.8882,
+      "step": 1441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855413453558781e-05,
+      "loss": 0.8597,
+      "step": 1442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551827928315257e-05,
+      "loss": 0.8978,
+      "step": 1443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854951962624269e-05,
+      "loss": 0.8503,
+      "step": 1444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547209629827576e-05,
+      "loss": 0.8845,
+      "step": 1445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854489793952771e-05,
+      "loss": 0.9068,
+      "step": 1446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854258455580122e-05,
+      "loss": 0.8739,
+      "step": 1447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540269479106573e-05,
+      "loss": 0.7831,
+      "step": 1448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853795270990257e-05,
+      "loss": 0.801,
+      "step": 1449
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853563424864835e-05,
+      "loss": 0.7944,
+      "step": 1450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8533314095803383e-05,
+      "loss": 0.3651,
+      "step": 1451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8530992251827477e-05,
+      "loss": 0.8084,
+      "step": 1452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8528668717180774e-05,
+      "loss": 0.8658,
+      "step": 1453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852634349232376e-05,
+      "loss": 0.9385,
+      "step": 1454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852401657771723e-05,
+      "loss": 0.861,
+      "step": 1455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8521687973822348e-05,
+      "loss": 0.8778,
+      "step": 1456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851935768110059e-05,
+      "loss": 0.7694,
+      "step": 1457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8517025700013775e-05,
+      "loss": 0.896,
+      "step": 1458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851469203102405e-05,
+      "loss": 0.9089,
+      "step": 1459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8512356674593904e-05,
+      "loss": 0.8969,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8510019631186163e-05,
+      "loss": 0.8319,
+      "step": 1461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507680901263975e-05,
+      "loss": 0.9083,
+      "step": 1462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505340485290827e-05,
+      "loss": 0.8668,
+      "step": 1463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502998383730554e-05,
+      "loss": 0.8632,
+      "step": 1464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85006545970473e-05,
+      "loss": 0.8166,
+      "step": 1465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849830912570556e-05,
+      "loss": 0.8662,
+      "step": 1466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849596197017017e-05,
+      "loss": 0.8698,
+      "step": 1467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8493613130906273e-05,
+      "loss": 0.8991,
+      "step": 1468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849126260837937e-05,
+      "loss": 0.8505,
+      "step": 1469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8488910403055287e-05,
+      "loss": 0.8818,
+      "step": 1470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486556515400185e-05,
+      "loss": 0.892,
+      "step": 1471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8484200945880552e-05,
+      "loss": 0.7791,
+      "step": 1472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481843694963215e-05,
+      "loss": 0.9785,
+      "step": 1473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847948476311534e-05,
+      "loss": 0.8308,
+      "step": 1474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8477124150804416e-05,
+      "loss": 0.7911,
+      "step": 1475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8474761858498268e-05,
+      "loss": 0.9074,
+      "step": 1476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847239788666505e-05,
+      "loss": 0.8237,
+      "step": 1477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470032235773266e-05,
+      "loss": 0.8551,
+      "step": 1478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467664906291726e-05,
+      "loss": 0.795,
+      "step": 1479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8465295898689598e-05,
+      "loss": 0.7996,
+      "step": 1480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846292521343637e-05,
+      "loss": 0.8109,
+      "step": 1481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8460552851001862e-05,
+      "loss": 0.8113,
+      "step": 1482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458178811856223e-05,
+      "loss": 0.8938,
+      "step": 1483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845580309646995e-05,
+      "loss": 0.8421,
+      "step": 1484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8453425705313856e-05,
+      "loss": 0.3448,
+      "step": 1485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845104663885909e-05,
+      "loss": 0.9403,
+      "step": 1486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448665897577137e-05,
+      "loss": 0.8736,
+      "step": 1487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844628348193982e-05,
+      "loss": 0.8335,
+      "step": 1488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844389939241927e-05,
+      "loss": 0.9231,
+      "step": 1489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844151362948798e-05,
+      "loss": 0.8556,
+      "step": 1490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8439126193618754e-05,
+      "loss": 0.84,
+      "step": 1491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8436737085284732e-05,
+      "loss": 0.8432,
+      "step": 1492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843434630495939e-05,
+      "loss": 0.8505,
+      "step": 1493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843195385311653e-05,
+      "loss": 0.9128,
+      "step": 1494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429559730230294e-05,
+      "loss": 0.8199,
+      "step": 1495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842716393677514e-05,
+      "loss": 0.8002,
+      "step": 1496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8424766473225873e-05,
+      "loss": 0.8413,
+      "step": 1497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8422367340057617e-05,
+      "loss": 0.8669,
+      "step": 1498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841996653774584e-05,
+      "loss": 0.8732,
+      "step": 1499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841756406676632e-05,
+      "loss": 0.8745,
+      "step": 1500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841515992759519e-05,
+      "loss": 0.8602,
+      "step": 1501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412754120708892e-05,
+      "loss": 0.8718,
+      "step": 1502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8410346646584218e-05,
+      "loss": 0.7819,
+      "step": 1503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407937505698275e-05,
+      "loss": 0.8732,
+      "step": 1504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8405526698528508e-05,
+      "loss": 0.7717,
+      "step": 1505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840311422555269e-05,
+      "loss": 0.8332,
+      "step": 1506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8400700087248925e-05,
+      "loss": 0.3296,
+      "step": 1507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8398284284095646e-05,
+      "loss": 0.8415,
+      "step": 1508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8395866816571618e-05,
+      "loss": 0.8876,
+      "step": 1509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8393447685155926e-05,
+      "loss": 0.8482,
+      "step": 1510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8391026890328008e-05,
+      "loss": 0.8003,
+      "step": 1511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8388604432567607e-05,
+      "loss": 0.8252,
+      "step": 1512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8386180312354806e-05,
+      "loss": 0.8227,
+      "step": 1513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838375453017002e-05,
+      "loss": 0.9108,
+      "step": 1514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381327086493987e-05,
+      "loss": 0.7899,
+      "step": 1515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837889798180778e-05,
+      "loss": 0.8004,
+      "step": 1516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376467216592795e-05,
+      "loss": 0.8261,
+      "step": 1517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8374034791330763e-05,
+      "loss": 0.7826,
+      "step": 1518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8371600706503746e-05,
+      "loss": 0.8764,
+      "step": 1519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369164962594118e-05,
+      "loss": 0.8966,
+      "step": 1520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836672756008461e-05,
+      "loss": 0.8324,
+      "step": 1521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364288499458256e-05,
+      "loss": 0.8345,
+      "step": 1522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8361847781198432e-05,
+      "loss": 0.9017,
+      "step": 1523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835940540578884e-05,
+      "loss": 0.874,
+      "step": 1524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356961373713505e-05,
+      "loss": 0.8455,
+      "step": 1525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8354515685456787e-05,
+      "loss": 0.9108,
+      "step": 1526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835206834150338e-05,
+      "loss": 0.8489,
+      "step": 1527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8349619342338286e-05,
+      "loss": 0.8662,
+      "step": 1528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834716868844685e-05,
+      "loss": 0.7652,
+      "step": 1529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8344716380314752e-05,
+      "loss": 0.8377,
+      "step": 1530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834226241842798e-05,
+      "loss": 0.8378,
+      "step": 1531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833980680327286e-05,
+      "loss": 0.8525,
+      "step": 1532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337349535336053e-05,
+      "loss": 0.819,
+      "step": 1533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833489061510453e-05,
+      "loss": 0.8497,
+      "step": 1534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833243004306561e-05,
+      "loss": 0.9032,
+      "step": 1535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832996781970692e-05,
+      "loss": 0.8468,
+      "step": 1536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832750394551643e-05,
+      "loss": 0.8877,
+      "step": 1537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8325038420982424e-05,
+      "loss": 0.9224,
+      "step": 1538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8322571246593524e-05,
+      "loss": 0.8875,
+      "step": 1539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832010242283867e-05,
+      "loss": 0.9128,
+      "step": 1540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8317631950207137e-05,
+      "loss": 0.8714,
+      "step": 1541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315159829188523e-05,
+      "loss": 0.7956,
+      "step": 1542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8312686060272747e-05,
+      "loss": 0.8238,
+      "step": 1543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831021064395007e-05,
+      "loss": 0.8844,
+      "step": 1544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830773358071106e-05,
+      "loss": 0.8263,
+      "step": 1545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305254871046626e-05,
+      "loss": 0.8516,
+      "step": 1546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8302774515447998e-05,
+      "loss": 0.7943,
+      "step": 1547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300292514406727e-05,
+      "loss": 0.8571,
+      "step": 1548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297808868414706e-05,
+      "loss": 0.8546,
+      "step": 1549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8295323577964136e-05,
+      "loss": 0.8862,
+      "step": 1550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829283664354755e-05,
+      "loss": 0.8942,
+      "step": 1551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290348065657814e-05,
+      "loss": 0.8506,
+      "step": 1552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287857844788114e-05,
+      "loss": 0.8918,
+      "step": 1553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8285365981431957e-05,
+      "loss": 0.82,
+      "step": 1554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8282872476083183e-05,
+      "loss": 0.9034,
+      "step": 1555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280377329235953e-05,
+      "loss": 0.8128,
+      "step": 1556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827788054138476e-05,
+      "loss": 0.9341,
+      "step": 1557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275382113024406e-05,
+      "loss": 0.9121,
+      "step": 1558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827288204465004e-05,
+      "loss": 0.8469,
+      "step": 1559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827038033675712e-05,
+      "loss": 0.8536,
+      "step": 1560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826787698984144e-05,
+      "loss": 0.7684,
+      "step": 1561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265372004399103e-05,
+      "loss": 0.845,
+      "step": 1562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8262865380926554e-05,
+      "loss": 0.8528,
+      "step": 1563
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826035711992055e-05,
+      "loss": 0.9169,
+      "step": 1564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8257847221878182e-05,
+      "loss": 0.8079,
+      "step": 1565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8255335687296862e-05,
+      "loss": 0.8538,
+      "step": 1566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8252822516674325e-05,
+      "loss": 0.8406,
+      "step": 1567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8250307710508627e-05,
+      "loss": 0.7922,
+      "step": 1568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8247791269298154e-05,
+      "loss": 0.8967,
+      "step": 1569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245273193541617e-05,
+      "loss": 0.8159,
+      "step": 1570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242753483738042e-05,
+      "loss": 0.9073,
+      "step": 1571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824023214038679e-05,
+      "loss": 0.8903,
+      "step": 1572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8237709163987538e-05,
+      "loss": 0.3681,
+      "step": 1573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8235184555040288e-05,
+      "loss": 0.7996,
+      "step": 1574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823265831404537e-05,
+      "loss": 0.8879,
+      "step": 1575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8230130441503434e-05,
+      "loss": 0.8738,
+      "step": 1576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8227600937915447e-05,
+      "loss": 0.7751,
+      "step": 1577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8225069803782714e-05,
+      "loss": 0.7958,
+      "step": 1578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822253703960685e-05,
+      "loss": 0.9077,
+      "step": 1579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82200026458898e-05,
+      "loss": 0.974,
+      "step": 1580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217466623133825e-05,
+      "loss": 0.795,
+      "step": 1581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821492897184152e-05,
+      "loss": 0.8737,
+      "step": 1582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8212389692515796e-05,
+      "loss": 0.8976,
+      "step": 1583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8209848785659882e-05,
+      "loss": 0.8608,
+      "step": 1584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207306251777336e-05,
+      "loss": 0.8905,
+      "step": 1585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204762091372038e-05,
+      "loss": 0.364,
+      "step": 1586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8202216304948192e-05,
+      "loss": 0.8478,
+      "step": 1587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819966889301032e-05,
+      "loss": 0.8032,
+      "step": 1588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197119856063265e-05,
+      "loss": 0.7913,
+      "step": 1589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194569194612195e-05,
+      "loss": 0.8972,
+      "step": 1590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8192016909162606e-05,
+      "loss": 0.8087,
+      "step": 1591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81894630002203e-05,
+      "loss": 0.9073,
+      "step": 1592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186907468291417e-05,
+      "loss": 0.843,
+      "step": 1593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8184350313882412e-05,
+      "loss": 0.4054,
+      "step": 1594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8181791537500055e-05,
+      "loss": 0.9202,
+      "step": 1595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8179231139651452e-05,
+      "loss": 0.8138,
+      "step": 1596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817666912084402e-05,
+      "loss": 0.8577,
+      "step": 1597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8174105481585495e-05,
+      "loss": 0.3496,
+      "step": 1598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171540222383943e-05,
+      "loss": 0.9924,
+      "step": 1599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8168973343747747e-05,
+      "loss": 0.3324,
+      "step": 1600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816640484618561e-05,
+      "loss": 0.8864,
+      "step": 1601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816383473020655e-05,
+      "loss": 0.828,
+      "step": 1602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8161262996319927e-05,
+      "loss": 0.8559,
+      "step": 1603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158689645035392e-05,
+      "loss": 0.8451,
+      "step": 1604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815611467686294e-05,
+      "loss": 0.8272,
+      "step": 1605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815353809231287e-05,
+      "loss": 0.8883,
+      "step": 1606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8150959891895818e-05,
+      "loss": 0.837,
+      "step": 1607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148380076122728e-05,
+      "loss": 0.9452,
+      "step": 1608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8145798645504866e-05,
+      "loss": 0.8087,
+      "step": 1609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8143215600553824e-05,
+      "loss": 0.7213,
+      "step": 1610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81406309417815e-05,
+      "loss": 0.858,
+      "step": 1611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8138044669700134e-05,
+      "loss": 0.8349,
+      "step": 1612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813545678482226e-05,
+      "loss": 0.8515,
+      "step": 1613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132867287660754e-05,
+      "loss": 0.8711,
+      "step": 1614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81302761787288e-05,
+      "loss": 0.8129,
+      "step": 1615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8127683458539903e-05,
+      "loss": 0.3526,
+      "step": 1616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8125089127607885e-05,
+      "loss": 0.9417,
+      "step": 1617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8122493186446896e-05,
+      "loss": 0.8209,
+      "step": 1618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8119895635571394e-05,
+      "loss": 0.8472,
+      "step": 1619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811729647549616e-05,
+      "loss": 0.7874,
+      "step": 1620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114695706736306e-05,
+      "loss": 0.9564,
+      "step": 1621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811209332980724e-05,
+      "loss": 0.3494,
+      "step": 1622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8109489345224703e-05,
+      "loss": 0.8732,
+      "step": 1623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810688375350476e-05,
+      "loss": 0.8034,
+      "step": 1624
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810427655516378e-05,
+      "loss": 0.864,
+      "step": 1625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8101667750718457e-05,
+      "loss": 0.8327,
+      "step": 1626
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099057340685803e-05,
+      "loss": 0.832,
+      "step": 1627
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096445325583153e-05,
+      "loss": 0.7949,
+      "step": 1628
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093831705928155e-05,
+      "loss": 0.8806,
+      "step": 1629
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8091216482238772e-05,
+      "loss": 0.8579,
+      "step": 1630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808859965503329e-05,
+      "loss": 0.9247,
+      "step": 1631
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085981224830314e-05,
+      "loss": 0.8354,
+      "step": 1632
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808336119214876e-05,
+      "loss": 0.7828,
+      "step": 1633
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8080739557507868e-05,
+      "loss": 0.8585,
+      "step": 1634
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8078116321427197e-05,
+      "loss": 0.3659,
+      "step": 1635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807549148442661e-05,
+      "loss": 0.8702,
+      "step": 1636
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807286504702631e-05,
+      "loss": 0.8446,
+      "step": 1637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807023700974679e-05,
+      "loss": 0.7952,
+      "step": 1638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806760737310888e-05,
+      "loss": 0.8066,
+      "step": 1639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8064976137633728e-05,
+      "loss": 0.7832,
+      "step": 1640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062343303842782e-05,
+      "loss": 0.8837,
+      "step": 1641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8059708872257817e-05,
+      "loss": 0.8855,
+      "step": 1642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805707284340093e-05,
+      "loss": 0.8327,
+      "step": 1643
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805443521779453e-05,
+      "loss": 0.8241,
+      "step": 1644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805179599596133e-05,
+      "loss": 0.8787,
+      "step": 1645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8049155178424384e-05,
+      "loss": 0.7849,
+      "step": 1646
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8046512765707037e-05,
+      "loss": 0.8685,
+      "step": 1647
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8043868758332973e-05,
+      "loss": 0.9044,
+      "step": 1648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804122315682617e-05,
+      "loss": 0.7675,
+      "step": 1649
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038575961710944e-05,
+      "loss": 0.3501,
+      "step": 1650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8035927173511913e-05,
+      "loss": 0.9475,
+      "step": 1651
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033276792754005e-05,
+      "loss": 0.8761,
+      "step": 1652
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8030624819962478e-05,
+      "loss": 0.9083,
+      "step": 1653
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80279712556629e-05,
+      "loss": 0.8415,
+      "step": 1654
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8025316100381155e-05,
+      "loss": 0.8863,
+      "step": 1655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802265935464344e-05,
+      "loss": 0.8647,
+      "step": 1656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802000101897627e-05,
+      "loss": 0.8038,
+      "step": 1657
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8017341093906467e-05,
+      "loss": 0.8426,
+      "step": 1658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801467957996118e-05,
+      "loss": 0.8656,
+      "step": 1659
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801201647766787e-05,
+      "loss": 0.8868,
+      "step": 1660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8009351787554304e-05,
+      "loss": 0.8566,
+      "step": 1661
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8006685510148575e-05,
+      "loss": 0.8415,
+      "step": 1662
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8004017645979085e-05,
+      "loss": 0.8404,
+      "step": 1663
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8001348195574545e-05,
+      "loss": 0.841,
+      "step": 1664
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7998677159463996e-05,
+      "loss": 0.8349,
+      "step": 1665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996004538176775e-05,
+      "loss": 0.7893,
+      "step": 1666
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7993330332242546e-05,
+      "loss": 0.8795,
+      "step": 1667
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799065454219128e-05,
+      "loss": 0.8066,
+      "step": 1668
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7987977168553266e-05,
+      "loss": 0.8728,
+      "step": 1669
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7985298211859108e-05,
+      "loss": 0.8942,
+      "step": 1670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798261767263972e-05,
+      "loss": 0.9508,
+      "step": 1671
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7979935551426326e-05,
+      "loss": 0.858,
+      "step": 1672
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977251848750473e-05,
+      "loss": 0.8494,
+      "step": 1673
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7974566565144017e-05,
+      "loss": 0.7939,
+      "step": 1674
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7971879701139127e-05,
+      "loss": 0.8823,
+      "step": 1675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7969191257268286e-05,
+      "loss": 0.9161,
+      "step": 1676
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7966501234064285e-05,
+      "loss": 0.8272,
+      "step": 1677
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7963809632060238e-05,
+      "loss": 0.9372,
+      "step": 1678
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796111645178956e-05,
+      "loss": 0.9799,
+      "step": 1679
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7958421693785993e-05,
+      "loss": 0.8255,
+      "step": 1680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795572535858358e-05,
+      "loss": 0.9454,
+      "step": 1681
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7953027446716675e-05,
+      "loss": 0.8728,
+      "step": 1682
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7950327958719954e-05,
+      "loss": 0.8675,
+      "step": 1683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7947626895128408e-05,
+      "loss": 0.8164,
+      "step": 1684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794492425647732e-05,
+      "loss": 0.8016,
+      "step": 1685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794222004330231e-05,
+      "loss": 0.9015,
+      "step": 1686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7939514256139295e-05,
+      "loss": 0.9001,
+      "step": 1687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7936806895524507e-05,
+      "loss": 0.8072,
+      "step": 1688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793409796199449e-05,
+      "loss": 0.8625,
+      "step": 1689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7931387456086097e-05,
+      "loss": 0.8413,
+      "step": 1690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79286753783365e-05,
+      "loss": 0.8717,
+      "step": 1691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7925961729283178e-05,
+      "loss": 0.8431,
+      "step": 1692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7923246509463915e-05,
+      "loss": 0.887,
+      "step": 1693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7920529719416824e-05,
+      "loss": 0.7993,
+      "step": 1694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791781135968031e-05,
+      "loss": 0.8373,
+      "step": 1695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7915091430793103e-05,
+      "loss": 0.9268,
+      "step": 1696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7912369933294234e-05,
+      "loss": 0.8544,
+      "step": 1697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7909646867723046e-05,
+      "loss": 0.8692,
+      "step": 1698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7906922234619202e-05,
+      "loss": 0.8631,
+      "step": 1699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7904196034522667e-05,
+      "loss": 0.7881,
+      "step": 1700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790146826797372e-05,
+      "loss": 0.8488,
+      "step": 1701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7898738935512944e-05,
+      "loss": 0.3943,
+      "step": 1702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789600803768124e-05,
+      "loss": 0.9323,
+      "step": 1703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7893275575019828e-05,
+      "loss": 0.3508,
+      "step": 1704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7890541548070212e-05,
+      "loss": 0.8426,
+      "step": 1705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7887805957374227e-05,
+      "loss": 0.9115,
+      "step": 1706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7885068803474012e-05,
+      "loss": 0.9078,
+      "step": 1707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7882330086912016e-05,
+      "loss": 0.3668,
+      "step": 1708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7879589808231e-05,
+      "loss": 0.3594,
+      "step": 1709
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7876847967974022e-05,
+      "loss": 0.8612,
+      "step": 1710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7874104566684472e-05,
+      "loss": 0.8216,
+      "step": 1711
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787135960490603e-05,
+      "loss": 0.4085,
+      "step": 1712
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7868613083182694e-05,
+      "loss": 0.859,
+      "step": 1713
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7865865002058767e-05,
+      "loss": 0.8721,
+      "step": 1714
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7863115362078866e-05,
+      "loss": 0.8607,
+      "step": 1715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7860364163787912e-05,
+      "loss": 0.8073,
+      "step": 1716
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785761140773114e-05,
+      "loss": 0.8347,
+      "step": 1717
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785485709445409e-05,
+      "loss": 0.9559,
+      "step": 1718
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785210122450261e-05,
+      "loss": 0.8684,
+      "step": 1719
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7849343798422858e-05,
+      "loss": 0.3371,
+      "step": 1720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78465848167613e-05,
+      "loss": 0.8257,
+      "step": 1721
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784382428006471e-05,
+      "loss": 0.8562,
+      "step": 1722
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784106218888018e-05,
+      "loss": 0.856,
+      "step": 1723
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7838298543755087e-05,
+      "loss": 0.9367,
+      "step": 1724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835533345237134e-05,
+      "loss": 0.8342,
+      "step": 1725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7832766593874332e-05,
+      "loss": 0.8474,
+      "step": 1726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7829998290214994e-05,
+      "loss": 0.7734,
+      "step": 1727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782722843480774e-05,
+      "loss": 0.7956,
+      "step": 1728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78244570282015e-05,
+      "loss": 0.9126,
+      "step": 1729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782168407094551e-05,
+      "loss": 0.8132,
+      "step": 1730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7818909563589317e-05,
+      "loss": 0.9562,
+      "step": 1731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781613350668277e-05,
+      "loss": 0.8235,
+      "step": 1732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781335590077603e-05,
+      "loss": 0.7958,
+      "step": 1733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781057674641956e-05,
+      "loss": 0.848,
+      "step": 1734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7807796044164133e-05,
+      "loss": 0.8922,
+      "step": 1735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7805013794560834e-05,
+      "loss": 0.7942,
+      "step": 1736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7802229998161036e-05,
+      "loss": 0.737,
+      "step": 1737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7799444655516445e-05,
+      "loss": 0.8157,
+      "step": 1738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779665776717905e-05,
+      "loss": 0.8817,
+      "step": 1739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779386933370116e-05,
+      "loss": 0.8401,
+      "step": 1740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7791079355635384e-05,
+      "loss": 0.9293,
+      "step": 1741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778828783353465e-05,
+      "loss": 0.8388,
+      "step": 1742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7785494767952164e-05,
+      "loss": 0.8262,
+      "step": 1743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782700159441465e-05,
+      "loss": 0.7933,
+      "step": 1744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777990400855639e-05,
+      "loss": 0.8505,
+      "step": 1745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7777106315851077e-05,
+      "loss": 0.8253,
+      "step": 1746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7774307081879974e-05,
+      "loss": 0.8525,
+      "step": 1747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777150630719783e-05,
+      "loss": 0.8324,
+      "step": 1748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7768703992359708e-05,
+      "loss": 0.854,
+      "step": 1749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7765900137920957e-05,
+      "loss": 0.9212,
+      "step": 1750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776309474443726e-05,
+      "loss": 0.8441,
+      "step": 1751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7760287812464582e-05,
+      "loss": 0.9121,
+      "step": 1752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77574793425592e-05,
+      "loss": 0.9261,
+      "step": 1753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7754669335277694e-05,
+      "loss": 0.8667,
+      "step": 1754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7751857791176956e-05,
+      "loss": 0.8332,
+      "step": 1755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774904471081418e-05,
+      "loss": 0.8393,
+      "step": 1756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774623009474685e-05,
+      "loss": 0.8025,
+      "step": 1757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743413943532777e-05,
+      "loss": 0.8566,
+      "step": 1758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7740596257730062e-05,
+      "loss": 0.7592,
+      "step": 1759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7737777037897113e-05,
+      "loss": 0.8098,
+      "step": 1760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7734956284592647e-05,
+      "loss": 0.8434,
+      "step": 1761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7732133998375674e-05,
+      "loss": 0.8491,
+      "step": 1762
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7729310179805515e-05,
+      "loss": 0.8715,
+      "step": 1763
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7726484829441798e-05,
+      "loss": 0.852,
+      "step": 1764
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7723657947844444e-05,
+      "loss": 0.8905,
+      "step": 1765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7720829535573695e-05,
+      "loss": 0.9148,
+      "step": 1766
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771799959319007e-05,
+      "loss": 0.8511,
+      "step": 1767
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7715168121254422e-05,
+      "loss": 0.3444,
+      "step": 1768
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7712335120327883e-05,
+      "loss": 0.9345,
+      "step": 1769
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77095005909719e-05,
+      "loss": 0.8847,
+      "step": 1770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7706664533748216e-05,
+      "loss": 0.8314,
+      "step": 1771
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7703826949218886e-05,
+      "loss": 0.9089,
+      "step": 1772
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770098783794626e-05,
+      "loss": 0.8831,
+      "step": 1773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7698147200492984e-05,
+      "loss": 0.7971,
+      "step": 1774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7695305037422028e-05,
+      "loss": 0.8891,
+      "step": 1775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7692461349296642e-05,
+      "loss": 0.836,
+      "step": 1776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7689616136680395e-05,
+      "loss": 0.8376,
+      "step": 1777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7686769400137148e-05,
+      "loss": 0.7684,
+      "step": 1778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768392114023106e-05,
+      "loss": 0.8115,
+      "step": 1779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7681071357526613e-05,
+      "loss": 0.8791,
+      "step": 1780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7678220052588563e-05,
+      "loss": 0.3292,
+      "step": 1781
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767536722598199e-05,
+      "loss": 0.8322,
+      "step": 1782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7672512878272264e-05,
+      "loss": 0.8199,
+      "step": 1783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7669657010025055e-05,
+      "loss": 0.8589,
+      "step": 1784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7666799621806347e-05,
+      "loss": 0.8392,
+      "step": 1785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663940714182407e-05,
+      "loss": 0.8701,
+      "step": 1786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7661080287719824e-05,
+      "loss": 0.9113,
+      "step": 1787
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7658218342985462e-05,
+      "loss": 0.8247,
+      "step": 1788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765535488054652e-05,
+      "loss": 0.7753,
+      "step": 1789
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765248990097046e-05,
+      "loss": 0.8755,
+      "step": 1790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7649623404825073e-05,
+      "loss": 0.8063,
+      "step": 1791
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7646755392678442e-05,
+      "loss": 0.809,
+      "step": 1792
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7643885865098945e-05,
+      "loss": 0.7933,
+      "step": 1793
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7641014822655265e-05,
+      "loss": 0.792,
+      "step": 1794
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763814226591639e-05,
+      "loss": 0.3557,
+      "step": 1795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635268195451595e-05,
+      "loss": 0.872,
+      "step": 1796
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7632392611830464e-05,
+      "loss": 0.874,
+      "step": 1797
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7629515515622886e-05,
+      "loss": 0.7806,
+      "step": 1798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762663690739904e-05,
+      "loss": 0.8144,
+      "step": 1799
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7623756787729408e-05,
+      "loss": 0.805,
+      "step": 1800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7620875157184766e-05,
+      "loss": 0.7877,
+      "step": 1801
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7617992016336207e-05,
+      "loss": 0.7917,
+      "step": 1802
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7615107365755104e-05,
+      "loss": 0.8502,
+      "step": 1803
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7612221206013136e-05,
+      "loss": 0.8383,
+      "step": 1804
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7609333537682286e-05,
+      "loss": 0.8981,
+      "step": 1805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7606444361334828e-05,
+      "loss": 0.3697,
+      "step": 1806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760355367754334e-05,
+      "loss": 0.7521,
+      "step": 1807
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7600661486880693e-05,
+      "loss": 0.861,
+      "step": 1808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759776778992007e-05,
+      "loss": 0.3694,
+      "step": 1809
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7594872587234935e-05,
+      "loss": 0.835,
+      "step": 1810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7591975879399067e-05,
+      "loss": 0.3692,
+      "step": 1811
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7589077666986524e-05,
+      "loss": 0.8541,
+      "step": 1812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7586177950571685e-05,
+      "loss": 0.8947,
+      "step": 1813
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7583276730729207e-05,
+      "loss": 0.855,
+      "step": 1814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7580374008034058e-05,
+      "loss": 0.8324,
+      "step": 1815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7577469783061496e-05,
+      "loss": 0.8322,
+      "step": 1816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7574564056387086e-05,
+      "loss": 0.8911,
+      "step": 1817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7571656828586673e-05,
+      "loss": 0.8653,
+      "step": 1818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756874810023642e-05,
+      "loss": 0.8036,
+      "step": 1819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756583787191278e-05,
+      "loss": 0.8614,
+      "step": 1820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7562926144192494e-05,
+      "loss": 0.8387,
+      "step": 1821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7560012917652612e-05,
+      "loss": 0.8533,
+      "step": 1822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7557098192870476e-05,
+      "loss": 0.8147,
+      "step": 1823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755418197042373e-05,
+      "loss": 0.8345,
+      "step": 1824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7551264250890302e-05,
+      "loss": 0.8302,
+      "step": 1825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7548345034848432e-05,
+      "loss": 0.8988,
+      "step": 1826
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7545424322876646e-05,
+      "loss": 0.9143,
+      "step": 1827
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7542502115553775e-05,
+      "loss": 0.8786,
+      "step": 1828
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7539578413458937e-05,
+      "loss": 0.8833,
+      "step": 1829
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7536653217171556e-05,
+      "loss": 0.89,
+      "step": 1830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753372652727134e-05,
+      "loss": 0.851,
+      "step": 1831
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7530798344338303e-05,
+      "loss": 0.8029,
+      "step": 1832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7527868668952752e-05,
+      "loss": 0.8225,
+      "step": 1833
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752493750169529e-05,
+      "loss": 0.8133,
+      "step": 1834
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7522004843146818e-05,
+      "loss": 0.8311,
+      "step": 1835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7519070693888524e-05,
+      "loss": 0.9144,
+      "step": 1836
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75161350545019e-05,
+      "loss": 0.7571,
+      "step": 1837
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7513197925568733e-05,
+      "loss": 0.8563,
+      "step": 1838
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7510259307671103e-05,
+      "loss": 0.8215,
+      "step": 1839
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750731920139138e-05,
+      "loss": 0.7713,
+      "step": 1840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7504377607312237e-05,
+      "loss": 0.9498,
+      "step": 1841
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7501434526016633e-05,
+      "loss": 0.8661,
+      "step": 1842
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7498489958087842e-05,
+      "loss": 0.7871,
+      "step": 1843
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74955439041094e-05,
+      "loss": 0.7907,
+      "step": 1844
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7492596364665168e-05,
+      "loss": 0.8435,
+      "step": 1845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7489647340339283e-05,
+      "loss": 0.8257,
+      "step": 1846
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748669683171619e-05,
+      "loss": 0.8638,
+      "step": 1847
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748374483938061e-05,
+      "loss": 0.8024,
+      "step": 1848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7480791363917574e-05,
+      "loss": 0.8481,
+      "step": 1849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7477836405912397e-05,
+      "loss": 0.8051,
+      "step": 1850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7474879965950698e-05,
+      "loss": 0.8525,
+      "step": 1851
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7471922044618383e-05,
+      "loss": 0.8326,
+      "step": 1852
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746896264250165e-05,
+      "loss": 0.8542,
+      "step": 1853
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7466001760186992e-05,
+      "loss": 0.8867,
+      "step": 1854
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74630393982612e-05,
+      "loss": 0.3788,
+      "step": 1855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7460075557311352e-05,
+      "loss": 0.8828,
+      "step": 1856
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745711023792482e-05,
+      "loss": 0.8446,
+      "step": 1857
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7454143440689274e-05,
+      "loss": 0.7891,
+      "step": 1858
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745117516619267e-05,
+      "loss": 0.8622,
+      "step": 1859
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7448205415023265e-05,
+      "loss": 0.8786,
+      "step": 1860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74452341877696e-05,
+      "loss": 0.8686,
+      "step": 1861
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442261485020512e-05,
+      "loss": 0.905,
+      "step": 1862
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743928730736513e-05,
+      "loss": 0.7827,
+      "step": 1863
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7436311655392883e-05,
+      "loss": 0.8422,
+      "step": 1864
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743333452969348e-05,
+      "loss": 0.8232,
+      "step": 1865
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7430355930856925e-05,
+      "loss": 0.811,
+      "step": 1866
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742737585947352e-05,
+      "loss": 0.8469,
+      "step": 1867
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7424394316133857e-05,
+      "loss": 0.8759,
+      "step": 1868
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7421411301428813e-05,
+      "loss": 0.8992,
+      "step": 1869
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7418426815949562e-05,
+      "loss": 0.8684,
+      "step": 1870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7415440860287575e-05,
+      "loss": 0.8514,
+      "step": 1871
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7412453435034598e-05,
+      "loss": 0.8792,
+      "step": 1872
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7409464540782684e-05,
+      "loss": 0.8867,
+      "step": 1873
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7406474178124177e-05,
+      "loss": 0.8301,
+      "step": 1874
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7403482347651698e-05,
+      "loss": 0.8735,
+      "step": 1875
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740048904995817e-05,
+      "loss": 0.8931,
+      "step": 1876
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739749428563681e-05,
+      "loss": 0.829,
+      "step": 1877
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739449805528111e-05,
+      "loss": 0.7125,
+      "step": 1878
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739150035948487e-05,
+      "loss": 0.8425,
+      "step": 1879
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7388501198842168e-05,
+      "loss": 0.8205,
+      "step": 1880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7385500573947382e-05,
+      "loss": 0.8752,
+      "step": 1881
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738249848539517e-05,
+      "loss": 0.9325,
+      "step": 1882
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7379494933780494e-05,
+      "loss": 0.8057,
+      "step": 1883
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7376489919698588e-05,
+      "loss": 0.8341,
+      "step": 1884
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737348344374499e-05,
+      "loss": 0.8088,
+      "step": 1885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7370475506515525e-05,
+      "loss": 0.8903,
+      "step": 1886
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73674661086063e-05,
+      "loss": 0.8425,
+      "step": 1887
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736445525061372e-05,
+      "loss": 0.9041,
+      "step": 1888
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7361442933134478e-05,
+      "loss": 0.8567,
+      "step": 1889
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7358429156765553e-05,
+      "loss": 0.9097,
+      "step": 1890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7355413922104212e-05,
+      "loss": 0.8607,
+      "step": 1891
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735239722974802e-05,
+      "loss": 0.8424,
+      "step": 1892
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734937908029482e-05,
+      "loss": 0.8342,
+      "step": 1893
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7346359474342745e-05,
+      "loss": 0.8514,
+      "step": 1894
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734333841249023e-05,
+      "loss": 0.8116,
+      "step": 1895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7340315895335982e-05,
+      "loss": 0.3786,
+      "step": 1896
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7337291923479002e-05,
+      "loss": 0.8406,
+      "step": 1897
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7334266497518586e-05,
+      "loss": 0.8244,
+      "step": 1898
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7331239618054305e-05,
+      "loss": 0.8768,
+      "step": 1899
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732821128568603e-05,
+      "loss": 0.8918,
+      "step": 1900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732518150101392e-05,
+      "loss": 0.8649,
+      "step": 1901
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7322150264638407e-05,
+      "loss": 0.8141,
+      "step": 1902
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731911757716023e-05,
+      "loss": 0.8402,
+      "step": 1903
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73160834391804e-05,
+      "loss": 0.8736,
+      "step": 1904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7313047851300225e-05,
+      "loss": 0.8905,
+      "step": 1905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7310010814121295e-05,
+      "loss": 0.828,
+      "step": 1906
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7306972328245496e-05,
+      "loss": 0.8724,
+      "step": 1907
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730393239427499e-05,
+      "loss": 0.8854,
+      "step": 1908
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7300891012812228e-05,
+      "loss": 0.8439,
+      "step": 1909
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7297848184459954e-05,
+      "loss": 0.8182,
+      "step": 1910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294803909821196e-05,
+      "loss": 0.9126,
+      "step": 1911
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729175818949927e-05,
+      "loss": 0.3547,
+      "step": 1912
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7288711024097765e-05,
+      "loss": 0.8916,
+      "step": 1913
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7285662414220584e-05,
+      "loss": 0.8252,
+      "step": 1914
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7282612360471888e-05,
+      "loss": 0.8154,
+      "step": 1915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727956086345614e-05,
+      "loss": 0.3576,
+      "step": 1916
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727650792377809e-05,
+      "loss": 0.8351,
+      "step": 1917
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727345354204276e-05,
+      "loss": 0.7514,
+      "step": 1918
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727039771885547e-05,
+      "loss": 0.8039,
+      "step": 1919
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726734045482183e-05,
+      "loss": 0.9326,
+      "step": 1920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726428175054772e-05,
+      "loss": 0.8736,
+      "step": 1921
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7261221606639316e-05,
+      "loss": 0.8321,
+      "step": 1922
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7258160023703076e-05,
+      "loss": 0.8288,
+      "step": 1923
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7255097002345745e-05,
+      "loss": 0.8698,
+      "step": 1924
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725203254317435e-05,
+      "loss": 0.7867,
+      "step": 1925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724896664679621e-05,
+      "loss": 0.7689,
+      "step": 1926
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724589931381892e-05,
+      "loss": 0.9417,
+      "step": 1927
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7242830544850364e-05,
+      "loss": 0.9131,
+      "step": 1928
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723976034049871e-05,
+      "loss": 0.3404,
+      "step": 1929
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723668870137241e-05,
+      "loss": 0.7909,
+      "step": 1930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7233615628080207e-05,
+      "loss": 0.8596,
+      "step": 1931
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7230541121231115e-05,
+      "loss": 0.8096,
+      "step": 1932
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7227465181434442e-05,
+      "loss": 0.8674,
+      "step": 1933
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7224387809299776e-05,
+      "loss": 0.8567,
+      "step": 1934
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722130900543699e-05,
+      "loss": 0.7911,
+      "step": 1935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7218228770456246e-05,
+      "loss": 0.7535,
+      "step": 1936
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721514710496798e-05,
+      "loss": 0.7871,
+      "step": 1937
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7212064009582916e-05,
+      "loss": 0.8621,
+      "step": 1938
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720897948491206e-05,
+      "loss": 0.8635,
+      "step": 1939
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7205893531566705e-05,
+      "loss": 0.4146,
+      "step": 1940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720280615015843e-05,
+      "loss": 0.8946,
+      "step": 1941
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719971734129908e-05,
+      "loss": 0.8516,
+      "step": 1942
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71966271056008e-05,
+      "loss": 0.8578,
+      "step": 1943
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719353544367602e-05,
+      "loss": 0.8614,
+      "step": 1944
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7190442356137434e-05,
+      "loss": 0.8738,
+      "step": 1945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7187347843598038e-05,
+      "loss": 0.8003,
+      "step": 1946
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7184251906671097e-05,
+      "loss": 0.8484,
+      "step": 1947
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718115454597016e-05,
+      "loss": 0.8817,
+      "step": 1948
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7178055762109073e-05,
+      "loss": 0.8522,
+      "step": 1949
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7174955555701943e-05,
+      "loss": 0.8432,
+      "step": 1950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717185392736317e-05,
+      "loss": 0.8188,
+      "step": 1951
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7168750877707436e-05,
+      "loss": 0.8027,
+      "step": 1952
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71656464073497e-05,
+      "loss": 0.8158,
+      "step": 1953
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716254051690521e-05,
+      "loss": 0.8728,
+      "step": 1954
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715943320698949e-05,
+      "loss": 0.8056,
+      "step": 1955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7156324478218342e-05,
+      "loss": 0.8571,
+      "step": 1956
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715321433120786e-05,
+      "loss": 0.3738,
+      "step": 1957
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7150102766574405e-05,
+      "loss": 0.8077,
+      "step": 1958
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714698978493463e-05,
+      "loss": 0.8961,
+      "step": 1959
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714387538690547e-05,
+      "loss": 0.8695,
+      "step": 1960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7140759573104127e-05,
+      "loss": 0.8022,
+      "step": 1961
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71376423441481e-05,
+      "loss": 0.8594,
+      "step": 1962
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7134523700655154e-05,
+      "loss": 0.8941,
+      "step": 1963
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7131403643243344e-05,
+      "loss": 0.8758,
+      "step": 1964
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712828217253101e-05,
+      "loss": 0.8681,
+      "step": 1965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7125159289136754e-05,
+      "loss": 0.7858,
+      "step": 1966
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7122034993679473e-05,
+      "loss": 0.825,
+      "step": 1967
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7118909286778343e-05,
+      "loss": 0.8522,
+      "step": 1968
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7115782169052808e-05,
+      "loss": 0.8369,
+      "step": 1969
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7112653641122607e-05,
+      "loss": 0.8107,
+      "step": 1970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7109523703607747e-05,
+      "loss": 0.8472,
+      "step": 1971
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710639235712852e-05,
+      "loss": 0.8094,
+      "step": 1972
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7103259602305495e-05,
+      "loss": 0.9229,
+      "step": 1973
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7100125439759524e-05,
+      "loss": 0.3772,
+      "step": 1974
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709698987011173e-05,
+      "loss": 0.8462,
+      "step": 1975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7093852893983524e-05,
+      "loss": 0.8885,
+      "step": 1976
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709071451199659e-05,
+      "loss": 0.8263,
+      "step": 1977
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708757472477289e-05,
+      "loss": 0.8742,
+      "step": 1978
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708443353293467e-05,
+      "loss": 0.8278,
+      "step": 1979
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708129093710445e-05,
+      "loss": 0.8085,
+      "step": 1980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7078146937905025e-05,
+      "loss": 0.9627,
+      "step": 1981
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707500153595948e-05,
+      "loss": 0.8752,
+      "step": 1982
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071854731891164e-05,
+      "loss": 0.8707,
+      "step": 1983
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7068706526323713e-05,
+      "loss": 0.8605,
+      "step": 1984
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7065556919881034e-05,
+      "loss": 0.8524,
+      "step": 1985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706240591318732e-05,
+      "loss": 0.3523,
+      "step": 1986
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705925350686704e-05,
+      "loss": 0.8661,
+      "step": 1987
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7056099701544927e-05,
+      "loss": 0.8436,
+      "step": 1988
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705294449784601e-05,
+      "loss": 0.8724,
+      "step": 1989
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7049787896395586e-05,
+      "loss": 0.8291,
+      "step": 1990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704662989781923e-05,
+      "loss": 0.8511,
+      "step": 1991
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704347050274279e-05,
+      "loss": 0.8269,
+      "step": 1992
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70403097117924e-05,
+      "loss": 0.8114,
+      "step": 1993
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7037147525594452e-05,
+      "loss": 0.8619,
+      "step": 1994
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7033983944775643e-05,
+      "loss": 0.8356,
+      "step": 1995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703081896996293e-05,
+      "loss": 0.7786,
+      "step": 1996
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7027652601783538e-05,
+      "loss": 0.8449,
+      "step": 1997
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702448484086498e-05,
+      "loss": 0.8451,
+      "step": 1998
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702131568783505e-05,
+      "loss": 0.8486,
+      "step": 1999
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7018145143321802e-05,
+      "loss": 0.3372,
+      "step": 2000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7014973207953573e-05,
+      "loss": 0.8347,
+      "step": 2001
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7011799882358984e-05,
+      "loss": 0.85,
+      "step": 2002
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7008625167166923e-05,
+      "loss": 0.7526,
+      "step": 2003
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7005449063006545e-05,
+      "loss": 0.7909,
+      "step": 2004
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7002271570507304e-05,
+      "loss": 0.8268,
+      "step": 2005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6999092690298905e-05,
+      "loss": 0.8675,
+      "step": 2006
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699591242301134e-05,
+      "loss": 0.8116,
+      "step": 2007
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6992730769274875e-05,
+      "loss": 0.8887,
+      "step": 2008
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6989547729720052e-05,
+      "loss": 0.8829,
+      "step": 2009
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698636330497768e-05,
+      "loss": 0.8135,
+      "step": 2010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6983177495678853e-05,
+      "loss": 0.8745,
+      "step": 2011
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6979990302454933e-05,
+      "loss": 0.831,
+      "step": 2012
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6976801725937552e-05,
+      "loss": 0.8427,
+      "step": 2013
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6973611766758628e-05,
+      "loss": 0.85,
+      "step": 2014
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6970420425550346e-05,
+      "loss": 0.8092,
+      "step": 2015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6967227702945165e-05,
+      "loss": 0.8273,
+      "step": 2016
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6964033599575815e-05,
+      "loss": 0.8328,
+      "step": 2017
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960838116075304e-05,
+      "loss": 0.8901,
+      "step": 2018
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695764125307692e-05,
+      "loss": 0.8419,
+      "step": 2019
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6954443011214207e-05,
+      "loss": 0.7812,
+      "step": 2020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6951243391120993e-05,
+      "loss": 0.3829,
+      "step": 2021
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948042393431384e-05,
+      "loss": 0.8628,
+      "step": 2022
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694484001877975e-05,
+      "loss": 0.7972,
+      "step": 2023
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6941636267800738e-05,
+      "loss": 0.9043,
+      "step": 2024
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6938431141129267e-05,
+      "loss": 0.3589,
+      "step": 2025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693522463940053e-05,
+      "loss": 0.9136,
+      "step": 2026
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6932016763249985e-05,
+      "loss": 0.8413,
+      "step": 2027
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6928807513313376e-05,
+      "loss": 0.4055,
+      "step": 2028
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6925596890226703e-05,
+      "loss": 0.8012,
+      "step": 2029
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6922384894626257e-05,
+      "loss": 0.8248,
+      "step": 2030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6919171527148583e-05,
+      "loss": 0.9294,
+      "step": 2031
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6915956788430507e-05,
+      "loss": 0.8363,
+      "step": 2032
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691274067910913e-05,
+      "loss": 0.8083,
+      "step": 2033
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6909523199821812e-05,
+      "loss": 0.8859,
+      "step": 2034
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69063043512062e-05,
+      "loss": 0.8438,
+      "step": 2035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690308413390021e-05,
+      "loss": 0.8599,
+      "step": 2036
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6899862548542006e-05,
+      "loss": 0.8463,
+      "step": 2037
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6896639595770056e-05,
+      "loss": 0.7741,
+      "step": 2038
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6893415276223083e-05,
+      "loss": 0.9498,
+      "step": 2039
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6890189590540074e-05,
+      "loss": 0.8545,
+      "step": 2040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6886962539360307e-05,
+      "loss": 0.7998,
+      "step": 2041
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6883734123323307e-05,
+      "loss": 0.8022,
+      "step": 2042
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6880504343068886e-05,
+      "loss": 0.8149,
+      "step": 2043
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6877273199237123e-05,
+      "loss": 0.8949,
+      "step": 2044
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874040692468365e-05,
+      "loss": 0.8802,
+      "step": 2045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6870806823403226e-05,
+      "loss": 0.8174,
+      "step": 2046
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6867571592682595e-05,
+      "loss": 0.8418,
+      "step": 2047
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6864335000947636e-05,
+      "loss": 0.7381,
+      "step": 2048
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6861097048839766e-05,
+      "loss": 0.7674,
+      "step": 2049
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685785773700069e-05,
+      "loss": 0.8678,
+      "step": 2050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685461706607237e-05,
+      "loss": 0.7635,
+      "step": 2051
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6851375036697043e-05,
+      "loss": 0.8325,
+      "step": 2052
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6848131649517217e-05,
+      "loss": 0.8743,
+      "step": 2053
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684488690517566e-05,
+      "loss": 0.8123,
+      "step": 2054
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6841640804315416e-05,
+      "loss": 0.7752,
+      "step": 2055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6838393347579806e-05,
+      "loss": 0.7852,
+      "step": 2056
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6835144535612395e-05,
+      "loss": 0.8036,
+      "step": 2057
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6831894369057043e-05,
+      "loss": 0.8447,
+      "step": 2058
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828642848557866e-05,
+      "loss": 0.8668,
+      "step": 2059
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682538997475925e-05,
+      "loss": 0.7932,
+      "step": 2060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6822135748305845e-05,
+      "loss": 0.8843,
+      "step": 2061
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681888016984258e-05,
+      "loss": 0.8239,
+      "step": 2062
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6815623240014634e-05,
+      "loss": 0.8673,
+      "step": 2063
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6812364959467474e-05,
+      "loss": 0.7899,
+      "step": 2064
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680910532884683e-05,
+      "loss": 0.8727,
+      "step": 2065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6805844348798682e-05,
+      "loss": 0.8591,
+      "step": 2066
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6802582019969298e-05,
+      "loss": 0.9229,
+      "step": 2067
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6799318343005205e-05,
+      "loss": 0.8528,
+      "step": 2068
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.67960533185532e-05,
+      "loss": 0.8974,
+      "step": 2069
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679278694726034e-05,
+      "loss": 0.7694,
+      "step": 2070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6789519229773958e-05,
+      "loss": 0.8071,
+      "step": 2071
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678625016674165e-05,
+      "loss": 0.7278,
+      "step": 2072
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678297975881128e-05,
+      "loss": 0.8672,
+      "step": 2073
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6779708006630967e-05,
+      "loss": 0.7385,
+      "step": 2074
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776434910849118e-05,
+      "loss": 0.7786,
+      "step": 2075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6773160472114387e-05,
+      "loss": 0.8342,
+      "step": 2076
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676988469107571e-05,
+      "loss": 0.7944,
+      "step": 2077
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6766607568382268e-05,
+      "loss": 0.8406,
+      "step": 2078
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6763329104683533e-05,
+      "loss": 0.8128,
+      "step": 2079
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6760049300629228e-05,
+      "loss": 0.8567,
+      "step": 2080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6756768156869337e-05,
+      "loss": 0.8132,
+      "step": 2081
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6753485674054126e-05,
+      "loss": 0.7915,
+      "step": 2082
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675020185283411e-05,
+      "loss": 0.851,
+      "step": 2083
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6746916693860084e-05,
+      "loss": 0.8799,
+      "step": 2084
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674363019778309e-05,
+      "loss": 0.7794,
+      "step": 2085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674034236525445e-05,
+      "loss": 0.8113,
+      "step": 2086
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673705319692575e-05,
+      "loss": 0.8611,
+      "step": 2087
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673376269344883e-05,
+      "loss": 0.7994,
+      "step": 2088
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6730470855475806e-05,
+      "loss": 0.8158,
+      "step": 2089
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6727177683659054e-05,
+      "loss": 0.9262,
+      "step": 2090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6723883178651213e-05,
+      "loss": 0.7985,
+      "step": 2091
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6720587341105184e-05,
+      "loss": 0.8655,
+      "step": 2092
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6717290171674143e-05,
+      "loss": 0.8261,
+      "step": 2093
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713991671011517e-05,
+      "loss": 0.826,
+      "step": 2094
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6710691839771006e-05,
+      "loss": 0.8194,
+      "step": 2095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670739067860657e-05,
+      "loss": 0.8104,
+      "step": 2096
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6704088188172427e-05,
+      "loss": 0.7569,
+      "step": 2097
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670078436912307e-05,
+      "loss": 0.8688,
+      "step": 2098
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6697479222113245e-05,
+      "loss": 0.8258,
+      "step": 2099
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6694172747797972e-05,
+      "loss": 0.8108,
+      "step": 2100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6690864946832522e-05,
+      "loss": 0.7791,
+      "step": 2101
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6687555819872436e-05,
+      "loss": 0.8733,
+      "step": 2102
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6684245367573514e-05,
+      "loss": 0.8402,
+      "step": 2103
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6680933590591827e-05,
+      "loss": 0.8263,
+      "step": 2104
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6677620489583697e-05,
+      "loss": 0.8821,
+      "step": 2105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6674306065205718e-05,
+      "loss": 0.8637,
+      "step": 2106
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6670990318114736e-05,
+      "loss": 0.8737,
+      "step": 2107
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666767324896787e-05,
+      "loss": 0.8029,
+      "step": 2108
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66643548584225e-05,
+      "loss": 0.8363,
+      "step": 2109
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6661035147136258e-05,
+      "loss": 0.3802,
+      "step": 2110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6657714115767042e-05,
+      "loss": 0.8169,
+      "step": 2111
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665439176497302e-05,
+      "loss": 0.925,
+      "step": 2112
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6651068095412606e-05,
+      "loss": 0.8379,
+      "step": 2113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66477431077445e-05,
+      "loss": 0.723,
+      "step": 2114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6644416802627628e-05,
+      "loss": 0.842,
+      "step": 2115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6641089180721214e-05,
+      "loss": 0.9231,
+      "step": 2116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637760242684713e-05,
+      "loss": 0.9057,
+      "step": 2117
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663442998917786e-05,
+      "loss": 0.8489,
+      "step": 2118
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6631098420860644e-05,
+      "loss": 0.7427,
+      "step": 2119
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6627765538393313e-05,
+      "loss": 0.8745,
+      "step": 2120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6624431342436377e-05,
+      "loss": 0.875,
+      "step": 2121
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6621095833650608e-05,
+      "loss": 0.8569,
+      "step": 2122
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6617759012697034e-05,
+      "loss": 0.871,
+      "step": 2123
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6614420880236947e-05,
+      "loss": 0.9062,
+      "step": 2124
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6611081436931902e-05,
+      "loss": 0.8404,
+      "step": 2125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6607740683443702e-05,
+      "loss": 0.7449,
+      "step": 2126
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6604398620434423e-05,
+      "loss": 0.8459,
+      "step": 2127
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660105524856639e-05,
+      "loss": 0.8861,
+      "step": 2128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6597710568502195e-05,
+      "loss": 0.8464,
+      "step": 2129
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6594364580904685e-05,
+      "loss": 0.8209,
+      "step": 2130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659101728643697e-05,
+      "loss": 0.7814,
+      "step": 2131
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6587668685762407e-05,
+      "loss": 0.8905,
+      "step": 2132
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6584318779544634e-05,
+      "loss": 0.8516,
+      "step": 2133
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6580967568447527e-05,
+      "loss": 0.8691,
+      "step": 2134
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657761505313523e-05,
+      "loss": 0.7813,
+      "step": 2135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6574261234272144e-05,
+      "loss": 0.8109,
+      "step": 2136
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657090611252293e-05,
+      "loss": 0.7168,
+      "step": 2137
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65675496885525e-05,
+      "loss": 0.7714,
+      "step": 2138
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656419196302604e-05,
+      "loss": 0.8604,
+      "step": 2139
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656083293660898e-05,
+      "loss": 0.8768,
+      "step": 2140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6557472609967e-05,
+      "loss": 0.7875,
+      "step": 2141
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655411098376606e-05,
+      "loss": 0.877,
+      "step": 2142
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6550748058672368e-05,
+      "loss": 0.7885,
+      "step": 2143
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6547383835352384e-05,
+      "loss": 0.874,
+      "step": 2144
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654401831447283e-05,
+      "loss": 0.8399,
+      "step": 2145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654065149670068e-05,
+      "loss": 0.8568,
+      "step": 2146
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653728338270318e-05,
+      "loss": 0.8121,
+      "step": 2147
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6533913973147808e-05,
+      "loss": 0.3333,
+      "step": 2148
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653054326870233e-05,
+      "loss": 0.8712,
+      "step": 2149
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6527171270034734e-05,
+      "loss": 0.86,
+      "step": 2150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6523797977813297e-05,
+      "loss": 0.8166,
+      "step": 2151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652042339270653e-05,
+      "loss": 0.8624,
+      "step": 2152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651704751538321e-05,
+      "loss": 0.8164,
+      "step": 2153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6513670346512368e-05,
+      "loss": 0.8289,
+      "step": 2154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6510291886763287e-05,
+      "loss": 0.7961,
+      "step": 2155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6506912136805513e-05,
+      "loss": 0.8984,
+      "step": 2156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6503531097308846e-05,
+      "loss": 0.9337,
+      "step": 2157
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650014876894334e-05,
+      "loss": 0.87,
+      "step": 2158
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6496765152379298e-05,
+      "loss": 0.8156,
+      "step": 2159
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6493380248287287e-05,
+      "loss": 0.3721,
+      "step": 2160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6489994057338133e-05,
+      "loss": 0.859,
+      "step": 2161
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6486606580202907e-05,
+      "loss": 0.8309,
+      "step": 2162
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6483217817552936e-05,
+      "loss": 0.8058,
+      "step": 2163
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647982777005981e-05,
+      "loss": 0.8805,
+      "step": 2164
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647643643839536e-05,
+      "loss": 0.798,
+      "step": 2165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6473043823231688e-05,
+      "loss": 0.8185,
+      "step": 2166
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646964992524114e-05,
+      "loss": 0.8328,
+      "step": 2167
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6466254745096313e-05,
+      "loss": 0.8663,
+      "step": 2168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646285828347007e-05,
+      "loss": 0.8181,
+      "step": 2169
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6459460541035513e-05,
+      "loss": 0.8733,
+      "step": 2170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6456061518466014e-05,
+      "loss": 0.9,
+      "step": 2171
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6452661216435193e-05,
+      "loss": 0.8772,
+      "step": 2172
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644925963561691e-05,
+      "loss": 0.7861,
+      "step": 2173
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6445856776685303e-05,
+      "loss": 0.9109,
+      "step": 2174
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644245264031474e-05,
+      "loss": 0.8272,
+      "step": 2175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439047227179852e-05,
+      "loss": 0.8404,
+      "step": 2176
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643564053795553e-05,
+      "loss": 0.7743,
+      "step": 2177
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643223257331691e-05,
+      "loss": 0.877,
+      "step": 2178
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6428823333939376e-05,
+      "loss": 0.8614,
+      "step": 2179
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642541282049858e-05,
+      "loss": 0.792,
+      "step": 2180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6422001033670406e-05,
+      "loss": 0.8034,
+      "step": 2181
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641858797413101e-05,
+      "loss": 0.8426,
+      "step": 2182
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415173642556786e-05,
+      "loss": 0.8117,
+      "step": 2183
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6411758039624394e-05,
+      "loss": 0.8222,
+      "step": 2184
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6408341166010728e-05,
+      "loss": 0.8776,
+      "step": 2185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6404923022392948e-05,
+      "loss": 0.7886,
+      "step": 2186
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640150360944846e-05,
+      "loss": 0.8361,
+      "step": 2187
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6398082927854922e-05,
+      "loss": 0.3617,
+      "step": 2188
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639466097829025e-05,
+      "loss": 0.8385,
+      "step": 2189
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63912377614326e-05,
+      "loss": 0.7373,
+      "step": 2190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6387813277960385e-05,
+      "loss": 0.7368,
+      "step": 2191
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638438752855227e-05,
+      "loss": 0.8642,
+      "step": 2192
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638096051388717e-05,
+      "loss": 0.8762,
+      "step": 2193
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637753223464425e-05,
+      "loss": 0.881,
+      "step": 2194
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637410269150292e-05,
+      "loss": 0.8548,
+      "step": 2195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6370671885142857e-05,
+      "loss": 0.3478,
+      "step": 2196
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636723981624397e-05,
+      "loss": 0.8647,
+      "step": 2197
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6363806485486428e-05,
+      "loss": 0.7767,
+      "step": 2198
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6360371893550646e-05,
+      "loss": 0.7617,
+      "step": 2199
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6356936041117294e-05,
+      "loss": 0.8222,
+      "step": 2200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635349892886729e-05,
+      "loss": 0.8809,
+      "step": 2201
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6350060557481797e-05,
+      "loss": 0.8284,
+      "step": 2202
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6346620927642235e-05,
+      "loss": 0.802,
+      "step": 2203
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634318004003026e-05,
+      "loss": 0.8723,
+      "step": 2204
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6339737895327798e-05,
+      "loss": 0.8501,
+      "step": 2205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6336294494217008e-05,
+      "loss": 0.8343,
+      "step": 2206
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63328498373803e-05,
+      "loss": 0.8327,
+      "step": 2207
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632940392550034e-05,
+      "loss": 0.9302,
+      "step": 2208
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6325956759260034e-05,
+      "loss": 0.3667,
+      "step": 2209
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6322508339342544e-05,
+      "loss": 0.7852,
+      "step": 2210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631905866643128e-05,
+      "loss": 0.9204,
+      "step": 2211
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6315607741209888e-05,
+      "loss": 0.8877,
+      "step": 2212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6312155564362284e-05,
+      "loss": 0.9409,
+      "step": 2213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630870213657261e-05,
+      "loss": 0.8132,
+      "step": 2214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6305247458525273e-05,
+      "loss": 0.8537,
+      "step": 2215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6301791530904916e-05,
+      "loss": 0.7581,
+      "step": 2216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6298334354396436e-05,
+      "loss": 0.8075,
+      "step": 2217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6294875929684976e-05,
+      "loss": 0.7761,
+      "step": 2218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6291416257455927e-05,
+      "loss": 0.906,
+      "step": 2219
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628795533839492e-05,
+      "loss": 0.7956,
+      "step": 2220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6284493173187847e-05,
+      "loss": 0.8309,
+      "step": 2221
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6281029762520836e-05,
+      "loss": 0.4111,
+      "step": 2222
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6277565107080268e-05,
+      "loss": 0.8513,
+      "step": 2223
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6274099207552763e-05,
+      "loss": 0.8558,
+      "step": 2224
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62706320646252e-05,
+      "loss": 0.8752,
+      "step": 2225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6267163678984688e-05,
+      "loss": 0.8688,
+      "step": 2226
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6263694051318597e-05,
+      "loss": 0.8199,
+      "step": 2227
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626022318231453e-05,
+      "loss": 0.805,
+      "step": 2228
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6256751072660356e-05,
+      "loss": 0.9045,
+      "step": 2229
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625327772304417e-05,
+      "loss": 0.7666,
+      "step": 2230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6249803134154318e-05,
+      "loss": 0.7637,
+      "step": 2231
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6246327306679396e-05,
+      "loss": 0.7608,
+      "step": 2232
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624285024130824e-05,
+      "loss": 0.9378,
+      "step": 2233
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6239371938729942e-05,
+      "loss": 0.8079,
+      "step": 2234
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6235892399633824e-05,
+      "loss": 0.8064,
+      "step": 2235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6232411624709463e-05,
+      "loss": 0.8891,
+      "step": 2236
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6228929614646676e-05,
+      "loss": 0.7893,
+      "step": 2237
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622544637013553e-05,
+      "loss": 0.8287,
+      "step": 2238
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622196189186634e-05,
+      "loss": 0.7423,
+      "step": 2239
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6218476180529645e-05,
+      "loss": 0.8191,
+      "step": 2240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6214989236816253e-05,
+      "loss": 0.8436,
+      "step": 2241
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6211501061417205e-05,
+      "loss": 0.8526,
+      "step": 2242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6208011655023783e-05,
+      "loss": 0.8462,
+      "step": 2243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620452101832752e-05,
+      "loss": 0.8447,
+      "step": 2244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6201029152020194e-05,
+      "loss": 0.8035,
+      "step": 2245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6197536056793817e-05,
+      "loss": 0.7471,
+      "step": 2246
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6194041733340648e-05,
+      "loss": 0.8344,
+      "step": 2247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6190546182353196e-05,
+      "loss": 0.8149,
+      "step": 2248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618704940452421e-05,
+      "loss": 0.7684,
+      "step": 2249
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6183551400546676e-05,
+      "loss": 0.8962,
+      "step": 2250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6180052171113828e-05,
+      "loss": 0.3804,
+      "step": 2251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6176551716919152e-05,
+      "loss": 0.9408,
+      "step": 2252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6173050038656354e-05,
+      "loss": 0.8027,
+      "step": 2253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6169547137019407e-05,
+      "loss": 0.8372,
+      "step": 2254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616604301270251e-05,
+      "loss": 0.7911,
+      "step": 2255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6162537666400116e-05,
+      "loss": 0.7884,
+      "step": 2256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6159031098806902e-05,
+      "loss": 0.3523,
+      "step": 2257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6155523310617808e-05,
+      "loss": 0.8729,
+      "step": 2258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6152014302528007e-05,
+      "loss": 0.8068,
+      "step": 2259
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614850407523291e-05,
+      "loss": 0.9233,
+      "step": 2260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6144992629428178e-05,
+      "loss": 0.8459,
+      "step": 2261
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6141479965809705e-05,
+      "loss": 0.3276,
+      "step": 2262
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6137966085073632e-05,
+      "loss": 0.3868,
+      "step": 2263
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6134450987916335e-05,
+      "loss": 0.8604,
+      "step": 2264
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6130934675034442e-05,
+      "loss": 0.8527,
+      "step": 2265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6127417147124812e-05,
+      "loss": 0.3771,
+      "step": 2266
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612389840488455e-05,
+      "loss": 0.7502,
+      "step": 2267
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6120378449010995e-05,
+      "loss": 0.8366,
+      "step": 2268
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6116857280201735e-05,
+      "loss": 0.8025,
+      "step": 2269
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6113334899154595e-05,
+      "loss": 0.8417,
+      "step": 2270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610981130656764e-05,
+      "loss": 0.8052,
+      "step": 2271
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6106286503139174e-05,
+      "loss": 0.3539,
+      "step": 2272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6102760489567742e-05,
+      "loss": 0.8202,
+      "step": 2273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609923326655213e-05,
+      "loss": 0.8548,
+      "step": 2274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6095704834791363e-05,
+      "loss": 0.7985,
+      "step": 2275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6092175194984707e-05,
+      "loss": 0.7829,
+      "step": 2276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6088644347831658e-05,
+      "loss": 0.8252,
+      "step": 2277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6085112294031962e-05,
+      "loss": 0.8968,
+      "step": 2278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6081579034285605e-05,
+      "loss": 0.742,
+      "step": 2279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6078044569292807e-05,
+      "loss": 0.8669,
+      "step": 2280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607450889975402e-05,
+      "loss": 0.8599,
+      "step": 2281
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6070972026369954e-05,
+      "loss": 0.7907,
+      "step": 2282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606743394984154e-05,
+      "loss": 0.8,
+      "step": 2283
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606389467086995e-05,
+      "loss": 0.875,
+      "step": 2284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60603541901566e-05,
+      "loss": 0.8234,
+      "step": 2285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605681250840315e-05,
+      "loss": 0.8667,
+      "step": 2286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605326962631148e-05,
+      "loss": 0.8219,
+      "step": 2287
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6049725544583718e-05,
+      "loss": 0.9077,
+      "step": 2288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604618026392224e-05,
+      "loss": 0.8122,
+      "step": 2289
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604263378502964e-05,
+      "loss": 0.8149,
+      "step": 2290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6039086108608755e-05,
+      "loss": 0.8206,
+      "step": 2291
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6035537235362674e-05,
+      "loss": 0.8215,
+      "step": 2292
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603198716599471e-05,
+      "loss": 0.7641,
+      "step": 2293
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602843590120841e-05,
+      "loss": 0.8017,
+      "step": 2294
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6024883441707563e-05,
+      "loss": 0.8451,
+      "step": 2295
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6021329788196197e-05,
+      "loss": 0.8517,
+      "step": 2296
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6017774941378577e-05,
+      "loss": 0.812,
+      "step": 2297
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.60142189019592e-05,
+      "loss": 0.8533,
+      "step": 2298
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.60106616706428e-05,
+      "loss": 0.8575,
+      "step": 2299
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600710324813435e-05,
+      "loss": 0.8303,
+      "step": 2300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6003543635139058e-05,
+      "loss": 0.795,
+      "step": 2301
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5999982832362367e-05,
+      "loss": 0.8173,
+      "step": 2302
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5996420840509956e-05,
+      "loss": 0.887,
+      "step": 2303
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5992857660287742e-05,
+      "loss": 0.7954,
+      "step": 2304
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5989293292401872e-05,
+      "loss": 0.3622,
+      "step": 2305
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5985727737558735e-05,
+      "loss": 0.8802,
+      "step": 2306
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598216099646495e-05,
+      "loss": 0.8106,
+      "step": 2307
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5978593069827374e-05,
+      "loss": 0.8805,
+      "step": 2308
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59750239583531e-05,
+      "loss": 0.3857,
+      "step": 2309
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597145366274945e-05,
+      "loss": 0.7667,
+      "step": 2310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596788218372399e-05,
+      "loss": 0.7538,
+      "step": 2311
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596430952198451e-05,
+      "loss": 0.8569,
+      "step": 2312
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5960735678239044e-05,
+      "loss": 0.7811,
+      "step": 2313
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5957160653195855e-05,
+      "loss": 0.8481,
+      "step": 2314
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5953584447563436e-05,
+      "loss": 0.8322,
+      "step": 2315
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595000706205052e-05,
+      "loss": 0.834,
+      "step": 2316
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594642849736608e-05,
+      "loss": 0.8726,
+      "step": 2317
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5942848754219313e-05,
+      "loss": 0.8564,
+      "step": 2318
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5939267833319645e-05,
+      "loss": 0.8109,
+      "step": 2319
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593568573537675e-05,
+      "loss": 0.8838,
+      "step": 2320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593210246110052e-05,
+      "loss": 0.8319,
+      "step": 2321
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5928518011201094e-05,
+      "loss": 0.8565,
+      "step": 2322
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592493238638884e-05,
+      "loss": 0.7622,
+      "step": 2323
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592134558737435e-05,
+      "loss": 0.8164,
+      "step": 2324
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5917757614868458e-05,
+      "loss": 0.7684,
+      "step": 2325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591416846958223e-05,
+      "loss": 0.37,
+      "step": 2326
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591057815222696e-05,
+      "loss": 0.8647,
+      "step": 2327
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5906986663514176e-05,
+      "loss": 0.911,
+      "step": 2328
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5903394004155636e-05,
+      "loss": 0.8903,
+      "step": 2329
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589980017486334e-05,
+      "loss": 0.7854,
+      "step": 2330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589620517634951e-05,
+      "loss": 0.8039,
+      "step": 2331
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5892609009326597e-05,
+      "loss": 0.7835,
+      "step": 2332
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5889011674507298e-05,
+      "loss": 0.8905,
+      "step": 2333
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5885413172604526e-05,
+      "loss": 0.7294,
+      "step": 2334
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588181350433143e-05,
+      "loss": 0.9082,
+      "step": 2335
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5878212670401393e-05,
+      "loss": 0.8293,
+      "step": 2336
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5874610671528035e-05,
+      "loss": 0.8289,
+      "step": 2337
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587100750842519e-05,
+      "loss": 0.818,
+      "step": 2338
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586740318180694e-05,
+      "loss": 0.7783,
+      "step": 2339
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5863797692387585e-05,
+      "loss": 0.8609,
+      "step": 2340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586019104088166e-05,
+      "loss": 0.963,
+      "step": 2341
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585658322800393e-05,
+      "loss": 0.8422,
+      "step": 2342
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5852974254469397e-05,
+      "loss": 0.8351,
+      "step": 2343
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584936412099328e-05,
+      "loss": 0.8107,
+      "step": 2344
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5845752828291038e-05,
+      "loss": 0.7949,
+      "step": 2345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5842140377078358e-05,
+      "loss": 0.8856,
+      "step": 2346
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5838526768071152e-05,
+      "loss": 0.7748,
+      "step": 2347
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5834912001985566e-05,
+      "loss": 0.7946,
+      "step": 2348
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5831296079537974e-05,
+      "loss": 0.8527,
+      "step": 2349
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5827679001444974e-05,
+      "loss": 0.8486,
+      "step": 2350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5824060768423408e-05,
+      "loss": 0.819,
+      "step": 2351
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582044138119033e-05,
+      "loss": 0.827,
+      "step": 2352
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5816820840463026e-05,
+      "loss": 0.8008,
+      "step": 2353
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5813199146959024e-05,
+      "loss": 0.857,
+      "step": 2354
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5809576301396067e-05,
+      "loss": 0.7475,
+      "step": 2355
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5805952304492125e-05,
+      "loss": 0.8607,
+      "step": 2356
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580232715696541e-05,
+      "loss": 0.872,
+      "step": 2357
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5798700859534346e-05,
+      "loss": 0.8682,
+      "step": 2358
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579507341291759e-05,
+      "loss": 0.7708,
+      "step": 2359
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5791444817834037e-05,
+      "loss": 0.8585,
+      "step": 2360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5787815075002798e-05,
+      "loss": 0.8546,
+      "step": 2361
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5784184185143218e-05,
+      "loss": 0.8668,
+      "step": 2362
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578055214897486e-05,
+      "loss": 0.8045,
+      "step": 2363
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5776918967217518e-05,
+      "loss": 0.9089,
+      "step": 2364
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5773284640591224e-05,
+      "loss": 0.843,
+      "step": 2365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5769649169816224e-05,
+      "loss": 0.8517,
+      "step": 2366
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5766012555613e-05,
+      "loss": 0.823,
+      "step": 2367
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5762374798702246e-05,
+      "loss": 0.9041,
+      "step": 2368
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57587358998049e-05,
+      "loss": 0.8273,
+      "step": 2369
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5755095859642118e-05,
+      "loss": 0.3841,
+      "step": 2370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5751454678935282e-05,
+      "loss": 0.8381,
+      "step": 2371
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5747812358405994e-05,
+      "loss": 0.8008,
+      "step": 2372
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5744168898776096e-05,
+      "loss": 0.8267,
+      "step": 2373
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574052430076765e-05,
+      "loss": 0.8032,
+      "step": 2374
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5736878565102935e-05,
+      "loss": 0.3374,
+      "step": 2375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5733231692504466e-05,
+      "loss": 0.868,
+      "step": 2376
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572958368369498e-05,
+      "loss": 0.8358,
+      "step": 2377
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5725934539397436e-05,
+      "loss": 0.8788,
+      "step": 2378
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5722284260335027e-05,
+      "loss": 0.8134,
+      "step": 2379
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571863284723116e-05,
+      "loss": 0.8392,
+      "step": 2380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5714980300809467e-05,
+      "loss": 0.8127,
+      "step": 2381
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571132662179382e-05,
+      "loss": 0.8268,
+      "step": 2382
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5707671810908296e-05,
+      "loss": 0.8993,
+      "step": 2383
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570401586887721e-05,
+      "loss": 0.3526,
+      "step": 2384
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570035879642509e-05,
+      "loss": 0.7863,
+      "step": 2385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5696700594276705e-05,
+      "loss": 0.7763,
+      "step": 2386
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5693041263157022e-05,
+      "loss": 0.8551,
+      "step": 2387
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5689380803791263e-05,
+      "loss": 0.8127,
+      "step": 2388
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5685719216904843e-05,
+      "loss": 0.7741,
+      "step": 2389
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682056503223422e-05,
+      "loss": 0.3268,
+      "step": 2390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5678392663472875e-05,
+      "loss": 0.3955,
+      "step": 2391
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56747276983793e-05,
+      "loss": 0.8932,
+      "step": 2392
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567106160866902e-05,
+      "loss": 0.8462,
+      "step": 2393
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5667394395068583e-05,
+      "loss": 0.7885,
+      "step": 2394
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5663726058304752e-05,
+      "loss": 0.8127,
+      "step": 2395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5660056599104522e-05,
+      "loss": 0.8996,
+      "step": 2396
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56563860181951e-05,
+      "loss": 0.8572,
+      "step": 2397
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565271431630393e-05,
+      "loss": 0.8088,
+      "step": 2398
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564904149415866e-05,
+      "loss": 0.379,
+      "step": 2399
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5645367552487175e-05,
+      "loss": 0.7972,
+      "step": 2400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5641692492017578e-05,
+      "loss": 0.7077,
+      "step": 2401
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5638016313478185e-05,
+      "loss": 0.8852,
+      "step": 2402
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563433901759754e-05,
+      "loss": 0.9231,
+      "step": 2403
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5630660605104417e-05,
+      "loss": 0.8163,
+      "step": 2404
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5626981076727805e-05,
+      "loss": 0.8666,
+      "step": 2405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5623300433196898e-05,
+      "loss": 0.8462,
+      "step": 2406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5619618675241138e-05,
+      "loss": 0.764,
+      "step": 2407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561593580359017e-05,
+      "loss": 0.8469,
+      "step": 2408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561225181897387e-05,
+      "loss": 0.805,
+      "step": 2409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5608566722122322e-05,
+      "loss": 0.8223,
+      "step": 2410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560488051376584e-05,
+      "loss": 0.7617,
+      "step": 2411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560119319463496e-05,
+      "loss": 0.9057,
+      "step": 2412
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5597504765460437e-05,
+      "loss": 0.8534,
+      "step": 2413
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5593815226973236e-05,
+      "loss": 0.7707,
+      "step": 2414
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559012457990455e-05,
+      "loss": 0.8997,
+      "step": 2415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5586432824985792e-05,
+      "loss": 0.8326,
+      "step": 2416
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55827399629486e-05,
+      "loss": 0.686,
+      "step": 2417
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5579045994524813e-05,
+      "loss": 0.7366,
+      "step": 2418
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557535092044651e-05,
+      "loss": 0.8126,
+      "step": 2419
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5571654741445977e-05,
+      "loss": 0.8272,
+      "step": 2420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5567957458255724e-05,
+      "loss": 0.7747,
+      "step": 2421
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5564259071608478e-05,
+      "loss": 0.8285,
+      "step": 2422
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5560559582237183e-05,
+      "loss": 0.7519,
+      "step": 2423
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5556858990875e-05,
+      "loss": 0.8876,
+      "step": 2424
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5553157298255318e-05,
+      "loss": 0.7651,
+      "step": 2425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554945450511174e-05,
+      "loss": 0.776,
+      "step": 2426
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5545750612178073e-05,
+      "loss": 0.8891,
+      "step": 2427
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5542045620188364e-05,
+      "loss": 0.8253,
+      "step": 2428
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5538339529876866e-05,
+      "loss": 0.779,
+      "step": 2429
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553463234197805e-05,
+      "loss": 0.3826,
+      "step": 2430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5530924057226606e-05,
+      "loss": 0.8373,
+      "step": 2431
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5527214676357442e-05,
+      "loss": 0.8221,
+      "step": 2432
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5523504200105685e-05,
+      "loss": 0.913,
+      "step": 2433
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551979262920667e-05,
+      "loss": 0.8056,
+      "step": 2434
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551607996439596e-05,
+      "loss": 0.8498,
+      "step": 2435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5512366206409333e-05,
+      "loss": 0.9191,
+      "step": 2436
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5508651355982776e-05,
+      "loss": 0.8016,
+      "step": 2437
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.55049354138525e-05,
+      "loss": 0.8659,
+      "step": 2438
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550121838075493e-05,
+      "loss": 0.8507,
+      "step": 2439
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5497500257426705e-05,
+      "loss": 0.8755,
+      "step": 2440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5493781044604682e-05,
+      "loss": 0.3694,
+      "step": 2441
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5490060743025938e-05,
+      "loss": 0.7545,
+      "step": 2442
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548633935342776e-05,
+      "loss": 0.8175,
+      "step": 2443
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548261687654765e-05,
+      "loss": 0.8174,
+      "step": 2444
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5478893313123333e-05,
+      "loss": 0.8539,
+      "step": 2445
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5475168663892738e-05,
+      "loss": 0.8876,
+      "step": 2446
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547144292959402e-05,
+      "loss": 0.88,
+      "step": 2447
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546771611096554e-05,
+      "loss": 0.8557,
+      "step": 2448
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5463988208745886e-05,
+      "loss": 0.9349,
+      "step": 2449
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5460259223673845e-05,
+      "loss": 0.8599,
+      "step": 2450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5456529156488432e-05,
+      "loss": 0.7955,
+      "step": 2451
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5452798007928872e-05,
+      "loss": 0.8506,
+      "step": 2452
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5449065778734595e-05,
+      "loss": 0.8227,
+      "step": 2453
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544533246964526e-05,
+      "loss": 0.797,
+      "step": 2454
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5441598081400736e-05,
+      "loss": 0.8753,
+      "step": 2455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5437862614741103e-05,
+      "loss": 0.8087,
+      "step": 2456
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543412607040665e-05,
+      "loss": 0.8235,
+      "step": 2457
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5430388449137886e-05,
+      "loss": 0.795,
+      "step": 2458
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5426649751675536e-05,
+      "loss": 0.8221,
+      "step": 2459
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542290997876053e-05,
+      "loss": 0.8507,
+      "step": 2460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541916913113402e-05,
+      "loss": 0.7386,
+      "step": 2461
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5415427209537366e-05,
+      "loss": 0.8351,
+      "step": 2462
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541168421471214e-05,
+      "loss": 0.8461,
+      "step": 2463
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5407940147400126e-05,
+      "loss": 0.3315,
+      "step": 2464
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540419500834333e-05,
+      "loss": 0.7846,
+      "step": 2465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5400448798283952e-05,
+      "loss": 0.8914,
+      "step": 2466
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5396701517964422e-05,
+      "loss": 0.8652,
+      "step": 2467
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539295316812738e-05,
+      "loss": 0.8247,
+      "step": 2468
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5389203749515663e-05,
+      "loss": 0.8144,
+      "step": 2469
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538545326287234e-05,
+      "loss": 0.767,
+      "step": 2470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5381701708940676e-05,
+      "loss": 0.7533,
+      "step": 2471
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5377949088464156e-05,
+      "loss": 0.8687,
+      "step": 2472
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537419540218647e-05,
+      "loss": 0.8111,
+      "step": 2473
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5370440650851525e-05,
+      "loss": 0.7725,
+      "step": 2474
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536668483520344e-05,
+      "loss": 0.8956,
+      "step": 2475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536292795598654e-05,
+      "loss": 0.7779,
+      "step": 2476
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5359170013945362e-05,
+      "loss": 0.9079,
+      "step": 2477
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5355411009824656e-05,
+      "loss": 0.7829,
+      "step": 2478
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535165094436938e-05,
+      "loss": 0.8296,
+      "step": 2479
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5347889818324703e-05,
+      "loss": 0.9149,
+      "step": 2480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5344127632436e-05,
+      "loss": 0.8522,
+      "step": 2481
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534036438744887e-05,
+      "loss": 0.8578,
+      "step": 2482
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5336600084109106e-05,
+      "loss": 0.9002,
+      "step": 2483
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533283472316272e-05,
+      "loss": 0.8271,
+      "step": 2484
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532906830535593e-05,
+      "loss": 0.7955,
+      "step": 2485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5325300831435158e-05,
+      "loss": 0.8162,
+      "step": 2486
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532153230214705e-05,
+      "loss": 0.8347,
+      "step": 2487
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531776271823845e-05,
+      "loss": 0.8522,
+      "step": 2488
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5313992080456408e-05,
+      "loss": 0.8222,
+      "step": 2489
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5310220389548195e-05,
+      "loss": 0.7931,
+      "step": 2490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5306447646261285e-05,
+      "loss": 0.3683,
+      "step": 2491
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530267385134336e-05,
+      "loss": 0.8796,
+      "step": 2492
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5298899005542303e-05,
+      "loss": 0.8757,
+      "step": 2493
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5295123109606214e-05,
+      "loss": 0.8624,
+      "step": 2494
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5291346164283405e-05,
+      "loss": 0.786,
+      "step": 2495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528756817032239e-05,
+      "loss": 0.8153,
+      "step": 2496
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5283789128471883e-05,
+      "loss": 0.9167,
+      "step": 2497
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5280009039480823e-05,
+      "loss": 0.8811,
+      "step": 2498
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5276227904098345e-05,
+      "loss": 0.8774,
+      "step": 2499
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5272445723073795e-05,
+      "loss": 0.771,
+      "step": 2500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526866249715672e-05,
+      "loss": 0.8498,
+      "step": 2501
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5264878227096885e-05,
+      "loss": 0.8023,
+      "step": 2502
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5261092913644252e-05,
+      "loss": 0.8394,
+      "step": 2503
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257306557548992e-05,
+      "loss": 0.355,
+      "step": 2504
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5253519159561493e-05,
+      "loss": 0.3835,
+      "step": 2505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5249730720432334e-05,
+      "loss": 0.7009,
+      "step": 2506
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524594124091231e-05,
+      "loss": 0.7916,
+      "step": 2507
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5242150721752418e-05,
+      "loss": 0.8455,
+      "step": 2508
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.523835916370386e-05,
+      "loss": 0.7835,
+      "step": 2509
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5234566567518055e-05,
+      "loss": 0.8105,
+      "step": 2510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.523077293394661e-05,
+      "loss": 0.7993,
+      "step": 2511
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5226978263741355e-05,
+      "loss": 0.8215,
+      "step": 2512
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5223182557654308e-05,
+      "loss": 0.8287,
+      "step": 2513
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5219385816437707e-05,
+      "loss": 0.8443,
+      "step": 2514
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5215588040843989e-05,
+      "loss": 0.8241,
+      "step": 2515
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5211789231625798e-05,
+      "loss": 0.8512,
+      "step": 2516
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5207989389535977e-05,
+      "loss": 0.7963,
+      "step": 2517
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5204188515327584e-05,
+      "loss": 0.7503,
+      "step": 2518
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520038660975387e-05,
+      "loss": 0.8544,
+      "step": 2519
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5196583673568302e-05,
+      "loss": 0.8009,
+      "step": 2520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519277970752454e-05,
+      "loss": 0.8167,
+      "step": 2521
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5188974712376456e-05,
+      "loss": 0.3604,
+      "step": 2522
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5185168688878122e-05,
+      "loss": 0.7896,
+      "step": 2523
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5181361637783815e-05,
+      "loss": 0.8416,
+      "step": 2524
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5177553559848019e-05,
+      "loss": 0.3903,
+      "step": 2525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517374445582542e-05,
+      "loss": 0.8774,
+      "step": 2526
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5169934326470896e-05,
+      "loss": 0.8291,
+      "step": 2527
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166123172539549e-05,
+      "loss": 0.793,
+      "step": 2528
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5162310994786664e-05,
+      "loss": 0.7437,
+      "step": 2529
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5158497793967742e-05,
+      "loss": 0.8418,
+      "step": 2530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5154683570838482e-05,
+      "loss": 0.7892,
+      "step": 2531
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5150868326154787e-05,
+      "loss": 0.8739,
+      "step": 2532
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514705206067276e-05,
+      "loss": 0.8231,
+      "step": 2533
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5143234775148709e-05,
+      "loss": 0.7059,
+      "step": 2534
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5139416470339145e-05,
+      "loss": 0.9329,
+      "step": 2535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5135597147000777e-05,
+      "loss": 0.7778,
+      "step": 2536
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5131776805890515e-05,
+      "loss": 0.9118,
+      "step": 2537
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5127955447765479e-05,
+      "loss": 0.3365,
+      "step": 2538
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5124133073382982e-05,
+      "loss": 0.7652,
+      "step": 2539
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5120309683500543e-05,
+      "loss": 0.8437,
+      "step": 2540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5116485278875881e-05,
+      "loss": 0.8442,
+      "step": 2541
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5112659860266913e-05,
+      "loss": 0.8101,
+      "step": 2542
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5108833428431764e-05,
+      "loss": 0.896,
+      "step": 2543
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5105005984128754e-05,
+      "loss": 0.7701,
+      "step": 2544
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5101177528116404e-05,
+      "loss": 0.8721,
+      "step": 2545
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509734806115344e-05,
+      "loss": 0.7784,
+      "step": 2546
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5093517583998782e-05,
+      "loss": 0.8526,
+      "step": 2547
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5089686097411558e-05,
+      "loss": 0.8203,
+      "step": 2548
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5085853602151084e-05,
+      "loss": 0.8472,
+      "step": 2549
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5082020098976886e-05,
+      "loss": 0.8578,
+      "step": 2550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507818558864869e-05,
+      "loss": 0.8249,
+      "step": 2551
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507435007192642e-05,
+      "loss": 0.8375,
+      "step": 2552
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5070513549570194e-05,
+      "loss": 0.7854,
+      "step": 2553
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5066676022340333e-05,
+      "loss": 0.7771,
+      "step": 2554
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5062837490997361e-05,
+      "loss": 0.8332,
+      "step": 2555
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5058997956301998e-05,
+      "loss": 0.8036,
+      "step": 2556
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5055157419015156e-05,
+      "loss": 0.8316,
+      "step": 2557
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5051315879897958e-05,
+      "loss": 0.8751,
+      "step": 2558
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.504747333971172e-05,
+      "loss": 0.8645,
+      "step": 2559
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043629799217953e-05,
+      "loss": 0.8085,
+      "step": 2560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5039785259178372e-05,
+      "loss": 0.8021,
+      "step": 2561
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5035939720354887e-05,
+      "loss": 0.8332,
+      "step": 2562
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5032093183509599e-05,
+      "loss": 0.8746,
+      "step": 2563
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5028245649404825e-05,
+      "loss": 0.7588,
+      "step": 2564
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5024397118803062e-05,
+      "loss": 0.7734,
+      "step": 2565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5020547592467019e-05,
+      "loss": 0.8907,
+      "step": 2566
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5016697071159583e-05,
+      "loss": 0.8149,
+      "step": 2567
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.501284555564386e-05,
+      "loss": 0.8801,
+      "step": 2568
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5008993046683136e-05,
+      "loss": 0.7816,
+      "step": 2569
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50051395450409e-05,
+      "loss": 0.8593,
+      "step": 2570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5001285051480845e-05,
+      "loss": 0.7964,
+      "step": 2571
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499742956676685e-05,
+      "loss": 0.8101,
+      "step": 2572
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4993573091662996e-05,
+      "loss": 0.7984,
+      "step": 2573
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4989715626933556e-05,
+      "loss": 0.8123,
+      "step": 2574
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4985857173343003e-05,
+      "loss": 0.9338,
+      "step": 2575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981997731656008e-05,
+      "loss": 0.8256,
+      "step": 2576
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4978137302637428e-05,
+      "loss": 0.8255,
+      "step": 2577
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4974275887052328e-05,
+      "loss": 0.8934,
+      "step": 2578
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497041348566596e-05,
+      "loss": 0.8338,
+      "step": 2579
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4966550099243776e-05,
+      "loss": 0.8312,
+      "step": 2580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4962685728551417e-05,
+      "loss": 0.8704,
+      "step": 2581
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.495882037435473e-05,
+      "loss": 0.8053,
+      "step": 2582
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4954954037419746e-05,
+      "loss": 0.7696,
+      "step": 2583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4951086718512697e-05,
+      "loss": 0.7666,
+      "step": 2584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4947218418400004e-05,
+      "loss": 0.3306,
+      "step": 2585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494334913784829e-05,
+      "loss": 0.8817,
+      "step": 2586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4939478877624369e-05,
+      "loss": 0.3639,
+      "step": 2587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4935607638495247e-05,
+      "loss": 0.8714,
+      "step": 2588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4931735421228126e-05,
+      "loss": 0.8059,
+      "step": 2589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4927862226590398e-05,
+      "loss": 0.7857,
+      "step": 2590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4923988055349656e-05,
+      "loss": 0.8345,
+      "step": 2591
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4920112908273683e-05,
+      "loss": 0.385,
+      "step": 2592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4916236786130453e-05,
+      "loss": 0.8565,
+      "step": 2593
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4912359689688138e-05,
+      "loss": 0.9216,
+      "step": 2594
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4908481619715099e-05,
+      "loss": 0.8357,
+      "step": 2595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4904602576979887e-05,
+      "loss": 0.8737,
+      "step": 2596
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4900722562251258e-05,
+      "loss": 0.7565,
+      "step": 2597
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4896841576298144e-05,
+      "loss": 0.8997,
+      "step": 2598
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489295961988969e-05,
+      "loss": 0.7653,
+      "step": 2599
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4889076693795208e-05,
+      "loss": 0.7578,
+      "step": 2600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4885192798784225e-05,
+      "loss": 0.8031,
+      "step": 2601
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488130793562645e-05,
+      "loss": 0.7886,
+      "step": 2602
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487742210509178e-05,
+      "loss": 0.8312,
+      "step": 2603
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4873535307950314e-05,
+      "loss": 0.7479,
+      "step": 2604
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486964754497233e-05,
+      "loss": 0.8933,
+      "step": 2605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4865758816928311e-05,
+      "loss": 0.8204,
+      "step": 2606
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861869124588922e-05,
+      "loss": 0.8278,
+      "step": 2607
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4857978468725021e-05,
+      "loss": 0.8768,
+      "step": 2608
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4854086850107656e-05,
+      "loss": 0.7811,
+      "step": 2609
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4850194269508077e-05,
+      "loss": 0.8487,
+      "step": 2610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4846300727697698e-05,
+      "loss": 0.8838,
+      "step": 2611
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4842406225448155e-05,
+      "loss": 0.8357,
+      "step": 2612
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4838510763531253e-05,
+      "loss": 0.7673,
+      "step": 2613
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4834614342718995e-05,
+      "loss": 0.8043,
+      "step": 2614
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4830716963783578e-05,
+      "loss": 0.7935,
+      "step": 2615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4826818627497379e-05,
+      "loss": 0.364,
+      "step": 2616
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482291933463297e-05,
+      "loss": 0.7443,
+      "step": 2617
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481901908596311e-05,
+      "loss": 0.7088,
+      "step": 2618
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4815117882260756e-05,
+      "loss": 0.8145,
+      "step": 2619
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4811215724299046e-05,
+      "loss": 0.7935,
+      "step": 2620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4807312612851306e-05,
+      "loss": 0.8758,
+      "step": 2621
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4803408548691058e-05,
+      "loss": 0.7843,
+      "step": 2622
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4799503532592006e-05,
+      "loss": 0.8692,
+      "step": 2623
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4795597565328045e-05,
+      "loss": 0.7995,
+      "step": 2624
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479169064767326e-05,
+      "loss": 0.878,
+      "step": 2625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4787782780401926e-05,
+      "loss": 0.9187,
+      "step": 2626
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.47838739642885e-05,
+      "loss": 0.3814,
+      "step": 2627
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4779964200107633e-05,
+      "loss": 0.8681,
+      "step": 2628
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4776053488634156e-05,
+      "loss": 0.8001,
+      "step": 2629
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4772141830643103e-05,
+      "loss": 0.8524,
+      "step": 2630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4768229226909679e-05,
+      "loss": 0.7745,
+      "step": 2631
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4764315678209284e-05,
+      "loss": 0.7606,
+      "step": 2632
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760401185317505e-05,
+      "loss": 0.813,
+      "step": 2633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.475648574901012e-05,
+      "loss": 0.7425,
+      "step": 2634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4752569370063079e-05,
+      "loss": 0.7397,
+      "step": 2635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4748652049252539e-05,
+      "loss": 0.8438,
+      "step": 2636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4744733787354831e-05,
+      "loss": 0.7916,
+      "step": 2637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4740814585146472e-05,
+      "loss": 0.8304,
+      "step": 2638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4736894443404177e-05,
+      "loss": 0.8296,
+      "step": 2639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4732973362904833e-05,
+      "loss": 0.7876,
+      "step": 2640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4729051344425518e-05,
+      "loss": 0.3511,
+      "step": 2641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4725128388743503e-05,
+      "loss": 0.8057,
+      "step": 2642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4721204496636235e-05,
+      "loss": 0.8265,
+      "step": 2643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471727966888135e-05,
+      "loss": 0.8462,
+      "step": 2644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471335390625667e-05,
+      "loss": 0.8418,
+      "step": 2645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4709427209540204e-05,
+      "loss": 0.8584,
+      "step": 2646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4705499579510143e-05,
+      "loss": 0.8405,
+      "step": 2647
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4701571016944863e-05,
+      "loss": 0.8714,
+      "step": 2648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4697641522622927e-05,
+      "loss": 0.8319,
+      "step": 2649
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4693711097323082e-05,
+      "loss": 0.7936,
+      "step": 2650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.468977974182426e-05,
+      "loss": 0.8321,
+      "step": 2651
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4685847456905573e-05,
+      "loss": 0.8395,
+      "step": 2652
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4681914243346324e-05,
+      "loss": 0.7447,
+      "step": 2653
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4677980101925995e-05,
+      "loss": 0.8042,
+      "step": 2654
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4674045033424254e-05,
+      "loss": 0.8106,
+      "step": 2655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4670109038620952e-05,
+      "loss": 0.3624,
+      "step": 2656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4666172118296125e-05,
+      "loss": 0.8915,
+      "step": 2657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4662234273229989e-05,
+      "loss": 0.8486,
+      "step": 2658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4658295504202948e-05,
+      "loss": 0.7488,
+      "step": 2659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4654355811995587e-05,
+      "loss": 0.7792,
+      "step": 2660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4650415197388672e-05,
+      "loss": 0.7712,
+      "step": 2661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4646473661163155e-05,
+      "loss": 0.8129,
+      "step": 2662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4642531204100165e-05,
+      "loss": 0.8282,
+      "step": 2663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4638587826981021e-05,
+      "loss": 0.8542,
+      "step": 2664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463464353058722e-05,
+      "loss": 0.3556,
+      "step": 2665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4630698315700441e-05,
+      "loss": 0.8951,
+      "step": 2666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4626752183102552e-05,
+      "loss": 0.8324,
+      "step": 2667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.462280513357559e-05,
+      "loss": 0.7519,
+      "step": 2668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4618857167901784e-05,
+      "loss": 0.8591,
+      "step": 2669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461490828686354e-05,
+      "loss": 0.3757,
+      "step": 2670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4610958491243447e-05,
+      "loss": 0.7834,
+      "step": 2671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4607007781824277e-05,
+      "loss": 0.8685,
+      "step": 2672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4603056159388977e-05,
+      "loss": 0.8864,
+      "step": 2673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459910362472068e-05,
+      "loss": 0.7114,
+      "step": 2674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4595150178602701e-05,
+      "loss": 0.6362,
+      "step": 2675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4591195821818533e-05,
+      "loss": 0.7465,
+      "step": 2676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4587240555151849e-05,
+      "loss": 0.7866,
+      "step": 2677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4583284379386502e-05,
+      "loss": 0.7178,
+      "step": 2678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4579327295306528e-05,
+      "loss": 0.8479,
+      "step": 2679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4575369303696138e-05,
+      "loss": 0.8988,
+      "step": 2680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4571410405339733e-05,
+      "loss": 0.8291,
+      "step": 2681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4567450601021881e-05,
+      "loss": 0.8294,
+      "step": 2682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4563489891527338e-05,
+      "loss": 0.8796,
+      "step": 2683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4559528277641037e-05,
+      "loss": 0.7769,
+      "step": 2684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455556576014809e-05,
+      "loss": 0.8033,
+      "step": 2685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4551602339833784e-05,
+      "loss": 0.864,
+      "step": 2686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4547638017483592e-05,
+      "loss": 0.7684,
+      "step": 2687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4543672793883166e-05,
+      "loss": 0.8022,
+      "step": 2688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4539706669818329e-05,
+      "loss": 0.8721,
+      "step": 2689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453573964607509e-05,
+      "loss": 0.7575,
+      "step": 2690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4531771723439633e-05,
+      "loss": 0.8166,
+      "step": 2691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4527802902698318e-05,
+      "loss": 0.8308,
+      "step": 2692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4523833184637685e-05,
+      "loss": 0.8282,
+      "step": 2693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4519862570044457e-05,
+      "loss": 0.8201,
+      "step": 2694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4515891059705523e-05,
+      "loss": 0.8603,
+      "step": 2695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4511918654407962e-05,
+      "loss": 0.8475,
+      "step": 2696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4507945354939021e-05,
+      "loss": 0.8611,
+      "step": 2697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450397116208613e-05,
+      "loss": 0.7574,
+      "step": 2698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449999607663689e-05,
+      "loss": 0.8375,
+      "step": 2699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4496020099379089e-05,
+      "loss": 0.8591,
+      "step": 2700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4492043231100684e-05,
+      "loss": 0.8502,
+      "step": 2701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4488065472589807e-05,
+      "loss": 0.3664,
+      "step": 2702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4484086824634771e-05,
+      "loss": 0.869,
+      "step": 2703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4480107288024065e-05,
+      "loss": 0.788,
+      "step": 2704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4476126863546354e-05,
+      "loss": 0.8265,
+      "step": 2705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4472145551990473e-05,
+      "loss": 0.8544,
+      "step": 2706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4468163354145443e-05,
+      "loss": 0.8749,
+      "step": 2707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464180270800455e-05,
+      "loss": 0.8388,
+      "step": 2708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4460196302744873e-05,
+      "loss": 0.7331,
+      "step": 2709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.445621145076824e-05,
+      "loss": 0.8255,
+      "step": 2710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4452225715660274e-05,
+      "loss": 0.8125,
+      "step": 2711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4448239098210868e-05,
+      "loss": 0.9003,
+      "step": 2712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4444251599210086e-05,
+      "loss": 0.8535,
+      "step": 2713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4440263219448175e-05,
+      "loss": 0.7507,
+      "step": 2714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443627395971555e-05,
+      "loss": 0.8472,
+      "step": 2715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44322838208028e-05,
+      "loss": 0.7951,
+      "step": 2716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4428292803500692e-05,
+      "loss": 0.8427,
+      "step": 2717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4424300908600162e-05,
+      "loss": 0.8511,
+      "step": 2718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4420308136892325e-05,
+      "loss": 0.373,
+      "step": 2719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441631448916847e-05,
+      "loss": 0.8261,
+      "step": 2720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4412319966220054e-05,
+      "loss": 0.8288,
+      "step": 2721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4408324568838716e-05,
+      "loss": 0.836,
+      "step": 2722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440432829781626e-05,
+      "loss": 0.8489,
+      "step": 2723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4400331153944662e-05,
+      "loss": 0.9205,
+      "step": 2724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4396333138016085e-05,
+      "loss": 0.3506,
+      "step": 2725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4392334250822847e-05,
+      "loss": 0.8259,
+      "step": 2726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438833449315745e-05,
+      "loss": 0.7992,
+      "step": 2727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4384333865812568e-05,
+      "loss": 0.7549,
+      "step": 2728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.438033236958104e-05,
+      "loss": 0.8166,
+      "step": 2729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4376330005255885e-05,
+      "loss": 0.849,
+      "step": 2730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4372326773630289e-05,
+      "loss": 0.7462,
+      "step": 2731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4368322675497614e-05,
+      "loss": 0.8059,
+      "step": 2732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4364317711651392e-05,
+      "loss": 0.718,
+      "step": 2733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436031188288532e-05,
+      "loss": 0.8572,
+      "step": 2734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4356305189993282e-05,
+      "loss": 0.735,
+      "step": 2735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435229763376932e-05,
+      "loss": 0.8082,
+      "step": 2736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4348289215007648e-05,
+      "loss": 0.8542,
+      "step": 2737
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4344279934502652e-05,
+      "loss": 0.7968,
+      "step": 2738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4340269793048896e-05,
+      "loss": 0.7907,
+      "step": 2739
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4336258791441108e-05,
+      "loss": 0.7849,
+      "step": 2740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4332246930474189e-05,
+      "loss": 0.8247,
+      "step": 2741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4328234210943209e-05,
+      "loss": 0.8788,
+      "step": 2742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324220633643403e-05,
+      "loss": 0.8376,
+      "step": 2743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4320206199370185e-05,
+      "loss": 0.7898,
+      "step": 2744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4316190908919136e-05,
+      "loss": 0.8222,
+      "step": 2745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4312174763086004e-05,
+      "loss": 0.8239,
+      "step": 2746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4308157762666708e-05,
+      "loss": 0.7934,
+      "step": 2747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304139908457338e-05,
+      "loss": 0.8192,
+      "step": 2748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4300121201254152e-05,
+      "loss": 0.8378,
+      "step": 2749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4296101641853573e-05,
+      "loss": 0.8855,
+      "step": 2750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42920812310522e-05,
+      "loss": 0.8495,
+      "step": 2751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42880599696468e-05,
+      "loss": 0.8829,
+      "step": 2752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4284037858434299e-05,
+      "loss": 0.8993,
+      "step": 2753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4280014898211803e-05,
+      "loss": 0.82,
+      "step": 2754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4275991089776579e-05,
+      "loss": 0.8029,
+      "step": 2755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4271966433926072e-05,
+      "loss": 0.8163,
+      "step": 2756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426794093145788e-05,
+      "loss": 0.3876,
+      "step": 2757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4263914583169778e-05,
+      "loss": 0.8315,
+      "step": 2758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4259887389859707e-05,
+      "loss": 0.8659,
+      "step": 2759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4255859352325777e-05,
+      "loss": 0.8737,
+      "step": 2760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4251830471366264e-05,
+      "loss": 0.819,
+      "step": 2761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424780074777961e-05,
+      "loss": 0.748,
+      "step": 2762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4243770182364427e-05,
+      "loss": 0.8407,
+      "step": 2763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4239738775919489e-05,
+      "loss": 0.3899,
+      "step": 2764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4235706529243742e-05,
+      "loss": 0.7634,
+      "step": 2765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4231673443136293e-05,
+      "loss": 0.7259,
+      "step": 2766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4227639518396421e-05,
+      "loss": 0.8771,
+      "step": 2767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422360475582357e-05,
+      "loss": 0.9075,
+      "step": 2768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4219569156217343e-05,
+      "loss": 0.7819,
+      "step": 2769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4215532720377522e-05,
+      "loss": 0.8221,
+      "step": 2770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4211495449104042e-05,
+      "loss": 0.381,
+      "step": 2771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4207457343197011e-05,
+      "loss": 0.8479,
+      "step": 2772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4203418403456697e-05,
+      "loss": 0.9574,
+      "step": 2773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4199378630683541e-05,
+      "loss": 0.8509,
+      "step": 2774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4195338025678141e-05,
+      "loss": 0.8289,
+      "step": 2775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4191296589241268e-05,
+      "loss": 0.7807,
+      "step": 2776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4187254322173848e-05,
+      "loss": 0.8555,
+      "step": 2777
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4183211225276981e-05,
+      "loss": 0.7655,
+      "step": 2778
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4179167299351926e-05,
+      "loss": 0.7688,
+      "step": 2779
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4175122545200105e-05,
+      "loss": 0.8013,
+      "step": 2780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4171076963623107e-05,
+      "loss": 0.8776,
+      "step": 2781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4167030555422692e-05,
+      "loss": 0.8326,
+      "step": 2782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4162983321400767e-05,
+      "loss": 0.8449,
+      "step": 2783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4158935262359421e-05,
+      "loss": 0.8148,
+      "step": 2784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154886379100892e-05,
+      "loss": 0.8731,
+      "step": 2785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4150836672427587e-05,
+      "loss": 0.3778,
+      "step": 2786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4146786143142075e-05,
+      "loss": 0.9037,
+      "step": 2787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4142734792047098e-05,
+      "loss": 0.8293,
+      "step": 2788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4138682619945543e-05,
+      "loss": 0.8017,
+      "step": 2789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4134629627640473e-05,
+      "loss": 0.7657,
+      "step": 2790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4130575815935108e-05,
+      "loss": 0.8035,
+      "step": 2791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4126521185632834e-05,
+      "loss": 0.916,
+      "step": 2792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4122465737537191e-05,
+      "loss": 0.8785,
+      "step": 2793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4118409472451894e-05,
+      "loss": 0.7297,
+      "step": 2794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4114352391180812e-05,
+      "loss": 0.7905,
+      "step": 2795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4110294494527973e-05,
+      "loss": 0.8296,
+      "step": 2796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4106235783297575e-05,
+      "loss": 0.8587,
+      "step": 2797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.410217625829397e-05,
+      "loss": 0.8527,
+      "step": 2798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4098115920321676e-05,
+      "loss": 0.7798,
+      "step": 2799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4094054770185363e-05,
+      "loss": 0.7747,
+      "step": 2800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.408999280868988e-05,
+      "loss": 0.7735,
+      "step": 2801
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4085930036640218e-05,
+      "loss": 0.8094,
+      "step": 2802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4081866454841541e-05,
+      "loss": 0.8246,
+      "step": 2803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4077802064099169e-05,
+      "loss": 0.3793,
+      "step": 2804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407373686521858e-05,
+      "loss": 0.9028,
+      "step": 2805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406967085900541e-05,
+      "loss": 0.7931,
+      "step": 2806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4065604046265469e-05,
+      "loss": 0.8551,
+      "step": 2807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4061536427804714e-05,
+      "loss": 0.7541,
+      "step": 2808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4057468004429262e-05,
+      "loss": 0.8704,
+      "step": 2809
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4053398776945395e-05,
+      "loss": 0.8633,
+      "step": 2810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4049328746159554e-05,
+      "loss": 0.7917,
+      "step": 2811
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4045257912878332e-05,
+      "loss": 0.7832,
+      "step": 2812
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4041186277908492e-05,
+      "loss": 0.8507,
+      "step": 2813
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4037113842056944e-05,
+      "loss": 0.7929,
+      "step": 2814
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4033040606130764e-05,
+      "loss": 0.7579,
+      "step": 2815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402896657093719e-05,
+      "loss": 0.8173,
+      "step": 2816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402489173728361e-05,
+      "loss": 0.8873,
+      "step": 2817
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4020816105977574e-05,
+      "loss": 0.797,
+      "step": 2818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016739677826794e-05,
+      "loss": 0.8343,
+      "step": 2819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4012662453639127e-05,
+      "loss": 0.8435,
+      "step": 2820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4008584434222607e-05,
+      "loss": 0.7755,
+      "step": 2821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4004505620385408e-05,
+      "loss": 0.8404,
+      "step": 2822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400042601293587e-05,
+      "loss": 0.8026,
+      "step": 2823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3996345612682493e-05,
+      "loss": 0.8471,
+      "step": 2824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3992264420433927e-05,
+      "loss": 0.783,
+      "step": 2825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3988182436998983e-05,
+      "loss": 0.842,
+      "step": 2826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3984099663186626e-05,
+      "loss": 0.8384,
+      "step": 2827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3980016099805977e-05,
+      "loss": 0.8101,
+      "step": 2828
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3975931747666324e-05,
+      "loss": 0.801,
+      "step": 2829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39718466075771e-05,
+      "loss": 0.8358,
+      "step": 2830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967760680347894e-05,
+      "loss": 0.7841,
+      "step": 2831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396367396678846e-05,
+      "loss": 0.8722,
+      "step": 2832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39595864677087e-05,
+      "loss": 0.6938,
+      "step": 2833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.395549818391867e-05,
+      "loss": 0.7873,
+      "step": 2834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3951409116228588e-05,
+      "loss": 0.759,
+      "step": 2835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394731926544883e-05,
+      "loss": 0.7987,
+      "step": 2836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3943228632389913e-05,
+      "loss": 0.8435,
+      "step": 2837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3939137217862527e-05,
+      "loss": 0.346,
+      "step": 2838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3935045022677502e-05,
+      "loss": 0.8689,
+      "step": 2839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3930952047645833e-05,
+      "loss": 0.8894,
+      "step": 2840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3926858293578657e-05,
+      "loss": 0.7812,
+      "step": 2841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392276376128728e-05,
+      "loss": 0.8451,
+      "step": 2842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391866845158316e-05,
+      "loss": 0.8137,
+      "step": 2843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3914572365277896e-05,
+      "loss": 0.9033,
+      "step": 2844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3910475503183255e-05,
+      "loss": 0.8105,
+      "step": 2845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390637786611115e-05,
+      "loss": 0.8267,
+      "step": 2846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390227945487365e-05,
+      "loss": 0.7812,
+      "step": 2847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389818027028298e-05,
+      "loss": 0.7993,
+      "step": 2848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894080313151514e-05,
+      "loss": 0.8075,
+      "step": 2849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3889979584291782e-05,
+      "loss": 0.9024,
+      "step": 2850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3885878084516467e-05,
+      "loss": 0.3539,
+      "step": 2851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38817758146384e-05,
+      "loss": 0.8564,
+      "step": 2852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3877672775470572e-05,
+      "loss": 0.8298,
+      "step": 2853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3873568967826118e-05,
+      "loss": 0.7857,
+      "step": 2854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3869464392518332e-05,
+      "loss": 0.8074,
+      "step": 2855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3865359050360663e-05,
+      "loss": 0.8478,
+      "step": 2856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3861252942166702e-05,
+      "loss": 0.7588,
+      "step": 2857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3857146068750197e-05,
+      "loss": 0.8389,
+      "step": 2858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385303843092505e-05,
+      "loss": 0.8148,
+      "step": 2859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384893002950531e-05,
+      "loss": 0.3645,
+      "step": 2860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3844820865305179e-05,
+      "loss": 0.8636,
+      "step": 2861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3840710939139012e-05,
+      "loss": 0.7842,
+      "step": 2862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3836600251821315e-05,
+      "loss": 0.9541,
+      "step": 2863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3832488804166742e-05,
+      "loss": 0.8098,
+      "step": 2864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3828376596990096e-05,
+      "loss": 0.8248,
+      "step": 2865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3824263631106339e-05,
+      "loss": 0.7977,
+      "step": 2866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3820149907330574e-05,
+      "loss": 0.8015,
+      "step": 2867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816035426478057e-05,
+      "loss": 0.7999,
+      "step": 2868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3811920189364199e-05,
+      "loss": 0.7973,
+      "step": 2869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3807804196804557e-05,
+      "loss": 0.796,
+      "step": 2870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3803687449614835e-05,
+      "loss": 0.7696,
+      "step": 2871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3799569948610894e-05,
+      "loss": 0.8503,
+      "step": 2872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3795451694608734e-05,
+      "loss": 0.8508,
+      "step": 2873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3791332688424514e-05,
+      "loss": 0.3623,
+      "step": 2874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3787212930874534e-05,
+      "loss": 0.3505,
+      "step": 2875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3783092422775252e-05,
+      "loss": 0.8656,
+      "step": 2876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3778971164943267e-05,
+      "loss": 0.7897,
+      "step": 2877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3774849158195332e-05,
+      "loss": 0.7832,
+      "step": 2878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3770726403348341e-05,
+      "loss": 0.8388,
+      "step": 2879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376660290121934e-05,
+      "loss": 0.8086,
+      "step": 2880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3762478652625532e-05,
+      "loss": 0.8128,
+      "step": 2881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3758353658384254e-05,
+      "loss": 0.8496,
+      "step": 2882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3754227919313e-05,
+      "loss": 0.7382,
+      "step": 2883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3750101436229403e-05,
+      "loss": 0.8693,
+      "step": 2884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3745974209951254e-05,
+      "loss": 0.7855,
+      "step": 2885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3741846241296483e-05,
+      "loss": 0.8439,
+      "step": 2886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3737717531083172e-05,
+      "loss": 0.8788,
+      "step": 2887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3733588080129548e-05,
+      "loss": 0.8238,
+      "step": 2888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3729457889253986e-05,
+      "loss": 0.8483,
+      "step": 2889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3725326959275003e-05,
+      "loss": 0.8818,
+      "step": 2890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3721195291011269e-05,
+      "loss": 0.8201,
+      "step": 2891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3717062885281598e-05,
+      "loss": 0.814,
+      "step": 2892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3712929742904947e-05,
+      "loss": 0.8718,
+      "step": 2893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3708795864700424e-05,
+      "loss": 0.8066,
+      "step": 2894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3704661251487282e-05,
+      "loss": 0.8129,
+      "step": 2895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.370052590408491e-05,
+      "loss": 0.785,
+      "step": 2896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3696389823312858e-05,
+      "loss": 0.7886,
+      "step": 2897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3692253009990812e-05,
+      "loss": 0.8359,
+      "step": 2898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3688115464938605e-05,
+      "loss": 0.8755,
+      "step": 2899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3683977188976217e-05,
+      "loss": 0.7846,
+      "step": 2900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3679838182923766e-05,
+      "loss": 0.7161,
+      "step": 2901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3675698447601521e-05,
+      "loss": 0.8399,
+      "step": 2902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36715579838299e-05,
+      "loss": 0.8812,
+      "step": 2903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3667416792429454e-05,
+      "loss": 0.8011,
+      "step": 2904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3663274874220884e-05,
+      "loss": 0.7882,
+      "step": 2905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3659132230025036e-05,
+      "loss": 0.7894,
+      "step": 2906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3654988860662896e-05,
+      "loss": 0.7434,
+      "step": 2907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3650844766955603e-05,
+      "loss": 0.7443,
+      "step": 2908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3646699949724423e-05,
+      "loss": 0.8086,
+      "step": 2909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3642554409790782e-05,
+      "loss": 0.7786,
+      "step": 2910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3638408147976242e-05,
+      "loss": 0.8658,
+      "step": 2911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363426116510251e-05,
+      "loss": 0.8275,
+      "step": 2912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363011346199143e-05,
+      "loss": 0.8199,
+      "step": 2913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3625965039464997e-05,
+      "loss": 0.8479,
+      "step": 2914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3621815898345343e-05,
+      "loss": 0.7576,
+      "step": 2915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361766603945474e-05,
+      "loss": 0.7595,
+      "step": 2916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3613515463615613e-05,
+      "loss": 0.9024,
+      "step": 2917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3609364171650522e-05,
+      "loss": 0.8656,
+      "step": 2918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3605212164382169e-05,
+      "loss": 0.8564,
+      "step": 2919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3601059442633394e-05,
+      "loss": 0.7606,
+      "step": 2920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3596906007227188e-05,
+      "loss": 0.8718,
+      "step": 2921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3592751858986675e-05,
+      "loss": 0.8125,
+      "step": 2922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3588596998735125e-05,
+      "loss": 0.8399,
+      "step": 2923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3584441427295945e-05,
+      "loss": 0.842,
+      "step": 2924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358028514549269e-05,
+      "loss": 0.788,
+      "step": 2925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3576128154149046e-05,
+      "loss": 0.8707,
+      "step": 2926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357197045408885e-05,
+      "loss": 0.8719,
+      "step": 2927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356781204613607e-05,
+      "loss": 0.7503,
+      "step": 2928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3563652931114822e-05,
+      "loss": 0.8688,
+      "step": 2929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3559493109849357e-05,
+      "loss": 0.7741,
+      "step": 2930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355533258316407e-05,
+      "loss": 0.8213,
+      "step": 2931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3551171351883493e-05,
+      "loss": 0.8497,
+      "step": 2932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3547009416832295e-05,
+      "loss": 0.843,
+      "step": 2933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.354284677883529e-05,
+      "loss": 0.7869,
+      "step": 2934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353868343871743e-05,
+      "loss": 0.7955,
+      "step": 2935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3534519397303803e-05,
+      "loss": 0.8625,
+      "step": 2936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353035465541964e-05,
+      "loss": 0.7641,
+      "step": 2937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3526189213890306e-05,
+      "loss": 0.8712,
+      "step": 2938
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3522023073541313e-05,
+      "loss": 0.8547,
+      "step": 2939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3517856235198301e-05,
+      "loss": 0.8441,
+      "step": 2940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3513688699687056e-05,
+      "loss": 0.8504,
+      "step": 2941
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3509520467833496e-05,
+      "loss": 0.8478,
+      "step": 2942
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3505351540463687e-05,
+      "loss": 0.7473,
+      "step": 2943
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350118191840382e-05,
+      "loss": 0.8815,
+      "step": 2944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3497011602480234e-05,
+      "loss": 0.8361,
+      "step": 2945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3492840593519401e-05,
+      "loss": 0.8208,
+      "step": 2946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3488668892347932e-05,
+      "loss": 0.8502,
+      "step": 2947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3484496499792568e-05,
+      "loss": 0.8175,
+      "step": 2948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.34803234166802e-05,
+      "loss": 0.8669,
+      "step": 2949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3476149643837846e-05,
+      "loss": 0.8137,
+      "step": 2950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3471975182092665e-05,
+      "loss": 0.8256,
+      "step": 2951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467800032271949e-05,
+      "loss": 0.8527,
+      "step": 2952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3463624195203134e-05,
+      "loss": 0.8298,
+      "step": 2953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3459447671713779e-05,
+      "loss": 0.8416,
+      "step": 2954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3455270462631594e-05,
+      "loss": 0.8774,
+      "step": 2955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3451092568784412e-05,
+      "loss": 0.8597,
+      "step": 2956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3446913991000207e-05,
+      "loss": 0.8795,
+      "step": 2957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3442734730107097e-05,
+      "loss": 0.8197,
+      "step": 2958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343855478693332e-05,
+      "loss": 0.7664,
+      "step": 2959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3434374162307258e-05,
+      "loss": 0.8569,
+      "step": 2960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3430192857057428e-05,
+      "loss": 0.7294,
+      "step": 2961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342601087201248e-05,
+      "loss": 0.8263,
+      "step": 2962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.34218282080012e-05,
+      "loss": 0.7989,
+      "step": 2963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3417644865852504e-05,
+      "loss": 0.7883,
+      "step": 2964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.341346084639545e-05,
+      "loss": 0.7778,
+      "step": 2965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3409276150459225e-05,
+      "loss": 0.7665,
+      "step": 2966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340509077887315e-05,
+      "loss": 0.7571,
+      "step": 2967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3400904732466686e-05,
+      "loss": 0.7921,
+      "step": 2968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3396718012069416e-05,
+      "loss": 0.7518,
+      "step": 2969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3392530618511067e-05,
+      "loss": 0.9262,
+      "step": 2970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33883425526215e-05,
+      "loss": 0.8388,
+      "step": 2971
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33841538152307e-05,
+      "loss": 0.8733,
+      "step": 2972
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3379964407168796e-05,
+      "loss": 0.8117,
+      "step": 2973
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3375774329266036e-05,
+      "loss": 0.8676,
+      "step": 2974
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3371583582352816e-05,
+      "loss": 0.7776,
+      "step": 2975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3367392167259654e-05,
+      "loss": 0.8567,
+      "step": 2976
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3363200084817203e-05,
+      "loss": 0.8309,
+      "step": 2977
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3359007335856254e-05,
+      "loss": 0.7822,
+      "step": 2978
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3354813921207721e-05,
+      "loss": 0.8033,
+      "step": 2979
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335061984170266e-05,
+      "loss": 0.8542,
+      "step": 2980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3346425098172245e-05,
+      "loss": 0.7977,
+      "step": 2981
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3342229691447796e-05,
+      "loss": 0.7141,
+      "step": 2982
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3338033622360755e-05,
+      "loss": 0.7753,
+      "step": 2983
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3333836891742698e-05,
+      "loss": 0.8127,
+      "step": 2984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3329639500425336e-05,
+      "loss": 0.8031,
+      "step": 2985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3325441449240507e-05,
+      "loss": 0.8427,
+      "step": 2986
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3321242739020175e-05,
+      "loss": 0.8002,
+      "step": 2987
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3317043370596443e-05,
+      "loss": 0.8125,
+      "step": 2988
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3312843344801543e-05,
+      "loss": 0.802,
+      "step": 2989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330864266246783e-05,
+      "loss": 0.7951,
+      "step": 2990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3304441324427801e-05,
+      "loss": 0.7475,
+      "step": 2991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3300239331514076e-05,
+      "loss": 0.9279,
+      "step": 2992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3296036684559402e-05,
+      "loss": 0.8138,
+      "step": 2993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3291833384396657e-05,
+      "loss": 0.7584,
+      "step": 2994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3287629431858858e-05,
+      "loss": 0.7342,
+      "step": 2995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3283424827779137e-05,
+      "loss": 0.7912,
+      "step": 2996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3279219572990762e-05,
+      "loss": 0.8761,
+      "step": 2997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3275013668327137e-05,
+      "loss": 0.8599,
+      "step": 2998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.327080711462178e-05,
+      "loss": 0.8484,
+      "step": 2999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3266599912708346e-05,
+      "loss": 0.8542,
+      "step": 3000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3262392063420623e-05,
+      "loss": 0.84,
+      "step": 3001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3258183567592517e-05,
+      "loss": 0.7834,
+      "step": 3002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325397442605807e-05,
+      "loss": 0.8897,
+      "step": 3003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3249764639651441e-05,
+      "loss": 0.7737,
+      "step": 3004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3245554209206936e-05,
+      "loss": 0.764,
+      "step": 3005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3241343135558976e-05,
+      "loss": 0.7397,
+      "step": 3006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3237131419542106e-05,
+      "loss": 0.8735,
+      "step": 3007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3232919061991003e-05,
+      "loss": 0.8603,
+      "step": 3008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3228706063740475e-05,
+      "loss": 0.8196,
+      "step": 3009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.322449242562545e-05,
+      "loss": 0.8334,
+      "step": 3010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3220278148480987e-05,
+      "loss": 0.8516,
+      "step": 3011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3216063233142276e-05,
+      "loss": 0.8413,
+      "step": 3012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321184768044462e-05,
+      "loss": 0.3427,
+      "step": 3013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3207631491223465e-05,
+      "loss": 0.8033,
+      "step": 3014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3203414666314366e-05,
+      "loss": 0.8349,
+      "step": 3015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3199197206553022e-05,
+      "loss": 0.8431,
+      "step": 3016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319497911277524e-05,
+      "loss": 0.8364,
+      "step": 3017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3190760385816965e-05,
+      "loss": 0.8259,
+      "step": 3018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3186541026514266e-05,
+      "loss": 0.7849,
+      "step": 3019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3182321035703331e-05,
+      "loss": 0.8713,
+      "step": 3020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317810041422048e-05,
+      "loss": 0.7394,
+      "step": 3021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173879162902154e-05,
+      "loss": 0.7411,
+      "step": 3022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316965728258492e-05,
+      "loss": 0.8073,
+      "step": 3023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316543477410547e-05,
+      "loss": 0.8001,
+      "step": 3024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3161211638300617e-05,
+      "loss": 0.7603,
+      "step": 3025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3156987876007307e-05,
+      "loss": 0.3991,
+      "step": 3026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3152763488062601e-05,
+      "loss": 0.7936,
+      "step": 3027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3148538475303692e-05,
+      "loss": 0.783,
+      "step": 3028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3144312838567885e-05,
+      "loss": 0.7945,
+      "step": 3029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3140086578692616e-05,
+      "loss": 0.8471,
+      "step": 3030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135859696515455e-05,
+      "loss": 0.7574,
+      "step": 3031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3131632192874075e-05,
+      "loss": 0.8045,
+      "step": 3032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3127404068606286e-05,
+      "loss": 0.7172,
+      "step": 3033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3123175324550016e-05,
+      "loss": 0.8031,
+      "step": 3034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3118945961543318e-05,
+      "loss": 0.688,
+      "step": 3035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3114715980424368e-05,
+      "loss": 0.8813,
+      "step": 3036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311048538203146e-05,
+      "loss": 0.8493,
+      "step": 3037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3106254167203009e-05,
+      "loss": 0.8347,
+      "step": 3038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3102022336777567e-05,
+      "loss": 0.8262,
+      "step": 3039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309778989159379e-05,
+      "loss": 0.3403,
+      "step": 3040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3093556832490468e-05,
+      "loss": 0.8638,
+      "step": 3041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3089323160306504e-05,
+      "loss": 0.7899,
+      "step": 3042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3085088875880928e-05,
+      "loss": 0.8938,
+      "step": 3043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308085398005289e-05,
+      "loss": 0.8284,
+      "step": 3044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3076618473661661e-05,
+      "loss": 0.812,
+      "step": 3045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3072382357546633e-05,
+      "loss": 0.8412,
+      "step": 3046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306814563254732e-05,
+      "loss": 0.3732,
+      "step": 3047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3063908299503356e-05,
+      "loss": 0.7624,
+      "step": 3048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3059670359254492e-05,
+      "loss": 0.8033,
+      "step": 3049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3055431812640606e-05,
+      "loss": 0.839,
+      "step": 3050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3051192660501685e-05,
+      "loss": 0.8368,
+      "step": 3051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3046952903677855e-05,
+      "loss": 0.7135,
+      "step": 3052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3042712543009342e-05,
+      "loss": 0.8009,
+      "step": 3053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3038471579336502e-05,
+      "loss": 0.77,
+      "step": 3054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3034230013499811e-05,
+      "loss": 0.8365,
+      "step": 3055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302998784633986e-05,
+      "loss": 0.7841,
+      "step": 3056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3025745078697361e-05,
+      "loss": 0.7717,
+      "step": 3057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302150171141314e-05,
+      "loss": 0.8352,
+      "step": 3058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3017257745328155e-05,
+      "loss": 0.8302,
+      "step": 3059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3013013181283473e-05,
+      "loss": 0.8944,
+      "step": 3060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300876802012028e-05,
+      "loss": 0.3849,
+      "step": 3061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3004522262679881e-05,
+      "loss": 0.8036,
+      "step": 3062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3000275909803701e-05,
+      "loss": 0.7973,
+      "step": 3063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2996028962333278e-05,
+      "loss": 0.8196,
+      "step": 3064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2991781421110274e-05,
+      "loss": 0.8171,
+      "step": 3065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2987533286976466e-05,
+      "loss": 0.8289,
+      "step": 3066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2983284560773747e-05,
+      "loss": 0.8571,
+      "step": 3067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2979035243344134e-05,
+      "loss": 0.8708,
+      "step": 3068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2974785335529751e-05,
+      "loss": 0.8628,
+      "step": 3069
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2970534838172846e-05,
+      "loss": 0.8696,
+      "step": 3070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.296628375211578e-05,
+      "loss": 0.6971,
+      "step": 3071
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2962032078201034e-05,
+      "loss": 0.7473,
+      "step": 3072
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2957779817271205e-05,
+      "loss": 0.7763,
+      "step": 3073
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2953526970169005e-05,
+      "loss": 0.8587,
+      "step": 3074
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2949273537737263e-05,
+      "loss": 0.7764,
+      "step": 3075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2945019520818922e-05,
+      "loss": 0.8188,
+      "step": 3076
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2940764920257043e-05,
+      "loss": 0.8488,
+      "step": 3077
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2936509736894802e-05,
+      "loss": 0.7718,
+      "step": 3078
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2932253971575493e-05,
+      "loss": 0.7109,
+      "step": 3079
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2927997625142518e-05,
+      "loss": 0.7958,
+      "step": 3080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2923740698439405e-05,
+      "loss": 0.7363,
+      "step": 3081
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2919483192309787e-05,
+      "loss": 0.3614,
+      "step": 3082
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2915225107597418e-05,
+      "loss": 0.7444,
+      "step": 3083
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2910966445146164e-05,
+      "loss": 0.8212,
+      "step": 3084
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2906707205800004e-05,
+      "loss": 0.8189,
+      "step": 3085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2902447390403037e-05,
+      "loss": 0.7593,
+      "step": 3086
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.289818699979947e-05,
+      "loss": 0.7819,
+      "step": 3087
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2893926034833628e-05,
+      "loss": 0.7787,
+      "step": 3088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288966449634995e-05,
+      "loss": 0.7336,
+      "step": 3089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2885402385192984e-05,
+      "loss": 0.8063,
+      "step": 3090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2881139702207394e-05,
+      "loss": 0.7398,
+      "step": 3091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.287687644823796e-05,
+      "loss": 0.8022,
+      "step": 3092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2872612624129575e-05,
+      "loss": 0.882,
+      "step": 3093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2868348230727236e-05,
+      "loss": 0.7953,
+      "step": 3094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2864083268876068e-05,
+      "loss": 0.8359,
+      "step": 3095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2859817739421296e-05,
+      "loss": 0.8542,
+      "step": 3096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2855551643208264e-05,
+      "loss": 0.7484,
+      "step": 3097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2851284981082421e-05,
+      "loss": 0.7388,
+      "step": 3098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2847017753889344e-05,
+      "loss": 0.7974,
+      "step": 3099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28427499624747e-05,
+      "loss": 0.8873,
+      "step": 3100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2838481607684288e-05,
+      "loss": 0.8447,
+      "step": 3101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2834212690364005e-05,
+      "loss": 0.7772,
+      "step": 3102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2829943211359867e-05,
+      "loss": 0.8601,
+      "step": 3103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2825673171517996e-05,
+      "loss": 0.879,
+      "step": 3104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2821402571684631e-05,
+      "loss": 0.3399,
+      "step": 3105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2817131412706112e-05,
+      "loss": 0.8717,
+      "step": 3106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2812859695428907e-05,
+      "loss": 0.7508,
+      "step": 3107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2808587420699578e-05,
+      "loss": 0.8144,
+      "step": 3108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2804314589364805e-05,
+      "loss": 0.7907,
+      "step": 3109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280004120227138e-05,
+      "loss": 0.9136,
+      "step": 3110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2795767260266196e-05,
+      "loss": 0.8415,
+      "step": 3111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2791492764196267e-05,
+      "loss": 0.7653,
+      "step": 3112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2787217714908708e-05,
+      "loss": 0.8438,
+      "step": 3113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2782942113250754e-05,
+      "loss": 0.7774,
+      "step": 3114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2778665960069737e-05,
+      "loss": 0.8063,
+      "step": 3115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2774389256213108e-05,
+      "loss": 0.8138,
+      "step": 3116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2770112002528422e-05,
+      "loss": 0.8086,
+      "step": 3117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2765834199863343e-05,
+      "loss": 0.7996,
+      "step": 3118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2761555849065647e-05,
+      "loss": 0.8141,
+      "step": 3119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2757276950983217e-05,
+      "loss": 0.905,
+      "step": 3120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2752997506464047e-05,
+      "loss": 0.7782,
+      "step": 3121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2748717516356233e-05,
+      "loss": 0.855,
+      "step": 3122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2744436981507984e-05,
+      "loss": 0.8438,
+      "step": 3123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2740155902767614e-05,
+      "loss": 0.8124,
+      "step": 3124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2735874280983551e-05,
+      "loss": 0.7967,
+      "step": 3125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2731592117004318e-05,
+      "loss": 0.8501,
+      "step": 3126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2727309411678564e-05,
+      "loss": 0.8242,
+      "step": 3127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272302616585503e-05,
+      "loss": 0.7504,
+      "step": 3128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2718742380382567e-05,
+      "loss": 0.7831,
+      "step": 3129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2714458056110139e-05,
+      "loss": 0.3394,
+      "step": 3130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271017319388681e-05,
+      "loss": 0.7298,
+      "step": 3131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2705887794561752e-05,
+      "loss": 0.8052,
+      "step": 3132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.270160185898425e-05,
+      "loss": 0.827,
+      "step": 3133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697315388003686e-05,
+      "loss": 0.838,
+      "step": 3134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2693028382469552e-05,
+      "loss": 0.757,
+      "step": 3135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2688740843231449e-05,
+      "loss": 0.8736,
+      "step": 3136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2684452771139077e-05,
+      "loss": 0.8152,
+      "step": 3137
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2680164167042249e-05,
+      "loss": 0.7855,
+      "step": 3138
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2675875031790876e-05,
+      "loss": 0.8241,
+      "step": 3139
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2671585366234984e-05,
+      "loss": 0.9071,
+      "step": 3140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2667295171224696e-05,
+      "loss": 0.7327,
+      "step": 3141
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266300444761024e-05,
+      "loss": 0.7673,
+      "step": 3142
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2658713196241952e-05,
+      "loss": 0.816,
+      "step": 3143
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265442141797027e-05,
+      "loss": 0.8893,
+      "step": 3144
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2650129113645743e-05,
+      "loss": 0.7469,
+      "step": 3145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2645836284119013e-05,
+      "loss": 0.8133,
+      "step": 3146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2641542930240838e-05,
+      "loss": 0.7885,
+      "step": 3147
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263724905286207e-05,
+      "loss": 0.849,
+      "step": 3148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2632954652833673e-05,
+      "loss": 0.8274,
+      "step": 3149
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2628659731006707e-05,
+      "loss": 0.8265,
+      "step": 3150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2624364288232341e-05,
+      "loss": 0.8134,
+      "step": 3151
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2620068325361846e-05,
+      "loss": 0.8393,
+      "step": 3152
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2615771843246589e-05,
+      "loss": 0.7173,
+      "step": 3153
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2611474842738055e-05,
+      "loss": 0.8474,
+      "step": 3154
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2607177324687816e-05,
+      "loss": 0.8543,
+      "step": 3155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2602879289947559e-05,
+      "loss": 0.8403,
+      "step": 3156
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2598580739369065e-05,
+      "loss": 0.8206,
+      "step": 3157
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2594281673804219e-05,
+      "loss": 0.7639,
+      "step": 3158
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2589982094105009e-05,
+      "loss": 0.791,
+      "step": 3159
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258568200112353e-05,
+      "loss": 0.8235,
+      "step": 3160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2581381395711968e-05,
+      "loss": 0.8509,
+      "step": 3161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2577080278722619e-05,
+      "loss": 0.7749,
+      "step": 3162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2572778651007877e-05,
+      "loss": 0.8195,
+      "step": 3163
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.256847651342024e-05,
+      "loss": 0.7182,
+      "step": 3164
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2564173866812305e-05,
+      "loss": 0.8171,
+      "step": 3165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2559870712036766e-05,
+      "loss": 0.8162,
+      "step": 3166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2555567049946423e-05,
+      "loss": 0.7841,
+      "step": 3167
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.255126288139418e-05,
+      "loss": 0.8111,
+      "step": 3168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2546958207233028e-05,
+      "loss": 0.8412,
+      "step": 3169
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2542653028316073e-05,
+      "loss": 0.8811,
+      "step": 3170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2538347345496515e-05,
+      "loss": 0.7912,
+      "step": 3171
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.253404115962765e-05,
+      "loss": 0.3722,
+      "step": 3172
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.252973447156288e-05,
+      "loss": 0.8288,
+      "step": 3173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2525427282155702e-05,
+      "loss": 0.6949,
+      "step": 3174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2521119592259714e-05,
+      "loss": 0.8277,
+      "step": 3175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2516811402728615e-05,
+      "loss": 0.847,
+      "step": 3176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2512502714416198e-05,
+      "loss": 0.9444,
+      "step": 3177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2508193528176365e-05,
+      "loss": 0.8341,
+      "step": 3178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.25038838448631e-05,
+      "loss": 0.7775,
+      "step": 3179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2499573665330507e-05,
+      "loss": 0.8504,
+      "step": 3180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2495262990432769e-05,
+      "loss": 0.4111,
+      "step": 3181
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2490951821024176e-05,
+      "loss": 0.8526,
+      "step": 3182
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486640157959116e-05,
+      "loss": 0.7506,
+      "step": 3183
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2482328002092072e-05,
+      "loss": 0.712,
+      "step": 3184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2478015354277631e-05,
+      "loss": 0.8081,
+      "step": 3185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2473702215370467e-05,
+      "loss": 0.818,
+      "step": 3186
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.246938858622536e-05,
+      "loss": 0.85,
+      "step": 3187
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2465074467697184e-05,
+      "loss": 0.7725,
+      "step": 3188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2460759860640915e-05,
+      "loss": 0.777,
+      "step": 3189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2456444765911613e-05,
+      "loss": 0.7993,
+      "step": 3190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2452129184364448e-05,
+      "loss": 0.8861,
+      "step": 3191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2447813116854678e-05,
+      "loss": 0.7472,
+      "step": 3192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2443496564237667e-05,
+      "loss": 0.8312,
+      "step": 3193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2439179527368861e-05,
+      "loss": 0.7809,
+      "step": 3194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2434862007103816e-05,
+      "loss": 0.7598,
+      "step": 3195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2430544004298173e-05,
+      "loss": 0.7033,
+      "step": 3196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2426225519807677e-05,
+      "loss": 0.8544,
+      "step": 3197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242190655448816e-05,
+      "loss": 0.8677,
+      "step": 3198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241758710919556e-05,
+      "loss": 0.9076,
+      "step": 3199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2413267184785894e-05,
+      "loss": 0.854,
+      "step": 3200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2408946782115293e-05,
+      "loss": 0.8031,
+      "step": 3201
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2404625902039975e-05,
+      "loss": 0.869,
+      "step": 3202
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2400304545416244e-05,
+      "loss": 0.7644,
+      "step": 3203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2395982713100511e-05,
+      "loss": 0.8511,
+      "step": 3204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2391660405949275e-05,
+      "loss": 0.8168,
+      "step": 3205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2387337624819131e-05,
+      "loss": 0.8643,
+      "step": 3206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2383014370566762e-05,
+      "loss": 0.845,
+      "step": 3207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2378690644048957e-05,
+      "loss": 0.8577,
+      "step": 3208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2374366446122585e-05,
+      "loss": 0.8188,
+      "step": 3209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2370041777644622e-05,
+      "loss": 0.7804,
+      "step": 3210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2365716639472124e-05,
+      "loss": 0.8828,
+      "step": 3211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2361391032462252e-05,
+      "loss": 0.665,
+      "step": 3212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2357064957472246e-05,
+      "loss": 0.8111,
+      "step": 3213
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2352738415359453e-05,
+      "loss": 0.808,
+      "step": 3214
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2348411406981307e-05,
+      "loss": 0.3579,
+      "step": 3215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2344083933195332e-05,
+      "loss": 0.7736,
+      "step": 3216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2339755994859147e-05,
+      "loss": 0.8898,
+      "step": 3217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233542759283046e-05,
+      "loss": 0.8278,
+      "step": 3218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2331098727967075e-05,
+      "loss": 0.8312,
+      "step": 3219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2326769401126881e-05,
+      "loss": 0.6885,
+      "step": 3220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2322439613167871e-05,
+      "loss": 0.7714,
+      "step": 3221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2318109364948119e-05,
+      "loss": 0.8855,
+      "step": 3222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2313778657325792e-05,
+      "loss": 0.775,
+      "step": 3223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.230944749115915e-05,
+      "loss": 0.7966,
+      "step": 3224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2305115867306541e-05,
+      "loss": 0.7857,
+      "step": 3225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2300783786626407e-05,
+      "loss": 0.765,
+      "step": 3226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2296451249977277e-05,
+      "loss": 0.8595,
+      "step": 3227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2292118258217778e-05,
+      "loss": 0.792,
+      "step": 3228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2287784812206616e-05,
+      "loss": 0.7659,
+      "step": 3229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283450912802594e-05,
+      "loss": 0.8227,
+      "step": 3230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2279116560864609e-05,
+      "loss": 0.8351,
+      "step": 3231
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2274781757251632e-05,
+      "loss": 0.8187,
+      "step": 3232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2270446502822742e-05,
+      "loss": 0.8423,
+      "step": 3233
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2266110798437095e-05,
+      "loss": 0.8038,
+      "step": 3234
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2261774644953941e-05,
+      "loss": 0.8192,
+      "step": 3235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2257438043232624e-05,
+      "loss": 0.8468,
+      "step": 3236
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2253100994132561e-05,
+      "loss": 0.8772,
+      "step": 3237
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2248763498513275e-05,
+      "loss": 0.755,
+      "step": 3238
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2244425557234366e-05,
+      "loss": 0.7974,
+      "step": 3239
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2240087171155527e-05,
+      "loss": 0.889,
+      "step": 3240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2235748341136545e-05,
+      "loss": 0.7597,
+      "step": 3241
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223140906803728e-05,
+      "loss": 0.814,
+      "step": 3242
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2227069352717696e-05,
+      "loss": 0.7967,
+      "step": 3243
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222272919603783e-05,
+      "loss": 0.8069,
+      "step": 3244
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221838859885782e-05,
+      "loss": 0.8096,
+      "step": 3245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221404756203788e-05,
+      "loss": 0.788,
+      "step": 3246
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2209706086438317e-05,
+      "loss": 0.7875,
+      "step": 3247
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220536417291953e-05,
+      "loss": 0.8902,
+      "step": 3248
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2201021822341992e-05,
+      "loss": 0.8627,
+      "step": 3249
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219667903556627e-05,
+      "loss": 0.8157,
+      "step": 3250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219233581345302e-05,
+      "loss": 0.8153,
+      "step": 3251
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2187992156862977e-05,
+      "loss": 0.8534,
+      "step": 3252
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2183648066656973e-05,
+      "loss": 0.7299,
+      "step": 3253
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217930354369591e-05,
+      "loss": 0.8363,
+      "step": 3254
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217495858884079e-05,
+      "loss": 0.6832,
+      "step": 3255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2170613202952696e-05,
+      "loss": 0.8224,
+      "step": 3256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2166267386892799e-05,
+      "loss": 0.8536,
+      "step": 3257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2161921141522346e-05,
+      "loss": 0.8496,
+      "step": 3258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2157574467702678e-05,
+      "loss": 0.3792,
+      "step": 3259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2153227366295213e-05,
+      "loss": 0.7304,
+      "step": 3260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2148879838161467e-05,
+      "loss": 0.8592,
+      "step": 3261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.214453188416303e-05,
+      "loss": 0.7102,
+      "step": 3262
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2140183505161577e-05,
+      "loss": 0.7321,
+      "step": 3263
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2135834702018872e-05,
+      "loss": 0.3622,
+      "step": 3264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2131485475596756e-05,
+      "loss": 0.8715,
+      "step": 3265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212713582675716e-05,
+      "loss": 0.7664,
+      "step": 3266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2122785756362101e-05,
+      "loss": 0.7728,
+      "step": 3267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2118435265273666e-05,
+      "loss": 0.8654,
+      "step": 3268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2114084354354042e-05,
+      "loss": 0.8197,
+      "step": 3269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2109733024465494e-05,
+      "loss": 0.8146,
+      "step": 3270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210538127647036e-05,
+      "loss": 0.8454,
+      "step": 3271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2101029111231075e-05,
+      "loss": 0.7986,
+      "step": 3272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2096676529610147e-05,
+      "loss": 0.8115,
+      "step": 3273
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2092323532470172e-05,
+      "loss": 0.8394,
+      "step": 3274
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2087970120673824e-05,
+      "loss": 0.8452,
+      "step": 3275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2083616295083864e-05,
+      "loss": 0.7272,
+      "step": 3276
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207926205656313e-05,
+      "loss": 0.8504,
+      "step": 3277
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2074907405974551e-05,
+      "loss": 0.8689,
+      "step": 3278
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2070552344181123e-05,
+      "loss": 0.7758,
+      "step": 3279
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2066196872045934e-05,
+      "loss": 0.3788,
+      "step": 3280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206184099043215e-05,
+      "loss": 0.8492,
+      "step": 3281
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2057484700203022e-05,
+      "loss": 0.8235,
+      "step": 3282
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2053128002221877e-05,
+      "loss": 0.8501,
+      "step": 3283
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2048770897352124e-05,
+      "loss": 0.8028,
+      "step": 3284
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2044413386457256e-05,
+      "loss": 0.8083,
+      "step": 3285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2040055470400842e-05,
+      "loss": 0.757,
+      "step": 3286
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2035697150046532e-05,
+      "loss": 0.7559,
+      "step": 3287
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2031338426258058e-05,
+      "loss": 0.7433,
+      "step": 3288
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2026979299899235e-05,
+      "loss": 0.7421,
+      "step": 3289
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202261977183395e-05,
+      "loss": 0.7377,
+      "step": 3290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2018259842926175e-05,
+      "loss": 0.7455,
+      "step": 3291
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2013899514039958e-05,
+      "loss": 0.7928,
+      "step": 3292
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2009538786039433e-05,
+      "loss": 0.8292,
+      "step": 3293
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2005177659788803e-05,
+      "loss": 0.7974,
+      "step": 3294
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2000816136152356e-05,
+      "loss": 0.7476,
+      "step": 3295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1996454215994462e-05,
+      "loss": 0.8233,
+      "step": 3296
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1992091900179566e-05,
+      "loss": 0.8169,
+      "step": 3297
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1987729189572185e-05,
+      "loss": 0.7494,
+      "step": 3298
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1983366085036929e-05,
+      "loss": 0.7645,
+      "step": 3299
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1979002587438471e-05,
+      "loss": 0.7906,
+      "step": 3300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1974638697641565e-05,
+      "loss": 0.794,
+      "step": 3301
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1970274416511054e-05,
+      "loss": 0.8994,
+      "step": 3302
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196590974491185e-05,
+      "loss": 0.8216,
+      "step": 3303
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1961544683708936e-05,
+      "loss": 0.8243,
+      "step": 3304
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1957179233767389e-05,
+      "loss": 0.8141,
+      "step": 3305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1952813395952343e-05,
+      "loss": 0.7304,
+      "step": 3306
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1948447171129027e-05,
+      "loss": 0.8033,
+      "step": 3307
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1944080560162732e-05,
+      "loss": 0.8241,
+      "step": 3308
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.193971356391884e-05,
+      "loss": 0.726,
+      "step": 3309
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1935346183262796e-05,
+      "loss": 0.8311,
+      "step": 3310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1930978419060128e-05,
+      "loss": 0.8951,
+      "step": 3311
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1926610272176442e-05,
+      "loss": 0.3779,
+      "step": 3312
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1922241743477414e-05,
+      "loss": 0.72,
+      "step": 3313
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1917872833828798e-05,
+      "loss": 0.3319,
+      "step": 3314
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1913503544096422e-05,
+      "loss": 0.8592,
+      "step": 3315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1909133875146196e-05,
+      "loss": 0.7813,
+      "step": 3316
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1904763827844097e-05,
+      "loss": 0.8331,
+      "step": 3317
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1900393403056182e-05,
+      "loss": 0.7875,
+      "step": 3318
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1896022601648583e-05,
+      "loss": 0.8794,
+      "step": 3319
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1891651424487503e-05,
+      "loss": 0.8021,
+      "step": 3320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1887279872439214e-05,
+      "loss": 0.8422,
+      "step": 3321
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188290794637008e-05,
+      "loss": 0.3625,
+      "step": 3322
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1878535647146525e-05,
+      "loss": 0.7871,
+      "step": 3323
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187416297563505e-05,
+      "loss": 0.8125,
+      "step": 3324
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.186978993270223e-05,
+      "loss": 0.792,
+      "step": 3325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1865416519214717e-05,
+      "loss": 0.801,
+      "step": 3326
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1861042736039227e-05,
+      "loss": 0.8017,
+      "step": 3327
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1856668584042562e-05,
+      "loss": 0.8021,
+      "step": 3328
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185229406409159e-05,
+      "loss": 0.8503,
+      "step": 3329
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1847919177053251e-05,
+      "loss": 0.7557,
+      "step": 3330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184354392379456e-05,
+      "loss": 0.8423,
+      "step": 3331
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1839168305182602e-05,
+      "loss": 0.8053,
+      "step": 3332
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1834792322084541e-05,
+      "loss": 0.8104,
+      "step": 3333
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1830415975367606e-05,
+      "loss": 0.7911,
+      "step": 3334
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18260392658991e-05,
+      "loss": 0.8428,
+      "step": 3335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1821662194546397e-05,
+      "loss": 0.8092,
+      "step": 3336
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181728476217695e-05,
+      "loss": 0.7385,
+      "step": 3337
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1812906969658276e-05,
+      "loss": 0.3729,
+      "step": 3338
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1808528817857963e-05,
+      "loss": 0.8122,
+      "step": 3339
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1804150307643672e-05,
+      "loss": 0.8233,
+      "step": 3340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1799771439883138e-05,
+      "loss": 0.7982,
+      "step": 3341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1795392215444164e-05,
+      "loss": 0.8,
+      "step": 3342
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1791012635194622e-05,
+      "loss": 0.8547,
+      "step": 3343
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178663270000246e-05,
+      "loss": 0.8413,
+      "step": 3344
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1782252410735691e-05,
+      "loss": 0.7942,
+      "step": 3345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1777871768262402e-05,
+      "loss": 0.7878,
+      "step": 3346
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1773490773450742e-05,
+      "loss": 0.7321,
+      "step": 3347
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176910942716894e-05,
+      "loss": 0.803,
+      "step": 3348
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1764727730285293e-05,
+      "loss": 0.7762,
+      "step": 3349
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1760345683668162e-05,
+      "loss": 0.8617,
+      "step": 3350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.175596328818598e-05,
+      "loss": 0.8992,
+      "step": 3351
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.175158054470725e-05,
+      "loss": 0.8819,
+      "step": 3352
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1747197454100548e-05,
+      "loss": 0.7224,
+      "step": 3353
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1742814017234507e-05,
+      "loss": 0.3788,
+      "step": 3354
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.173843023497784e-05,
+      "loss": 0.7762,
+      "step": 3355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1734046108199322e-05,
+      "loss": 0.7588,
+      "step": 3356
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1729661637767802e-05,
+      "loss": 0.8182,
+      "step": 3357
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172527682455219e-05,
+      "loss": 0.795,
+      "step": 3358
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172089166942147e-05,
+      "loss": 0.8677,
+      "step": 3359
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1716506173244692e-05,
+      "loss": 0.8685,
+      "step": 3360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1712120336890974e-05,
+      "loss": 0.7011,
+      "step": 3361
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1707734161229494e-05,
+      "loss": 0.7561,
+      "step": 3362
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1703347647129508e-05,
+      "loss": 0.7898,
+      "step": 3363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1698960795460336e-05,
+      "loss": 0.7457,
+      "step": 3364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1694573607091364e-05,
+      "loss": 0.873,
+      "step": 3365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169018608289204e-05,
+      "loss": 0.7597,
+      "step": 3366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1685798223731888e-05,
+      "loss": 0.8191,
+      "step": 3367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168141003048049e-05,
+      "loss": 0.7926,
+      "step": 3368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1677021504007493e-05,
+      "loss": 0.7625,
+      "step": 3369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1672632645182624e-05,
+      "loss": 0.7551,
+      "step": 3370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1668243454875663e-05,
+      "loss": 0.8025,
+      "step": 3371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1663853933956458e-05,
+      "loss": 0.7783,
+      "step": 3372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1659464083294922e-05,
+      "loss": 0.803,
+      "step": 3373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1655073903761038e-05,
+      "loss": 0.8219,
+      "step": 3374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165068339622485e-05,
+      "loss": 0.8799,
+      "step": 3375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1646292561556468e-05,
+      "loss": 0.7632,
+      "step": 3376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1641901400626066e-05,
+      "loss": 0.8398,
+      "step": 3377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1637509914303887e-05,
+      "loss": 0.8071,
+      "step": 3378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1633118103460233e-05,
+      "loss": 0.7729,
+      "step": 3379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.162872596896547e-05,
+      "loss": 0.7364,
+      "step": 3380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1624333511690034e-05,
+      "loss": 0.8635,
+      "step": 3381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.161994073250442e-05,
+      "loss": 0.8524,
+      "step": 3382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.161554763227919e-05,
+      "loss": 0.8295,
+      "step": 3383
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1611154211884966e-05,
+      "loss": 0.7707,
+      "step": 3384
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1606760472192436e-05,
+      "loss": 0.7808,
+      "step": 3385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160236641407235e-05,
+      "loss": 0.8465,
+      "step": 3386
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1597972038395524e-05,
+      "loss": 0.7701,
+      "step": 3387
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1593577346032831e-05,
+      "loss": 0.7803,
+      "step": 3388
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1589182337855213e-05,
+      "loss": 0.8094,
+      "step": 3389
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1584787014733675e-05,
+      "loss": 0.8254,
+      "step": 3390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1580391377539275e-05,
+      "loss": 0.8215,
+      "step": 3391
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1575995427143145e-05,
+      "loss": 0.804,
+      "step": 3392
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1571599164416471e-05,
+      "loss": 0.8358,
+      "step": 3393
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1567202590230507e-05,
+      "loss": 0.8695,
+      "step": 3394
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1562805705456561e-05,
+      "loss": 0.7064,
+      "step": 3395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1558408510966008e-05,
+      "loss": 0.8583,
+      "step": 3396
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1554011007630286e-05,
+      "loss": 0.7939,
+      "step": 3397
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1549613196320888e-05,
+      "loss": 0.8362,
+      "step": 3398
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1545215077909376e-05,
+      "loss": 0.7981,
+      "step": 3399
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1540816653267366e-05,
+      "loss": 0.834,
+      "step": 3400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1536417923266536e-05,
+      "loss": 0.787,
+      "step": 3401
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1532018888778627e-05,
+      "loss": 0.8122,
+      "step": 3402
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152761955067544e-05,
+      "loss": 0.8264,
+      "step": 3403
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1523219909828831e-05,
+      "loss": 0.7778,
+      "step": 3404
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1518819967110726e-05,
+      "loss": 0.8047,
+      "step": 3405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1514419723393103e-05,
+      "loss": 0.741,
+      "step": 3406
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1510019179547998e-05,
+      "loss": 0.8205,
+      "step": 3407
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1505618336447515e-05,
+      "loss": 0.8741,
+      "step": 3408
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1501217194963807e-05,
+      "loss": 0.8337,
+      "step": 3409
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1496815755969097e-05,
+      "loss": 0.8074,
+      "step": 3410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149241402033566e-05,
+      "loss": 0.7975,
+      "step": 3411
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1488011988935832e-05,
+      "loss": 0.7704,
+      "step": 3412
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1483609662642005e-05,
+      "loss": 0.8131,
+      "step": 3413
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1479207042326634e-05,
+      "loss": 0.672,
+      "step": 3414
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1474804128862224e-05,
+      "loss": 0.8313,
+      "step": 3415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1470400923121352e-05,
+      "loss": 0.8394,
+      "step": 3416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1465997425976635e-05,
+      "loss": 0.8382,
+      "step": 3417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1461593638300766e-05,
+      "loss": 0.8367,
+      "step": 3418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1457189560966481e-05,
+      "loss": 0.786,
+      "step": 3419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1452785194846588e-05,
+      "loss": 0.8157,
+      "step": 3420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1448380540813932e-05,
+      "loss": 0.8103,
+      "step": 3421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1443975599741435e-05,
+      "loss": 0.8441,
+      "step": 3422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1439570372502064e-05,
+      "loss": 0.8042,
+      "step": 3423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1435164859968848e-05,
+      "loss": 0.7994,
+      "step": 3424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1430759063014869e-05,
+      "loss": 0.7308,
+      "step": 3425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1426352982513272e-05,
+      "loss": 0.8594,
+      "step": 3426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1421946619337246e-05,
+      "loss": 0.7898,
+      "step": 3427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1417539974360047e-05,
+      "loss": 0.7954,
+      "step": 3428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1413133048454985e-05,
+      "loss": 0.7928,
+      "step": 3429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1408725842495419e-05,
+      "loss": 0.7529,
+      "step": 3430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1404318357354774e-05,
+      "loss": 0.842,
+      "step": 3431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1399910593906522e-05,
+      "loss": 0.8287,
+      "step": 3432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1395502553024196e-05,
+      "loss": 0.7741,
+      "step": 3433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1391094235581374e-05,
+      "loss": 0.7554,
+      "step": 3434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1386685642451703e-05,
+      "loss": 0.7631,
+      "step": 3435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1382276774508874e-05,
+      "loss": 0.7961,
+      "step": 3436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1377867632626633e-05,
+      "loss": 0.847,
+      "step": 3437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137345821767879e-05,
+      "loss": 0.8222,
+      "step": 3438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1369048530539197e-05,
+      "loss": 0.7995,
+      "step": 3439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1364638572081767e-05,
+      "loss": 0.8479,
+      "step": 3440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1360228343180464e-05,
+      "loss": 0.799,
+      "step": 3441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135581784470931e-05,
+      "loss": 0.8413,
+      "step": 3442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1351407077542368e-05,
+      "loss": 0.8208,
+      "step": 3443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1346996042553772e-05,
+      "loss": 0.7228,
+      "step": 3444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1342584740617701e-05,
+      "loss": 0.8079,
+      "step": 3445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133817317260838e-05,
+      "loss": 0.8538,
+      "step": 3446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1333761339400096e-05,
+      "loss": 0.8979,
+      "step": 3447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1329349241867184e-05,
+      "loss": 0.8126,
+      "step": 3448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1324936880884034e-05,
+      "loss": 0.8325,
+      "step": 3449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1320524257325086e-05,
+      "loss": 0.8083,
+      "step": 3450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1316111372064833e-05,
+      "loss": 0.8676,
+      "step": 3451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1311698225977823e-05,
+      "loss": 0.8044,
+      "step": 3452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.130728481993865e-05,
+      "loss": 0.8175,
+      "step": 3453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1302871154821963e-05,
+      "loss": 0.7702,
+      "step": 3454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1298457231502462e-05,
+      "loss": 0.7608,
+      "step": 3455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1294043050854894e-05,
+      "loss": 0.757,
+      "step": 3456
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1289628613754064e-05,
+      "loss": 0.8255,
+      "step": 3457
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1285213921074827e-05,
+      "loss": 0.8675,
+      "step": 3458
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1280798973692083e-05,
+      "loss": 0.7921,
+      "step": 3459
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1276383772480783e-05,
+      "loss": 0.7869,
+      "step": 3460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1271968318315936e-05,
+      "loss": 0.7818,
+      "step": 3461
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1267552612072595e-05,
+      "loss": 0.3623,
+      "step": 3462
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1263136654625862e-05,
+      "loss": 0.7587,
+      "step": 3463
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1258720446850889e-05,
+      "loss": 0.7999,
+      "step": 3464
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1254303989622882e-05,
+      "loss": 0.7073,
+      "step": 3465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1249887283817098e-05,
+      "loss": 0.7526,
+      "step": 3466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1245470330308831e-05,
+      "loss": 0.3659,
+      "step": 3467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1241053129973441e-05,
+      "loss": 0.7774,
+      "step": 3468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1236635683686319e-05,
+      "loss": 0.7513,
+      "step": 3469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1232217992322918e-05,
+      "loss": 0.8176,
+      "step": 3470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1227800056758737e-05,
+      "loss": 0.8397,
+      "step": 3471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1223381877869322e-05,
+      "loss": 0.7799,
+      "step": 3472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1218963456530263e-05,
+      "loss": 0.807,
+      "step": 3473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1214544793617206e-05,
+      "loss": 0.7605,
+      "step": 3474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121012589000584e-05,
+      "loss": 0.7395,
+      "step": 3475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1205706746571902e-05,
+      "loss": 0.7751,
+      "step": 3476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1201287364191175e-05,
+      "loss": 0.8831,
+      "step": 3477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1196867743739496e-05,
+      "loss": 0.8564,
+      "step": 3478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1192447886092743e-05,
+      "loss": 0.8417,
+      "step": 3479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1188027792126841e-05,
+      "loss": 0.8177,
+      "step": 3480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1183607462717767e-05,
+      "loss": 0.8274,
+      "step": 3481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1179186898741539e-05,
+      "loss": 0.807,
+      "step": 3482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1174766101074223e-05,
+      "loss": 0.8004,
+      "step": 3483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1170345070591934e-05,
+      "loss": 0.7829,
+      "step": 3484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1165923808170827e-05,
+      "loss": 0.77,
+      "step": 3485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1161502314687114e-05,
+      "loss": 0.7267,
+      "step": 3486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1157080591017041e-05,
+      "loss": 0.8257,
+      "step": 3487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1152658638036908e-05,
+      "loss": 0.757,
+      "step": 3488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1148236456623054e-05,
+      "loss": 0.8645,
+      "step": 3489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1143814047651864e-05,
+      "loss": 0.3813,
+      "step": 3490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1139391411999774e-05,
+      "loss": 0.8477,
+      "step": 3491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1134968550543264e-05,
+      "loss": 0.6645,
+      "step": 3492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113054546415885e-05,
+      "loss": 0.8837,
+      "step": 3493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1126122153723102e-05,
+      "loss": 0.7926,
+      "step": 3494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1121698620112631e-05,
+      "loss": 0.8348,
+      "step": 3495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111727486420409e-05,
+      "loss": 0.735,
+      "step": 3496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1112850886874182e-05,
+      "loss": 0.7914,
+      "step": 3497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108426688999645e-05,
+      "loss": 0.6651,
+      "step": 3498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1104002271457273e-05,
+      "loss": 0.7628,
+      "step": 3499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1099577635123891e-05,
+      "loss": 0.7559,
+      "step": 3500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1095152780876373e-05,
+      "loss": 0.8589,
+      "step": 3501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1090727709591637e-05,
+      "loss": 0.8037,
+      "step": 3502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086302422146647e-05,
+      "loss": 0.8106,
+      "step": 3503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10818769194184e-05,
+      "loss": 0.7846,
+      "step": 3504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1077451202283939e-05,
+      "loss": 0.8376,
+      "step": 3505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1073025271620361e-05,
+      "loss": 0.7428,
+      "step": 3506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1068599128304792e-05,
+      "loss": 0.3464,
+      "step": 3507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1064172773214404e-05,
+      "loss": 0.7535,
+      "step": 3508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105974620722641e-05,
+      "loss": 0.8079,
+      "step": 3509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1055319431218067e-05,
+      "loss": 0.7484,
+      "step": 3510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1050892446066674e-05,
+      "loss": 0.7385,
+      "step": 3511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1046465252649572e-05,
+      "loss": 0.8022,
+      "step": 3512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1042037851844137e-05,
+      "loss": 0.7826,
+      "step": 3513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1037610244527795e-05,
+      "loss": 0.8153,
+      "step": 3514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1033182431578005e-05,
+      "loss": 0.7244,
+      "step": 3515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1028754413872273e-05,
+      "loss": 0.8383,
+      "step": 3516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1024326192288141e-05,
+      "loss": 0.3323,
+      "step": 3517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1019897767703193e-05,
+      "loss": 0.808,
+      "step": 3518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1015469140995056e-05,
+      "loss": 0.821,
+      "step": 3519
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1011040313041395e-05,
+      "loss": 0.7076,
+      "step": 3520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.100661128471991e-05,
+      "loss": 0.7956,
+      "step": 3521
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1002182056908351e-05,
+      "loss": 0.7467,
+      "step": 3522
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0997752630484496e-05,
+      "loss": 0.8114,
+      "step": 3523
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099332300632617e-05,
+      "loss": 0.7938,
+      "step": 3524
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0988893185311236e-05,
+      "loss": 0.7716,
+      "step": 3525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0984463168317598e-05,
+      "loss": 0.8246,
+      "step": 3526
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0980032956223193e-05,
+      "loss": 0.8421,
+      "step": 3527
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0975602549906e-05,
+      "loss": 0.8356,
+      "step": 3528
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0971171950244037e-05,
+      "loss": 0.7694,
+      "step": 3529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0966741158115357e-05,
+      "loss": 0.7587,
+      "step": 3530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0962310174398054e-05,
+      "loss": 0.842,
+      "step": 3531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0957878999970264e-05,
+      "loss": 0.8023,
+      "step": 3532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0953447635710154e-05,
+      "loss": 0.8289,
+      "step": 3533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0949016082495934e-05,
+      "loss": 0.3854,
+      "step": 3534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0944584341205842e-05,
+      "loss": 0.3536,
+      "step": 3535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0940152412718163e-05,
+      "loss": 0.7805,
+      "step": 3536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0935720297911219e-05,
+      "loss": 0.8475,
+      "step": 3537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093128799766336e-05,
+      "loss": 0.8312,
+      "step": 3538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0926855512852983e-05,
+      "loss": 0.8488,
+      "step": 3539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0922422844358518e-05,
+      "loss": 0.7252,
+      "step": 3540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0917989993058428e-05,
+      "loss": 0.7488,
+      "step": 3541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0913556959831215e-05,
+      "loss": 0.4029,
+      "step": 3542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0909123745555418e-05,
+      "loss": 0.8316,
+      "step": 3543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090469035110961e-05,
+      "loss": 0.7465,
+      "step": 3544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0900256777372398e-05,
+      "loss": 0.871,
+      "step": 3545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0895823025222433e-05,
+      "loss": 0.8075,
+      "step": 3546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0891389095538391e-05,
+      "loss": 0.3748,
+      "step": 3547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0886954989198988e-05,
+      "loss": 0.8062,
+      "step": 3548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0882520707082978e-05,
+      "loss": 0.7518,
+      "step": 3549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0878086250069142e-05,
+      "loss": 0.8473,
+      "step": 3550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0873651619036298e-05,
+      "loss": 0.7032,
+      "step": 3551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0869216814863308e-05,
+      "loss": 0.795,
+      "step": 3552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0864781838429055e-05,
+      "loss": 0.7236,
+      "step": 3553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0860346690612465e-05,
+      "loss": 0.8361,
+      "step": 3554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0855911372292492e-05,
+      "loss": 0.7163,
+      "step": 3555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0851475884348131e-05,
+      "loss": 0.8055,
+      "step": 3556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0847040227658401e-05,
+      "loss": 0.7868,
+      "step": 3557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0842604403102364e-05,
+      "loss": 0.8625,
+      "step": 3558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083816841155911e-05,
+      "loss": 0.7276,
+      "step": 3559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083373225390776e-05,
+      "loss": 0.3587,
+      "step": 3560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0829295931027476e-05,
+      "loss": 0.8238,
+      "step": 3561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.082485944379745e-05,
+      "loss": 0.8043,
+      "step": 3562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0820422793096894e-05,
+      "loss": 0.8088,
+      "step": 3563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0815985979805072e-05,
+      "loss": 0.7954,
+      "step": 3564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0811549004801268e-05,
+      "loss": 0.8013,
+      "step": 3565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0807111868964799e-05,
+      "loss": 0.8518,
+      "step": 3566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0802674573175018e-05,
+      "loss": 0.7846,
+      "step": 3567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.079823711831131e-05,
+      "loss": 0.7878,
+      "step": 3568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0793799505253087e-05,
+      "loss": 0.7667,
+      "step": 3569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078936173487979e-05,
+      "loss": 0.812,
+      "step": 3570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0784923808070904e-05,
+      "loss": 0.7998,
+      "step": 3571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780485725705932e-05,
+      "loss": 0.7806,
+      "step": 3572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0776047488664412e-05,
+      "loss": 0.831,
+      "step": 3573
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0771609097825915e-05,
+      "loss": 0.7419,
+      "step": 3574
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0767170554070038e-05,
+      "loss": 0.8648,
+      "step": 3575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0762731858276416e-05,
+      "loss": 0.7172,
+      "step": 3576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0758293011324706e-05,
+      "loss": 0.7996,
+      "step": 3577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0753854014094596e-05,
+      "loss": 0.8035,
+      "step": 3578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0749414867465807e-05,
+      "loss": 0.7794,
+      "step": 3579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0744975572318088e-05,
+      "loss": 0.7876,
+      "step": 3580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0740536129531223e-05,
+      "loss": 0.7705,
+      "step": 3581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0736096539985014e-05,
+      "loss": 0.8623,
+      "step": 3582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0731656804559302e-05,
+      "loss": 0.367,
+      "step": 3583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0727216924133952e-05,
+      "loss": 0.8329,
+      "step": 3584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0722776899588856e-05,
+      "loss": 0.8428,
+      "step": 3585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071833673180394e-05,
+      "loss": 0.7603,
+      "step": 3586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0713896421659156e-05,
+      "loss": 0.8775,
+      "step": 3587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0709455970034486e-05,
+      "loss": 0.8244,
+      "step": 3588
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0705015377809933e-05,
+      "loss": 0.8338,
+      "step": 3589
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0700574645865537e-05,
+      "loss": 0.7845,
+      "step": 3590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0696133775081364e-05,
+      "loss": 0.7749,
+      "step": 3591
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0691692766337493e-05,
+      "loss": 0.8229,
+      "step": 3592
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0687251620514056e-05,
+      "loss": 0.8393,
+      "step": 3593
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0682810338491192e-05,
+      "loss": 0.7689,
+      "step": 3594
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0678368921149076e-05,
+      "loss": 0.7526,
+      "step": 3595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0673927369367905e-05,
+      "loss": 0.3858,
+      "step": 3596
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066948568402791e-05,
+      "loss": 0.3395,
+      "step": 3597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0665043866009338e-05,
+      "loss": 0.7596,
+      "step": 3598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0660601916192466e-05,
+      "loss": 0.73,
+      "step": 3599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0656159835457606e-05,
+      "loss": 0.8159,
+      "step": 3600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0651717624685087e-05,
+      "loss": 0.8746,
+      "step": 3601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0647275284755262e-05,
+      "loss": 0.7029,
+      "step": 3602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0642832816548521e-05,
+      "loss": 0.7653,
+      "step": 3603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0638390220945264e-05,
+      "loss": 0.817,
+      "step": 3604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0633947498825922e-05,
+      "loss": 0.8062,
+      "step": 3605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0629504651070962e-05,
+      "loss": 0.8161,
+      "step": 3606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062506167856086e-05,
+      "loss": 0.8601,
+      "step": 3607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0620618582176128e-05,
+      "loss": 0.3574,
+      "step": 3608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0616175362797296e-05,
+      "loss": 0.7823,
+      "step": 3609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0611732021304921e-05,
+      "loss": 0.821,
+      "step": 3610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0607288558579583e-05,
+      "loss": 0.7369,
+      "step": 3611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0602844975501887e-05,
+      "loss": 0.8282,
+      "step": 3612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0598401272952464e-05,
+      "loss": 0.8309,
+      "step": 3613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0593957451811963e-05,
+      "loss": 0.7731,
+      "step": 3614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058951351296106e-05,
+      "loss": 0.9262,
+      "step": 3615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0585069457280457e-05,
+      "loss": 0.7425,
+      "step": 3616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0580625285650874e-05,
+      "loss": 0.7708,
+      "step": 3617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0576180998953056e-05,
+      "loss": 0.8744,
+      "step": 3618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.057173659806777e-05,
+      "loss": 0.8221,
+      "step": 3619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056729208387581e-05,
+      "loss": 0.8856,
+      "step": 3620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0562847457257992e-05,
+      "loss": 0.8287,
+      "step": 3621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0558402719095144e-05,
+      "loss": 0.8176,
+      "step": 3622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0553957870268128e-05,
+      "loss": 0.7882,
+      "step": 3623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054951291165782e-05,
+      "loss": 0.7826,
+      "step": 3624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0545067844145125e-05,
+      "loss": 0.8294,
+      "step": 3625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0540622668610963e-05,
+      "loss": 0.8389,
+      "step": 3626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0536177385936281e-05,
+      "loss": 0.8376,
+      "step": 3627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0531731997002044e-05,
+      "loss": 0.7398,
+      "step": 3628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0527286502689238e-05,
+      "loss": 0.7881,
+      "step": 3629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0522840903878868e-05,
+      "loss": 0.347,
+      "step": 3630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0518395201451966e-05,
+      "loss": 0.8529,
+      "step": 3631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0513949396289577e-05,
+      "loss": 0.7782,
+      "step": 3632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0509503489272774e-05,
+      "loss": 0.7161,
+      "step": 3633
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0505057481282643e-05,
+      "loss": 0.8519,
+      "step": 3634
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0500611373200295e-05,
+      "loss": 0.8539,
+      "step": 3635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0496165165906858e-05,
+      "loss": 0.8462,
+      "step": 3636
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0491718860283479e-05,
+      "loss": 0.7887,
+      "step": 3637
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0487272457211331e-05,
+      "loss": 0.807,
+      "step": 3638
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0482825957571596e-05,
+      "loss": 0.8657,
+      "step": 3639
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0478379362245485e-05,
+      "loss": 0.8428,
+      "step": 3640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0473932672114222e-05,
+      "loss": 0.8924,
+      "step": 3641
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0469485888059052e-05,
+      "loss": 0.6897,
+      "step": 3642
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0465039010961237e-05,
+      "loss": 0.8553,
+      "step": 3643
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0460592041702059e-05,
+      "loss": 0.8285,
+      "step": 3644
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0456144981162817e-05,
+      "loss": 0.3375,
+      "step": 3645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0451697830224829e-05,
+      "loss": 0.8386,
+      "step": 3646
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0447250589769431e-05,
+      "loss": 0.7644,
+      "step": 3647
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0442803260677978e-05,
+      "loss": 0.8311,
+      "step": 3648
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0438355843831837e-05,
+      "loss": 0.7692,
+      "step": 3649
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0433908340112402e-05,
+      "loss": 0.7976,
+      "step": 3650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0429460750401075e-05,
+      "loss": 0.8291,
+      "step": 3651
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.042501307557928e-05,
+      "loss": 0.7353,
+      "step": 3652
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0420565316528454e-05,
+      "loss": 0.731,
+      "step": 3653
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0416117474130057e-05,
+      "loss": 0.8495,
+      "step": 3654
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.041166954926556e-05,
+      "loss": 0.8459,
+      "step": 3655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040722154281645e-05,
+      "loss": 0.8088,
+      "step": 3656
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0402773455664234e-05,
+      "loss": 0.7896,
+      "step": 3657
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0398325288690436e-05,
+      "loss": 0.888,
+      "step": 3658
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039387704277659e-05,
+      "loss": 0.8282,
+      "step": 3659
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0389428718804247e-05,
+      "loss": 0.8203,
+      "step": 3660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0384980317654979e-05,
+      "loss": 0.772,
+      "step": 3661
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0380531840210368e-05,
+      "loss": 0.7998,
+      "step": 3662
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0376083287352013e-05,
+      "loss": 0.8066,
+      "step": 3663
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0371634659961527e-05,
+      "loss": 0.7783,
+      "step": 3664
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0367185958920538e-05,
+      "loss": 0.8202,
+      "step": 3665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036273718511069e-05,
+      "loss": 0.8073,
+      "step": 3666
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0358288339413638e-05,
+      "loss": 0.3299,
+      "step": 3667
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0353839422711056e-05,
+      "loss": 0.3452,
+      "step": 3668
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0349390435884629e-05,
+      "loss": 0.719,
+      "step": 3669
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0344941379816059e-05,
+      "loss": 0.814,
+      "step": 3670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0340492255387054e-05,
+      "loss": 0.7981,
+      "step": 3671
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0336043063479346e-05,
+      "loss": 0.73,
+      "step": 3672
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0331593804974672e-05,
+      "loss": 0.7758,
+      "step": 3673
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0327144480754783e-05,
+      "loss": 0.7898,
+      "step": 3674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0322695091701454e-05,
+      "loss": 0.8047,
+      "step": 3675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0318245638696456e-05,
+      "loss": 0.7423,
+      "step": 3676
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0313796122621584e-05,
+      "loss": 0.7719,
+      "step": 3677
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0309346544358645e-05,
+      "loss": 0.7947,
+      "step": 3678
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.030489690478945e-05,
+      "loss": 0.8252,
+      "step": 3679
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.030044720479583e-05,
+      "loss": 0.3312,
+      "step": 3680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029599744525963e-05,
+      "loss": 0.7755,
+      "step": 3681
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0291547627062697e-05,
+      "loss": 0.858,
+      "step": 3682
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0287097751086901e-05,
+      "loss": 0.7986,
+      "step": 3683
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0282647818214114e-05,
+      "loss": 0.7426,
+      "step": 3684
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0278197829326223e-05,
+      "loss": 0.7171,
+      "step": 3685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0273747785305126e-05,
+      "loss": 0.7847,
+      "step": 3686
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0269297687032732e-05,
+      "loss": 0.79,
+      "step": 3687
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0264847535390963e-05,
+      "loss": 0.7214,
+      "step": 3688
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0260397331261749e-05,
+      "loss": 0.7633,
+      "step": 3689
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.025594707552703e-05,
+      "loss": 0.8163,
+      "step": 3690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0251496769068756e-05,
+      "loss": 0.7926,
+      "step": 3691
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024704641276889e-05,
+      "loss": 0.7674,
+      "step": 3692
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0242596007509403e-05,
+      "loss": 0.8134,
+      "step": 3693
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0238145554172274e-05,
+      "loss": 0.8666,
+      "step": 3694
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0233695053639493e-05,
+      "loss": 0.8547,
+      "step": 3695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022924450679306e-05,
+      "loss": 0.7675,
+      "step": 3696
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224793914514988e-05,
+      "loss": 0.7769,
+      "step": 3697
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0220343277687287e-05,
+      "loss": 0.7522,
+      "step": 3698
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021589259719199e-05,
+      "loss": 0.6827,
+      "step": 3699
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0211441873911126e-05,
+      "loss": 0.7389,
+      "step": 3700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0206991108726744e-05,
+      "loss": 0.7399,
+      "step": 3701
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0202540302520896e-05,
+      "loss": 0.7832,
+      "step": 3702
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019808945617564e-05,
+      "loss": 0.7725,
+      "step": 3703
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0193638570573044e-05,
+      "loss": 0.8632,
+      "step": 3704
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0189187646595184e-05,
+      "loss": 0.6759,
+      "step": 3705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0184736685124145e-05,
+      "loss": 0.8505,
+      "step": 3706
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0180285687042015e-05,
+      "loss": 0.7823,
+      "step": 3707
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0175834653230896e-05,
+      "loss": 0.81,
+      "step": 3708
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.017138358457289e-05,
+      "loss": 0.8059,
+      "step": 3709
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166932481950112e-05,
+      "loss": 0.8377,
+      "step": 3710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0162481346244678e-05,
+      "loss": 0.8374,
+      "step": 3711
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0158030178338713e-05,
+      "loss": 0.8151,
+      "step": 3712
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0153578979114351e-05,
+      "loss": 0.865,
+      "step": 3713
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0149127749453732e-05,
+      "loss": 0.7888,
+      "step": 3714
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0144676490238994e-05,
+      "loss": 0.7746,
+      "step": 3715
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0140225202352292e-05,
+      "loss": 0.3616,
+      "step": 3716
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0135773886675781e-05,
+      "loss": 0.8637,
+      "step": 3717
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.013132254409162e-05,
+      "loss": 0.7421,
+      "step": 3718
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0126871175481976e-05,
+      "loss": 0.755,
+      "step": 3719
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012241978172902e-05,
+      "loss": 0.8297,
+      "step": 3720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0117968363714932e-05,
+      "loss": 0.7866,
+      "step": 3721
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0113516922321889e-05,
+      "loss": 0.6811,
+      "step": 3722
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0109065458432078e-05,
+      "loss": 0.3811,
+      "step": 3723
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0104613972927695e-05,
+      "loss": 0.7801,
+      "step": 3724
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0100162466690925e-05,
+      "loss": 0.7961,
+      "step": 3725
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0095710940603975e-05,
+      "loss": 0.7345,
+      "step": 3726
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0091259395549047e-05,
+      "loss": 0.7615,
+      "step": 3727
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0086807832408338e-05,
+      "loss": 0.8098,
+      "step": 3728
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0082356252064072e-05,
+      "loss": 0.8105,
+      "step": 3729
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0077904655398455e-05,
+      "loss": 0.8854,
+      "step": 3730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0073453043293701e-05,
+      "loss": 0.8926,
+      "step": 3731
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0069001416632038e-05,
+      "loss": 0.8086,
+      "step": 3732
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0064549776295682e-05,
+      "loss": 0.8442,
+      "step": 3733
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0060098123166864e-05,
+      "loss": 0.7666,
+      "step": 3734
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0055646458127803e-05,
+      "loss": 0.8728,
+      "step": 3735
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0051194782060738e-05,
+      "loss": 0.7277,
+      "step": 3736
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.00467430958479e-05,
+      "loss": 0.824,
+      "step": 3737
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0042291400371523e-05,
+      "loss": 0.829,
+      "step": 3738
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003783969651384e-05,
+      "loss": 0.8012,
+      "step": 3739
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0033387985157093e-05,
+      "loss": 0.7423,
+      "step": 3740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0028936267183517e-05,
+      "loss": 0.8346,
+      "step": 3741
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002448454347536e-05,
+      "loss": 0.8066,
+      "step": 3742
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0020032814914855e-05,
+      "loss": 0.7585,
+      "step": 3743
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0015581082384253e-05,
+      "loss": 0.7974,
+      "step": 3744
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0011129346765794e-05,
+      "loss": 0.7933,
+      "step": 3745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0006677608941722e-05,
+      "loss": 0.7725,
+      "step": 3746
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0002225869794283e-05,
+      "loss": 0.3708,
+      "step": 3747
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997774130205722e-06,
+      "loss": 0.8099,
+      "step": 3748
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993322391058283e-06,
+      "loss": 0.3653,
+      "step": 3749
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.988870653234211e-06,
+      "loss": 0.783,
+      "step": 3750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.984418917615752e-06,
+      "loss": 0.7521,
+      "step": 3751
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.979967185085148e-06,
+      "loss": 0.8268,
+      "step": 3752
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.975515456524647e-06,
+      "loss": 0.7892,
+      "step": 3753
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.971063732816486e-06,
+      "loss": 0.8021,
+      "step": 3754
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.966612014842912e-06,
+      "loss": 0.7579,
+      "step": 3755
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.962160303486164e-06,
+      "loss": 0.8045,
+      "step": 3756
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.957708599628482e-06,
+      "loss": 0.7918,
+      "step": 3757
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.953256904152104e-06,
+      "loss": 0.7305,
+      "step": 3758
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.948805217939264e-06,
+      "loss": 0.8643,
+      "step": 3759
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9443535418722e-06,
+      "loss": 0.7385,
+      "step": 3760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93990187683314e-06,
+      "loss": 0.8586,
+      "step": 3761
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93545022370432e-06,
+      "loss": 0.7659,
+      "step": 3762
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.930998583367964e-06,
+      "loss": 0.8871,
+      "step": 3763
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9265469567063e-06,
+      "loss": 0.8485,
+      "step": 3764
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.922095344601549e-06,
+      "loss": 0.7888,
+      "step": 3765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.917643747935931e-06,
+      "loss": 0.8241,
+      "step": 3766
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.913192167591664e-06,
+      "loss": 0.7842,
+      "step": 3767
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.908740604450958e-06,
+      "loss": 0.7744,
+      "step": 3768
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.904289059396026e-06,
+      "loss": 0.8257,
+      "step": 3769
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.899837533309077e-06,
+      "loss": 0.6951,
+      "step": 3770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895386027072308e-06,
+      "loss": 0.8161,
+      "step": 3771
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.890934541567923e-06,
+      "loss": 0.768,
+      "step": 3772
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.886483077678115e-06,
+      "loss": 0.7702,
+      "step": 3773
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.882031636285073e-06,
+      "loss": 0.7199,
+      "step": 3774
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.877580218270982e-06,
+      "loss": 0.8274,
+      "step": 3775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.873128824518026e-06,
+      "loss": 0.7689,
+      "step": 3776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.868677455908382e-06,
+      "loss": 0.7909,
+      "step": 3777
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.864226113324222e-06,
+      "loss": 0.8727,
+      "step": 3778
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85977479764771e-06,
+      "loss": 0.774,
+      "step": 3779
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.855323509761007e-06,
+      "loss": 0.7285,
+      "step": 3780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850872250546272e-06,
+      "loss": 0.8136,
+      "step": 3781
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84642102088565e-06,
+      "loss": 0.8054,
+      "step": 3782
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.841969821661288e-06,
+      "loss": 0.7537,
+      "step": 3783
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.837518653755324e-06,
+      "loss": 0.8269,
+      "step": 3784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83306751804989e-06,
+      "loss": 0.8296,
+      "step": 3785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82861641542711e-06,
+      "loss": 0.8108,
+      "step": 3786
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.824165346769106e-06,
+      "loss": 0.8096,
+      "step": 3787
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.819714312957985e-06,
+      "loss": 0.8078,
+      "step": 3788
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.815263314875856e-06,
+      "loss": 0.8447,
+      "step": 3789
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81081235340482e-06,
+      "loss": 0.8104,
+      "step": 3790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.806361429426961e-06,
+      "loss": 0.8121,
+      "step": 3791
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.801910543824365e-06,
+      "loss": 0.8159,
+      "step": 3792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79745969747911e-06,
+      "loss": 0.782,
+      "step": 3793
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.793008891273257e-06,
+      "loss": 0.822,
+      "step": 3794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.788558126088877e-06,
+      "loss": 0.7873,
+      "step": 3795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.784107402808015e-06,
+      "loss": 0.8039,
+      "step": 3796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.779656722312716e-06,
+      "loss": 0.8359,
+      "step": 3797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775206085485017e-06,
+      "loss": 0.7634,
+      "step": 3798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.770755493206943e-06,
+      "loss": 0.7242,
+      "step": 3799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76630494636051e-06,
+      "loss": 0.7799,
+      "step": 3800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76185444582773e-06,
+      "loss": 0.8682,
+      "step": 3801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.757403992490599e-06,
+      "loss": 0.7922,
+      "step": 3802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.752953587231113e-06,
+      "loss": 0.7349,
+      "step": 3803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.748503230931246e-06,
+      "loss": 0.8102,
+      "step": 3804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.744052924472973e-06,
+      "loss": 0.7892,
+      "step": 3805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.739602668738254e-06,
+      "loss": 0.8096,
+      "step": 3806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73515246460904e-06,
+      "loss": 0.8333,
+      "step": 3807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.730702312967272e-06,
+      "loss": 0.7735,
+      "step": 3808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.726252214694877e-06,
+      "loss": 0.745,
+      "step": 3809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72180217067378e-06,
+      "loss": 0.7934,
+      "step": 3810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.717352181785891e-06,
+      "loss": 0.7911,
+      "step": 3811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.712902248913102e-06,
+      "loss": 0.8314,
+      "step": 3812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.708452372937305e-06,
+      "loss": 0.739,
+      "step": 3813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.704002554740374e-06,
+      "loss": 0.8117,
+      "step": 3814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.699552795204172e-06,
+      "loss": 0.6976,
+      "step": 3815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.695103095210553e-06,
+      "loss": 0.797,
+      "step": 3816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.690653455641359e-06,
+      "loss": 0.706,
+      "step": 3817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.686203877378418e-06,
+      "loss": 0.7764,
+      "step": 3818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.681754361303545e-06,
+      "loss": 0.785,
+      "step": 3819
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.67730490829855e-06,
+      "loss": 0.7535,
+      "step": 3820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.672855519245217e-06,
+      "loss": 0.8138,
+      "step": 3821
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.668406195025332e-06,
+      "loss": 0.7892,
+      "step": 3822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663956936520656e-06,
+      "loss": 0.8141,
+      "step": 3823
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.659507744612946e-06,
+      "loss": 0.8371,
+      "step": 3824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.655058620183943e-06,
+      "loss": 0.7236,
+      "step": 3825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.650609564115371e-06,
+      "loss": 0.8479,
+      "step": 3826
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.646160577288944e-06,
+      "loss": 0.8516,
+      "step": 3827
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.641711660586364e-06,
+      "loss": 0.8572,
+      "step": 3828
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.637262814889313e-06,
+      "loss": 0.8251,
+      "step": 3829
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.632814041079465e-06,
+      "loss": 0.8035,
+      "step": 3830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.628365340038475e-06,
+      "loss": 0.7596,
+      "step": 3831
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.623916712647992e-06,
+      "loss": 0.8232,
+      "step": 3832
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.619468159789637e-06,
+      "loss": 0.8706,
+      "step": 3833
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.615019682345026e-06,
+      "loss": 0.8277,
+      "step": 3834
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.610571281195756e-06,
+      "loss": 0.8431,
+      "step": 3835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.606122957223416e-06,
+      "loss": 0.3971,
+      "step": 3836
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.601674711309569e-06,
+      "loss": 0.8017,
+      "step": 3837
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.597226544335769e-06,
+      "loss": 0.7368,
+      "step": 3838
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.592778457183555e-06,
+      "loss": 0.8646,
+      "step": 3839
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588330450734446e-06,
+      "loss": 0.6689,
+      "step": 3840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.583882525869947e-06,
+      "loss": 0.8456,
+      "step": 3841
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.579434683471547e-06,
+      "loss": 0.7541,
+      "step": 3842
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.574986924420722e-06,
+      "loss": 0.8097,
+      "step": 3843
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.570539249598926e-06,
+      "loss": 0.8372,
+      "step": 3844
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.5660916598876e-06,
+      "loss": 0.7784,
+      "step": 3845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.561644156168165e-06,
+      "loss": 0.8048,
+      "step": 3846
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557196739322025e-06,
+      "loss": 0.759,
+      "step": 3847
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.552749410230572e-06,
+      "loss": 0.7845,
+      "step": 3848
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.548302169775173e-06,
+      "loss": 0.8181,
+      "step": 3849
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.543855018837187e-06,
+      "loss": 0.7811,
+      "step": 3850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.539407958297943e-06,
+      "loss": 0.7902,
+      "step": 3851
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.534960989038766e-06,
+      "loss": 0.7811,
+      "step": 3852
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53051411194095e-06,
+      "loss": 0.7841,
+      "step": 3853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.526067327885781e-06,
+      "loss": 0.8528,
+      "step": 3854
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.521620637754516e-06,
+      "loss": 0.7965,
+      "step": 3855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.517174042428407e-06,
+      "loss": 0.7535,
+      "step": 3856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51272754278867e-06,
+      "loss": 0.3533,
+      "step": 3857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.508281139716522e-06,
+      "loss": 0.7434,
+      "step": 3858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.503834834093145e-06,
+      "loss": 0.7938,
+      "step": 3859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.499388626799708e-06,
+      "loss": 0.756,
+      "step": 3860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.494942518717359e-06,
+      "loss": 0.3553,
+      "step": 3861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.490496510727229e-06,
+      "loss": 0.7948,
+      "step": 3862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.486050603710425e-06,
+      "loss": 0.8026,
+      "step": 3863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.481604798548036e-06,
+      "loss": 0.8419,
+      "step": 3864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.477159096121132e-06,
+      "loss": 0.8448,
+      "step": 3865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.472713497310764e-06,
+      "loss": 0.7275,
+      "step": 3866
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.468268002997958e-06,
+      "loss": 0.7212,
+      "step": 3867
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.463822614063719e-06,
+      "loss": 0.7743,
+      "step": 3868
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.459377331389037e-06,
+      "loss": 0.7846,
+      "step": 3869
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454932155854876e-06,
+      "loss": 0.7832,
+      "step": 3870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45048708834218e-06,
+      "loss": 0.8228,
+      "step": 3871
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.446042129731873e-06,
+      "loss": 0.8187,
+      "step": 3872
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.441597280904861e-06,
+      "loss": 0.6701,
+      "step": 3873
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.437152542742014e-06,
+      "loss": 0.8324,
+      "step": 3874
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.432707916124191e-06,
+      "loss": 0.8162,
+      "step": 3875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.428263401932232e-06,
+      "loss": 0.8157,
+      "step": 3876
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42381900104695e-06,
+      "loss": 0.879,
+      "step": 3877
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.419374714349131e-06,
+      "loss": 0.8369,
+      "step": 3878
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41493054271955e-06,
+      "loss": 0.7483,
+      "step": 3879
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.410486487038945e-06,
+      "loss": 0.7747,
+      "step": 3880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.406042548188044e-06,
+      "loss": 0.762,
+      "step": 3881
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.401598727047541e-06,
+      "loss": 0.7733,
+      "step": 3882
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397155024498115e-06,
+      "loss": 0.782,
+      "step": 3883
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.39271144142042e-06,
+      "loss": 0.8242,
+      "step": 3884
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388267978695082e-06,
+      "loss": 0.7123,
+      "step": 3885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.383824637202707e-06,
+      "loss": 0.7919,
+      "step": 3886
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.379381417823874e-06,
+      "loss": 0.8711,
+      "step": 3887
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.374938321439142e-06,
+      "loss": 0.3538,
+      "step": 3888
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.370495348929042e-06,
+      "loss": 0.7702,
+      "step": 3889
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.36605250117408e-06,
+      "loss": 0.8194,
+      "step": 3890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.361609779054741e-06,
+      "loss": 0.8213,
+      "step": 3891
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.357167183451484e-06,
+      "loss": 0.7809,
+      "step": 3892
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.35272471524474e-06,
+      "loss": 0.8388,
+      "step": 3893
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.348282375314916e-06,
+      "loss": 0.7325,
+      "step": 3894
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.343840164542397e-06,
+      "loss": 0.7809,
+      "step": 3895
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.339398083807537e-06,
+      "loss": 0.7351,
+      "step": 3896
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.334956133990667e-06,
+      "loss": 0.8478,
+      "step": 3897
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.330514315972095e-06,
+      "loss": 0.3559,
+      "step": 3898
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326072630632096e-06,
+      "loss": 0.8239,
+      "step": 3899
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321631078850928e-06,
+      "loss": 0.9047,
+      "step": 3900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.317189661508811e-06,
+      "loss": 0.7559,
+      "step": 3901
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.312748379485947e-06,
+      "loss": 0.8655,
+      "step": 3902
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.308307233662508e-06,
+      "loss": 0.7718,
+      "step": 3903
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.303866224918641e-06,
+      "loss": 0.3316,
+      "step": 3904
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.299425354134464e-06,
+      "loss": 0.7552,
+      "step": 3905
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.294984622190067e-06,
+      "loss": 0.3375,
+      "step": 3906
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.290544029965516e-06,
+      "loss": 0.783,
+      "step": 3907
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.286103578340844e-06,
+      "loss": 0.7074,
+      "step": 3908
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28166326819606e-06,
+      "loss": 0.8046,
+      "step": 3909
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.277223100411146e-06,
+      "loss": 0.79,
+      "step": 3910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.272783075866051e-06,
+      "loss": 0.7816,
+      "step": 3911
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2683431954407e-06,
+      "loss": 0.7576,
+      "step": 3912
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.263903460014985e-06,
+      "loss": 0.8132,
+      "step": 3913
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.259463870468777e-06,
+      "loss": 0.7444,
+      "step": 3914
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255024427681915e-06,
+      "loss": 0.8541,
+      "step": 3915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.250585132534198e-06,
+      "loss": 0.8653,
+      "step": 3916
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24614598590541e-06,
+      "loss": 0.795,
+      "step": 3917
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2417069886753e-06,
+      "loss": 0.9218,
+      "step": 3918
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.237268141723589e-06,
+      "loss": 0.7638,
+      "step": 3919
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.232829445929965e-06,
+      "loss": 0.806,
+      "step": 3920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22839090217409e-06,
+      "loss": 0.8889,
+      "step": 3921
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.223952511335593e-06,
+      "loss": 0.7296,
+      "step": 3922
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.219514274294072e-06,
+      "loss": 0.803,
+      "step": 3923
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2150761919291e-06,
+      "loss": 0.3536,
+      "step": 3924
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.210638265120212e-06,
+      "loss": 0.7517,
+      "step": 3925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.206200494746918e-06,
+      "loss": 0.7488,
+      "step": 3926
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.201762881688692e-06,
+      "loss": 0.7064,
+      "step": 3927
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.197325426824985e-06,
+      "loss": 0.8084,
+      "step": 3928
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.192888131035204e-06,
+      "loss": 0.7228,
+      "step": 3929
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.188450995198734e-06,
+      "loss": 0.8636,
+      "step": 3930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18401402019493e-06,
+      "loss": 0.7789,
+      "step": 3931
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179577206903108e-06,
+      "loss": 0.3802,
+      "step": 3932
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.175140556202554e-06,
+      "loss": 0.7997,
+      "step": 3933
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.170704068972525e-06,
+      "loss": 0.8573,
+      "step": 3934
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.166267746092241e-06,
+      "loss": 0.7452,
+      "step": 3935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.161831588440894e-06,
+      "loss": 0.7044,
+      "step": 3936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.157395596897641e-06,
+      "loss": 0.8182,
+      "step": 3937
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.1529597723416e-06,
+      "loss": 0.7945,
+      "step": 3938
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.148524115651874e-06,
+      "loss": 0.7539,
+      "step": 3939
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14408862770751e-06,
+      "loss": 0.8677,
+      "step": 3940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.139653309387539e-06,
+      "loss": 0.7961,
+      "step": 3941
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.135218161570948e-06,
+      "loss": 0.7124,
+      "step": 3942
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.130783185136695e-06,
+      "loss": 0.7577,
+      "step": 3943
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.126348380963704e-06,
+      "loss": 0.8975,
+      "step": 3944
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121913749930861e-06,
+      "loss": 0.8049,
+      "step": 3945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.117479292917023e-06,
+      "loss": 0.836,
+      "step": 3946
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.113045010801012e-06,
+      "loss": 0.7527,
+      "step": 3947
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10861090446161e-06,
+      "loss": 0.779,
+      "step": 3948
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.104176974777568e-06,
+      "loss": 0.7588,
+      "step": 3949
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.099743222627602e-06,
+      "loss": 0.7254,
+      "step": 3950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.095309648890393e-06,
+      "loss": 0.7821,
+      "step": 3951
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.090876254444583e-06,
+      "loss": 0.7745,
+      "step": 3952
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.086443040168785e-06,
+      "loss": 0.6577,
+      "step": 3953
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.082010006941573e-06,
+      "loss": 0.7465,
+      "step": 3954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.077577155641482e-06,
+      "loss": 0.7563,
+      "step": 3955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.073144487147017e-06,
+      "loss": 0.8055,
+      "step": 3956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.068712002336643e-06,
+      "loss": 0.8112,
+      "step": 3957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.064279702088786e-06,
+      "loss": 0.7988,
+      "step": 3958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.059847587281842e-06,
+      "loss": 0.3322,
+      "step": 3959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.055415658794163e-06,
+      "loss": 0.747,
+      "step": 3960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.050983917504073e-06,
+      "loss": 0.8356,
+      "step": 3961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04655236428985e-06,
+      "loss": 0.7557,
+      "step": 3962
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04212100002974e-06,
+      "loss": 0.7814,
+      "step": 3963
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.037689825601948e-06,
+      "loss": 0.7361,
+      "step": 3964
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.033258841884648e-06,
+      "loss": 0.7875,
+      "step": 3965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02882804975597e-06,
+      "loss": 0.8021,
+      "step": 3966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.024397450094004e-06,
+      "loss": 0.7533,
+      "step": 3967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01996704377681e-06,
+      "loss": 0.781,
+      "step": 3968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.015536831682405e-06,
+      "loss": 0.7434,
+      "step": 3969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.011106814688766e-06,
+      "loss": 0.7774,
+      "step": 3970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.006676993673831e-06,
+      "loss": 0.7348,
+      "step": 3971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.002247369515506e-06,
+      "loss": 0.7939,
+      "step": 3972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.997817943091654e-06,
+      "loss": 0.8464,
+      "step": 3973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.993388715280093e-06,
+      "loss": 0.3646,
+      "step": 3974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988959686958609e-06,
+      "loss": 0.8536,
+      "step": 3975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984530859004947e-06,
+      "loss": 0.7862,
+      "step": 3976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98010223229681e-06,
+      "loss": 0.8639,
+      "step": 3977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97567380771186e-06,
+      "loss": 0.7888,
+      "step": 3978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97124558612773e-06,
+      "loss": 0.8121,
+      "step": 3979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.966817568421997e-06,
+      "loss": 0.7859,
+      "step": 3980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.962389755472208e-06,
+      "loss": 0.7954,
+      "step": 3981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.957962148155865e-06,
+      "loss": 0.8306,
+      "step": 3982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.953534747350432e-06,
+      "loss": 0.7836,
+      "step": 3983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.949107553933328e-06,
+      "loss": 0.7296,
+      "step": 3984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.944680568781934e-06,
+      "loss": 0.7794,
+      "step": 3985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.940253792773592e-06,
+      "loss": 0.8446,
+      "step": 3986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9358272267856e-06,
+      "loss": 0.7424,
+      "step": 3987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93140087169521e-06,
+      "loss": 0.858,
+      "step": 3988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92697472837964e-06,
+      "loss": 0.7285,
+      "step": 3989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.922548797716063e-06,
+      "loss": 0.3579,
+      "step": 3990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.918123080581605e-06,
+      "loss": 0.7981,
+      "step": 3991
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913697577853357e-06,
+      "loss": 0.8125,
+      "step": 3992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.909272290408361e-06,
+      "loss": 0.8349,
+      "step": 3993
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.904847219123627e-06,
+      "loss": 0.739,
+      "step": 3994
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90042236487611e-06,
+      "loss": 0.8024,
+      "step": 3995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.895997728542727e-06,
+      "loss": 0.7845,
+      "step": 3996
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.891573311000355e-06,
+      "loss": 0.6854,
+      "step": 3997
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88714911312582e-06,
+      "loss": 0.8011,
+      "step": 3998
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.882725135795913e-06,
+      "loss": 0.7562,
+      "step": 3999
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.878301379887374e-06,
+      "loss": 0.767,
+      "step": 4000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.873877846276903e-06,
+      "loss": 0.7387,
+      "step": 4001
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.869454535841155e-06,
+      "loss": 0.8,
+      "step": 4002
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.865031449456743e-06,
+      "loss": 0.8045,
+      "step": 4003
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.860608588000228e-06,
+      "loss": 0.8114,
+      "step": 4004
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.856185952348139e-06,
+      "loss": 0.8534,
+      "step": 4005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.851763543376953e-06,
+      "loss": 0.7903,
+      "step": 4006
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.847341361963097e-06,
+      "loss": 0.7876,
+      "step": 4007
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.842919408982962e-06,
+      "loss": 0.7685,
+      "step": 4008
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.838497685312889e-06,
+      "loss": 0.7902,
+      "step": 4009
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.834076191829174e-06,
+      "loss": 0.7492,
+      "step": 4010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.829654929408069e-06,
+      "loss": 0.8225,
+      "step": 4011
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.825233898925779e-06,
+      "loss": 0.7896,
+      "step": 4012
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.820813101258465e-06,
+      "loss": 0.7644,
+      "step": 4013
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.816392537282235e-06,
+      "loss": 0.7548,
+      "step": 4014
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81197220787316e-06,
+      "loss": 0.3548,
+      "step": 4015
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80755211390726e-06,
+      "loss": 0.7449,
+      "step": 4016
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.803132256260507e-06,
+      "loss": 0.8042,
+      "step": 4017
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.798712635808828e-06,
+      "loss": 0.812,
+      "step": 4018
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.794293253428101e-06,
+      "loss": 0.7186,
+      "step": 4019
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.789874109994163e-06,
+      "loss": 0.8458,
+      "step": 4020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.785455206382796e-06,
+      "loss": 0.8028,
+      "step": 4021
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.781036543469739e-06,
+      "loss": 0.8357,
+      "step": 4022
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.776618122130681e-06,
+      "loss": 0.7628,
+      "step": 4023
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.772199943241265e-06,
+      "loss": 0.7384,
+      "step": 4024
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.767782007677085e-06,
+      "loss": 0.8458,
+      "step": 4025
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.763364316313683e-06,
+      "loss": 0.7356,
+      "step": 4026
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.758946870026562e-06,
+      "loss": 0.8211,
+      "step": 4027
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75452966969117e-06,
+      "loss": 0.8581,
+      "step": 4028
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.750112716182905e-06,
+      "loss": 0.7836,
+      "step": 4029
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.745696010377118e-06,
+      "loss": 0.8139,
+      "step": 4030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.741279553149113e-06,
+      "loss": 0.7341,
+      "step": 4031
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.736863345374142e-06,
+      "loss": 0.7395,
+      "step": 4032
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.732447387927407e-06,
+      "loss": 0.8157,
+      "step": 4033
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.728031681684064e-06,
+      "loss": 0.7675,
+      "step": 4034
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.723616227519217e-06,
+      "loss": 0.8271,
+      "step": 4035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.719201026307919e-06,
+      "loss": 0.7933,
+      "step": 4036
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.714786078925173e-06,
+      "loss": 0.782,
+      "step": 4037
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.710371386245934e-06,
+      "loss": 0.8062,
+      "step": 4038
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.705956949145107e-06,
+      "loss": 0.7668,
+      "step": 4039
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.701542768497544e-06,
+      "loss": 0.8634,
+      "step": 4040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.697128845178042e-06,
+      "loss": 0.7244,
+      "step": 4041
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.692715180061356e-06,
+      "loss": 0.7905,
+      "step": 4042
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.688301774022182e-06,
+      "loss": 0.8224,
+      "step": 4043
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.683888627935172e-06,
+      "loss": 0.8428,
+      "step": 4044
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67947574267492e-06,
+      "loss": 0.8136,
+      "step": 4045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675063119115971e-06,
+      "loss": 0.7916,
+      "step": 4046
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67065075813282e-06,
+      "loss": 0.7596,
+      "step": 4047
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66623866059991e-06,
+      "loss": 0.7605,
+      "step": 4048
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.661826827391626e-06,
+      "loss": 0.7837,
+      "step": 4049
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.657415259382304e-06,
+      "loss": 0.7344,
+      "step": 4050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65300395744623e-06,
+      "loss": 0.7979,
+      "step": 4051
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.648592922457633e-06,
+      "loss": 0.7304,
+      "step": 4052
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.644182155290694e-06,
+      "loss": 0.8076,
+      "step": 4053
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.639771656819538e-06,
+      "loss": 0.8486,
+      "step": 4054
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.635361427918235e-06,
+      "loss": 0.3582,
+      "step": 4055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.630951469460805e-06,
+      "loss": 0.7809,
+      "step": 4056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.626541782321212e-06,
+      "loss": 0.835,
+      "step": 4057
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.622132367373369e-06,
+      "loss": 0.835,
+      "step": 4058
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61772322549113e-06,
+      "loss": 0.8224,
+      "step": 4059
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.613314357548298e-06,
+      "loss": 0.7822,
+      "step": 4060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.608905764418627e-06,
+      "loss": 0.7778,
+      "step": 4061
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.604497446975809e-06,
+      "loss": 0.81,
+      "step": 4062
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.600089406093481e-06,
+      "loss": 0.8063,
+      "step": 4063
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.59568164264523e-06,
+      "loss": 0.8438,
+      "step": 4064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.591274157504585e-06,
+      "loss": 0.3347,
+      "step": 4065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.586866951545018e-06,
+      "loss": 0.7248,
+      "step": 4066
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.582460025639955e-06,
+      "loss": 0.7377,
+      "step": 4067
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.578053380662757e-06,
+      "loss": 0.7979,
+      "step": 4068
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.573647017486733e-06,
+      "loss": 0.3249,
+      "step": 4069
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.569240936985133e-06,
+      "loss": 0.3308,
+      "step": 4070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.564835140031155e-06,
+      "loss": 0.6977,
+      "step": 4071
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.560429627497938e-06,
+      "loss": 0.7699,
+      "step": 4072
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.556024400258568e-06,
+      "loss": 0.8074,
+      "step": 4073
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.551619459186068e-06,
+      "loss": 0.8244,
+      "step": 4074
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.547214805153414e-06,
+      "loss": 0.7974,
+      "step": 4075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.542810439033517e-06,
+      "loss": 0.8385,
+      "step": 4076
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.538406361699236e-06,
+      "loss": 0.7894,
+      "step": 4077
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.534002574023366e-06,
+      "loss": 0.7877,
+      "step": 4078
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.529599076878651e-06,
+      "loss": 0.8344,
+      "step": 4079
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.525195871137778e-06,
+      "loss": 0.7745,
+      "step": 4080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.520792957673368e-06,
+      "loss": 0.7739,
+      "step": 4081
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.516390337358e-06,
+      "loss": 0.8564,
+      "step": 4082
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.511988011064171e-06,
+      "loss": 0.7743,
+      "step": 4083
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.507585979664343e-06,
+      "loss": 0.818,
+      "step": 4084
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.503184244030905e-06,
+      "loss": 0.8066,
+      "step": 4085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.498782805036195e-06,
+      "loss": 0.8561,
+      "step": 4086
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49438166355249e-06,
+      "loss": 0.7782,
+      "step": 4087
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.489980820452007e-06,
+      "loss": 0.7189,
+      "step": 4088
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.485580276606902e-06,
+      "loss": 0.7712,
+      "step": 4089
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.481180032889278e-06,
+      "loss": 0.7267,
+      "step": 4090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.476780090171172e-06,
+      "loss": 0.76,
+      "step": 4091
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472380449324564e-06,
+      "loss": 0.7503,
+      "step": 4092
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.467981111221376e-06,
+      "loss": 0.8637,
+      "step": 4093
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.463582076733467e-06,
+      "loss": 0.734,
+      "step": 4094
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.459183346732638e-06,
+      "loss": 0.7939,
+      "step": 4095
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.454784922090626e-06,
+      "loss": 0.7437,
+      "step": 4096
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.450386803679114e-06,
+      "loss": 0.8143,
+      "step": 4097
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.445988992369719e-06,
+      "loss": 0.7913,
+      "step": 4098
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.441591489033995e-06,
+      "loss": 0.7382,
+      "step": 4099
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.437194294543442e-06,
+      "loss": 0.7584,
+      "step": 4100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.432797409769497e-06,
+      "loss": 0.7802,
+      "step": 4101
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42840083558353e-06,
+      "loss": 0.8324,
+      "step": 4102
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.424004572856856e-06,
+      "loss": 0.7739,
+      "step": 4103
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.419608622460727e-06,
+      "loss": 0.7413,
+      "step": 4104
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.415212985266329e-06,
+      "loss": 0.719,
+      "step": 4105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.410817662144788e-06,
+      "loss": 0.7458,
+      "step": 4106
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40642265396717e-06,
+      "loss": 0.7782,
+      "step": 4107
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40202796160448e-06,
+      "loss": 0.868,
+      "step": 4108
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.397633585927653e-06,
+      "loss": 0.8527,
+      "step": 4109
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.393239527807567e-06,
+      "loss": 0.7944,
+      "step": 4110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.388845788115036e-06,
+      "loss": 0.7575,
+      "step": 4111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.384452367720814e-06,
+      "loss": 0.7613,
+      "step": 4112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.380059267495582e-06,
+      "loss": 0.791,
+      "step": 4113
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.375666488309966e-06,
+      "loss": 0.8318,
+      "step": 4114
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.371274031034531e-06,
+      "loss": 0.7683,
+      "step": 4115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.366881896539769e-06,
+      "loss": 0.828,
+      "step": 4116
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.362490085696115e-06,
+      "loss": 0.7253,
+      "step": 4117
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358098599373934e-06,
+      "loss": 0.7797,
+      "step": 4118
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.353707438443534e-06,
+      "loss": 0.7343,
+      "step": 4119
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.349316603775153e-06,
+      "loss": 0.7903,
+      "step": 4120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.344926096238962e-06,
+      "loss": 0.6589,
+      "step": 4121
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.340535916705078e-06,
+      "loss": 0.7209,
+      "step": 4122
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.336146066043544e-06,
+      "loss": 0.8438,
+      "step": 4123
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.331756545124343e-06,
+      "loss": 0.7952,
+      "step": 4124
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.327367354817381e-06,
+      "loss": 0.8083,
+      "step": 4125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32297849599251e-06,
+      "loss": 0.8033,
+      "step": 4126
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.318589969519517e-06,
+      "loss": 0.7697,
+      "step": 4127
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.314201776268119e-06,
+      "loss": 0.8421,
+      "step": 4128
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.309813917107965e-06,
+      "loss": 0.834,
+      "step": 4129
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.305426392908641e-06,
+      "loss": 0.6917,
+      "step": 4130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.301039204539667e-06,
+      "loss": 0.8041,
+      "step": 4131
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.296652352870495e-06,
+      "loss": 0.3619,
+      "step": 4132
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29226583877051e-06,
+      "loss": 0.8337,
+      "step": 4133
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.287879663109031e-06,
+      "loss": 0.7532,
+      "step": 4134
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28349382675531e-06,
+      "loss": 0.7775,
+      "step": 4135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.279108330578532e-06,
+      "loss": 0.8029,
+      "step": 4136
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.274723175447812e-06,
+      "loss": 0.7503,
+      "step": 4137
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.270338362232202e-06,
+      "loss": 0.8122,
+      "step": 4138
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.265953891800681e-06,
+      "loss": 0.7696,
+      "step": 4139
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.261569765022163e-06,
+      "loss": 0.7672,
+      "step": 4140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.257185982765495e-06,
+      "loss": 0.8236,
+      "step": 4141
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.252802545899455e-06,
+      "loss": 0.8367,
+      "step": 4142
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.248419455292751e-06,
+      "loss": 0.8138,
+      "step": 4143
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.244036711814023e-06,
+      "loss": 0.8113,
+      "step": 4144
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.239654316331843e-06,
+      "loss": 0.7331,
+      "step": 4145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.235272269714712e-06,
+      "loss": 0.7638,
+      "step": 4146
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23089057283106e-06,
+      "loss": 0.6995,
+      "step": 4147
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.226509226549261e-06,
+      "loss": 0.7978,
+      "step": 4148
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.222128231737603e-06,
+      "loss": 0.8434,
+      "step": 4149
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21774758926431e-06,
+      "loss": 0.7324,
+      "step": 4150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.213367299997541e-06,
+      "loss": 0.8634,
+      "step": 4151
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.20898736480538e-06,
+      "loss": 0.8698,
+      "step": 4152
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.204607784555838e-06,
+      "loss": 0.7629,
+      "step": 4153
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.200228560116864e-06,
+      "loss": 0.7495,
+      "step": 4154
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.195849692356328e-06,
+      "loss": 0.8097,
+      "step": 4155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.191471182142038e-06,
+      "loss": 0.7985,
+      "step": 4156
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.187093030341726e-06,
+      "loss": 0.7761,
+      "step": 4157
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.18271523782305e-06,
+      "loss": 0.8334,
+      "step": 4158
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.178337805453603e-06,
+      "loss": 0.7633,
+      "step": 4159
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.173960734100903e-06,
+      "loss": 0.7524,
+      "step": 4160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.169584024632396e-06,
+      "loss": 0.3511,
+      "step": 4161
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.165207677915459e-06,
+      "loss": 0.7848,
+      "step": 4162
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.160831694817398e-06,
+      "loss": 0.3599,
+      "step": 4163
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.156456076205442e-06,
+      "loss": 0.7902,
+      "step": 4164
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.15208082294675e-06,
+      "loss": 0.7386,
+      "step": 4165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.147705935908413e-06,
+      "loss": 0.7569,
+      "step": 4166
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14333141595744e-06,
+      "loss": 0.8226,
+      "step": 4167
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.138957263960776e-06,
+      "loss": 0.7894,
+      "step": 4168
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13458348078529e-06,
+      "loss": 0.8269,
+      "step": 4169
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.130210067297775e-06,
+      "loss": 0.7713,
+      "step": 4170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.125837024364956e-06,
+      "loss": 0.8257,
+      "step": 4171
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12146435285348e-06,
+      "loss": 0.6819,
+      "step": 4172
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.117092053629924e-06,
+      "loss": 0.728,
+      "step": 4173
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11272012756079e-06,
+      "loss": 0.7645,
+      "step": 4174
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.108348575512504e-06,
+      "loss": 0.8914,
+      "step": 4175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10397739835142e-06,
+      "loss": 0.7553,
+      "step": 4176
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09960659694382e-06,
+      "loss": 0.8199,
+      "step": 4177
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.095236172155904e-06,
+      "loss": 0.3369,
+      "step": 4178
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.090866124853806e-06,
+      "loss": 0.775,
+      "step": 4179
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08649645590358e-06,
+      "loss": 0.8473,
+      "step": 4180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.082127166171205e-06,
+      "loss": 0.8079,
+      "step": 4181
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.077758256522589e-06,
+      "loss": 0.8424,
+      "step": 4182
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07338972782356e-06,
+      "loss": 0.7602,
+      "step": 4183
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.069021580939874e-06,
+      "loss": 0.7832,
+      "step": 4184
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.064653816737207e-06,
+      "loss": 0.7263,
+      "step": 4185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.060286436081164e-06,
+      "loss": 0.8026,
+      "step": 4186
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05591943983727e-06,
+      "loss": 0.8439,
+      "step": 4187
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051552828870977e-06,
+      "loss": 0.8337,
+      "step": 4188
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.047186604047658e-06,
+      "loss": 0.779,
+      "step": 4189
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042820766232616e-06,
+      "loss": 0.7252,
+      "step": 4190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.038455316291065e-06,
+      "loss": 0.9026,
+      "step": 4191
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.034090255088154e-06,
+      "loss": 0.8195,
+      "step": 4192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.029725583488948e-06,
+      "loss": 0.7288,
+      "step": 4193
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.025361302358437e-06,
+      "loss": 0.7984,
+      "step": 4194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.020997412561532e-06,
+      "loss": 0.7753,
+      "step": 4195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.016633914963073e-06,
+      "loss": 0.7595,
+      "step": 4196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.012270810427814e-06,
+      "loss": 0.8201,
+      "step": 4197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.007908099820436e-06,
+      "loss": 0.8243,
+      "step": 4198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.003545784005538e-06,
+      "loss": 0.7732,
+      "step": 4199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999183863847645e-06,
+      "loss": 0.8069,
+      "step": 4200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9948223402112e-06,
+      "loss": 0.7301,
+      "step": 4201
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.990461213960569e-06,
+      "loss": 0.8052,
+      "step": 4202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.986100485960042e-06,
+      "loss": 0.7798,
+      "step": 4203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.981740157073827e-06,
+      "loss": 0.7963,
+      "step": 4204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.977380228166052e-06,
+      "loss": 0.7609,
+      "step": 4205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.973020700100767e-06,
+      "loss": 0.7948,
+      "step": 4206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.968661573741945e-06,
+      "loss": 0.8311,
+      "step": 4207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.964302849953473e-06,
+      "loss": 0.8884,
+      "step": 4208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.959944529599165e-06,
+      "loss": 0.7838,
+      "step": 4209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95558661354275e-06,
+      "loss": 0.7826,
+      "step": 4210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.951229102647881e-06,
+      "loss": 0.3316,
+      "step": 4211
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.946871997778128e-06,
+      "loss": 0.776,
+      "step": 4212
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.942515299796983e-06,
+      "loss": 0.7408,
+      "step": 4213
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.938159009567855e-06,
+      "loss": 0.8117,
+      "step": 4214
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.933803127954071e-06,
+      "loss": 0.7761,
+      "step": 4215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.929447655818882e-06,
+      "loss": 0.8559,
+      "step": 4216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.925092594025454e-06,
+      "loss": 0.821,
+      "step": 4217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.920737943436872e-06,
+      "loss": 0.7203,
+      "step": 4218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91638370491614e-06,
+      "loss": 0.3778,
+      "step": 4219
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.912029879326179e-06,
+      "loss": 0.8245,
+      "step": 4220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90767646752983e-06,
+      "loss": 0.744,
+      "step": 4221
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.903323470389855e-06,
+      "loss": 0.782,
+      "step": 4222
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.898970888768928e-06,
+      "loss": 0.7788,
+      "step": 4223
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.894618723529643e-06,
+      "loss": 0.893,
+      "step": 4224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89026697553451e-06,
+      "loss": 0.6854,
+      "step": 4225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88591564564596e-06,
+      "loss": 0.7551,
+      "step": 4226
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.881564734726335e-06,
+      "loss": 0.7064,
+      "step": 4227
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.877214243637902e-06,
+      "loss": 0.7194,
+      "step": 4228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.872864173242841e-06,
+      "loss": 0.7539,
+      "step": 4229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.868514524403245e-06,
+      "loss": 0.7418,
+      "step": 4230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.864165297981132e-06,
+      "loss": 0.8177,
+      "step": 4231
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.859816494838426e-06,
+      "loss": 0.756,
+      "step": 4232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.855468115836972e-06,
+      "loss": 0.7809,
+      "step": 4233
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.851120161838535e-06,
+      "loss": 0.8401,
+      "step": 4234
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84677263370479e-06,
+      "loss": 0.8641,
+      "step": 4235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.842425532297327e-06,
+      "loss": 0.6885,
+      "step": 4236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.838078858477657e-06,
+      "loss": 0.8245,
+      "step": 4237
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.833732613107203e-06,
+      "loss": 0.7129,
+      "step": 4238
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829386797047304e-06,
+      "loss": 0.8293,
+      "step": 4239
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.82504141115921e-06,
+      "loss": 0.6464,
+      "step": 4240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.820696456304093e-06,
+      "loss": 0.7935,
+      "step": 4241
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81635193334303e-06,
+      "loss": 0.7626,
+      "step": 4242
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.812007843137022e-06,
+      "loss": 0.7713,
+      "step": 4243
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.807664186546982e-06,
+      "loss": 0.8084,
+      "step": 4244
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80332096443373e-06,
+      "loss": 0.8092,
+      "step": 4245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79897817765801e-06,
+      "loss": 0.7998,
+      "step": 4246
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.794635827080472e-06,
+      "loss": 0.7684,
+      "step": 4247
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.790293913561681e-06,
+      "loss": 0.7758,
+      "step": 4248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.785952437962123e-06,
+      "loss": 0.8509,
+      "step": 4249
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.781611401142185e-06,
+      "loss": 0.8708,
+      "step": 4250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777270803962173e-06,
+      "loss": 0.7685,
+      "step": 4251
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77293064728231e-06,
+      "loss": 0.7817,
+      "step": 4252
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.768590931962725e-06,
+      "loss": 0.801,
+      "step": 4253
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76425165886346e-06,
+      "loss": 0.8309,
+      "step": 4254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.759912828844476e-06,
+      "loss": 0.8254,
+      "step": 4255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75557444276564e-06,
+      "loss": 0.7609,
+      "step": 4256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.75123650148673e-06,
+      "loss": 0.7928,
+      "step": 4257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.746899005867444e-06,
+      "loss": 0.7978,
+      "step": 4258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.742561956767383e-06,
+      "loss": 0.7729,
+      "step": 4259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.738225355046062e-06,
+      "loss": 0.7966,
+      "step": 4260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73388920156291e-06,
+      "loss": 0.7766,
+      "step": 4261
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.729553497177262e-06,
+      "loss": 0.7786,
+      "step": 4262
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.725218242748371e-06,
+      "loss": 0.8436,
+      "step": 4263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.720883439135396e-06,
+      "loss": 0.7907,
+      "step": 4264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.716549087197407e-06,
+      "loss": 0.7861,
+      "step": 4265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.712215187793388e-06,
+      "loss": 0.8133,
+      "step": 4266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707881741782227e-06,
+      "loss": 0.759,
+      "step": 4267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.703548750022724e-06,
+      "loss": 0.6924,
+      "step": 4268
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.699216213373594e-06,
+      "loss": 0.8025,
+      "step": 4269
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.694884132693462e-06,
+      "loss": 0.8699,
+      "step": 4270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.690552508840853e-06,
+      "loss": 0.8231,
+      "step": 4271
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.686221342674211e-06,
+      "loss": 0.7961,
+      "step": 4272
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681890635051883e-06,
+      "loss": 0.8534,
+      "step": 4273
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.677560386832132e-06,
+      "loss": 0.7832,
+      "step": 4274
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.673230598873122e-06,
+      "loss": 0.893,
+      "step": 4275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66890127203293e-06,
+      "loss": 0.8167,
+      "step": 4276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.664572407169544e-06,
+      "loss": 0.7332,
+      "step": 4277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660244005140857e-06,
+      "loss": 0.8432,
+      "step": 4278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.65591606680467e-06,
+      "loss": 0.7759,
+      "step": 4279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651588593018695e-06,
+      "loss": 0.7912,
+      "step": 4280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.647261584640548e-06,
+      "loss": 0.7692,
+      "step": 4281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.642935042527756e-06,
+      "loss": 0.8078,
+      "step": 4282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63860896753775e-06,
+      "loss": 0.7569,
+      "step": 4283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.634283360527877e-06,
+      "loss": 0.7926,
+      "step": 4284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.62995822235538e-06,
+      "loss": 0.7481,
+      "step": 4285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.625633553877414e-06,
+      "loss": 0.377,
+      "step": 4286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6213093559510456e-06,
+      "loss": 0.7664,
+      "step": 4287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.616985629433239e-06,
+      "loss": 0.88,
+      "step": 4288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.612662375180872e-06,
+      "loss": 0.7843,
+      "step": 4289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.608339594050726e-06,
+      "loss": 0.73,
+      "step": 4290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.604017286899492e-06,
+      "loss": 0.7845,
+      "step": 4291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.59969545458376e-06,
+      "loss": 0.658,
+      "step": 4292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.595374097960031e-06,
+      "loss": 0.8187,
+      "step": 4293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.59105321788471e-06,
+      "loss": 0.7507,
+      "step": 4294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.586732815214109e-06,
+      "loss": 0.7381,
+      "step": 4295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.582412890804447e-06,
+      "loss": 0.7528,
+      "step": 4296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.578093445511845e-06,
+      "loss": 0.77,
+      "step": 4297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.573774480192328e-06,
+      "loss": 0.8038,
+      "step": 4298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.569455995701831e-06,
+      "loss": 0.8069,
+      "step": 4299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.565137992896189e-06,
+      "loss": 0.7733,
+      "step": 4300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.560820472631143e-06,
+      "loss": 0.7816,
+      "step": 4301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.556503435762336e-06,
+      "loss": 0.802,
+      "step": 4302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.552186883145322e-06,
+      "loss": 0.8552,
+      "step": 4303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.547870815635555e-06,
+      "loss": 0.7832,
+      "step": 4304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.54355523408839e-06,
+      "loss": 0.7808,
+      "step": 4305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.53924013935909e-06,
+      "loss": 0.779,
+      "step": 4306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.534925532302817e-06,
+      "loss": 0.8529,
+      "step": 4307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.530611413774643e-06,
+      "loss": 0.689,
+      "step": 4308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.526297784629535e-06,
+      "loss": 0.8177,
+      "step": 4309
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.521984645722372e-06,
+      "loss": 0.7745,
+      "step": 4310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5176719979079295e-06,
+      "loss": 0.7888,
+      "step": 4311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513359842040887e-06,
+      "loss": 0.7453,
+      "step": 4312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.509048178975827e-06,
+      "loss": 0.7816,
+      "step": 4313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5047370095672345e-06,
+      "loss": 0.8423,
+      "step": 4314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.500426334669498e-06,
+      "loss": 0.7593,
+      "step": 4315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.496116155136901e-06,
+      "loss": 0.7207,
+      "step": 4316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.491806471823637e-06,
+      "loss": 0.8044,
+      "step": 4317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.487497285583802e-06,
+      "loss": 0.8092,
+      "step": 4318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.483188597271389e-06,
+      "loss": 0.7451,
+      "step": 4319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.47888040774029e-06,
+      "loss": 0.6635,
+      "step": 4320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.474572717844302e-06,
+      "loss": 0.8215,
+      "step": 4321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.470265528437124e-06,
+      "loss": 0.7792,
+      "step": 4322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.465958840372352e-06,
+      "loss": 0.7319,
+      "step": 4323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.461652654503486e-06,
+      "loss": 0.8099,
+      "step": 4324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4573469716839266e-06,
+      "loss": 0.8428,
+      "step": 4325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.453041792766972e-06,
+      "loss": 0.7853,
+      "step": 4326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.448737118605822e-06,
+      "loss": 0.8335,
+      "step": 4327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.444432950053577e-06,
+      "loss": 0.7877,
+      "step": 4328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.440129287963236e-06,
+      "loss": 0.7095,
+      "step": 4329
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.435826133187698e-06,
+      "loss": 0.8608,
+      "step": 4330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.431523486579759e-06,
+      "loss": 0.8218,
+      "step": 4331
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.427221348992122e-06,
+      "loss": 0.7479,
+      "step": 4332
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.422919721277386e-06,
+      "loss": 0.8308,
+      "step": 4333
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.418618604288038e-06,
+      "loss": 0.8348,
+      "step": 4334
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.414317998876475e-06,
+      "loss": 0.7845,
+      "step": 4335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.410017905894995e-06,
+      "loss": 0.7567,
+      "step": 4336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4057183261957865e-06,
+      "loss": 0.7658,
+      "step": 4337
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.401419260630941e-06,
+      "loss": 0.7589,
+      "step": 4338
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3971207100524456e-06,
+      "loss": 0.8679,
+      "step": 4339
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3928226753121875e-06,
+      "loss": 0.3527,
+      "step": 4340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.38852515726195e-06,
+      "loss": 0.8086,
+      "step": 4341
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.384228156753414e-06,
+      "loss": 0.7902,
+      "step": 4342
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.379931674638159e-06,
+      "loss": 0.8107,
+      "step": 4343
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3756357117676615e-06,
+      "loss": 0.7855,
+      "step": 4344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.371340268993294e-06,
+      "loss": 0.8316,
+      "step": 4345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3670453471663295e-06,
+      "loss": 0.7466,
+      "step": 4346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.362750947137931e-06,
+      "loss": 0.8102,
+      "step": 4347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.358457069759164e-06,
+      "loss": 0.7295,
+      "step": 4348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354163715880988e-06,
+      "loss": 0.8172,
+      "step": 4349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.34987088635426e-06,
+      "loss": 0.826,
+      "step": 4350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.345578582029731e-06,
+      "loss": 0.8,
+      "step": 4351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.341286803758052e-06,
+      "loss": 0.7384,
+      "step": 4352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.336995552389764e-06,
+      "loss": 0.8148,
+      "step": 4353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.332704828775308e-06,
+      "loss": 0.3657,
+      "step": 4354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.328414633765019e-06,
+      "loss": 0.3625,
+      "step": 4355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3241249682091255e-06,
+      "loss": 0.8123,
+      "step": 4356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.319835832957754e-06,
+      "loss": 0.7477,
+      "step": 4357
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3155472288609254e-06,
+      "loss": 0.8116,
+      "step": 4358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.311259156768556e-06,
+      "loss": 0.7734,
+      "step": 4359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.306971617530451e-06,
+      "loss": 0.8103,
+      "step": 4360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302684611996318e-06,
+      "loss": 0.8661,
+      "step": 4361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.298398141015753e-06,
+      "loss": 0.8138,
+      "step": 4362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.294112205438251e-06,
+      "loss": 0.8344,
+      "step": 4363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.289826806113192e-06,
+      "loss": 0.8122,
+      "step": 4364
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2855419438898625e-06,
+      "loss": 0.7192,
+      "step": 4365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.281257619617433e-06,
+      "loss": 0.7394,
+      "step": 4366
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.276973834144971e-06,
+      "loss": 0.8221,
+      "step": 4367
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.272690588321436e-06,
+      "loss": 0.8041,
+      "step": 4368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2684078829956815e-06,
+      "loss": 0.7687,
+      "step": 4369
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.264125719016453e-06,
+      "loss": 0.8211,
+      "step": 4370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.259844097232388e-06,
+      "loss": 0.7327,
+      "step": 4371
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.255563018492018e-06,
+      "loss": 0.7697,
+      "step": 4372
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2512824836437675e-06,
+      "loss": 0.3283,
+      "step": 4373
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.247002493535958e-06,
+      "loss": 0.832,
+      "step": 4374
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.242723049016787e-06,
+      "loss": 0.836,
+      "step": 4375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.238444150934356e-06,
+      "loss": 0.7457,
+      "step": 4376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.234165800136661e-06,
+      "loss": 0.8214,
+      "step": 4377
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.229887997471583e-06,
+      "loss": 0.721,
+      "step": 4378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2256107437868974e-06,
+      "loss": 0.7748,
+      "step": 4379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.221334039930268e-06,
+      "loss": 0.7768,
+      "step": 4380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2170578867492504e-06,
+      "loss": 0.7985,
+      "step": 4381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2127822850912955e-06,
+      "loss": 0.8083,
+      "step": 4382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.208507235803737e-06,
+      "loss": 0.8403,
+      "step": 4383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204232739733807e-06,
+      "loss": 0.7925,
+      "step": 4384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.199958797728624e-06,
+      "loss": 0.8194,
+      "step": 4385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.195685410635196e-06,
+      "loss": 0.825,
+      "step": 4386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.191412579300423e-06,
+      "loss": 0.781,
+      "step": 4387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.187140304571095e-06,
+      "loss": 0.8174,
+      "step": 4388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.18286858729389e-06,
+      "loss": 0.8005,
+      "step": 4389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.178597428315372e-06,
+      "loss": 0.7355,
+      "step": 4390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.174326828482007e-06,
+      "loss": 0.7691,
+      "step": 4391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.170056788640136e-06,
+      "loss": 0.7639,
+      "step": 4392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.165787309635998e-06,
+      "loss": 0.3153,
+      "step": 4393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.161518392315715e-06,
+      "loss": 0.859,
+      "step": 4394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.157250037525302e-06,
+      "loss": 0.8243,
+      "step": 4395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.15298224611066e-06,
+      "loss": 0.7576,
+      "step": 4396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1487150189175805e-06,
+      "loss": 0.7827,
+      "step": 4397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.144448356791738e-06,
+      "loss": 0.8031,
+      "step": 4398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.140182260578705e-06,
+      "loss": 0.8067,
+      "step": 4399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.135916731123933e-06,
+      "loss": 0.7195,
+      "step": 4400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.131651769272765e-06,
+      "loss": 0.7903,
+      "step": 4401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.127387375870429e-06,
+      "loss": 0.8665,
+      "step": 4402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.123123551762041e-06,
+      "loss": 0.713,
+      "step": 4403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.118860297792609e-06,
+      "loss": 0.7721,
+      "step": 4404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.114597614807017e-06,
+      "loss": 0.8347,
+      "step": 4405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.110335503650051e-06,
+      "loss": 0.8183,
+      "step": 4406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.106073965166372e-06,
+      "loss": 0.8229,
+      "step": 4407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.10181300020053e-06,
+      "loss": 0.7686,
+      "step": 4408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.097552609596964e-06,
+      "loss": 0.8066,
+      "step": 4409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0932927941999974e-06,
+      "loss": 0.8218,
+      "step": 4410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0890335548538395e-06,
+      "loss": 0.7629,
+      "step": 4411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0847748924025815e-06,
+      "loss": 0.7457,
+      "step": 4412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.080516807690213e-06,
+      "loss": 0.7607,
+      "step": 4413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.076259301560595e-06,
+      "loss": 0.8071,
+      "step": 4414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0720023748574825e-06,
+      "loss": 0.7345,
+      "step": 4415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0677460284245115e-06,
+      "loss": 0.8061,
+      "step": 4416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.063490263105202e-06,
+      "loss": 0.8195,
+      "step": 4417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.059235079742962e-06,
+      "loss": 0.7924,
+      "step": 4418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.054980479181084e-06,
+      "loss": 0.8339,
+      "step": 4419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.050726462262743e-06,
+      "loss": 0.7982,
+      "step": 4420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.046473029831e-06,
+      "loss": 0.775,
+      "step": 4421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0422201827288e-06,
+      "loss": 0.7934,
+      "step": 4422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.037967921798971e-06,
+      "loss": 0.7755,
+      "step": 4423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.033716247884224e-06,
+      "loss": 0.7151,
+      "step": 4424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0294651618271585e-06,
+      "loss": 0.8487,
+      "step": 4425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.025214664470252e-06,
+      "loss": 0.8385,
+      "step": 4426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.020964756655869e-06,
+      "loss": 0.8167,
+      "step": 4427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.016715439226253e-06,
+      "loss": 0.8014,
+      "step": 4428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.012466713023538e-06,
+      "loss": 0.7414,
+      "step": 4429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00821857888973e-06,
+      "loss": 0.7129,
+      "step": 4430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.003971037666723e-06,
+      "loss": 0.8562,
+      "step": 4431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9997240901963024e-06,
+      "loss": 0.7388,
+      "step": 4432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.99547773732012e-06,
+      "loss": 0.7605,
+      "step": 4433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.991231979879721e-06,
+      "loss": 0.782,
+      "step": 4434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.986986818716528e-06,
+      "loss": 0.7924,
+      "step": 4435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.982742254671846e-06,
+      "loss": 0.7408,
+      "step": 4436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9784982885868624e-06,
+      "loss": 0.7423,
+      "step": 4437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.974254921302644e-06,
+      "loss": 0.8474,
+      "step": 4438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.970012153660144e-06,
+      "loss": 0.8014,
+      "step": 4439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.965769986500191e-06,
+      "loss": 0.8138,
+      "step": 4440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.961528420663499e-06,
+      "loss": 0.7322,
+      "step": 4441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.95728745699066e-06,
+      "loss": 0.7555,
+      "step": 4442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.953047096322148e-06,
+      "loss": 0.3221,
+      "step": 4443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.948807339498316e-06,
+      "loss": 0.7421,
+      "step": 4444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.944568187359397e-06,
+      "loss": 0.8383,
+      "step": 4445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.940329640745509e-06,
+      "loss": 0.8622,
+      "step": 4446
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.936091700496645e-06,
+      "loss": 0.7336,
+      "step": 4447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.93185436745268e-06,
+      "loss": 0.9191,
+      "step": 4448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.927617642453367e-06,
+      "loss": 0.8605,
+      "step": 4449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9233815263383394e-06,
+      "loss": 0.7887,
+      "step": 4450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.919146019947111e-06,
+      "loss": 0.7512,
+      "step": 4451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.914911124119075e-06,
+      "loss": 0.8656,
+      "step": 4452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.910676839693497e-06,
+      "loss": 0.7632,
+      "step": 4453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9064431675095336e-06,
+      "loss": 0.348,
+      "step": 4454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.902210108406211e-06,
+      "loss": 0.7317,
+      "step": 4455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.897977663222435e-06,
+      "loss": 0.7801,
+      "step": 4456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.893745832796991e-06,
+      "loss": 0.8226,
+      "step": 4457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8895146179685464e-06,
+      "loss": 0.8072,
+      "step": 4458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.885284019575637e-06,
+      "loss": 0.8439,
+      "step": 4459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.881054038456685e-06,
+      "loss": 0.7216,
+      "step": 4460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.876824675449987e-06,
+      "loss": 0.7744,
+      "step": 4461
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.872595931393718e-06,
+      "loss": 0.7922,
+      "step": 4462
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868367807125928e-06,
+      "loss": 0.791,
+      "step": 4463
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.864140303484548e-06,
+      "loss": 0.808,
+      "step": 4464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8599134213073845e-06,
+      "loss": 0.8012,
+      "step": 4465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.85568716143212e-06,
+      "loss": 0.7566,
+      "step": 4466
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.851461524696314e-06,
+      "loss": 0.8421,
+      "step": 4467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.847236511937401e-06,
+      "loss": 0.732,
+      "step": 4468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.843012123992696e-06,
+      "loss": 0.7792,
+      "step": 4469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.838788361699384e-06,
+      "loss": 0.8462,
+      "step": 4470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.834565225894531e-06,
+      "loss": 0.7749,
+      "step": 4471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.830342717415081e-06,
+      "loss": 0.8087,
+      "step": 4472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.826120837097848e-06,
+      "loss": 0.8617,
+      "step": 4473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.821899585779522e-06,
+      "loss": 0.7746,
+      "step": 4474
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.817678964296671e-06,
+      "loss": 0.8119,
+      "step": 4475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.813458973485737e-06,
+      "loss": 0.7398,
+      "step": 4476
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.809239614183038e-06,
+      "loss": 0.8153,
+      "step": 4477
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8050208872247646e-06,
+      "loss": 0.8051,
+      "step": 4478
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8008027934469814e-06,
+      "loss": 0.7381,
+      "step": 4479
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7965853336856345e-06,
+      "loss": 0.7171,
+      "step": 4480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.792368508776539e-06,
+      "loss": 0.7913,
+      "step": 4481
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.788152319555381e-06,
+      "loss": 0.792,
+      "step": 4482
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7839367668577286e-06,
+      "loss": 0.7769,
+      "step": 4483
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.779721851519015e-06,
+      "loss": 0.7748,
+      "step": 4484
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.775507574374553e-06,
+      "loss": 0.7673,
+      "step": 4485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.771293936259527e-06,
+      "loss": 0.8067,
+      "step": 4486
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.767080938008999e-06,
+      "loss": 0.8113,
+      "step": 4487
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7628685804578976e-06,
+      "loss": 0.7977,
+      "step": 4488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.758656864441027e-06,
+      "loss": 0.7997,
+      "step": 4489
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.754445790793063e-06,
+      "loss": 0.7737,
+      "step": 4490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.750235360348558e-06,
+      "loss": 0.7204,
+      "step": 4491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.746025573941934e-06,
+      "loss": 0.7337,
+      "step": 4492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.741816432407483e-06,
+      "loss": 0.8056,
+      "step": 4493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.737607936579378e-06,
+      "loss": 0.707,
+      "step": 4494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.733400087291653e-06,
+      "loss": 0.7927,
+      "step": 4495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.729192885378222e-06,
+      "loss": 0.7959,
+      "step": 4496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.724986331672865e-06,
+      "loss": 0.7329,
+      "step": 4497
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.720780427009237e-06,
+      "loss": 0.7036,
+      "step": 4498
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.716575172220865e-06,
+      "loss": 0.7472,
+      "step": 4499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.712370568141148e-06,
+      "loss": 0.7671,
+      "step": 4500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.708166615603347e-06,
+      "loss": 0.7432,
+      "step": 4501
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.703963315440605e-06,
+      "loss": 0.3179,
+      "step": 4502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.699760668485931e-06,
+      "loss": 0.7591,
+      "step": 4503
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695558675572204e-06,
+      "loss": 0.7593,
+      "step": 4504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.691357337532173e-06,
+      "loss": 0.7171,
+      "step": 4505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.687156655198463e-06,
+      "loss": 0.7768,
+      "step": 4506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6829566294035616e-06,
+      "loss": 0.6513,
+      "step": 4507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6787572609798315e-06,
+      "loss": 0.8136,
+      "step": 4508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6745585507595e-06,
+      "loss": 0.8133,
+      "step": 4509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.670360499574668e-06,
+      "loss": 0.3252,
+      "step": 4510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.666163108257305e-06,
+      "loss": 0.7397,
+      "step": 4511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.661966377639248e-06,
+      "loss": 0.8323,
+      "step": 4512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.657770308552207e-06,
+      "loss": 0.7116,
+      "step": 4513
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6535749018277575e-06,
+      "loss": 0.7265,
+      "step": 4514
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.649380158297345e-06,
+      "loss": 0.7777,
+      "step": 4515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.645186078792281e-06,
+      "loss": 0.3404,
+      "step": 4516
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.640992664143749e-06,
+      "loss": 0.7439,
+      "step": 4517
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6367999151828e-06,
+      "loss": 0.7732,
+      "step": 4518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6326078327403494e-06,
+      "loss": 0.8287,
+      "step": 4519
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.628416417647188e-06,
+      "loss": 0.6719,
+      "step": 4520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.624225670733966e-06,
+      "loss": 0.7834,
+      "step": 4521
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.620035592831209e-06,
+      "loss": 0.8138,
+      "step": 4522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.615846184769301e-06,
+      "loss": 0.3538,
+      "step": 4523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611657447378501e-06,
+      "loss": 0.8374,
+      "step": 4524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.607469381488933e-06,
+      "loss": 0.6596,
+      "step": 4525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.603281987930588e-06,
+      "loss": 0.8014,
+      "step": 4526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.599095267533317e-06,
+      "loss": 0.7698,
+      "step": 4527
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5949092211268504e-06,
+      "loss": 0.7751,
+      "step": 4528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.590723849540778e-06,
+      "loss": 0.8303,
+      "step": 4529
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.586539153604552e-06,
+      "loss": 0.762,
+      "step": 4530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.582355134147498e-06,
+      "loss": 0.7713,
+      "step": 4531
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.578171791998804e-06,
+      "loss": 0.7839,
+      "step": 4532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.573989127987521e-06,
+      "loss": 0.8286,
+      "step": 4533
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5698071429425715e-06,
+      "loss": 0.7579,
+      "step": 4534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5656258376927416e-06,
+      "loss": 0.8298,
+      "step": 4535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.561445213066681e-06,
+      "loss": 0.8132,
+      "step": 4536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557265269892904e-06,
+      "loss": 0.3688,
+      "step": 4537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.553086008999793e-06,
+      "loss": 0.8387,
+      "step": 4538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.548907431215592e-06,
+      "loss": 0.7814,
+      "step": 4539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.54472953736841e-06,
+      "loss": 0.8191,
+      "step": 4540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.540552328286225e-06,
+      "loss": 0.7585,
+      "step": 4541
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.536375804796873e-06,
+      "loss": 0.8705,
+      "step": 4542
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.532199967728055e-06,
+      "loss": 0.7315,
+      "step": 4543
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.528024817907341e-06,
+      "loss": 0.7692,
+      "step": 4544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.523850356162158e-06,
+      "loss": 0.7775,
+      "step": 4545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.519676583319804e-06,
+      "loss": 0.8148,
+      "step": 4546
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515503500207436e-06,
+      "loss": 0.7384,
+      "step": 4547
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.511331107652075e-06,
+      "loss": 0.7753,
+      "step": 4548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.507159406480604e-06,
+      "loss": 0.3753,
+      "step": 4549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5029883975197695e-06,
+      "loss": 0.7795,
+      "step": 4550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.498818081596184e-06,
+      "loss": 0.826,
+      "step": 4551
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.494648459536317e-06,
+      "loss": 0.8477,
+      "step": 4552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.490479532166506e-06,
+      "loss": 0.802,
+      "step": 4553
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.486311300312948e-06,
+      "loss": 0.7475,
+      "step": 4554
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.482143764801702e-06,
+      "loss": 0.7523,
+      "step": 4555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.477976926458691e-06,
+      "loss": 0.7806,
+      "step": 4556
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.473810786109696e-06,
+      "loss": 0.7457,
+      "step": 4557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.469645344580364e-06,
+      "loss": 0.7702,
+      "step": 4558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.465480602696202e-06,
+      "loss": 0.863,
+      "step": 4559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4613165612825724e-06,
+      "loss": 0.7979,
+      "step": 4560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.457153221164712e-06,
+      "loss": 0.7691,
+      "step": 4561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.452990583167708e-06,
+      "loss": 0.8194,
+      "step": 4562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.448828648116511e-06,
+      "loss": 0.775,
+      "step": 4563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4446674168359324e-06,
+      "loss": 0.7421,
+      "step": 4564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4405068901506455e-06,
+      "loss": 0.7874,
+      "step": 4565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4363470688851805e-06,
+      "loss": 0.8533,
+      "step": 4566
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.432187953863931e-06,
+      "loss": 0.7675,
+      "step": 4567
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.428029545911153e-06,
+      "loss": 0.8132,
+      "step": 4568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.423871845850956e-06,
+      "loss": 0.7849,
+      "step": 4569
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.419714854507314e-06,
+      "loss": 0.74,
+      "step": 4570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.415558572704058e-06,
+      "loss": 0.7346,
+      "step": 4571
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.411403001264879e-06,
+      "loss": 0.7742,
+      "step": 4572
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.407248141013329e-06,
+      "loss": 0.7306,
+      "step": 4573
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.403093992772813e-06,
+      "loss": 0.7768,
+      "step": 4574
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.398940557366606e-06,
+      "loss": 0.8254,
+      "step": 4575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.394787835617833e-06,
+      "loss": 0.7955,
+      "step": 4576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3906358283494775e-06,
+      "loss": 0.7589,
+      "step": 4577
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.386484536384386e-06,
+      "loss": 0.7206,
+      "step": 4578
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.382333960545259e-06,
+      "loss": 0.7396,
+      "step": 4579
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.378184101654661e-06,
+      "loss": 0.7791,
+      "step": 4580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.374034960535006e-06,
+      "loss": 0.7456,
+      "step": 4581
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.369886538008569e-06,
+      "loss": 0.8117,
+      "step": 4582
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365738834897495e-06,
+      "loss": 0.7122,
+      "step": 4583
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.361591852023761e-06,
+      "loss": 0.7568,
+      "step": 4584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357445590209221e-06,
+      "loss": 0.3409,
+      "step": 4585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.35330005027558e-06,
+      "loss": 0.7347,
+      "step": 4586
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.349155233044403e-06,
+      "loss": 0.8559,
+      "step": 4587
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.345011139337107e-06,
+      "loss": 0.7503,
+      "step": 4588
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.34086776997497e-06,
+      "loss": 0.7905,
+      "step": 4589
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3367251257791215e-06,
+      "loss": 0.8103,
+      "step": 4590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.33258320757055e-06,
+      "loss": 0.795,
+      "step": 4591
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.328442016170105e-06,
+      "loss": 0.7758,
+      "step": 4592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.324301552398479e-06,
+      "loss": 0.7635,
+      "step": 4593
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.320161817076237e-06,
+      "loss": 0.7939,
+      "step": 4594
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.316022811023787e-06,
+      "loss": 0.8277,
+      "step": 4595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311884535061396e-06,
+      "loss": 0.7549,
+      "step": 4596
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.307746990009191e-06,
+      "loss": 0.7773,
+      "step": 4597
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.303610176687145e-06,
+      "loss": 0.8102,
+      "step": 4598
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.299474095915094e-06,
+      "loss": 0.7692,
+      "step": 4599
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.295338748512723e-06,
+      "loss": 0.7983,
+      "step": 4600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.291204135299578e-06,
+      "loss": 0.823,
+      "step": 4601
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.287070257095055e-06,
+      "loss": 0.7637,
+      "step": 4602
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.282937114718405e-06,
+      "loss": 0.7668,
+      "step": 4603
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.278804708988733e-06,
+      "loss": 0.7641,
+      "step": 4604
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.274673040725e-06,
+      "loss": 0.7489,
+      "step": 4605
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.270542110746018e-06,
+      "loss": 0.8018,
+      "step": 4606
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2664119198704555e-06,
+      "loss": 0.6871,
+      "step": 4607
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.262282468916829e-06,
+      "loss": 0.8267,
+      "step": 4608
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.258153758703519e-06,
+      "loss": 0.8159,
+      "step": 4609
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.254025790048749e-06,
+      "loss": 0.3549,
+      "step": 4610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2498985637706e-06,
+      "loss": 0.864,
+      "step": 4611
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.245772080687005e-06,
+      "loss": 0.8067,
+      "step": 4612
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.241646341615748e-06,
+      "loss": 0.7749,
+      "step": 4613
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.23752134737447e-06,
+      "loss": 0.7898,
+      "step": 4614
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.233397098780659e-06,
+      "loss": 0.7769,
+      "step": 4615
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2292735966516616e-06,
+      "loss": 0.705,
+      "step": 4616
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.22515084180467e-06,
+      "loss": 0.7499,
+      "step": 4617
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2210288350567325e-06,
+      "loss": 0.857,
+      "step": 4618
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.216907577224749e-06,
+      "loss": 0.7836,
+      "step": 4619
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.212787069125466e-06,
+      "loss": 0.7027,
+      "step": 4620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2086673115754895e-06,
+      "loss": 0.7311,
+      "step": 4621
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204548305391267e-06,
+      "loss": 0.8096,
+      "step": 4622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.200430051389108e-06,
+      "loss": 0.3552,
+      "step": 4623
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.196312550385164e-06,
+      "loss": 0.7018,
+      "step": 4624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1921958031954485e-06,
+      "loss": 0.7204,
+      "step": 4625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.188079810635803e-06,
+      "loss": 0.6483,
+      "step": 4626
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1839645735219465e-06,
+      "loss": 0.7229,
+      "step": 4627
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.179850092669433e-06,
+      "loss": 0.8121,
+      "step": 4628
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.175736368893667e-06,
+      "loss": 0.7543,
+      "step": 4629
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1716234030099075e-06,
+      "loss": 0.8201,
+      "step": 4630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1675111958332635e-06,
+      "loss": 0.8103,
+      "step": 4631
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.163399748178689e-06,
+      "loss": 0.8027,
+      "step": 4632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.159289060860989e-06,
+      "loss": 0.8536,
+      "step": 4633
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.155179134694822e-06,
+      "loss": 0.7443,
+      "step": 4634
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.151069970494693e-06,
+      "loss": 0.7336,
+      "step": 4635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.146961569074953e-06,
+      "loss": 0.8082,
+      "step": 4636
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.142853931249806e-06,
+      "loss": 0.7509,
+      "step": 4637
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.138747057833301e-06,
+      "loss": 0.8013,
+      "step": 4638
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13464094963934e-06,
+      "loss": 0.7583,
+      "step": 4639
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.130535607481669e-06,
+      "loss": 0.7092,
+      "step": 4640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.126431032173884e-06,
+      "loss": 0.3561,
+      "step": 4641
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1223272245294315e-06,
+      "loss": 0.8523,
+      "step": 4642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.118224185361603e-06,
+      "loss": 0.8225,
+      "step": 4643
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.114121915483535e-06,
+      "loss": 0.7344,
+      "step": 4644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.11002041570822e-06,
+      "loss": 0.7992,
+      "step": 4645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105919686848488e-06,
+      "loss": 0.6976,
+      "step": 4646
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.101819729717022e-06,
+      "loss": 0.6848,
+      "step": 4647
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0977205451263514e-06,
+      "loss": 0.385,
+      "step": 4648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.093622133888853e-06,
+      "loss": 0.7906,
+      "step": 4649
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.089524496816748e-06,
+      "loss": 0.3652,
+      "step": 4650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.085427634722108e-06,
+      "loss": 0.7322,
+      "step": 4651
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.081331548416844e-06,
+      "loss": 0.7168,
+      "step": 4652
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.07723623871272e-06,
+      "loss": 0.7536,
+      "step": 4653
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.073141706421346e-06,
+      "loss": 0.831,
+      "step": 4654
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.06904795235417e-06,
+      "loss": 0.7596,
+      "step": 4655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.064954977322499e-06,
+      "loss": 0.7453,
+      "step": 4656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.060862782137473e-06,
+      "loss": 0.7801,
+      "step": 4657
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056771367610086e-06,
+      "loss": 0.8396,
+      "step": 4658
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.052680734551171e-06,
+      "loss": 0.7408,
+      "step": 4659
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0485908837714124e-06,
+      "loss": 0.7459,
+      "step": 4660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.044501816081331e-06,
+      "loss": 0.821,
+      "step": 4661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.040413532291304e-06,
+      "loss": 0.7518,
+      "step": 4662
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.036326033211539e-06,
+      "loss": 0.7756,
+      "step": 4663
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032239319652105e-06,
+      "loss": 0.7322,
+      "step": 4664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0281533924228995e-06,
+      "loss": 0.8122,
+      "step": 4665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.024068252333674e-06,
+      "loss": 0.7327,
+      "step": 4666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.019983900194024e-06,
+      "loss": 0.6992,
+      "step": 4667
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.015900336813379e-06,
+      "loss": 0.6992,
+      "step": 4668
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.011817563001023e-06,
+      "loss": 0.7581,
+      "step": 4669
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.007735579566078e-06,
+      "loss": 0.78,
+      "step": 4670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.003654387317512e-06,
+      "loss": 0.8143,
+      "step": 4671
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.999573987064133e-06,
+      "loss": 0.7378,
+      "step": 4672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9954943796145974e-06,
+      "loss": 0.7991,
+      "step": 4673
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.991415565777396e-06,
+      "loss": 0.8047,
+      "step": 4674
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.987337546360874e-06,
+      "loss": 0.7999,
+      "step": 4675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9832603221732096e-06,
+      "loss": 0.7358,
+      "step": 4676
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.979183894022429e-06,
+      "loss": 0.7461,
+      "step": 4677
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.975108262716394e-06,
+      "loss": 0.7738,
+      "step": 4678
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9710334290628135e-06,
+      "loss": 0.7526,
+      "step": 4679
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.966959393869238e-06,
+      "loss": 0.7583,
+      "step": 4680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96288615794306e-06,
+      "loss": 0.8306,
+      "step": 4681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.958813722091512e-06,
+      "loss": 0.7883,
+      "step": 4682
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.95474208712167e-06,
+      "loss": 0.7787,
+      "step": 4683
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.95067125384045e-06,
+      "loss": 0.7984,
+      "step": 4684
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.946601223054606e-06,
+      "loss": 0.7177,
+      "step": 4685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.942531995570741e-06,
+      "loss": 0.6686,
+      "step": 4686
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.93846357219529e-06,
+      "loss": 0.3724,
+      "step": 4687
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.934395953734534e-06,
+      "loss": 0.7468,
+      "step": 4688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.93032914099459e-06,
+      "loss": 0.7299,
+      "step": 4689
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.926263134781425e-06,
+      "loss": 0.8255,
+      "step": 4690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922197935900835e-06,
+      "loss": 0.7845,
+      "step": 4691
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9181335451584606e-06,
+      "loss": 0.6897,
+      "step": 4692
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9140699633597834e-06,
+      "loss": 0.7517,
+      "step": 4693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.910007191310123e-06,
+      "loss": 0.8987,
+      "step": 4694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.905945229814638e-06,
+      "loss": 0.3621,
+      "step": 4695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.901884079678327e-06,
+      "loss": 0.8112,
+      "step": 4696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.897823741706032e-06,
+      "loss": 0.8509,
+      "step": 4697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.893764216702426e-06,
+      "loss": 0.8883,
+      "step": 4698
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8897055054720266e-06,
+      "loss": 0.7443,
+      "step": 4699
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.885647608819189e-06,
+      "loss": 0.7558,
+      "step": 4700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.881590527548106e-06,
+      "loss": 0.3468,
+      "step": 4701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8775342624628095e-06,
+      "loss": 0.3641,
+      "step": 4702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.873478814367169e-06,
+      "loss": 0.8122,
+      "step": 4703
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8694241840648935e-06,
+      "loss": 0.7053,
+      "step": 4704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.865370372359529e-06,
+      "loss": 0.8538,
+      "step": 4705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.861317380054458e-06,
+      "loss": 0.8014,
+      "step": 4706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.857265207952905e-06,
+      "loss": 0.7115,
+      "step": 4707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.853213856857927e-06,
+      "loss": 0.764,
+      "step": 4708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.849163327572418e-06,
+      "loss": 0.7903,
+      "step": 4709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8451136208991145e-06,
+      "loss": 0.6948,
+      "step": 4710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.841064737640585e-06,
+      "loss": 0.6924,
+      "step": 4711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.837016678599234e-06,
+      "loss": 0.7868,
+      "step": 4712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8329694445773135e-06,
+      "loss": 0.7908,
+      "step": 4713
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8289230363768944e-06,
+      "loss": 0.7611,
+      "step": 4714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.824877454799902e-06,
+      "loss": 0.3344,
+      "step": 4715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.820832700648079e-06,
+      "loss": 0.7755,
+      "step": 4716
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.816788774723026e-06,
+      "loss": 0.7768,
+      "step": 4717
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.812745677826155e-06,
+      "loss": 0.7517,
+      "step": 4718
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.808703410758733e-06,
+      "loss": 0.7255,
+      "step": 4719
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8046619743218615e-06,
+      "loss": 0.7786,
+      "step": 4720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8006213693164616e-06,
+      "loss": 0.8002,
+      "step": 4721
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.796581596543307e-06,
+      "loss": 0.6734,
+      "step": 4722
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.792542656802993e-06,
+      "loss": 0.7615,
+      "step": 4723
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.788504550895964e-06,
+      "loss": 0.7791,
+      "step": 4724
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.784467279622481e-06,
+      "loss": 0.8325,
+      "step": 4725
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.780430843782656e-06,
+      "loss": 0.7721,
+      "step": 4726
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.776395244176434e-06,
+      "loss": 0.334,
+      "step": 4727
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7723604816035805e-06,
+      "loss": 0.8113,
+      "step": 4728
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.768326556863711e-06,
+      "loss": 0.7685,
+      "step": 4729
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.764293470756262e-06,
+      "loss": 0.3405,
+      "step": 4730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.760261224080515e-06,
+      "loss": 0.6969,
+      "step": 4731
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7562298176355755e-06,
+      "loss": 0.8817,
+      "step": 4732
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7521992522203895e-06,
+      "loss": 0.7534,
+      "step": 4733
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.748169528633739e-06,
+      "loss": 0.7823,
+      "step": 4734
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.744140647674223e-06,
+      "loss": 0.775,
+      "step": 4735
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.740112610140297e-06,
+      "loss": 0.7632,
+      "step": 4736
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.736085416830225e-06,
+      "loss": 0.7447,
+      "step": 4737
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.732059068542126e-06,
+      "loss": 0.7382,
+      "step": 4738
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.728033566073932e-06,
+      "loss": 0.8168,
+      "step": 4739
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7240089102234194e-06,
+      "loss": 0.7699,
+      "step": 4740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.719985101788199e-06,
+      "loss": 0.3587,
+      "step": 4741
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.715962141565702e-06,
+      "loss": 0.8525,
+      "step": 4742
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.711940030353204e-06,
+      "loss": 0.8268,
+      "step": 4743
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7079187689478e-06,
+      "loss": 0.8395,
+      "step": 4744
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.703898358146429e-06,
+      "loss": 0.8522,
+      "step": 4745
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6998787987458504e-06,
+      "loss": 0.8359,
+      "step": 4746
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.695860091542661e-06,
+      "loss": 0.7736,
+      "step": 4747
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.691842237333292e-06,
+      "loss": 0.7444,
+      "step": 4748
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.687825236913996e-06,
+      "loss": 0.7608,
+      "step": 4749
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.683809091080869e-06,
+      "loss": 0.7911,
+      "step": 4750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679793800629818e-06,
+      "loss": 0.774,
+      "step": 4751
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6757793663566e-06,
+      "loss": 0.7592,
+      "step": 4752
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.671765789056797e-06,
+      "loss": 0.7606,
+      "step": 4753
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.667753069525812e-06,
+      "loss": 0.8,
+      "step": 4754
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.663741208558895e-06,
+      "loss": 0.7733,
+      "step": 4755
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.659730206951106e-06,
+      "loss": 0.7739,
+      "step": 4756
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.655720065497354e-06,
+      "loss": 0.8035,
+      "step": 4757
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651710784992359e-06,
+      "loss": 0.7632,
+      "step": 4758
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.647702366230684e-06,
+      "loss": 0.8212,
+      "step": 4759
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6436948100067215e-06,
+      "loss": 0.8028,
+      "step": 4760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.63968811711468e-06,
+      "loss": 0.77,
+      "step": 4761
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.635682288348613e-06,
+      "loss": 0.7599,
+      "step": 4762
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.631677324502387e-06,
+      "loss": 0.7646,
+      "step": 4763
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.627673226369714e-06,
+      "loss": 0.8005,
+      "step": 4764
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.623669994744118e-06,
+      "loss": 0.7982,
+      "step": 4765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.619667630418961e-06,
+      "loss": 0.3443,
+      "step": 4766
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.615666134187435e-06,
+      "loss": 0.797,
+      "step": 4767
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.61166550684255e-06,
+      "loss": 0.3614,
+      "step": 4768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607665749177156e-06,
+      "loss": 0.7511,
+      "step": 4769
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603666861983919e-06,
+      "loss": 0.8197,
+      "step": 4770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5996688460553415e-06,
+      "loss": 0.7661,
+      "step": 4771
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.595671702183745e-06,
+      "loss": 0.8437,
+      "step": 4772
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.591675431161285e-06,
+      "loss": 0.791,
+      "step": 4773
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.587680033779948e-06,
+      "loss": 0.784,
+      "step": 4774
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.583685510831531e-06,
+      "loss": 0.7751,
+      "step": 4775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.579691863107679e-06,
+      "loss": 0.7502,
+      "step": 4776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.575699091399841e-06,
+      "loss": 0.7545,
+      "step": 4777
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.571707196499314e-06,
+      "loss": 0.7564,
+      "step": 4778
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.567716179197202e-06,
+      "loss": 0.7958,
+      "step": 4779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.563726040284451e-06,
+      "loss": 0.7786,
+      "step": 4780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.559736780551825e-06,
+      "loss": 0.7364,
+      "step": 4781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5557484007899135e-06,
+      "loss": 0.8097,
+      "step": 4782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.551760901789135e-06,
+      "loss": 0.3563,
+      "step": 4783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.547774284339726e-06,
+      "loss": 0.8494,
+      "step": 4784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5437885492317635e-06,
+      "loss": 0.7162,
+      "step": 4785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539803697255128e-06,
+      "loss": 0.7403,
+      "step": 4786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535819729199548e-06,
+      "loss": 0.7726,
+      "step": 4787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.531836645854557e-06,
+      "loss": 0.8267,
+      "step": 4788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.527854448009526e-06,
+      "loss": 0.7312,
+      "step": 4789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.523873136453648e-06,
+      "loss": 0.709,
+      "step": 4790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.519892711975934e-06,
+      "loss": 0.7505,
+      "step": 4791
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5159131753652305e-06,
+      "loss": 0.7146,
+      "step": 4792
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.511934527410198e-06,
+      "loss": 0.7628,
+      "step": 4793
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.50795676889932e-06,
+      "loss": 0.7907,
+      "step": 4794
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.503979900620915e-06,
+      "loss": 0.7208,
+      "step": 4795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.500003923363113e-06,
+      "loss": 0.7619,
+      "step": 4796
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.496028837913877e-06,
+      "loss": 0.782,
+      "step": 4797
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.492054645060985e-06,
+      "loss": 0.7684,
+      "step": 4798
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.488081345592041e-06,
+      "loss": 0.691,
+      "step": 4799
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4841089402944816e-06,
+      "loss": 0.7517,
+      "step": 4800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.480137429955547e-06,
+      "loss": 0.7483,
+      "step": 4801
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.476166815362318e-06,
+      "loss": 0.7698,
+      "step": 4802
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.472197097301685e-06,
+      "loss": 0.7592,
+      "step": 4803
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.468228276560371e-06,
+      "loss": 0.7749,
+      "step": 4804
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.464260353924911e-06,
+      "loss": 0.6628,
+      "step": 4805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.460293330181669e-06,
+      "loss": 0.7948,
+      "step": 4806
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.456327206116836e-06,
+      "loss": 0.7647,
+      "step": 4807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.452361982516407e-06,
+      "loss": 0.7615,
+      "step": 4808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.448397660166221e-06,
+      "loss": 0.8516,
+      "step": 4809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4444342398519145e-06,
+      "loss": 0.7179,
+      "step": 4810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.440471722358969e-06,
+      "loss": 0.8116,
+      "step": 4811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.436510108472663e-06,
+      "loss": 0.7423,
+      "step": 4812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432549398978123e-06,
+      "loss": 0.7557,
+      "step": 4813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.42858959466027e-06,
+      "loss": 0.7149,
+      "step": 4814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.424630696303862e-06,
+      "loss": 0.8108,
+      "step": 4815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4206727046934774e-06,
+      "loss": 0.8255,
+      "step": 4816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4167156206135005e-06,
+      "loss": 0.7251,
+      "step": 4817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.412759444848157e-06,
+      "loss": 0.7472,
+      "step": 4818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4088041781814695e-06,
+      "loss": 0.8629,
+      "step": 4819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.404849821397302e-06,
+      "loss": 0.7562,
+      "step": 4820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.400896375279322e-06,
+      "loss": 0.344,
+      "step": 4821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.396943840611024e-06,
+      "loss": 0.7481,
+      "step": 4822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3929922181757265e-06,
+      "loss": 0.7347,
+      "step": 4823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.389041508756554e-06,
+      "loss": 0.79,
+      "step": 4824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.385091713136463e-06,
+      "loss": 0.3344,
+      "step": 4825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.381142832098216e-06,
+      "loss": 0.7725,
+      "step": 4826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.377194866424412e-06,
+      "loss": 0.7711,
+      "step": 4827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.373247816897449e-06,
+      "loss": 0.8221,
+      "step": 4828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.369301684299556e-06,
+      "loss": 0.7866,
+      "step": 4829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.365356469412781e-06,
+      "loss": 0.7653,
+      "step": 4830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3614121730189795e-06,
+      "loss": 0.7072,
+      "step": 4831
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.357468795899838e-06,
+      "loss": 0.8042,
+      "step": 4832
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3535263388368475e-06,
+      "loss": 0.7904,
+      "step": 4833
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.349584802611332e-06,
+      "loss": 0.7831,
+      "step": 4834
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.345644188004417e-06,
+      "loss": 0.7178,
+      "step": 4835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.341704495797053e-06,
+      "loss": 0.8319,
+      "step": 4836
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.337765726770016e-06,
+      "loss": 0.7243,
+      "step": 4837
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.333827881703879e-06,
+      "loss": 0.738,
+      "step": 4838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.329890961379054e-06,
+      "loss": 0.8053,
+      "step": 4839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32595496657575e-06,
+      "loss": 0.7288,
+      "step": 4840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.322019898074008e-06,
+      "loss": 0.8251,
+      "step": 4841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.318085756653681e-06,
+      "loss": 0.7963,
+      "step": 4842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.31415254309443e-06,
+      "loss": 0.7669,
+      "step": 4843
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.310220258175746e-06,
+      "loss": 0.8478,
+      "step": 4844
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.30628890267692e-06,
+      "loss": 0.7089,
+      "step": 4845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.302358477377077e-06,
+      "loss": 0.7666,
+      "step": 4846
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.298428983055141e-06,
+      "loss": 0.6639,
+      "step": 4847
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.294500420489859e-06,
+      "loss": 0.7568,
+      "step": 4848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2905727904598e-06,
+      "loss": 0.7731,
+      "step": 4849
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.286646093743332e-06,
+      "loss": 0.8044,
+      "step": 4850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.282720331118654e-06,
+      "loss": 0.7198,
+      "step": 4851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.278795503363767e-06,
+      "loss": 0.7499,
+      "step": 4852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274871611256502e-06,
+      "loss": 0.7965,
+      "step": 4853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.270948655574484e-06,
+      "loss": 0.8184,
+      "step": 4854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.26702663709517e-06,
+      "loss": 0.8133,
+      "step": 4855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2631055565958266e-06,
+      "loss": 0.8552,
+      "step": 4856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.259185414853528e-06,
+      "loss": 0.7684,
+      "step": 4857
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.255266212645173e-06,
+      "loss": 0.3151,
+      "step": 4858
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.251347950747463e-06,
+      "loss": 0.7997,
+      "step": 4859
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.247430629936925e-06,
+      "loss": 0.8121,
+      "step": 4860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.243514250989885e-06,
+      "loss": 0.7846,
+      "step": 4861
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239598814682494e-06,
+      "loss": 0.6934,
+      "step": 4862
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.235684321790719e-06,
+      "loss": 0.8671,
+      "step": 4863
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.231770773090321e-06,
+      "loss": 0.8161,
+      "step": 4864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.227858169356899e-06,
+      "loss": 0.7659,
+      "step": 4865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.223946511365843e-06,
+      "loss": 0.7113,
+      "step": 4866
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.220035799892371e-06,
+      "loss": 0.7857,
+      "step": 4867
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.216126035711502e-06,
+      "loss": 0.8367,
+      "step": 4868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2122172195980744e-06,
+      "loss": 0.7584,
+      "step": 4869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.208309352326741e-06,
+      "loss": 0.7733,
+      "step": 4870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.204402434671956e-06,
+      "loss": 0.8348,
+      "step": 4871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.200496467407998e-06,
+      "loss": 0.7364,
+      "step": 4872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.196591451308943e-06,
+      "loss": 0.7592,
+      "step": 4873
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.192687387148696e-06,
+      "loss": 0.7503,
+      "step": 4874
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.188784275700959e-06,
+      "loss": 0.7967,
+      "step": 4875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.184882117739245e-06,
+      "loss": 0.7066,
+      "step": 4876
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.180980914036894e-06,
+      "loss": 0.7938,
+      "step": 4877
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.177080665367036e-06,
+      "loss": 0.7506,
+      "step": 4878
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.173181372502628e-06,
+      "loss": 0.7365,
+      "step": 4879
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.169283036216426e-06,
+      "loss": 0.7147,
+      "step": 4880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.165385657281005e-06,
+      "loss": 0.7805,
+      "step": 4881
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.161489236468752e-06,
+      "loss": 0.7538,
+      "step": 4882
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.157593774551849e-06,
+      "loss": 0.7932,
+      "step": 4883
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.153699272302307e-06,
+      "loss": 0.7961,
+      "step": 4884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.14980573049193e-06,
+      "loss": 0.7599,
+      "step": 4885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.145913149892348e-06,
+      "loss": 0.8612,
+      "step": 4886
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.142021531274982e-06,
+      "loss": 0.7406,
+      "step": 4887
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138130875411079e-06,
+      "loss": 0.7867,
+      "step": 4888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.134241183071692e-06,
+      "loss": 0.7937,
+      "step": 4889
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130352455027671e-06,
+      "loss": 0.7677,
+      "step": 4890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1264646920496905e-06,
+      "loss": 0.7631,
+      "step": 4891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.122577894908221e-06,
+      "loss": 0.7979,
+      "step": 4892
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.118692064373555e-06,
+      "loss": 0.361,
+      "step": 4893
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1148072012157765e-06,
+      "loss": 0.3392,
+      "step": 4894
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.110923306204792e-06,
+      "loss": 0.785,
+      "step": 4895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.107040380110315e-06,
+      "loss": 0.7835,
+      "step": 4896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.103158423701856e-06,
+      "loss": 0.778,
+      "step": 4897
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.099277437748747e-06,
+      "loss": 0.7637,
+      "step": 4898
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.095397423020115e-06,
+      "loss": 0.7945,
+      "step": 4899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.091518380284908e-06,
+      "loss": 0.7881,
+      "step": 4900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.087640310311866e-06,
+      "loss": 0.75,
+      "step": 4901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.083763213869547e-06,
+      "loss": 0.7515,
+      "step": 4902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.07988709172632e-06,
+      "loss": 0.7892,
+      "step": 4903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.076011944650345e-06,
+      "loss": 0.8156,
+      "step": 4904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072137773409605e-06,
+      "loss": 0.7985,
+      "step": 4905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.068264578771878e-06,
+      "loss": 0.3433,
+      "step": 4906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.064392361504756e-06,
+      "loss": 0.8054,
+      "step": 4907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.060521122375632e-06,
+      "loss": 0.6875,
+      "step": 4908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.056650862151709e-06,
+      "loss": 0.8022,
+      "step": 4909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.052781581599997e-06,
+      "loss": 0.8107,
+      "step": 4910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.048913281487304e-06,
+      "loss": 0.7729,
+      "step": 4911
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045045962580255e-06,
+      "loss": 0.7418,
+      "step": 4912
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.041179625645269e-06,
+      "loss": 0.7361,
+      "step": 4913
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.037314271448583e-06,
+      "loss": 0.7205,
+      "step": 4914
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.033449900756225e-06,
+      "loss": 0.7782,
+      "step": 4915
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.029586514334043e-06,
+      "loss": 0.8241,
+      "step": 4916
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.025724112947677e-06,
+      "loss": 0.7274,
+      "step": 4917
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.021862697362575e-06,
+      "loss": 0.6691,
+      "step": 4918
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.018002268343999e-06,
+      "loss": 0.7588,
+      "step": 4919
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.014142826657e-06,
+      "loss": 0.7519,
+      "step": 4920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.010284373066446e-06,
+      "loss": 0.718,
+      "step": 4921
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.006426908337008e-06,
+      "loss": 0.6765,
+      "step": 4922
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.002570433233152e-06,
+      "loss": 0.2949,
+      "step": 4923
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.998714948519159e-06,
+      "loss": 0.7165,
+      "step": 4924
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9948604549591015e-06,
+      "loss": 0.7737,
+      "step": 4925
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.99100695331687e-06,
+      "loss": 0.886,
+      "step": 4926
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.987154444356145e-06,
+      "loss": 0.7115,
+      "step": 4927
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9833029288404174e-06,
+      "loss": 0.7821,
+      "step": 4928
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9794524075329855e-06,
+      "loss": 0.6949,
+      "step": 4929
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.975602881196937e-06,
+      "loss": 0.7382,
+      "step": 4930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.971754350595178e-06,
+      "loss": 0.8267,
+      "step": 4931
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.967906816490401e-06,
+      "loss": 0.82,
+      "step": 4932
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.96406027964512e-06,
+      "loss": 0.7682,
+      "step": 4933
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9602147408216305e-06,
+      "loss": 0.8143,
+      "step": 4934
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.956370200782047e-06,
+      "loss": 0.782,
+      "step": 4935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.952526660288283e-06,
+      "loss": 0.7746,
+      "step": 4936
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.948684120102042e-06,
+      "loss": 0.7722,
+      "step": 4937
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9448425809848464e-06,
+      "loss": 0.7517,
+      "step": 4938
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.941002043698006e-06,
+      "loss": 0.3162,
+      "step": 4939
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.937162509002642e-06,
+      "loss": 0.7841,
+      "step": 4940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.933323977659668e-06,
+      "loss": 0.7539,
+      "step": 4941
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.929486450429811e-06,
+      "loss": 0.7113,
+      "step": 4942
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.925649928073582e-06,
+      "loss": 0.8582,
+      "step": 4943
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.92181441135131e-06,
+      "loss": 0.7876,
+      "step": 4944
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.917979901023116e-06,
+      "loss": 0.7465,
+      "step": 4945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.914146397848919e-06,
+      "loss": 0.7444,
+      "step": 4946
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.910313902588448e-06,
+      "loss": 0.6726,
+      "step": 4947
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.906482416001219e-06,
+      "loss": 0.7857,
+      "step": 4948
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.902651938846563e-06,
+      "loss": 0.7868,
+      "step": 4949
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.898822471883596e-06,
+      "loss": 0.722,
+      "step": 4950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.894994015871245e-06,
+      "loss": 0.8134,
+      "step": 4951
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.891166571568237e-06,
+      "loss": 0.3158,
+      "step": 4952
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8873401397330865e-06,
+      "loss": 0.7718,
+      "step": 4953
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.883514721124122e-06,
+      "loss": 0.7869,
+      "step": 4954
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.879690316499457e-06,
+      "loss": 0.7417,
+      "step": 4955
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.87586692661702e-06,
+      "loss": 0.7535,
+      "step": 4956
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.872044552234521e-06,
+      "loss": 0.6803,
+      "step": 4957
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8682231941094836e-06,
+      "loss": 0.7663,
+      "step": 4958
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.86440285299923e-06,
+      "loss": 0.7583,
+      "step": 4959
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.86058352966086e-06,
+      "loss": 0.6185,
+      "step": 4960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856765224851292e-06,
+      "loss": 0.8097,
+      "step": 4961
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852947939327244e-06,
+      "loss": 0.7209,
+      "step": 4962
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.849131673845217e-06,
+      "loss": 0.7881,
+      "step": 4963
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.845316429161523e-06,
+      "loss": 0.3101,
+      "step": 4964
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.841502206032262e-06,
+      "loss": 0.8547,
+      "step": 4965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8376890052133414e-06,
+      "loss": 0.7916,
+      "step": 4966
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.833876827460457e-06,
+      "loss": 0.6997,
+      "step": 4967
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.830065673529105e-06,
+      "loss": 0.7967,
+      "step": 4968
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.826255544174585e-06,
+      "loss": 0.8195,
+      "step": 4969
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.82244644015198e-06,
+      "loss": 0.7508,
+      "step": 4970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.818638362216186e-06,
+      "loss": 0.7669,
+      "step": 4971
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.814831311121879e-06,
+      "loss": 0.8497,
+      "step": 4972
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8110252876235485e-06,
+      "loss": 0.7798,
+      "step": 4973
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.807220292475463e-06,
+      "loss": 0.7974,
+      "step": 4974
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.803416326431704e-06,
+      "loss": 0.6958,
+      "step": 4975
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.799613390246132e-06,
+      "loss": 0.7705,
+      "step": 4976
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7958114846724176e-06,
+      "loss": 0.7007,
+      "step": 4977
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.792010610464025e-06,
+      "loss": 0.3546,
+      "step": 4978
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.788210768374204e-06,
+      "loss": 0.7957,
+      "step": 4979
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.784411959156014e-06,
+      "loss": 0.7529,
+      "step": 4980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.780614183562295e-06,
+      "loss": 0.7892,
+      "step": 4981
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.776817442345697e-06,
+      "loss": 0.7807,
+      "step": 4982
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.77302173625865e-06,
+      "loss": 0.7963,
+      "step": 4983
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.76922706605339e-06,
+      "loss": 0.8142,
+      "step": 4984
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7654334324819485e-06,
+      "loss": 0.7478,
+      "step": 4985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.761640836296139e-06,
+      "loss": 0.706,
+      "step": 4986
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.757849278247586e-06,
+      "loss": 0.7996,
+      "step": 4987
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.754058759087693e-06,
+      "loss": 0.7235,
+      "step": 4988
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.75026927956767e-06,
+      "loss": 0.7843,
+      "step": 4989
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.746480840438509e-06,
+      "loss": 0.7597,
+      "step": 4990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742693442451006e-06,
+      "loss": 0.8146,
+      "step": 4991
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7389070863557514e-06,
+      "loss": 0.6598,
+      "step": 4992
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.735121772903117e-06,
+      "loss": 0.7793,
+      "step": 4993
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.731337502843283e-06,
+      "loss": 0.7989,
+      "step": 4994
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.727554276926208e-06,
+      "loss": 0.7558,
+      "step": 4995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.723772095901656e-06,
+      "loss": 0.7678,
+      "step": 4996
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.719990960519176e-06,
+      "loss": 0.7309,
+      "step": 4997
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.716210871528115e-06,
+      "loss": 0.8171,
+      "step": 4998
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.712431829677614e-06,
+      "loss": 0.3469,
+      "step": 4999
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.708653835716595e-06,
+      "loss": 0.8226,
+      "step": 5000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.704876890393791e-06,
+      "loss": 0.7924,
+      "step": 5001
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.701100994457703e-06,
+      "loss": 0.7847,
+      "step": 5002
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.697326148656645e-06,
+      "loss": 0.6801,
+      "step": 5003
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.693552353738718e-06,
+      "loss": 0.7646,
+      "step": 5004
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.689779610451806e-06,
+      "loss": 0.7454,
+      "step": 5005
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.686007919543596e-06,
+      "loss": 0.6668,
+      "step": 5006
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.682237281761556e-06,
+      "loss": 0.7699,
+      "step": 5007
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6784676978529565e-06,
+      "loss": 0.7784,
+      "step": 5008
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6746991685648455e-06,
+      "loss": 0.7632,
+      "step": 5009
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.670931694644074e-06,
+      "loss": 0.7692,
+      "step": 5010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667165276837284e-06,
+      "loss": 0.8312,
+      "step": 5011
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663399915890895e-06,
+      "loss": 0.8309,
+      "step": 5012
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659635612551133e-06,
+      "loss": 0.8114,
+      "step": 5013
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.655872367564e-06,
+      "loss": 0.8131,
+      "step": 5014
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.652110181675303e-06,
+      "loss": 0.81,
+      "step": 5015
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6483490556306234e-06,
+      "loss": 0.8213,
+      "step": 5016
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.644588990175345e-06,
+      "loss": 0.7543,
+      "step": 5017
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64082998605464e-06,
+      "loss": 0.3629,
+      "step": 5018
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637072044013461e-06,
+      "loss": 0.8221,
+      "step": 5019
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.633315164796564e-06,
+      "loss": 0.798,
+      "step": 5020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629559349148477e-06,
+      "loss": 0.3466,
+      "step": 5021
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.625804597813536e-06,
+      "loss": 0.7068,
+      "step": 5022
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.622050911535849e-06,
+      "loss": 0.8049,
+      "step": 5023
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.618298291059326e-06,
+      "loss": 0.772,
+      "step": 5024
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.614546737127664e-06,
+      "loss": 0.6974,
+      "step": 5025
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610796250484337e-06,
+      "loss": 0.7715,
+      "step": 5026
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6070468318726255e-06,
+      "loss": 0.8271,
+      "step": 5027
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.603298482035579e-06,
+      "loss": 0.7655,
+      "step": 5028
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599551201716052e-06,
+      "loss": 0.8202,
+      "step": 5029
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595804991656676e-06,
+      "loss": 0.8049,
+      "step": 5030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.592059852599874e-06,
+      "loss": 0.7125,
+      "step": 5031
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.588315785287864e-06,
+      "loss": 0.7861,
+      "step": 5032
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.584572790462635e-06,
+      "loss": 0.744,
+      "step": 5033
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.580830868865982e-06,
+      "loss": 0.6962,
+      "step": 5034
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.57709002123947e-06,
+      "loss": 0.7916,
+      "step": 5035
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.573350248324467e-06,
+      "loss": 0.8029,
+      "step": 5036
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.569611550862115e-06,
+      "loss": 0.7613,
+      "step": 5037
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.565873929593352e-06,
+      "loss": 0.7335,
+      "step": 5038
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5621373852589e-06,
+      "loss": 0.7734,
+      "step": 5039
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.558401918599262e-06,
+      "loss": 0.7706,
+      "step": 5040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5546675303547395e-06,
+      "loss": 0.7139,
+      "step": 5041
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.55093422126541e-06,
+      "loss": 0.7664,
+      "step": 5042
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.547201992071133e-06,
+      "loss": 0.7726,
+      "step": 5043
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543470843511572e-06,
+      "loss": 0.8088,
+      "step": 5044
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.539740776326157e-06,
+      "loss": 0.7411,
+      "step": 5045
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.536011791254119e-06,
+      "loss": 0.7653,
+      "step": 5046
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.532283889034461e-06,
+      "loss": 0.6699,
+      "step": 5047
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.528557070405986e-06,
+      "loss": 0.7005,
+      "step": 5048
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.524831336107266e-06,
+      "loss": 0.6978,
+      "step": 5049
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.52110668687667e-06,
+      "loss": 0.3781,
+      "step": 5050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.517383123452354e-06,
+      "loss": 0.705,
+      "step": 5051
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.513660646572242e-06,
+      "loss": 0.8448,
+      "step": 5052
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.509939256974065e-06,
+      "loss": 0.8289,
+      "step": 5053
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.506218955395319e-06,
+      "loss": 0.7636,
+      "step": 5054
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5024997425733005e-06,
+      "loss": 0.8189,
+      "step": 5055
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.498781619245075e-06,
+      "loss": 0.7494,
+      "step": 5056
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.495064586147502e-06,
+      "loss": 0.7971,
+      "step": 5057
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491348644017228e-06,
+      "loss": 0.8336,
+      "step": 5058
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.48763379359067e-06,
+      "loss": 0.7585,
+      "step": 5059
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.483920035604042e-06,
+      "loss": 0.7949,
+      "step": 5060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.480207370793332e-06,
+      "loss": 0.7881,
+      "step": 5061
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.476495799894322e-06,
+      "loss": 0.3848,
+      "step": 5062
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.47278532364256e-06,
+      "loss": 0.7376,
+      "step": 5063
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.469075942773394e-06,
+      "loss": 0.8165,
+      "step": 5064
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.465367658021953e-06,
+      "loss": 0.72,
+      "step": 5065
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.461660470123135e-06,
+      "loss": 0.7868,
+      "step": 5066
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.457954379811639e-06,
+      "loss": 0.7494,
+      "step": 5067
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454249387821929e-06,
+      "loss": 0.7364,
+      "step": 5068
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4505454948882666e-06,
+      "loss": 0.8321,
+      "step": 5069
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.446842701744683e-06,
+      "loss": 0.8238,
+      "step": 5070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.443141009124998e-06,
+      "loss": 0.8184,
+      "step": 5071
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.439440417762821e-06,
+      "loss": 0.7097,
+      "step": 5072
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435740928391523e-06,
+      "loss": 0.8162,
+      "step": 5073
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.432042541744277e-06,
+      "loss": 0.7269,
+      "step": 5074
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428345258554022e-06,
+      "loss": 0.7752,
+      "step": 5075
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.424649079553493e-06,
+      "loss": 0.8117,
+      "step": 5076
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.420954005475188e-06,
+      "loss": 0.7115,
+      "step": 5077
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.417260037051405e-06,
+      "loss": 0.7991,
+      "step": 5078
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4135671750142086e-06,
+      "loss": 0.8337,
+      "step": 5079
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4098754200954506e-06,
+      "loss": 0.7695,
+      "step": 5080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.406184773026768e-06,
+      "loss": 0.7567,
+      "step": 5081
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.402495234539565e-06,
+      "loss": 0.8105,
+      "step": 5082
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.39880680536504e-06,
+      "loss": 0.7683,
+      "step": 5083
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.395119486234164e-06,
+      "loss": 0.6959,
+      "step": 5084
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.391433277877683e-06,
+      "loss": 0.8127,
+      "step": 5085
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.387748181026138e-06,
+      "loss": 0.7515,
+      "step": 5086
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.384064196409834e-06,
+      "loss": 0.7754,
+      "step": 5087
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.380381324758869e-06,
+      "loss": 0.3385,
+      "step": 5088
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.376699566803106e-06,
+      "loss": 0.7821,
+      "step": 5089
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.373018923272201e-06,
+      "loss": 0.764,
+      "step": 5090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.369339394895585e-06,
+      "loss": 0.7915,
+      "step": 5091
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.365660982402461e-06,
+      "loss": 0.762,
+      "step": 5092
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3619836865218225e-06,
+      "loss": 0.7485,
+      "step": 5093
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3583075079824274e-06,
+      "loss": 0.7432,
+      "step": 5094
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.354632447512829e-06,
+      "loss": 0.7293,
+      "step": 5095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.350958505841342e-06,
+      "loss": 0.7474,
+      "step": 5096
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.347285683696072e-06,
+      "loss": 0.7849,
+      "step": 5097
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.343613981804901e-06,
+      "loss": 0.7483,
+      "step": 5098
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.339943400895479e-06,
+      "loss": 0.2934,
+      "step": 5099
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.33627394169525e-06,
+      "loss": 0.7237,
+      "step": 5100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.332605604931418e-06,
+      "loss": 0.7264,
+      "step": 5101
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3289383913309826e-06,
+      "loss": 0.7605,
+      "step": 5102
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3252723016207025e-06,
+      "loss": 0.7983,
+      "step": 5103
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.32160733652713e-06,
+      "loss": 0.7927,
+      "step": 5104
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.317943496776581e-06,
+      "loss": 0.7747,
+      "step": 5105
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31428078309516e-06,
+      "loss": 0.8624,
+      "step": 5106
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.310619196208743e-06,
+      "loss": 0.7251,
+      "step": 5107
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.306958736842978e-06,
+      "loss": 0.8733,
+      "step": 5108
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3032994057233e-06,
+      "loss": 0.7474,
+      "step": 5109
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.29964120357491e-06,
+      "loss": 0.8042,
+      "step": 5110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2959841311227945e-06,
+      "loss": 0.7566,
+      "step": 5111
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292328189091706e-06,
+      "loss": 0.8059,
+      "step": 5112
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.288673378206182e-06,
+      "loss": 0.7736,
+      "step": 5113
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.285019699190535e-06,
+      "loss": 0.7245,
+      "step": 5114
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.281367152768844e-06,
+      "loss": 0.8036,
+      "step": 5115
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277715739664977e-06,
+      "loss": 0.7481,
+      "step": 5116
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.274065460602565e-06,
+      "loss": 0.3632,
+      "step": 5117
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2704163163050235e-06,
+      "loss": 0.7972,
+      "step": 5118
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.266768307495535e-06,
+      "loss": 0.7741,
+      "step": 5119
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.263121434897065e-06,
+      "loss": 0.8549,
+      "step": 5120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.259475699232353e-06,
+      "loss": 0.7175,
+      "step": 5121
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.255831101223903e-06,
+      "loss": 0.886,
+      "step": 5122
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.252187641594007e-06,
+      "loss": 0.7919,
+      "step": 5123
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2485453210647206e-06,
+      "loss": 0.3637,
+      "step": 5124
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.244904140357885e-06,
+      "loss": 0.6977,
+      "step": 5125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.241264100195104e-06,
+      "loss": 0.7735,
+      "step": 5126
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237625201297757e-06,
+      "loss": 0.7598,
+      "step": 5127
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.233987444387008e-06,
+      "loss": 0.7008,
+      "step": 5128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230350830183779e-06,
+      "loss": 0.7103,
+      "step": 5129
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2267153594087815e-06,
+      "loss": 0.364,
+      "step": 5130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2230810327824866e-06,
+      "loss": 0.6745,
+      "step": 5131
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2194478510251465e-06,
+      "loss": 0.3333,
+      "step": 5132
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.215815814856789e-06,
+      "loss": 0.7525,
+      "step": 5133
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2121849249972035e-06,
+      "loss": 0.3571,
+      "step": 5134
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208555182165967e-06,
+      "loss": 0.7773,
+      "step": 5135
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2049265870824105e-06,
+      "loss": 0.3442,
+      "step": 5136
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.201299140465661e-06,
+      "loss": 0.7386,
+      "step": 5137
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.197672843034595e-06,
+      "loss": 0.7902,
+      "step": 5138
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.194047695507875e-06,
+      "loss": 0.8469,
+      "step": 5139
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.190423698603937e-06,
+      "loss": 0.6935,
+      "step": 5140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.186800853040977e-06,
+      "loss": 0.7868,
+      "step": 5141
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.183179159536975e-06,
+      "loss": 0.7981,
+      "step": 5142
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.179558618809674e-06,
+      "loss": 0.7771,
+      "step": 5143
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.175939231576597e-06,
+      "loss": 0.7531,
+      "step": 5144
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172320998555028e-06,
+      "loss": 0.8755,
+      "step": 5145
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.16870392046203e-06,
+      "loss": 0.774,
+      "step": 5146
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.165087998014438e-06,
+      "loss": 0.7378,
+      "step": 5147
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.161473231928849e-06,
+      "loss": 0.7561,
+      "step": 5148
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.157859622921646e-06,
+      "loss": 0.7009,
+      "step": 5149
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.154247171708964e-06,
+      "loss": 0.8488,
+      "step": 5150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1506358790067245e-06,
+      "loss": 0.7326,
+      "step": 5151
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.147025745530608e-06,
+      "loss": 0.7674,
+      "step": 5152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.143416771996071e-06,
+      "loss": 0.7882,
+      "step": 5153
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1398089591183455e-06,
+      "loss": 0.7346,
+      "step": 5154
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.136202307612418e-06,
+      "loss": 0.7904,
+      "step": 5155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.132596818193064e-06,
+      "loss": 0.6874,
+      "step": 5156
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.128992491574811e-06,
+      "loss": 0.7892,
+      "step": 5157
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.125389328471968e-06,
+      "loss": 0.7209,
+      "step": 5158
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.121787329598605e-06,
+      "loss": 0.8225,
+      "step": 5159
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.11818649566857e-06,
+      "loss": 0.7348,
+      "step": 5160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.114586827395478e-06,
+      "loss": 0.3525,
+      "step": 5161
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.110988325492703e-06,
+      "loss": 0.7731,
+      "step": 5162
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1073909906734035e-06,
+      "loss": 0.7929,
+      "step": 5163
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.103794823650491e-06,
+      "loss": 0.7583,
+      "step": 5164
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.100199825136662e-06,
+      "loss": 0.8165,
+      "step": 5165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.096605995844364e-06,
+      "loss": 0.6996,
+      "step": 5166
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093013336485825e-06,
+      "loss": 0.7083,
+      "step": 5167
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.089421847773047e-06,
+      "loss": 0.752,
+      "step": 5168
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.085831530417774e-06,
+      "loss": 0.8235,
+      "step": 5169
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.082242385131546e-06,
+      "loss": 0.7176,
+      "step": 5170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.078654412625653e-06,
+      "loss": 0.7391,
+      "step": 5171
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0750676136111625e-06,
+      "loss": 0.7776,
+      "step": 5172
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.071481988798907e-06,
+      "loss": 0.7443,
+      "step": 5173
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.067897538899483e-06,
+      "loss": 0.8161,
+      "step": 5174
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.064314264623257e-06,
+      "loss": 0.7731,
+      "step": 5175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.060732166680359e-06,
+      "loss": 0.7185,
+      "step": 5176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.057151245780694e-06,
+      "loss": 0.7738,
+      "step": 5177
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.053571502633922e-06,
+      "loss": 0.8087,
+      "step": 5178
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.049992937949479e-06,
+      "loss": 0.7132,
+      "step": 5179
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.046415552436569e-06,
+      "loss": 0.774,
+      "step": 5180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.042839346804149e-06,
+      "loss": 0.7228,
+      "step": 5181
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039264321760959e-06,
+      "loss": 0.8408,
+      "step": 5182
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.035690478015491e-06,
+      "loss": 0.818,
+      "step": 5183
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.032117816276014e-06,
+      "loss": 0.7516,
+      "step": 5184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.028546337250552e-06,
+      "loss": 0.8256,
+      "step": 5185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.024976041646902e-06,
+      "loss": 0.7588,
+      "step": 5186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.021406930172629e-06,
+      "loss": 0.7863,
+      "step": 5187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.017839003535052e-06,
+      "loss": 0.8373,
+      "step": 5188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.01427226244127e-06,
+      "loss": 0.7587,
+      "step": 5189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.010706707598131e-06,
+      "loss": 0.7542,
+      "step": 5190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.007142339712263e-06,
+      "loss": 0.8116,
+      "step": 5191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.003579159490046e-06,
+      "loss": 0.8066,
+      "step": 5192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0000171676376345e-06,
+      "loss": 0.7311,
+      "step": 5193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.996456364860944e-06,
+      "loss": 0.8042,
+      "step": 5194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.992896751865652e-06,
+      "loss": 0.8097,
+      "step": 5195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.989338329357203e-06,
+      "loss": 0.8045,
+      "step": 5196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9857810980408014e-06,
+      "loss": 0.856,
+      "step": 5197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982225058621426e-06,
+      "loss": 0.8798,
+      "step": 5198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978670211803805e-06,
+      "loss": 0.334,
+      "step": 5199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.975116558292438e-06,
+      "loss": 0.7194,
+      "step": 5200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.971564098791595e-06,
+      "loss": 0.749,
+      "step": 5201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.968012834005291e-06,
+      "loss": 0.7923,
+      "step": 5202
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964462764637327e-06,
+      "loss": 0.7642,
+      "step": 5203
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.960913891391244e-06,
+      "loss": 0.7649,
+      "step": 5204
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.957366214970365e-06,
+      "loss": 0.789,
+      "step": 5205
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.953819736077763e-06,
+      "loss": 0.7248,
+      "step": 5206
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.95027445541628e-06,
+      "loss": 0.8116,
+      "step": 5207
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.946730373688523e-06,
+      "loss": 0.8033,
+      "step": 5208
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.943187491596852e-06,
+      "loss": 0.8117,
+      "step": 5209
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.939645809843404e-06,
+      "loss": 0.7507,
+      "step": 5210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.936105329130056e-06,
+      "loss": 0.7326,
+      "step": 5211
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932566050158466e-06,
+      "loss": 0.7484,
+      "step": 5212
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.929027973630052e-06,
+      "loss": 0.8269,
+      "step": 5213
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925491100245982e-06,
+      "loss": 0.8123,
+      "step": 5214
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.921955430707199e-06,
+      "loss": 0.7676,
+      "step": 5215
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.918420965714398e-06,
+      "loss": 0.7493,
+      "step": 5216
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.914887705968041e-06,
+      "loss": 0.7371,
+      "step": 5217
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911355652168346e-06,
+      "loss": 0.3384,
+      "step": 5218
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907824805015297e-06,
+      "loss": 0.7279,
+      "step": 5219
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.904295165208639e-06,
+      "loss": 0.7518,
+      "step": 5220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.900766733447869e-06,
+      "loss": 0.7974,
+      "step": 5221
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8972395104322605e-06,
+      "loss": 0.7392,
+      "step": 5222
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893713496860827e-06,
+      "loss": 0.3554,
+      "step": 5223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.890188693432364e-06,
+      "loss": 0.7071,
+      "step": 5224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.886665100845408e-06,
+      "loss": 0.6888,
+      "step": 5225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8831427197982665e-06,
+      "loss": 0.7185,
+      "step": 5226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.87962155098901e-06,
+      "loss": 0.8148,
+      "step": 5227
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.876101595115455e-06,
+      "loss": 0.8001,
+      "step": 5228
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.872582852875193e-06,
+      "loss": 0.9016,
+      "step": 5229
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.869065324965561e-06,
+      "loss": 0.8109,
+      "step": 5230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.865549012083669e-06,
+      "loss": 0.364,
+      "step": 5231
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.862033914926373e-06,
+      "loss": 0.765,
+      "step": 5232
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.858520034190297e-06,
+      "loss": 0.7873,
+      "step": 5233
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855007370571825e-06,
+      "loss": 0.7917,
+      "step": 5234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.85149592476709e-06,
+      "loss": 0.7152,
+      "step": 5235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.847985697471995e-06,
+      "loss": 0.732,
+      "step": 5236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.844476689382193e-06,
+      "loss": 0.781,
+      "step": 5237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.840968901193101e-06,
+      "loss": 0.7299,
+      "step": 5238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.837462333599889e-06,
+      "loss": 0.7461,
+      "step": 5239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8339569872974915e-06,
+      "loss": 0.8075,
+      "step": 5240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8304528629805934e-06,
+      "loss": 0.7452,
+      "step": 5241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.826949961343645e-06,
+      "loss": 0.7619,
+      "step": 5242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.823448283080851e-06,
+      "loss": 0.7755,
+      "step": 5243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.81994782888617e-06,
+      "loss": 0.7621,
+      "step": 5244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816448599453327e-06,
+      "loss": 0.7738,
+      "step": 5245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.812950595475793e-06,
+      "loss": 0.7471,
+      "step": 5246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8094538176468064e-06,
+      "loss": 0.738,
+      "step": 5247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805958266659353e-06,
+      "loss": 0.7006,
+      "step": 5248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802463943206185e-06,
+      "loss": 0.6971,
+      "step": 5249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.798970847979808e-06,
+      "loss": 0.7583,
+      "step": 5250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7954789816724813e-06,
+      "loss": 0.7584,
+      "step": 5251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7919883449762185e-06,
+      "loss": 0.7719,
+      "step": 5252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.788498938582801e-06,
+      "loss": 0.7819,
+      "step": 5253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7850107631837498e-06,
+      "loss": 0.7675,
+      "step": 5254
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7815238194703606e-06,
+      "loss": 0.695,
+      "step": 5255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.778038108133667e-06,
+      "loss": 0.7933,
+      "step": 5256
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7745536298644734e-06,
+      "loss": 0.7414,
+      "step": 5257
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.771070385353327e-06,
+      "loss": 0.7985,
+      "step": 5258
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7675883752905408e-06,
+      "loss": 0.7565,
+      "step": 5259
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7641076003661814e-06,
+      "loss": 0.7696,
+      "step": 5260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.760628061270061e-06,
+      "loss": 0.8393,
+      "step": 5261
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7571497586917615e-06,
+      "loss": 0.8178,
+      "step": 5262
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7536726933206067e-06,
+      "loss": 0.7134,
+      "step": 5263
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7501968658456867e-06,
+      "loss": 0.8267,
+      "step": 5264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.746722276955833e-06,
+      "loss": 0.7868,
+      "step": 5265
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.743248927339647e-06,
+      "loss": 0.7867,
+      "step": 5266
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7397768176854697e-06,
+      "loss": 0.7917,
+      "step": 5267
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.736305948681406e-06,
+      "loss": 0.8075,
+      "step": 5268
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.732836321015316e-06,
+      "loss": 0.7837,
+      "step": 5269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7293679353748037e-06,
+      "loss": 0.7774,
+      "step": 5270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7259007924472403e-06,
+      "loss": 0.6903,
+      "step": 5271
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.722434892919735e-06,
+      "loss": 0.7933,
+      "step": 5272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7189702374791668e-06,
+      "loss": 0.7683,
+      "step": 5273
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7155068268121553e-06,
+      "loss": 0.8011,
+      "step": 5274
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.712044661605081e-06,
+      "loss": 0.7249,
+      "step": 5275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7085837425440785e-06,
+      "loss": 0.7278,
+      "step": 5276
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.705124070315026e-06,
+      "loss": 0.7764,
+      "step": 5277
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.701665645603567e-06,
+      "loss": 0.7426,
+      "step": 5278
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.698208469095086e-06,
+      "loss": 0.7818,
+      "step": 5279
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.694752541474731e-06,
+      "loss": 0.7062,
+      "step": 5280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.691297863427391e-06,
+      "loss": 0.8058,
+      "step": 5281
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6878444356377176e-06,
+      "loss": 0.747,
+      "step": 5282
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.684392258790114e-06,
+      "loss": 0.8433,
+      "step": 5283
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6809413335687226e-06,
+      "loss": 0.8489,
+      "step": 5284
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6774916606574583e-06,
+      "loss": 0.7847,
+      "step": 5285
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.674043240739967e-06,
+      "loss": 0.6889,
+      "step": 5286
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6705960744996647e-06,
+      "loss": 0.7328,
+      "step": 5287
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6671501626197025e-06,
+      "loss": 0.7777,
+      "step": 5288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6637055057829927e-06,
+      "loss": 0.7096,
+      "step": 5289
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.660262104672203e-06,
+      "loss": 0.7858,
+      "step": 5290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6568199599697383e-06,
+      "loss": 0.8177,
+      "step": 5291
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.653379072357769e-06,
+      "loss": 0.7082,
+      "step": 5292
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6499394425182056e-06,
+      "loss": 0.7715,
+      "step": 5293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.646501071132712e-06,
+      "loss": 0.8429,
+      "step": 5294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.643063958882709e-06,
+      "loss": 0.7252,
+      "step": 5295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6396281064493556e-06,
+      "loss": 0.7628,
+      "step": 5296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6361935145135772e-06,
+      "loss": 0.7951,
+      "step": 5297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6327601837560344e-06,
+      "loss": 0.7825,
+      "step": 5298
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.629328114857149e-06,
+      "loss": 0.7826,
+      "step": 5299
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.625897308497083e-06,
+      "loss": 0.736,
+      "step": 5300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6224677653557537e-06,
+      "loss": 0.7247,
+      "step": 5301
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.619039486112834e-06,
+      "loss": 0.8466,
+      "step": 5302
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6156124714477312e-06,
+      "loss": 0.8029,
+      "step": 5303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6121867220396177e-06,
+      "loss": 0.7785,
+      "step": 5304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6087622385674025e-06,
+      "loss": 0.7268,
+      "step": 5305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.605339021709754e-06,
+      "loss": 0.6869,
+      "step": 5306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.601917072145078e-06,
+      "loss": 0.7567,
+      "step": 5307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5984963905515415e-06,
+      "loss": 0.7503,
+      "step": 5308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.595076977607056e-06,
+      "loss": 0.3437,
+      "step": 5309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5916588339892743e-06,
+      "loss": 0.8099,
+      "step": 5310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5882419603756112e-06,
+      "loss": 0.8378,
+      "step": 5311
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.584826357443214e-06,
+      "loss": 0.7881,
+      "step": 5312
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5814120258689944e-06,
+      "loss": 0.7978,
+      "step": 5313
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5779989663295955e-06,
+      "loss": 0.6737,
+      "step": 5314
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.574587179501423e-06,
+      "loss": 0.7923,
+      "step": 5315
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.571176666060625e-06,
+      "loss": 0.7303,
+      "step": 5316
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.567767426683092e-06,
+      "loss": 0.7766,
+      "step": 5317
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.564359462044472e-06,
+      "loss": 0.7645,
+      "step": 5318
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5609527728201486e-06,
+      "loss": 0.7152,
+      "step": 5319
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.557547359685265e-06,
+      "loss": 0.762,
+      "step": 5320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5541432233147e-06,
+      "loss": 0.7432,
+      "step": 5321
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.550740364383087e-06,
+      "loss": 0.74,
+      "step": 5322
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5473387835648097e-06,
+      "loss": 0.6264,
+      "step": 5323
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.543938481533983e-06,
+      "loss": 0.8147,
+      "step": 5324
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.540539458964487e-06,
+      "loss": 0.7189,
+      "step": 5325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5371417165299327e-06,
+      "loss": 0.7605,
+      "step": 5326
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.53374525490369e-06,
+      "loss": 0.7602,
+      "step": 5327
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.530350074758863e-06,
+      "loss": 0.7806,
+      "step": 5328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5269561767683103e-06,
+      "loss": 0.7405,
+      "step": 5329
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5235635616046404e-06,
+      "loss": 0.7965,
+      "step": 5330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5201722299401906e-06,
+      "loss": 0.351,
+      "step": 5331
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5167821824470647e-06,
+      "loss": 0.7675,
+      "step": 5332
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5133934197970932e-06,
+      "loss": 0.724,
+      "step": 5333
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.510005942661868e-06,
+      "loss": 0.8457,
+      "step": 5334
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5066197517127153e-06,
+      "loss": 0.725,
+      "step": 5335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5032348476207067e-06,
+      "loss": 0.7471,
+      "step": 5336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4998512310566677e-06,
+      "loss": 0.7893,
+      "step": 5337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.496468902691157e-06,
+      "loss": 0.7345,
+      "step": 5338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.493087863194492e-06,
+      "loss": 0.7404,
+      "step": 5339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4897081132367172e-06,
+      "loss": 0.7398,
+      "step": 5340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4863296534876367e-06,
+      "loss": 0.7711,
+      "step": 5341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.482952484616795e-06,
+      "loss": 0.8137,
+      "step": 5342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4795766072934724e-06,
+      "loss": 0.7754,
+      "step": 5343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.476202022186708e-06,
+      "loss": 0.7634,
+      "step": 5344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4728287299652673e-06,
+      "loss": 0.6701,
+      "step": 5345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.469456731297677e-06,
+      "loss": 0.7346,
+      "step": 5346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4660860268521936e-06,
+      "loss": 0.7448,
+      "step": 5347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.462716617296824e-06,
+      "loss": 0.7876,
+      "step": 5348
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4593485032993223e-06,
+      "loss": 0.7965,
+      "step": 5349
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4559816855271742e-06,
+      "loss": 0.7331,
+      "step": 5350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.452616164647621e-06,
+      "loss": 0.7464,
+      "step": 5351
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.449251941327634e-06,
+      "loss": 0.7056,
+      "step": 5352
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.445889016233942e-06,
+      "loss": 0.7065,
+      "step": 5353
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4425273900330026e-06,
+      "loss": 0.7782,
+      "step": 5354
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4391670633910257e-06,
+      "loss": 0.906,
+      "step": 5355
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4358080369739622e-06,
+      "loss": 0.7074,
+      "step": 5356
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.432450311447498e-06,
+      "loss": 0.7723,
+      "step": 5357
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4290938874770742e-06,
+      "loss": 0.7923,
+      "step": 5358
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4257387657278584e-06,
+      "loss": 0.7723,
+      "step": 5359
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.422384946864775e-06,
+      "loss": 0.7948,
+      "step": 5360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.419032431552477e-06,
+      "loss": 0.7688,
+      "step": 5361
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.415681220455368e-06,
+      "loss": 0.7077,
+      "step": 5362
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4123313142375947e-06,
+      "loss": 0.6746,
+      "step": 5363
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.408982713563034e-06,
+      "loss": 0.707,
+      "step": 5364
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4056354190953178e-06,
+      "loss": 0.6719,
+      "step": 5365
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4022894314978062e-06,
+      "loss": 0.7523,
+      "step": 5366
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3989447514336127e-06,
+      "loss": 0.7896,
+      "step": 5367
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3956013795655785e-06,
+      "loss": 0.8587,
+      "step": 5368
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.392259316556297e-06,
+      "loss": 0.6994,
+      "step": 5369
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3889185630680997e-06,
+      "loss": 0.8217,
+      "step": 5370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3855791197630505e-06,
+      "loss": 0.7603,
+      "step": 5371
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3822409873029673e-06,
+      "loss": 0.7305,
+      "step": 5372
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3789041663493927e-06,
+      "loss": 0.7208,
+      "step": 5373
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.375568657563626e-06,
+      "loss": 0.7667,
+      "step": 5374
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3722344616066882e-06,
+      "loss": 0.6838,
+      "step": 5375
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3689015791393585e-06,
+      "loss": 0.7488,
+      "step": 5376
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3655700108221444e-06,
+      "loss": 0.7516,
+      "step": 5377
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.362239757315291e-06,
+      "loss": 0.8037,
+      "step": 5378
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3589108192787934e-06,
+      "loss": 0.8084,
+      "step": 5379
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.355583197372375e-06,
+      "loss": 0.7583,
+      "step": 5380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3522568922555053e-06,
+      "loss": 0.7728,
+      "step": 5381
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3489319045873957e-06,
+      "loss": 0.7204,
+      "step": 5382
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.345608235026985e-06,
+      "loss": 0.797,
+      "step": 5383
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.342285884232963e-06,
+      "loss": 0.7966,
+      "step": 5384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3389648528637476e-06,
+      "loss": 0.8129,
+      "step": 5385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.335645141577505e-06,
+      "loss": 0.7515,
+      "step": 5386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.33232675103213e-06,
+      "loss": 0.8382,
+      "step": 5387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3290096818852634e-06,
+      "loss": 0.8455,
+      "step": 5388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3256939347942852e-06,
+      "loss": 0.7134,
+      "step": 5389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3223795104163025e-06,
+      "loss": 0.8123,
+      "step": 5390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3190664094081747e-06,
+      "loss": 0.851,
+      "step": 5391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3157546324264856e-06,
+      "loss": 0.709,
+      "step": 5392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3124441801275674e-06,
+      "loss": 0.7961,
+      "step": 5393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3091350531674802e-06,
+      "loss": 0.7452,
+      "step": 5394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3058272522020283e-06,
+      "loss": 0.7475,
+      "step": 5395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.302520777886755e-06,
+      "loss": 0.7738,
+      "step": 5396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2992156308769318e-06,
+      "loss": 0.8167,
+      "step": 5397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2959118118275756e-06,
+      "loss": 0.7649,
+      "step": 5398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.292609321393434e-06,
+      "loss": 0.7835,
+      "step": 5399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2893081602289966e-06,
+      "loss": 0.6748,
+      "step": 5400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2860083289884837e-06,
+      "loss": 0.7478,
+      "step": 5401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2827098283258608e-06,
+      "loss": 0.7407,
+      "step": 5402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.279412658894818e-06,
+      "loss": 0.7944,
+      "step": 5403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2761168213487906e-06,
+      "loss": 0.7813,
+      "step": 5404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.272822316340951e-06,
+      "loss": 0.7968,
+      "step": 5405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.269529144524196e-06,
+      "loss": 0.8931,
+      "step": 5406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2662373065511743e-06,
+      "loss": 0.8072,
+      "step": 5407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.262946803074254e-06,
+      "loss": 0.7501,
+      "step": 5408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2596576347455535e-06,
+      "loss": 0.8241,
+      "step": 5409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.256369802216913e-06,
+      "loss": 0.7905,
+      "step": 5410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.253083306139918e-06,
+      "loss": 0.7025,
+      "step": 5411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.24979814716589e-06,
+      "loss": 0.7239,
+      "step": 5412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.246514325945873e-06,
+      "loss": 0.7365,
+      "step": 5413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.243231843130663e-06,
+      "loss": 0.6927,
+      "step": 5414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.239950699370773e-06,
+      "loss": 0.8241,
+      "step": 5415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2366708953164683e-06,
+      "loss": 0.7261,
+      "step": 5416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2333924316177314e-06,
+      "loss": 0.734,
+      "step": 5417
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2301153089242966e-06,
+      "loss": 0.7684,
+      "step": 5418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.226839527885618e-06,
+      "loss": 0.7503,
+      "step": 5419
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2235650891508873e-06,
+      "loss": 0.6985,
+      "step": 5420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2202919933690357e-06,
+      "loss": 0.7327,
+      "step": 5421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2170202411887273e-06,
+      "loss": 0.7808,
+      "step": 5422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2137498332583527e-06,
+      "loss": 0.767,
+      "step": 5423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2104807702260455e-06,
+      "loss": 0.7114,
+      "step": 5424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2072130527396617e-06,
+      "loss": 0.7958,
+      "step": 5425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.203946681446806e-06,
+      "loss": 0.7202,
+      "step": 5426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2006816569947975e-06,
+      "loss": 0.7125,
+      "step": 5427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.197417980030706e-06,
+      "loss": 0.8532,
+      "step": 5428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1941556512013215e-06,
+      "loss": 0.7685,
+      "step": 5429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.190894671153173e-06,
+      "loss": 0.6777,
+      "step": 5430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.187635040532526e-06,
+      "loss": 0.685,
+      "step": 5431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.184376759985367e-06,
+      "loss": 0.7402,
+      "step": 5432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1811198301574262e-06,
+      "loss": 0.7456,
+      "step": 5433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.177864251694157e-06,
+      "loss": 0.7286,
+      "step": 5434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.174610025240754e-06,
+      "loss": 0.7681,
+      "step": 5435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.171357151442136e-06,
+      "loss": 0.7883,
+      "step": 5436
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.168105630942957e-06,
+      "loss": 0.7854,
+      "step": 5437
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.164855464387607e-06,
+      "loss": 0.7387,
+      "step": 5438
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.161606652420198e-06,
+      "loss": 0.8174,
+      "step": 5439
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1583591956845847e-06,
+      "loss": 0.742,
+      "step": 5440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1551130948243426e-06,
+      "loss": 0.7603,
+      "step": 5441
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1518683504827885e-06,
+      "loss": 0.7475,
+      "step": 5442
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1486249633029585e-06,
+      "loss": 0.807,
+      "step": 5443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.145382933927631e-06,
+      "loss": 0.8179,
+      "step": 5444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1421422629993127e-06,
+      "loss": 0.6998,
+      "step": 5445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1389029511602353e-06,
+      "loss": 0.7297,
+      "step": 5446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.135664999052369e-06,
+      "loss": 0.7597,
+      "step": 5447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.132428407317406e-06,
+      "loss": 0.7727,
+      "step": 5448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1291931765967786e-06,
+      "loss": 0.8099,
+      "step": 5449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1259593075316386e-06,
+      "loss": 0.8418,
+      "step": 5450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1227268007628777e-06,
+      "loss": 0.8178,
+      "step": 5451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1194956569311165e-06,
+      "loss": 0.7425,
+      "step": 5452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.116265876676695e-06,
+      "loss": 0.7247,
+      "step": 5453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1130374606396975e-06,
+      "loss": 0.709,
+      "step": 5454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1098104094599256e-06,
+      "loss": 0.7565,
+      "step": 5455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.106584723776921e-06,
+      "loss": 0.7279,
+      "step": 5456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1033604042299438e-06,
+      "loss": 0.7143,
+      "step": 5457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.100137451457993e-06,
+      "loss": 0.7414,
+      "step": 5458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0969158660997946e-06,
+      "loss": 0.689,
+      "step": 5459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0936956487937995e-06,
+      "loss": 0.7537,
+      "step": 5460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0904768001781916e-06,
+      "loss": 0.7073,
+      "step": 5461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0872593208908753e-06,
+      "loss": 0.8144,
+      "step": 5462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.084043211569495e-06,
+      "loss": 0.7528,
+      "step": 5463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0808284728514224e-06,
+      "loss": 0.7098,
+      "step": 5464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.077615105373748e-06,
+      "loss": 0.693,
+      "step": 5465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0744031097733007e-06,
+      "loss": 0.6768,
+      "step": 5466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0711924866866295e-06,
+      "loss": 0.8737,
+      "step": 5467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0679832367500196e-06,
+      "loss": 0.7459,
+      "step": 5468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0647753605994747e-06,
+      "loss": 0.875,
+      "step": 5469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0615688588707337e-06,
+      "loss": 0.8326,
+      "step": 5470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.058363732199263e-06,
+      "loss": 0.7512,
+      "step": 5471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0551599812202506e-06,
+      "loss": 0.7475,
+      "step": 5472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0519576065686186e-06,
+      "loss": 0.8218,
+      "step": 5473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0487566088790077e-06,
+      "loss": 0.8224,
+      "step": 5474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0455569887857983e-06,
+      "loss": 0.3688,
+      "step": 5475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.042358746923084e-06,
+      "loss": 0.7347,
+      "step": 5476
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.039161883924695e-06,
+      "loss": 0.7675,
+      "step": 5477
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0359664004241885e-06,
+      "loss": 0.8037,
+      "step": 5478
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.032772297054838e-06,
+      "loss": 0.756,
+      "step": 5479
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.029579574449657e-06,
+      "loss": 0.6518,
+      "step": 5480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.026388233241373e-06,
+      "loss": 0.7824,
+      "step": 5481
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.023198274062451e-06,
+      "loss": 0.7687,
+      "step": 5482
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.020009697545071e-06,
+      "loss": 0.3547,
+      "step": 5483
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0168225043211487e-06,
+      "loss": 0.7929,
+      "step": 5484
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0136366950223215e-06,
+      "loss": 0.8113,
+      "step": 5485
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.01045227027995e-06,
+      "loss": 0.686,
+      "step": 5486
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.007269230725127e-06,
+      "loss": 0.6967,
+      "step": 5487
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.004087576988661e-06,
+      "loss": 0.7601,
+      "step": 5488
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0009073097011e-06,
+      "loss": 0.7669,
+      "step": 5489
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.997728429492699e-06,
+      "loss": 0.8255,
+      "step": 5490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.994550936993453e-06,
+      "loss": 0.8742,
+      "step": 5491
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9913748328330815e-06,
+      "loss": 0.8062,
+      "step": 5492
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.988200117641016e-06,
+      "loss": 0.8135,
+      "step": 5493
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.985026792046428e-06,
+      "loss": 0.6625,
+      "step": 5494
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9818548566782e-06,
+      "loss": 0.8055,
+      "step": 5495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9786843121649535e-06,
+      "loss": 0.6616,
+      "step": 5496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9755151591350193e-06,
+      "loss": 0.7286,
+      "step": 5497
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.972347398216462e-06,
+      "loss": 0.7746,
+      "step": 5498
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.969181030037073e-06,
+      "loss": 0.7154,
+      "step": 5499
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9660160552243544e-06,
+      "loss": 0.6792,
+      "step": 5500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9628524744055477e-06,
+      "loss": 0.7486,
+      "step": 5501
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9596902882076083e-06,
+      "loss": 0.7067,
+      "step": 5502
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9565294972572136e-06,
+      "loss": 0.7246,
+      "step": 5503
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9533701021807748e-06,
+      "loss": 0.7738,
+      "step": 5504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9502121036044153e-06,
+      "loss": 0.7336,
+      "step": 5505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9470555021539926e-06,
+      "loss": 0.7445,
+      "step": 5506
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9439002984550747e-06,
+      "loss": 0.7089,
+      "step": 5507
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9407464931329664e-06,
+      "loss": 0.7667,
+      "step": 5508
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.937594086812682e-06,
+      "loss": 0.7628,
+      "step": 5509
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.934443080118968e-06,
+      "loss": 0.7779,
+      "step": 5510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9312934736762933e-06,
+      "loss": 0.7602,
+      "step": 5511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9281452681088398e-06,
+      "loss": 0.7688,
+      "step": 5512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.924998464040525e-06,
+      "loss": 0.7425,
+      "step": 5513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.921853062094977e-06,
+      "loss": 0.7188,
+      "step": 5514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.918709062895555e-06,
+      "loss": 0.7363,
+      "step": 5515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9155664670653326e-06,
+      "loss": 0.7761,
+      "step": 5516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.912425275227111e-06,
+      "loss": 0.7764,
+      "step": 5517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.909285488003413e-06,
+      "loss": 0.7194,
+      "step": 5518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.906147106016477e-06,
+      "loss": 0.8021,
+      "step": 5519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.903010129888272e-06,
+      "loss": 0.7314,
+      "step": 5520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8998745602404777e-06,
+      "loss": 0.7907,
+      "step": 5521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8967403976945073e-06,
+      "loss": 0.806,
+      "step": 5522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8936076428714822e-06,
+      "loss": 0.7762,
+      "step": 5523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.890476296392254e-06,
+      "loss": 0.6959,
+      "step": 5524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8873463588773966e-06,
+      "loss": 0.7364,
+      "step": 5525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8842178309471924e-06,
+      "loss": 0.7679,
+      "step": 5526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8810907132216617e-06,
+      "loss": 0.7482,
+      "step": 5527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.877965006320528e-06,
+      "loss": 0.7247,
+      "step": 5528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.874840710863249e-06,
+      "loss": 0.7405,
+      "step": 5529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8717178274689928e-06,
+      "loss": 0.732,
+      "step": 5530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8685963567566544e-06,
+      "loss": 0.7998,
+      "step": 5531
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.865476299344849e-06,
+      "loss": 0.7337,
+      "step": 5532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.862357655851904e-06,
+      "loss": 0.7626,
+      "step": 5533
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8592404268958762e-06,
+      "loss": 0.7564,
+      "step": 5534
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8561246130945343e-06,
+      "loss": 0.6935,
+      "step": 5535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8530102150653727e-06,
+      "loss": 0.3296,
+      "step": 5536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.849897233425597e-06,
+      "loss": 0.7539,
+      "step": 5537
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.846785668792145e-06,
+      "loss": 0.6769,
+      "step": 5538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8436755217816594e-06,
+      "loss": 0.7521,
+      "step": 5539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.840566793010511e-06,
+      "loss": 0.7818,
+      "step": 5540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.837459483094791e-06,
+      "loss": 0.6987,
+      "step": 5541
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8343535926502995e-06,
+      "loss": 0.728,
+      "step": 5542
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8312491222925675e-06,
+      "loss": 0.8157,
+      "step": 5543
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.828146072636835e-06,
+      "loss": 0.8005,
+      "step": 5544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8250444442980606e-06,
+      "loss": 0.7467,
+      "step": 5545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.821944237890931e-06,
+      "loss": 0.739,
+      "step": 5546
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8188454540298405e-06,
+      "loss": 0.7801,
+      "step": 5547
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8157480933289083e-06,
+      "loss": 0.7654,
+      "step": 5548
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.812652156401965e-06,
+      "loss": 0.7755,
+      "step": 5549
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.809557643862565e-06,
+      "loss": 0.7466,
+      "step": 5550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8064645563239824e-06,
+      "loss": 0.8138,
+      "step": 5551
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.803372894399198e-06,
+      "loss": 0.7971,
+      "step": 5552
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8002826587009237e-06,
+      "loss": 0.7607,
+      "step": 5553
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.797193849841574e-06,
+      "loss": 0.8086,
+      "step": 5554
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7941064684332974e-06,
+      "loss": 0.7105,
+      "step": 5555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.791020515087942e-06,
+      "loss": 0.7398,
+      "step": 5556
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7879359904170866e-06,
+      "loss": 0.7791,
+      "step": 5557
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7848528950320243e-06,
+      "loss": 0.8334,
+      "step": 5558
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.781771229543756e-06,
+      "loss": 0.7712,
+      "step": 5559
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7786909945630113e-06,
+      "loss": 0.7249,
+      "step": 5560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7756121907002266e-06,
+      "loss": 0.7864,
+      "step": 5561
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.772534818565563e-06,
+      "loss": 0.6782,
+      "step": 5562
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7694588787688883e-06,
+      "loss": 0.7864,
+      "step": 5563
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.766384371919797e-06,
+      "loss": 0.7424,
+      "step": 5564
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.76331129862759e-06,
+      "loss": 0.6641,
+      "step": 5565
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7602396595012913e-06,
+      "loss": 0.7913,
+      "step": 5566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7571694551496397e-06,
+      "loss": 0.8216,
+      "step": 5567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7541006861810827e-06,
+      "loss": 0.7419,
+      "step": 5568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7510333532037938e-06,
+      "loss": 0.7634,
+      "step": 5569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7479674568256508e-06,
+      "loss": 0.7366,
+      "step": 5570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7449029976542596e-06,
+      "loss": 0.7232,
+      "step": 5571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.741839976296926e-06,
+      "loss": 0.6854,
+      "step": 5572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7387783933606846e-06,
+      "loss": 0.7269,
+      "step": 5573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7357182494522825e-06,
+      "loss": 0.7691,
+      "step": 5574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.73265954517817e-06,
+      "loss": 0.7814,
+      "step": 5575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.729602281144529e-06,
+      "loss": 0.7955,
+      "step": 5576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7265464579572408e-06,
+      "loss": 0.3231,
+      "step": 5577
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.723492076221914e-06,
+      "loss": 0.7275,
+      "step": 5578
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.72043913654386e-06,
+      "loss": 0.7676,
+      "step": 5579
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7173876395281117e-06,
+      "loss": 0.8009,
+      "step": 5580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7143375857794184e-06,
+      "loss": 0.7996,
+      "step": 5581
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7112889759022334e-06,
+      "loss": 0.7969,
+      "step": 5582
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7082418105007345e-06,
+      "loss": 0.6982,
+      "step": 5583
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7051960901788044e-06,
+      "loss": 0.7289,
+      "step": 5584
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7021518155400483e-06,
+      "loss": 0.7417,
+      "step": 5585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6991089871877773e-06,
+      "loss": 0.7198,
+      "step": 5586
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6960676057250158e-06,
+      "loss": 0.3353,
+      "step": 5587
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.693027671754509e-06,
+      "loss": 0.7597,
+      "step": 5588
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6899891858787075e-06,
+      "loss": 0.7659,
+      "step": 5589
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.686952148699781e-06,
+      "loss": 0.7288,
+      "step": 5590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.683916560819605e-06,
+      "loss": 0.7913,
+      "step": 5591
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6808824228397743e-06,
+      "loss": 0.6989,
+      "step": 5592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6778497353615964e-06,
+      "loss": 0.7186,
+      "step": 5593
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6748184989860838e-06,
+      "loss": 0.3493,
+      "step": 5594
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.671788714313972e-06,
+      "loss": 0.7801,
+      "step": 5595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6687603819456965e-06,
+      "loss": 0.8464,
+      "step": 5596
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.665733502481419e-06,
+      "loss": 0.7652,
+      "step": 5597
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6627080765210002e-06,
+      "loss": 0.6911,
+      "step": 5598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6596841046640197e-06,
+      "loss": 0.7408,
+      "step": 5599
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6566615875097724e-06,
+      "loss": 0.6975,
+      "step": 5600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.653640525657255e-06,
+      "loss": 0.7711,
+      "step": 5601
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.650620919705185e-06,
+      "loss": 0.7758,
+      "step": 5602
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6476027702519825e-06,
+      "loss": 0.7383,
+      "step": 5603
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.64458607789579e-06,
+      "loss": 0.7135,
+      "step": 5604
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.64157084323445e-06,
+      "loss": 0.7622,
+      "step": 5605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6385570668655223e-06,
+      "loss": 0.7798,
+      "step": 5606
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6355447493862805e-06,
+      "loss": 0.7074,
+      "step": 5607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6325338913937005e-06,
+      "loss": 0.7775,
+      "step": 5608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6295244934844776e-06,
+      "loss": 0.7608,
+      "step": 5609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.62651655625501e-06,
+      "loss": 0.7448,
+      "step": 5610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6235100803014134e-06,
+      "loss": 0.7175,
+      "step": 5611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6205050662195074e-06,
+      "loss": 0.7252,
+      "step": 5612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.617501514604828e-06,
+      "loss": 0.8278,
+      "step": 5613
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.61449942605262e-06,
+      "loss": 0.7594,
+      "step": 5614
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6114988011578324e-06,
+      "loss": 0.6691,
+      "step": 5615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.608499640515134e-06,
+      "loss": 0.8054,
+      "step": 5616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6055019447188923e-06,
+      "loss": 0.8313,
+      "step": 5617
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.602505714363196e-06,
+      "loss": 0.8129,
+      "step": 5618
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5995109500418303e-06,
+      "loss": 0.7585,
+      "step": 5619
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5965176523483026e-06,
+      "loss": 0.7618,
+      "step": 5620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.593525821875825e-06,
+      "loss": 0.8046,
+      "step": 5621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5905354592173148e-06,
+      "loss": 0.7607,
+      "step": 5622
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5875465649654042e-06,
+      "loss": 0.7337,
+      "step": 5623
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5845591397124283e-06,
+      "loss": 0.7593,
+      "step": 5624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.58157318405044e-06,
+      "loss": 0.7635,
+      "step": 5625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5785886985711882e-06,
+      "loss": 0.7078,
+      "step": 5626
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5756056838661457e-06,
+      "loss": 0.7371,
+      "step": 5627
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5726241405264827e-06,
+      "loss": 0.7145,
+      "step": 5628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.569644069143077e-06,
+      "loss": 0.7588,
+      "step": 5629
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5666654703065254e-06,
+      "loss": 0.7227,
+      "step": 5630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.563688344607119e-06,
+      "loss": 0.7925,
+      "step": 5631
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.560712692634869e-06,
+      "loss": 0.7863,
+      "step": 5632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.557738514979492e-06,
+      "loss": 0.6864,
+      "step": 5633
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5547658122304033e-06,
+      "loss": 0.7632,
+      "step": 5634
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.551794584976739e-06,
+      "loss": 0.7741,
+      "step": 5635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.548824833807332e-06,
+      "loss": 0.7153,
+      "step": 5636
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5458565593107308e-06,
+      "loss": 0.7698,
+      "step": 5637
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5428897620751824e-06,
+      "loss": 0.6872,
+      "step": 5638
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5399244426886505e-06,
+      "loss": 0.7786,
+      "step": 5639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.536960601738804e-06,
+      "loss": 0.8277,
+      "step": 5640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.533998239813009e-06,
+      "loss": 0.841,
+      "step": 5641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.531037357498354e-06,
+      "loss": 0.34,
+      "step": 5642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5280779553816194e-06,
+      "loss": 0.7781,
+      "step": 5643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5251200340493043e-06,
+      "loss": 0.3619,
+      "step": 5644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5221635940876044e-06,
+      "loss": 0.7162,
+      "step": 5645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5192086360824285e-06,
+      "loss": 0.7789,
+      "step": 5646
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.516255160619394e-06,
+      "loss": 0.7511,
+      "step": 5647
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5133031682838126e-06,
+      "loss": 0.8247,
+      "step": 5648
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5103526596607175e-06,
+      "loss": 0.7107,
+      "step": 5649
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.507403635334833e-06,
+      "loss": 0.6796,
+      "step": 5650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5044560958906018e-06,
+      "loss": 0.7811,
+      "step": 5651
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5015100419121617e-06,
+      "loss": 0.7673,
+      "step": 5652
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4985654739833643e-06,
+      "loss": 0.7432,
+      "step": 5653
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4956223926877664e-06,
+      "loss": 0.7461,
+      "step": 5654
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4926807986086223e-06,
+      "loss": 0.8482,
+      "step": 5655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4897406923289002e-06,
+      "loss": 0.7891,
+      "step": 5656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.486802074431266e-06,
+      "loss": 0.7416,
+      "step": 5657
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4838649454981e-06,
+      "loss": 0.859,
+      "step": 5658
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4809293061114768e-06,
+      "loss": 0.7697,
+      "step": 5659
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4779951568531824e-06,
+      "loss": 0.7015,
+      "step": 5660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.47506249830471e-06,
+      "loss": 0.7798,
+      "step": 5661
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.472131331047248e-06,
+      "loss": 0.7286,
+      "step": 5662
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4692016556616994e-06,
+      "loss": 0.7061,
+      "step": 5663
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.466273472728662e-06,
+      "loss": 0.8289,
+      "step": 5664
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4633467828284485e-06,
+      "loss": 0.826,
+      "step": 5665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.460421586541063e-06,
+      "loss": 0.8444,
+      "step": 5666
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.457497884446226e-06,
+      "loss": 0.7758,
+      "step": 5667
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.454575677123353e-06,
+      "loss": 0.7664,
+      "step": 5668
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.45165496515157e-06,
+      "loss": 0.8212,
+      "step": 5669
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4487357491097017e-06,
+      "loss": 0.6968,
+      "step": 5670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4458180295762746e-06,
+      "loss": 0.773,
+      "step": 5671
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.442901807129524e-06,
+      "loss": 0.7573,
+      "step": 5672
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4399870823473914e-06,
+      "loss": 0.8337,
+      "step": 5673
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.437073855807509e-06,
+      "loss": 0.7799,
+      "step": 5674
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.434162128087225e-06,
+      "loss": 0.7246,
+      "step": 5675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4312518997635816e-06,
+      "loss": 0.6999,
+      "step": 5676
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.428343171413331e-06,
+      "loss": 0.6863,
+      "step": 5677
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4254359436129204e-06,
+      "loss": 0.8163,
+      "step": 5678
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4225302169385045e-06,
+      "loss": 0.672,
+      "step": 5679
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.419625991965945e-06,
+      "loss": 0.7444,
+      "step": 5680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4167232692707944e-06,
+      "loss": 0.7278,
+      "step": 5681
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4138220494283195e-06,
+      "loss": 0.6722,
+      "step": 5682
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.410922333013477e-06,
+      "loss": 0.7145,
+      "step": 5683
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.408024120600938e-06,
+      "loss": 0.753,
+      "step": 5684
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.405127412765066e-06,
+      "loss": 0.7963,
+      "step": 5685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4022322100799313e-06,
+      "loss": 0.7384,
+      "step": 5686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3993385131193083e-06,
+      "loss": 0.6704,
+      "step": 5687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3964463224566636e-06,
+      "loss": 0.6682,
+      "step": 5688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.393555638665177e-06,
+      "loss": 0.8109,
+      "step": 5689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.390666462317717e-06,
+      "loss": 0.7934,
+      "step": 5690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.387778793986867e-06,
+      "loss": 0.7854,
+      "step": 5691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3848926342448984e-06,
+      "loss": 0.7439,
+      "step": 5692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3820079836637965e-06,
+      "loss": 0.7066,
+      "step": 5693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.379124842815235e-06,
+      "loss": 0.7933,
+      "step": 5694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3762432122705947e-06,
+      "loss": 0.8086,
+      "step": 5695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.373363092600963e-06,
+      "loss": 0.7632,
+      "step": 5696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3704844843771148e-06,
+      "loss": 0.76,
+      "step": 5697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3676073881695384e-06,
+      "loss": 0.7951,
+      "step": 5698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.364731804548408e-06,
+      "loss": 0.7221,
+      "step": 5699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3618577340836157e-06,
+      "loss": 0.716,
+      "step": 5700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3589851773447357e-06,
+      "loss": 0.7382,
+      "step": 5701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.356114134901055e-06,
+      "loss": 0.7979,
+      "step": 5702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.35324460732156e-06,
+      "loss": 0.7728,
+      "step": 5703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.350376595174927e-06,
+      "loss": 0.7216,
+      "step": 5704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.347510099029543e-06,
+      "loss": 0.7336,
+      "step": 5705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3446451194534848e-06,
+      "loss": 0.8204,
+      "step": 5706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.341781657014539e-06,
+      "loss": 0.7976,
+      "step": 5707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.338919712280181e-06,
+      "loss": 0.757,
+      "step": 5708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.336059285817592e-06,
+      "loss": 0.8165,
+      "step": 5709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.333200378193656e-06,
+      "loss": 0.687,
+      "step": 5710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3303429899749476e-06,
+      "loss": 0.7781,
+      "step": 5711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3274871217277393e-06,
+      "loss": 0.7641,
+      "step": 5712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.324632774018013e-06,
+      "loss": 0.7179,
+      "step": 5713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.321779947411438e-06,
+      "loss": 0.6817,
+      "step": 5714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3189286424733913e-06,
+      "loss": 0.7346,
+      "step": 5715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3160788597689397e-06,
+      "loss": 0.712,
+      "step": 5716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.313230599862858e-06,
+      "loss": 0.7389,
+      "step": 5717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3103838633196084e-06,
+      "loss": 0.6976,
+      "step": 5718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3075386507033594e-06,
+      "loss": 0.7063,
+      "step": 5719
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.304694962577977e-06,
+      "loss": 0.7038,
+      "step": 5720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.301852799507018e-06,
+      "loss": 0.7558,
+      "step": 5721
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2990121620537474e-06,
+      "loss": 0.7171,
+      "step": 5722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2961730507811164e-06,
+      "loss": 0.7497,
+      "step": 5723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.293335466251786e-06,
+      "loss": 0.6724,
+      "step": 5724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.290499409028102e-06,
+      "loss": 0.7287,
+      "step": 5725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.287664879672119e-06,
+      "loss": 0.7166,
+      "step": 5726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.284831878745579e-06,
+      "loss": 0.7118,
+      "step": 5727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.282000406809929e-06,
+      "loss": 0.7136,
+      "step": 5728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.27917046442631e-06,
+      "loss": 0.7301,
+      "step": 5729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.276342052155557e-06,
+      "loss": 0.727,
+      "step": 5730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.273515170558207e-06,
+      "loss": 0.7486,
+      "step": 5731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.270689820194488e-06,
+      "loss": 0.8173,
+      "step": 5732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2678660016243316e-06,
+      "loss": 0.3288,
+      "step": 5733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.265043715407357e-06,
+      "loss": 0.7133,
+      "step": 5734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2622229621028868e-06,
+      "loss": 0.7372,
+      "step": 5735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.25940374226994e-06,
+      "loss": 0.7375,
+      "step": 5736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2565860564672237e-06,
+      "loss": 0.7091,
+      "step": 5737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2537699052531526e-06,
+      "loss": 0.805,
+      "step": 5738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2509552891858242e-06,
+      "loss": 0.8133,
+      "step": 5739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2481422088230466e-06,
+      "loss": 0.7613,
+      "step": 5740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2453306647223084e-06,
+      "loss": 0.7454,
+      "step": 5741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.242520657440803e-06,
+      "loss": 0.8069,
+      "step": 5742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.239712187535422e-06,
+      "loss": 0.7702,
+      "step": 5743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2369052555627414e-06,
+      "loss": 0.7564,
+      "step": 5744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2340998620790435e-06,
+      "loss": 0.3554,
+      "step": 5745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2312960076402957e-06,
+      "loss": 0.7681,
+      "step": 5746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2284936928021706e-06,
+      "loss": 0.7893,
+      "step": 5747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.225692918120026e-06,
+      "loss": 0.3775,
+      "step": 5748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.22289368414892e-06,
+      "loss": 0.6821,
+      "step": 5749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2200959914436095e-06,
+      "loss": 0.7959,
+      "step": 5750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2172998405585333e-06,
+      "loss": 0.7592,
+      "step": 5751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.214505232047841e-06,
+      "loss": 0.7741,
+      "step": 5752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.211712166465357e-06,
+      "loss": 0.7643,
+      "step": 5753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.208920644364616e-06,
+      "loss": 0.8097,
+      "step": 5754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2061306662988447e-06,
+      "loss": 0.7056,
+      "step": 5755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.203342232820954e-06,
+      "loss": 0.7487,
+      "step": 5756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2005553444835613e-06,
+      "loss": 0.7335,
+      "step": 5757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1977700018389657e-06,
+      "loss": 0.7644,
+      "step": 5758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1949862054391726e-06,
+      "loss": 0.8151,
+      "step": 5759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1922039558358678e-06,
+      "loss": 0.7152,
+      "step": 5760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1894232535804405e-06,
+      "loss": 0.3403,
+      "step": 5761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1866440992239724e-06,
+      "loss": 0.7621,
+      "step": 5762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1838664933172306e-06,
+      "loss": 0.7311,
+      "step": 5763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1810904364106867e-06,
+      "loss": 0.7439,
+      "step": 5764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.178315929054492e-06,
+      "loss": 0.8029,
+      "step": 5765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.175542971798504e-06,
+      "loss": 0.7203,
+      "step": 5766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.172771565192263e-06,
+      "loss": 0.7638,
+      "step": 5767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.170001709785008e-06,
+      "loss": 0.8507,
+      "step": 5768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1672334061256697e-06,
+      "loss": 0.802,
+      "step": 5769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.164466654762867e-06,
+      "loss": 0.7686,
+      "step": 5770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.161701456244918e-06,
+      "loss": 0.7178,
+      "step": 5771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1589378111198232e-06,
+      "loss": 0.7863,
+      "step": 5772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1561757199352907e-06,
+      "loss": 0.797,
+      "step": 5773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1534151832387016e-06,
+      "loss": 0.7748,
+      "step": 5774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1506562015771435e-06,
+      "loss": 0.7705,
+      "step": 5775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1478987754973935e-06,
+      "loss": 0.7531,
+      "step": 5776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1451429055459116e-06,
+      "loss": 0.7322,
+      "step": 5777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1423885922688624e-06,
+      "loss": 0.725,
+      "step": 5778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1396358362120896e-06,
+      "loss": 0.7475,
+      "step": 5779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1368846379211385e-06,
+      "loss": 0.7755,
+      "step": 5780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1341349979412364e-06,
+      "loss": 0.7706,
+      "step": 5781
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1313869168173085e-06,
+      "loss": 0.7818,
+      "step": 5782
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.128640395093974e-06,
+      "loss": 0.7247,
+      "step": 5783
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.12589543331553e-06,
+      "loss": 0.7672,
+      "step": 5784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1231520320259793e-06,
+      "loss": 0.8573,
+      "step": 5785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1204101917690046e-06,
+      "loss": 0.7607,
+      "step": 5786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1176699130879865e-06,
+      "loss": 0.8146,
+      "step": 5787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1149311965259888e-06,
+      "loss": 0.8129,
+      "step": 5788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1121940426257723e-06,
+      "loss": 0.7154,
+      "step": 5789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1094584519297902e-06,
+      "loss": 0.7786,
+      "step": 5790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.106724424980173e-06,
+      "loss": 0.7748,
+      "step": 5791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1039919623187575e-06,
+      "loss": 0.735,
+      "step": 5792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1012610644870557e-06,
+      "loss": 0.8132,
+      "step": 5793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0985317320262845e-06,
+      "loss": 0.6992,
+      "step": 5794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0958039654773367e-06,
+      "loss": 0.3678,
+      "step": 5795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0930777653808e-06,
+      "loss": 0.772,
+      "step": 5796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0903531322769566e-06,
+      "loss": 0.7497,
+      "step": 5797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0876300667057704e-06,
+      "loss": 0.7586,
+      "step": 5798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0849085692069016e-06,
+      "loss": 0.7532,
+      "step": 5799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.082188640319691e-06,
+      "loss": 0.7267,
+      "step": 5800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0794702805831766e-06,
+      "loss": 0.7356,
+      "step": 5801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0767534905360865e-06,
+      "loss": 0.8062,
+      "step": 5802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.074038270716827e-06,
+      "loss": 0.7594,
+      "step": 5803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0713246216635053e-06,
+      "loss": 0.6806,
+      "step": 5804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.068612543913907e-06,
+      "loss": 0.8017,
+      "step": 5805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.065902038005517e-06,
+      "loss": 0.7527,
+      "step": 5806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.063193104475497e-06,
+      "loss": 0.6438,
+      "step": 5807
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0604857438607062e-06,
+      "loss": 0.8244,
+      "step": 5808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.057779956697692e-06,
+      "loss": 0.716,
+      "step": 5809
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0550757435226797e-06,
+      "loss": 0.7412,
+      "step": 5810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0523731048715968e-06,
+      "loss": 0.7414,
+      "step": 5811
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.049672041280046e-06,
+      "loss": 0.8094,
+      "step": 5812
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0469725532833295e-06,
+      "loss": 0.7601,
+      "step": 5813
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0442746414164263e-06,
+      "loss": 0.7542,
+      "step": 5814
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.041578306214009e-06,
+      "loss": 0.8357,
+      "step": 5815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.038883548210442e-06,
+      "loss": 0.7527,
+      "step": 5816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0361903679397644e-06,
+      "loss": 0.7068,
+      "step": 5817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0334987659357175e-06,
+      "loss": 0.3567,
+      "step": 5818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.030808742731717e-06,
+      "loss": 0.7945,
+      "step": 5819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0281202988608763e-06,
+      "loss": 0.7873,
+      "step": 5820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0254334348559845e-06,
+      "loss": 0.7326,
+      "step": 5821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0227481512495274e-06,
+      "loss": 0.808,
+      "step": 5822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0200644485736778e-06,
+      "loss": 0.796,
+      "step": 5823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0173823273602845e-06,
+      "loss": 0.7128,
+      "step": 5824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0147017881408947e-06,
+      "loss": 0.7212,
+      "step": 5825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.012022831446735e-06,
+      "loss": 0.7491,
+      "step": 5826
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0093454578087234e-06,
+      "loss": 0.7201,
+      "step": 5827
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.006669667757457e-06,
+      "loss": 0.75,
+      "step": 5828
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0039954618232294e-06,
+      "loss": 0.7888,
+      "step": 5829
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0013228405360076e-06,
+      "loss": 0.7394,
+      "step": 5830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.998651804425454e-06,
+      "loss": 0.7206,
+      "step": 5831
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9959823540209177e-06,
+      "loss": 0.7359,
+      "step": 5832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9933144898514244e-06,
+      "loss": 0.8123,
+      "step": 5833
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9906482124456973e-06,
+      "loss": 0.77,
+      "step": 5834
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.987983522332131e-06,
+      "loss": 0.7334,
+      "step": 5835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.985320420038821e-06,
+      "loss": 0.7179,
+      "step": 5836
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9826589060935373e-06,
+      "loss": 0.7915,
+      "step": 5837
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9799989810237365e-06,
+      "loss": 0.7785,
+      "step": 5838
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9773406453565645e-06,
+      "loss": 0.6937,
+      "step": 5839
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.974683899618848e-06,
+      "loss": 0.7276,
+      "step": 5840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.972028744337101e-06,
+      "loss": 0.3683,
+      "step": 5841
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.969375180037526e-06,
+      "loss": 0.8004,
+      "step": 5842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9667232072459984e-06,
+      "loss": 0.6808,
+      "step": 5843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9640728264880937e-06,
+      "loss": 0.743,
+      "step": 5844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.961424038289057e-06,
+      "loss": 0.7963,
+      "step": 5845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.958776843173831e-06,
+      "loss": 0.9264,
+      "step": 5846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9561312416670308e-06,
+      "loss": 0.7424,
+      "step": 5847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9534872342929634e-06,
+      "loss": 0.7511,
+      "step": 5848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9508448215756214e-06,
+      "loss": 0.8021,
+      "step": 5849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.948204004038671e-06,
+      "loss": 0.7549,
+      "step": 5850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9455647822054758e-06,
+      "loss": 0.7587,
+      "step": 5851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9429271565990703e-06,
+      "loss": 0.7241,
+      "step": 5852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9402911277421855e-06,
+      "loss": 0.7455,
+      "step": 5853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9376566961572217e-06,
+      "loss": 0.7891,
+      "step": 5854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9350238623662766e-06,
+      "loss": 0.7535,
+      "step": 5855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.93239262689112e-06,
+      "loss": 0.7594,
+      "step": 5856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.929762990253211e-06,
+      "loss": 0.8108,
+      "step": 5857
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9271349529736947e-06,
+      "loss": 0.8075,
+      "step": 5858
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.92450851557339e-06,
+      "loss": 0.7736,
+      "step": 5859
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.921883678572807e-06,
+      "loss": 0.7524,
+      "step": 5860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9192604424921323e-06,
+      "loss": 0.776,
+      "step": 5861
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9166388078512434e-06,
+      "loss": 0.7136,
+      "step": 5862
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9140187751696893e-06,
+      "loss": 0.7337,
+      "step": 5863
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9114003449667107e-06,
+      "loss": 0.7249,
+      "step": 5864
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908783517761231e-06,
+      "loss": 0.7997,
+      "step": 5865
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9061682940718473e-06,
+      "loss": 0.7348,
+      "step": 5866
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9035546744168486e-06,
+      "loss": 0.7256,
+      "step": 5867
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9009426593141966e-06,
+      "loss": 0.7558,
+      "step": 5868
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.898332249281546e-06,
+      "loss": 0.7388,
+      "step": 5869
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8957234448362227e-06,
+      "loss": 0.6897,
+      "step": 5870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8931162464952402e-06,
+      "loss": 0.7386,
+      "step": 5871
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.890510654775296e-06,
+      "loss": 0.8053,
+      "step": 5872
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8879066701927606e-06,
+      "loss": 0.7772,
+      "step": 5873
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8853042932636966e-06,
+      "loss": 0.7309,
+      "step": 5874
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8827035245038383e-06,
+      "loss": 0.8984,
+      "step": 5875
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8801043644286099e-06,
+      "loss": 0.7617,
+      "step": 5876
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.877506813553107e-06,
+      "loss": 0.7481,
+      "step": 5877
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8749108723921183e-06,
+      "loss": 0.7545,
+      "step": 5878
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.872316541460103e-06,
+      "loss": 0.7562,
+      "step": 5879
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8697238212712043e-06,
+      "loss": 0.8147,
+      "step": 5880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8671327123392502e-06,
+      "loss": 0.7558,
+      "step": 5881
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8645432151777431e-06,
+      "loss": 0.8254,
+      "step": 5882
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8619553302998716e-06,
+      "loss": 0.6965,
+      "step": 5883
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.859369058218503e-06,
+      "loss": 0.7344,
+      "step": 5884
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.856784399446181e-06,
+      "loss": 0.7125,
+      "step": 5885
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8542013544951376e-06,
+      "loss": 0.7315,
+      "step": 5886
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8516199238772747e-06,
+      "loss": 0.6783,
+      "step": 5887
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8490401081041853e-06,
+      "loss": 0.7991,
+      "step": 5888
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8464619076871316e-06,
+      "loss": 0.7722,
+      "step": 5889
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8438853231370635e-06,
+      "loss": 0.7459,
+      "step": 5890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8413103549646116e-06,
+      "loss": 0.7392,
+      "step": 5891
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8387370036800768e-06,
+      "loss": 0.7639,
+      "step": 5892
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8361652697934507e-06,
+      "loss": 0.8003,
+      "step": 5893
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8335951538143936e-06,
+      "loss": 0.7809,
+      "step": 5894
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8310266562522573e-06,
+      "loss": 0.337,
+      "step": 5895
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8284597776160583e-06,
+      "loss": 0.7898,
+      "step": 5896
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8258945184145061e-06,
+      "loss": 0.7782,
+      "step": 5897
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8233308791559846e-06,
+      "loss": 0.8169,
+      "step": 5898
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8207688603485495e-06,
+      "loss": 0.7079,
+      "step": 5899
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8182084624999474e-06,
+      "loss": 0.6933,
+      "step": 5900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8156496861175922e-06,
+      "loss": 0.7186,
+      "step": 5901
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8130925317085869e-06,
+      "loss": 0.6698,
+      "step": 5902
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8105369997797029e-06,
+      "loss": 0.7756,
+      "step": 5903
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8079830908373975e-06,
+      "loss": 0.8016,
+      "step": 5904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8054308053878068e-06,
+      "loss": 0.7214,
+      "step": 5905
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8028801439367371e-06,
+      "loss": 0.7119,
+      "step": 5906
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8003311069896844e-06,
+      "loss": 0.763,
+      "step": 5907
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.797783695051809e-06,
+      "loss": 0.7305,
+      "step": 5908
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7952379086279636e-06,
+      "loss": 0.7943,
+      "step": 5909
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7926937482226659e-06,
+      "loss": 0.8512,
+      "step": 5910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.79015121434012e-06,
+      "loss": 0.7931,
+      "step": 5911
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7876103074842077e-06,
+      "loss": 0.7805,
+      "step": 5912
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7850710281584792e-06,
+      "loss": 0.7529,
+      "step": 5913
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7825333768661757e-06,
+      "loss": 0.6651,
+      "step": 5914
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7799973541102022e-06,
+      "loss": 0.7696,
+      "step": 5915
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7774629603931525e-06,
+      "loss": 0.7568,
+      "step": 5916
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.774930196217287e-06,
+      "loss": 0.7581,
+      "step": 5917
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.772399062084552e-06,
+      "loss": 0.7721,
+      "step": 5918
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7698695584965708e-06,
+      "loss": 0.7888,
+      "step": 5919
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.767341685954632e-06,
+      "loss": 0.7537,
+      "step": 5920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7648154449597154e-06,
+      "loss": 0.729,
+      "step": 5921
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7622908360124658e-06,
+      "loss": 0.7113,
+      "step": 5922
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7597678596132117e-06,
+      "loss": 0.7833,
+      "step": 5923
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.757246516261961e-06,
+      "loss": 0.7509,
+      "step": 5924
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7547268064583866e-06,
+      "loss": 0.7953,
+      "step": 5925
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7522087307018488e-06,
+      "loss": 0.7774,
+      "step": 5926
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.749692289491376e-06,
+      "loss": 0.3351,
+      "step": 5927
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7471774833256794e-06,
+      "loss": 0.7239,
+      "step": 5928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7446643127031393e-06,
+      "loss": 0.8242,
+      "step": 5929
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7421527781218184e-06,
+      "loss": 0.7609,
+      "step": 5930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.739642880079453e-06,
+      "loss": 0.6955,
+      "step": 5931
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7371346190734496e-06,
+      "loss": 0.7528,
+      "step": 5932
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7346279956009016e-06,
+      "loss": 0.782,
+      "step": 5933
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.732123010158564e-06,
+      "loss": 0.7114,
+      "step": 5934
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7296196632428818e-06,
+      "loss": 0.7608,
+      "step": 5935
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7271179553499607e-06,
+      "loss": 0.769,
+      "step": 5936
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7246178869755935e-06,
+      "loss": 0.7985,
+      "step": 5937
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7221194586152446e-06,
+      "loss": 0.8073,
+      "step": 5938
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7196226707640473e-06,
+      "loss": 0.7633,
+      "step": 5939
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7171275239168194e-06,
+      "loss": 0.7552,
+      "step": 5940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7146340185680444e-06,
+      "loss": 0.7256,
+      "step": 5941
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.712142155211889e-06,
+      "loss": 0.6743,
+      "step": 5942
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709651934342187e-06,
+      "loss": 0.7389,
+      "step": 5943
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7071633564524503e-06,
+      "loss": 0.7786,
+      "step": 5944
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704676422035868e-06,
+      "loss": 0.7692,
+      "step": 5945
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702191131585297e-06,
+      "loss": 0.7469,
+      "step": 5946
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.699707485593275e-06,
+      "loss": 0.6867,
+      "step": 5947
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6972254845520064e-06,
+      "loss": 0.7552,
+      "step": 5948
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6947451289533778e-06,
+      "loss": 0.685,
+      "step": 5949
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6922664192889426e-06,
+      "loss": 0.8248,
+      "step": 5950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6897893560499324e-06,
+      "loss": 0.7856,
+      "step": 5951
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6873139397272532e-06,
+      "loss": 0.6354,
+      "step": 5952
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.684840170811478e-06,
+      "loss": 0.7256,
+      "step": 5953
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6823680497928641e-06,
+      "loss": 0.7332,
+      "step": 5954
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6798975771613301e-06,
+      "loss": 0.7666,
+      "step": 5955
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6774287534064793e-06,
+      "loss": 0.7348,
+      "step": 5956
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6749615790175778e-06,
+      "loss": 0.7892,
+      "step": 5957
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.672496054483571e-06,
+      "loss": 0.7441,
+      "step": 5958
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6700321802930808e-06,
+      "loss": 0.75,
+      "step": 5959
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6675699569343906e-06,
+      "loss": 0.7419,
+      "step": 5960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6651093848954725e-06,
+      "loss": 0.7398,
+      "step": 5961
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.662650464663952e-06,
+      "loss": 0.3274,
+      "step": 5962
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6601931967271412e-06,
+      "loss": 0.791,
+      "step": 5963
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6577375815720254e-06,
+      "loss": 0.3121,
+      "step": 5964
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6552836196852518e-06,
+      "loss": 0.7699,
+      "step": 5965
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6528313115531524e-06,
+      "loss": 0.7329,
+      "step": 5966
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6503806576617188e-06,
+      "loss": 0.8261,
+      "step": 5967
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6479316584966264e-06,
+      "loss": 0.7172,
+      "step": 5968
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6454843145432143e-06,
+      "loss": 0.7115,
+      "step": 5969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6430386262864972e-06,
+      "loss": 0.6793,
+      "step": 5970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6405945942111646e-06,
+      "loss": 0.7473,
+      "step": 5971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6381522188015697e-06,
+      "loss": 0.7889,
+      "step": 5972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6357115005417467e-06,
+      "loss": 0.7619,
+      "step": 5973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.633272439915392e-06,
+      "loss": 0.7493,
+      "step": 5974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6308350374058824e-06,
+      "loss": 0.7882,
+      "step": 5975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6283992934962588e-06,
+      "loss": 0.7434,
+      "step": 5976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.625965208669237e-06,
+      "loss": 0.7691,
+      "step": 5977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.623532783407208e-06,
+      "loss": 0.7985,
+      "step": 5978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6211020181922233e-06,
+      "loss": 0.7254,
+      "step": 5979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6186729135060164e-06,
+      "loss": 0.6694,
+      "step": 5980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6162454698299823e-06,
+      "loss": 0.7159,
+      "step": 5981
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.613819687645196e-06,
+      "loss": 0.7317,
+      "step": 5982
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6113955674323956e-06,
+      "loss": 0.7535,
+      "step": 5983
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6089731096719929e-06,
+      "loss": 0.7798,
+      "step": 5984
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6065523148440732e-06,
+      "loss": 0.7672,
+      "step": 5985
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6041331834283857e-06,
+      "loss": 0.739,
+      "step": 5986
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6017157159043572e-06,
+      "loss": 0.8603,
+      "step": 5987
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5992999127510766e-06,
+      "loss": 0.7071,
+      "step": 5988
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5968857744473132e-06,
+      "loss": 0.7395,
+      "step": 5989
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.594473301471494e-06,
+      "loss": 0.7339,
+      "step": 5990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5920624943017282e-06,
+      "loss": 0.7472,
+      "step": 5991
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5896533534157842e-06,
+      "loss": 0.7254,
+      "step": 5992
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5872458792911071e-06,
+      "loss": 0.7148,
+      "step": 5993
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5848400724048141e-06,
+      "loss": 0.6743,
+      "step": 5994
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5824359332336802e-06,
+      "loss": 0.7446,
+      "step": 5995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5800334622541647e-06,
+      "loss": 0.7779,
+      "step": 5996
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.577632659942382e-06,
+      "loss": 0.7023,
+      "step": 5997
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.575233526774128e-06,
+      "loss": 0.7855,
+      "step": 5998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5728360632248596e-06,
+      "loss": 0.7235,
+      "step": 5999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5704402697697064e-06,
+      "loss": 0.8157,
+      "step": 6000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5680461468834695e-06,
+      "loss": 0.743,
+      "step": 6001
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5656536950406099e-06,
+      "loss": 0.7433,
+      "step": 6002
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.56326291471527e-06,
+      "loss": 0.7862,
+      "step": 6003
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5608738063812502e-06,
+      "loss": 0.7438,
+      "step": 6004
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5584863705120211e-06,
+      "loss": 0.818,
+      "step": 6005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5561006075807317e-06,
+      "loss": 0.3628,
+      "step": 6006
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5537165180601853e-06,
+      "loss": 0.8285,
+      "step": 6007
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5513341024228646e-06,
+      "loss": 0.6922,
+      "step": 6008
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5489533611409135e-06,
+      "loss": 0.6873,
+      "step": 6009
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5465742946861484e-06,
+      "loss": 0.751,
+      "step": 6010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5441969035300542e-06,
+      "loss": 0.8401,
+      "step": 6011
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5418211881437783e-06,
+      "loss": 0.8086,
+      "step": 6012
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5394471489981434e-06,
+      "loss": 0.7339,
+      "step": 6013
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5370747865636315e-06,
+      "loss": 0.6842,
+      "step": 6014
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5347041013104025e-06,
+      "loss": 0.7553,
+      "step": 6015
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5323350937082738e-06,
+      "loss": 0.8061,
+      "step": 6016
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5299677642267396e-06,
+      "loss": 0.6594,
+      "step": 6017
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5276021133349506e-06,
+      "loss": 0.7759,
+      "step": 6018
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.525238141501736e-06,
+      "loss": 0.7529,
+      "step": 6019
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5228758491955875e-06,
+      "loss": 0.7766,
+      "step": 6020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.520515236884661e-06,
+      "loss": 0.7229,
+      "step": 6021
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5181563050367866e-06,
+      "loss": 0.6483,
+      "step": 6022
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5157990541194513e-06,
+      "loss": 0.724,
+      "step": 6023
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5134434845998191e-06,
+      "loss": 0.7503,
+      "step": 6024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5110895969447136e-06,
+      "loss": 0.7843,
+      "step": 6025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5087373916206304e-06,
+      "loss": 0.735,
+      "step": 6026
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5063868690937289e-06,
+      "loss": 0.7763,
+      "step": 6027
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.504038029829833e-06,
+      "loss": 0.8477,
+      "step": 6028
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5016908742944391e-06,
+      "loss": 0.7382,
+      "step": 6029
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4993454029527021e-06,
+      "loss": 0.8345,
+      "step": 6030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.497001616269451e-06,
+      "loss": 0.8035,
+      "step": 6031
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4946595147091726e-06,
+      "loss": 0.7597,
+      "step": 6032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.492319098736027e-06,
+      "loss": 0.7985,
+      "step": 6033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4899803688138393e-06,
+      "loss": 0.7132,
+      "step": 6034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.487643325406094e-06,
+      "loss": 0.6815,
+      "step": 6035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4853079689759519e-06,
+      "loss": 0.7608,
+      "step": 6036
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4829742999862272e-06,
+      "loss": 0.7489,
+      "step": 6037
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4806423188994122e-06,
+      "loss": 0.7821,
+      "step": 6038
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.478312026177653e-06,
+      "loss": 0.3048,
+      "step": 6039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4759834222827685e-06,
+      "loss": 0.6762,
+      "step": 6040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4736565076762455e-06,
+      "loss": 0.7668,
+      "step": 6041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.471331282819225e-06,
+      "loss": 0.7758,
+      "step": 6042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.469007748172524e-06,
+      "loss": 0.7903,
+      "step": 6043
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4666859041966175e-06,
+      "loss": 0.7487,
+      "step": 6044
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4643657513516517e-06,
+      "loss": 0.7821,
+      "step": 6045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4620472900974325e-06,
+      "loss": 0.7933,
+      "step": 6046
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4597305208934297e-06,
+      "loss": 0.8427,
+      "step": 6047
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4574154441987842e-06,
+      "loss": 0.7742,
+      "step": 6048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4551020604722932e-06,
+      "loss": 0.3425,
+      "step": 6049
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4527903701724278e-06,
+      "loss": 0.365,
+      "step": 6050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4504803737573137e-06,
+      "loss": 0.6926,
+      "step": 6051
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.448172071684747e-06,
+      "loss": 0.8231,
+      "step": 6052
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4458654644121905e-06,
+      "loss": 0.8207,
+      "step": 6053
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4435605523967622e-06,
+      "loss": 0.7937,
+      "step": 6054
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.441257336095253e-06,
+      "loss": 0.7462,
+      "step": 6055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4389558159641103e-06,
+      "loss": 0.7135,
+      "step": 6056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4366559924594537e-06,
+      "loss": 0.7129,
+      "step": 6057
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.434357866037057e-06,
+      "loss": 0.794,
+      "step": 6058
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4320614371523645e-06,
+      "loss": 0.6782,
+      "step": 6059
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4297667062604849e-06,
+      "loss": 0.697,
+      "step": 6060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.427473673816182e-06,
+      "loss": 0.7599,
+      "step": 6061
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4251823402738951e-06,
+      "loss": 0.7502,
+      "step": 6062
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.422892706087714e-06,
+      "loss": 0.7996,
+      "step": 6063
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4206047717114035e-06,
+      "loss": 0.7217,
+      "step": 6064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.418318537598382e-06,
+      "loss": 0.3558,
+      "step": 6065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4160340042017361e-06,
+      "loss": 0.7649,
+      "step": 6066
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.413751171974217e-06,
+      "loss": 0.7435,
+      "step": 6067
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4114700413682326e-06,
+      "loss": 0.7192,
+      "step": 6068
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4091906128358602e-06,
+      "loss": 0.7055,
+      "step": 6069
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.406912886828833e-06,
+      "loss": 0.7489,
+      "step": 6070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4046368637985541e-06,
+      "loss": 0.7195,
+      "step": 6071
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.402362544196081e-06,
+      "loss": 0.6543,
+      "step": 6072
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4000899284721402e-06,
+      "loss": 0.7915,
+      "step": 6073
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3978190170771222e-06,
+      "loss": 0.8013,
+      "step": 6074
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3955498104610698e-06,
+      "loss": 0.7107,
+      "step": 6075
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.393282309073699e-06,
+      "loss": 0.8173,
+      "step": 6076
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3910165133643772e-06,
+      "loss": 0.7252,
+      "step": 6077
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3887524237821458e-06,
+      "loss": 0.7728,
+      "step": 6078
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3864900407756965e-06,
+      "loss": 0.8723,
+      "step": 6079
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3842293647933902e-06,
+      "loss": 0.7694,
+      "step": 6080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3819703962832498e-06,
+      "loss": 0.7869,
+      "step": 6081
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3797131356929528e-06,
+      "loss": 0.734,
+      "step": 6082
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3774575834698478e-06,
+      "loss": 0.7666,
+      "step": 6083
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.375203740060934e-06,
+      "loss": 0.8408,
+      "step": 6084
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3729516059128823e-06,
+      "loss": 0.7677,
+      "step": 6085
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3707011814720195e-06,
+      "loss": 0.6982,
+      "step": 6086
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3684524671843313e-06,
+      "loss": 0.7518,
+      "step": 6087
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3662054634954725e-06,
+      "loss": 0.8066,
+      "step": 6088
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3639601708507477e-06,
+      "loss": 0.7855,
+      "step": 6089
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3617165896951358e-06,
+      "loss": 0.7954,
+      "step": 6090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3594747204732638e-06,
+      "loss": 0.7188,
+      "step": 6091
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.357234563629426e-06,
+      "loss": 0.7813,
+      "step": 6092
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.35499611960758e-06,
+      "loss": 0.6755,
+      "step": 6093
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3527593888513357e-06,
+      "loss": 0.7229,
+      "step": 6094
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.350524371803973e-06,
+      "loss": 0.7804,
+      "step": 6095
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3482910689084217e-06,
+      "loss": 0.7665,
+      "step": 6096
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3460594806072825e-06,
+      "loss": 0.7192,
+      "step": 6097
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.343829607342806e-06,
+      "loss": 0.7439,
+      "step": 6098
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.341601449556912e-06,
+      "loss": 0.7273,
+      "step": 6099
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.339375007691177e-06,
+      "loss": 0.7392,
+      "step": 6100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3371502821868332e-06,
+      "loss": 0.6601,
+      "step": 6101
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3349272734847818e-06,
+      "loss": 0.8247,
+      "step": 6102
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3327059820255716e-06,
+      "loss": 0.714,
+      "step": 6103
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.330486408249424e-06,
+      "loss": 0.7242,
+      "step": 6104
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.328268552596209e-06,
+      "loss": 0.7925,
+      "step": 6105
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3260524155054633e-06,
+      "loss": 0.8229,
+      "step": 6106
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3238379974163828e-06,
+      "loss": 0.7741,
+      "step": 6107
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3216252987678157e-06,
+      "loss": 0.7555,
+      "step": 6108
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.319414319998279e-06,
+      "loss": 0.6756,
+      "step": 6109
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3172050615459397e-06,
+      "loss": 0.8365,
+      "step": 6110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3149975238486323e-06,
+      "loss": 0.7685,
+      "step": 6111
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3127917073438433e-06,
+      "loss": 0.7428,
+      "step": 6112
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.310587612468721e-06,
+      "loss": 0.7503,
+      "step": 6113
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.308385239660077e-06,
+      "loss": 0.7114,
+      "step": 6114
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3061845893543723e-06,
+      "loss": 0.8146,
+      "step": 6115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3039856619877345e-06,
+      "loss": 0.7483,
+      "step": 6116
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3017884579959438e-06,
+      "loss": 0.7858,
+      "step": 6117
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2995929778144455e-06,
+      "loss": 0.7298,
+      "step": 6118
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.297399221878336e-06,
+      "loss": 0.7939,
+      "step": 6119
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.295207190622374e-06,
+      "loss": 0.7193,
+      "step": 6120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2930168844809798e-06,
+      "loss": 0.8002,
+      "step": 6121
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.290828303888223e-06,
+      "loss": 0.7559,
+      "step": 6122
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2886414492778399e-06,
+      "loss": 0.738,
+      "step": 6123
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2864563210832171e-06,
+      "loss": 0.3407,
+      "step": 6124
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2842729197374083e-06,
+      "loss": 0.8183,
+      "step": 6125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2820912456731139e-06,
+      "loss": 0.7745,
+      "step": 6126
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2799112993227037e-06,
+      "loss": 0.7497,
+      "step": 6127
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2777330811181942e-06,
+      "loss": 0.3333,
+      "step": 6128
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2755565914912648e-06,
+      "loss": 0.7457,
+      "step": 6129
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.273381830873256e-06,
+      "loss": 0.7727,
+      "step": 6130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2712087996951562e-06,
+      "loss": 0.7454,
+      "step": 6131
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2690374983876185e-06,
+      "loss": 0.7462,
+      "step": 6132
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2668679273809547e-06,
+      "loss": 0.7824,
+      "step": 6133
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2647000871051241e-06,
+      "loss": 0.707,
+      "step": 6134
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2625339779897539e-06,
+      "loss": 0.7787,
+      "step": 6135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2603696004641187e-06,
+      "loss": 0.7968,
+      "step": 6136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2582069549571596e-06,
+      "loss": 0.8278,
+      "step": 6137
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.256046041897464e-06,
+      "loss": 0.7778,
+      "step": 6138
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2538868617132848e-06,
+      "loss": 0.821,
+      "step": 6139
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2517294148325288e-06,
+      "loss": 0.3356,
+      "step": 6140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2495737016827558e-06,
+      "loss": 0.7481,
+      "step": 6141
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2474197226911877e-06,
+      "loss": 0.7716,
+      "step": 6142
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2452674782846951e-06,
+      "loss": 0.6974,
+      "step": 6143
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2431169688898148e-06,
+      "loss": 0.6947,
+      "step": 6144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2409681949327301e-06,
+      "loss": 0.8053,
+      "step": 6145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2388211568392872e-06,
+      "loss": 0.7685,
+      "step": 6146
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2366758550349866e-06,
+      "loss": 0.761,
+      "step": 6147
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2345322899449797e-06,
+      "loss": 0.8061,
+      "step": 6148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.232390461994084e-06,
+      "loss": 0.7421,
+      "step": 6149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2302503716067604e-06,
+      "loss": 0.7434,
+      "step": 6150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2281120192071373e-06,
+      "loss": 0.7817,
+      "step": 6151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2259754052189877e-06,
+      "loss": 0.3806,
+      "step": 6152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2238405300657508e-06,
+      "loss": 0.6915,
+      "step": 6153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2217073941705116e-06,
+      "loss": 0.7793,
+      "step": 6154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2195759979560152e-06,
+      "loss": 0.7664,
+      "step": 6155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2174463418446647e-06,
+      "loss": 0.3238,
+      "step": 6156
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2153184262585117e-06,
+      "loss": 0.7521,
+      "step": 6157
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2131922516192684e-06,
+      "loss": 0.7062,
+      "step": 6158
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.211067818348297e-06,
+      "loss": 0.7917,
+      "step": 6159
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2089451268666208e-06,
+      "loss": 0.6909,
+      "step": 6160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2068241775949108e-06,
+      "loss": 0.7987,
+      "step": 6161
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2047049709534985e-06,
+      "loss": 0.7251,
+      "step": 6162
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2025875073623694e-06,
+      "loss": 0.3433,
+      "step": 6163
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2004717872411586e-06,
+      "loss": 0.7863,
+      "step": 6164
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1983578110091643e-06,
+      "loss": 0.8056,
+      "step": 6165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1962455790853277e-06,
+      "loss": 0.7163,
+      "step": 6166
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1941350918882556e-06,
+      "loss": 0.8117,
+      "step": 6167
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1920263498362006e-06,
+      "loss": 0.7167,
+      "step": 6168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1899193533470743e-06,
+      "loss": 0.7097,
+      "step": 6169
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.187814102838446e-06,
+      "loss": 0.8206,
+      "step": 6170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.185710598727524e-06,
+      "loss": 0.7431,
+      "step": 6171
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1836088414311863e-06,
+      "loss": 0.8181,
+      "step": 6172
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1815088313659595e-06,
+      "loss": 0.8266,
+      "step": 6173
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1794105689480195e-06,
+      "loss": 0.7228,
+      "step": 6174
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.177314054593205e-06,
+      "loss": 0.7235,
+      "step": 6175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1752192887169978e-06,
+      "loss": 0.8453,
+      "step": 6176
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1731262717345416e-06,
+      "loss": 0.7401,
+      "step": 6177
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1710350040606277e-06,
+      "loss": 0.7669,
+      "step": 6178
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.168945486109706e-06,
+      "loss": 0.7308,
+      "step": 6179
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.166857718295874e-06,
+      "loss": 0.7977,
+      "step": 6180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1647717010328863e-06,
+      "loss": 0.7522,
+      "step": 6181
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.162687434734151e-06,
+      "loss": 0.7617,
+      "step": 6182
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1606049198127257e-06,
+      "loss": 0.8045,
+      "step": 6183
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1585241566813255e-06,
+      "loss": 0.7537,
+      "step": 6184
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1564451457523118e-06,
+      "loss": 0.7038,
+      "step": 6185
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1543678874377062e-06,
+      "loss": 0.7878,
+      "step": 6186
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1522923821491761e-06,
+      "loss": 0.7583,
+      "step": 6187
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1502186302980479e-06,
+      "loss": 0.7086,
+      "step": 6188
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1481466322952973e-06,
+      "loss": 0.3328,
+      "step": 6189
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1460763885515502e-06,
+      "loss": 0.7588,
+      "step": 6190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1440078994770898e-06,
+      "loss": 0.7749,
+      "step": 6191
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.141941165481847e-06,
+      "loss": 0.7623,
+      "step": 6192
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1398761869754083e-06,
+      "loss": 0.7516,
+      "step": 6193
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1378129643670087e-06,
+      "loss": 0.7544,
+      "step": 6194
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1357514980655394e-06,
+      "loss": 0.7473,
+      "step": 6195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1336917884795417e-06,
+      "loss": 0.741,
+      "step": 6196
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1316338360172064e-06,
+      "loss": 0.7354,
+      "step": 6197
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1295776410863813e-06,
+      "loss": 0.7177,
+      "step": 6198
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1275232040945595e-06,
+      "loss": 0.6559,
+      "step": 6199
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1254705254488919e-06,
+      "loss": 0.3122,
+      "step": 6200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1234196055561752e-06,
+      "loss": 0.7112,
+      "step": 6201
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1213704448228613e-06,
+      "loss": 0.7594,
+      "step": 6202
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1193230436550562e-06,
+      "loss": 0.7412,
+      "step": 6203
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1172774024585077e-06,
+      "loss": 0.3534,
+      "step": 6204
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1152335216386257e-06,
+      "loss": 0.7675,
+      "step": 6205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.113191401600462e-06,
+      "loss": 0.7119,
+      "step": 6206
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1111510427487281e-06,
+      "loss": 0.8368,
+      "step": 6207
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1091124454877777e-06,
+      "loss": 0.7756,
+      "step": 6208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1070756102216207e-06,
+      "loss": 0.8132,
+      "step": 6209
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1050405373539208e-06,
+      "loss": 0.7738,
+      "step": 6210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1030072272879833e-06,
+      "loss": 0.7247,
+      "step": 6211
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1009756804267746e-06,
+      "loss": 0.7094,
+      "step": 6212
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0989458971728995e-06,
+      "loss": 0.785,
+      "step": 6213
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0969178779286238e-06,
+      "loss": 0.7636,
+      "step": 6214
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0948916230958618e-06,
+      "loss": 0.7497,
+      "step": 6215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.092867133076173e-06,
+      "loss": 0.7204,
+      "step": 6216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0908444082707737e-06,
+      "loss": 0.833,
+      "step": 6217
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.088823449080524e-06,
+      "loss": 0.3323,
+      "step": 6218
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0868042559059411e-06,
+      "loss": 0.7629,
+      "step": 6219
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0847868291471842e-06,
+      "loss": 0.6336,
+      "step": 6220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0827711692040688e-06,
+      "loss": 0.7841,
+      "step": 6221
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0807572764760587e-06,
+      "loss": 0.7265,
+      "step": 6222
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.078745151362266e-06,
+      "loss": 0.7895,
+      "step": 6223
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0767347942614547e-06,
+      "loss": 0.7769,
+      "step": 6224
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0747262055720343e-06,
+      "loss": 0.8294,
+      "step": 6225
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0727193856920691e-06,
+      "loss": 0.7343,
+      "step": 6226
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0707143350192683e-06,
+      "loss": 0.766,
+      "step": 6227
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0687110539509937e-06,
+      "loss": 0.7619,
+      "step": 6228
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0667095428842577e-06,
+      "loss": 0.7352,
+      "step": 6229
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0647098022157142e-06,
+      "loss": 0.7203,
+      "step": 6230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0627118323416763e-06,
+      "loss": 0.8122,
+      "step": 6231
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0607156336580981e-06,
+      "loss": 0.7469,
+      "step": 6232
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.058721206560589e-06,
+      "loss": 0.7663,
+      "step": 6233
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0567285514444004e-06,
+      "loss": 0.74,
+      "step": 6234
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0547376687044397e-06,
+      "loss": 0.8202,
+      "step": 6235
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0527485587352592e-06,
+      "loss": 0.704,
+      "step": 6236
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0507612219310582e-06,
+      "loss": 0.3546,
+      "step": 6237
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0487756586856912e-06,
+      "loss": 0.6861,
+      "step": 6238
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0467918693926516e-06,
+      "loss": 0.7304,
+      "step": 6239
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0448098544450903e-06,
+      "loss": 0.771,
+      "step": 6240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0428296142358007e-06,
+      "loss": 0.7782,
+      "step": 6241
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.040851149157226e-06,
+      "loss": 0.7669,
+      "step": 6242
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0388744596014621e-06,
+      "loss": 0.7785,
+      "step": 6243
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0368995459602437e-06,
+      "loss": 0.6652,
+      "step": 6244
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0349264086249643e-06,
+      "loss": 0.7339,
+      "step": 6245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0329550479866547e-06,
+      "loss": 0.6756,
+      "step": 6246
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.030985464436003e-06,
+      "loss": 0.7443,
+      "step": 6247
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.029017658363337e-06,
+      "loss": 0.7997,
+      "step": 6248
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0270516301586397e-06,
+      "loss": 0.6968,
+      "step": 6249
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.025087380211538e-06,
+      "loss": 0.7734,
+      "step": 6250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0231249089113038e-06,
+      "loss": 0.7192,
+      "step": 6251
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0211642166468639e-06,
+      "loss": 0.7327,
+      "step": 6252
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.019205303806785e-06,
+      "loss": 0.7855,
+      "step": 6253
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0172481707792837e-06,
+      "loss": 0.8673,
+      "step": 6254
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.015292817952226e-06,
+      "loss": 0.7441,
+      "step": 6255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.013339245713121e-06,
+      "loss": 0.779,
+      "step": 6256
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.011387454449131e-06,
+      "loss": 0.7439,
+      "step": 6257
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0094374445470578e-06,
+      "loss": 0.3256,
+      "step": 6258
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.007489216393358e-06,
+      "loss": 0.8553,
+      "step": 6259
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0055427703741271e-06,
+      "loss": 0.7427,
+      "step": 6260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0035981068751143e-06,
+      "loss": 0.8174,
+      "step": 6261
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0016552262817124e-06,
+      "loss": 0.8251,
+      "step": 6262
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.997141289789602e-07,
+      "loss": 0.7007,
+      "step": 6263
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.977748153515464e-07,
+      "loss": 0.3287,
+      "step": 6264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.958372857838006e-07,
+      "loss": 0.7627,
+      "step": 6265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.939015406597042e-07,
+      "loss": 0.7369,
+      "step": 6266
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.919675803628815e-07,
+      "loss": 0.6977,
+      "step": 6267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.900354052766048e-07,
+      "loss": 0.7453,
+      "step": 6268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.881050157837958e-07,
+      "loss": 0.717,
+      "step": 6269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.861764122670125e-07,
+      "loss": 0.7971,
+      "step": 6270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.84249595108472e-07,
+      "loss": 0.7661,
+      "step": 6271
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.823245646900236e-07,
+      "loss": 0.7823,
+      "step": 6272
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.804013213931762e-07,
+      "loss": 0.7925,
+      "step": 6273
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.784798655990725e-07,
+      "loss": 0.7641,
+      "step": 6274
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.765601976885097e-07,
+      "loss": 0.821,
+      "step": 6275
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.746423180419274e-07,
+      "loss": 0.8191,
+      "step": 6276
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.727262270394078e-07,
+      "loss": 0.3226,
+      "step": 6277
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.70811925060685e-07,
+      "loss": 0.7611,
+      "step": 6278
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.688994124851315e-07,
+      "loss": 0.8096,
+      "step": 6279
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.669886896917724e-07,
+      "loss": 0.7449,
+      "step": 6280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.650797570592707e-07,
+      "loss": 0.7497,
+      "step": 6281
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.631726149659415e-07,
+      "loss": 0.8241,
+      "step": 6282
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.612672637897423e-07,
+      "loss": 0.8035,
+      "step": 6283
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.593637039082715e-07,
+      "loss": 0.7729,
+      "step": 6284
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.574619356987812e-07,
+      "loss": 0.7381,
+      "step": 6285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.555619595381581e-07,
+      "loss": 0.7038,
+      "step": 6286
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.536637758029455e-07,
+      "loss": 0.7818,
+      "step": 6287
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.517673848693187e-07,
+      "loss": 0.7032,
+      "step": 6288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.498727871131097e-07,
+      "loss": 0.7935,
+      "step": 6289
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.479799829097858e-07,
+      "loss": 0.7493,
+      "step": 6290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.460889726344635e-07,
+      "loss": 0.7391,
+      "step": 6291
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.441997566619055e-07,
+      "loss": 0.6994,
+      "step": 6292
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.423123353665131e-07,
+      "loss": 0.6867,
+      "step": 6293
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.404267091223373e-07,
+      "loss": 0.73,
+      "step": 6294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.385428783030703e-07,
+      "loss": 0.7301,
+      "step": 6295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.36660843282049e-07,
+      "loss": 0.7163,
+      "step": 6296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.34780604432256e-07,
+      "loss": 0.7426,
+      "step": 6297
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.329021621263145e-07,
+      "loss": 0.7369,
+      "step": 6298
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.310255167364968e-07,
+      "loss": 0.7044,
+      "step": 6299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.29150668634714e-07,
+      "loss": 0.6432,
+      "step": 6300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.272776181925236e-07,
+      "loss": 0.7742,
+      "step": 6301
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.254063657811286e-07,
+      "loss": 0.695,
+      "step": 6302
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.235369117713699e-07,
+      "loss": 0.7618,
+      "step": 6303
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.216692565337404e-07,
+      "loss": 0.7983,
+      "step": 6304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.19803400438366e-07,
+      "loss": 0.3564,
+      "step": 6305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.179393438550277e-07,
+      "loss": 0.7977,
+      "step": 6306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.160770871531388e-07,
+      "loss": 0.767,
+      "step": 6307
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.14216630701763e-07,
+      "loss": 0.7389,
+      "step": 6308
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.123579748696076e-07,
+      "loss": 0.7482,
+      "step": 6309
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.105011200250168e-07,
+      "loss": 0.7651,
+      "step": 6310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.086460665359853e-07,
+      "loss": 0.7106,
+      "step": 6311
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.067928147701444e-07,
+      "loss": 0.7971,
+      "step": 6312
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.049413650947747e-07,
+      "loss": 0.7791,
+      "step": 6313
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.030917178767918e-07,
+      "loss": 0.6904,
+      "step": 6314
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.012438734827623e-07,
+      "loss": 0.7095,
+      "step": 6315
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.993978322788888e-07,
+      "loss": 0.8005,
+      "step": 6316
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.97553594631021e-07,
+      "loss": 0.7748,
+      "step": 6317
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.95711160904652e-07,
+      "loss": 0.6837,
+      "step": 6318
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.938705314649099e-07,
+      "loss": 0.3025,
+      "step": 6319
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.920317066765761e-07,
+      "loss": 0.7244,
+      "step": 6320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.901946869040634e-07,
+      "loss": 0.7678,
+      "step": 6321
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.883594725114353e-07,
+      "loss": 0.7332,
+      "step": 6322
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.865260638623929e-07,
+      "loss": 0.7216,
+      "step": 6323
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.846944613202801e-07,
+      "loss": 0.7805,
+      "step": 6324
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.828646652480877e-07,
+      "loss": 0.7301,
+      "step": 6325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.810366760084387e-07,
+      "loss": 0.687,
+      "step": 6326
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.792104939636093e-07,
+      "loss": 0.7502,
+      "step": 6327
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.773861194755074e-07,
+      "loss": 0.7634,
+      "step": 6328
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.755635529056916e-07,
+      "loss": 0.6606,
+      "step": 6329
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.737427946153521e-07,
+      "loss": 0.7743,
+      "step": 6330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.719238449653311e-07,
+      "loss": 0.729,
+      "step": 6331
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.701067043161083e-07,
+      "loss": 0.8201,
+      "step": 6332
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.682913730277998e-07,
+      "loss": 0.7685,
+      "step": 6333
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.664778514601735e-07,
+      "loss": 0.7882,
+      "step": 6334
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.646661399726274e-07,
+      "loss": 0.7572,
+      "step": 6335
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.628562389242112e-07,
+      "loss": 0.7201,
+      "step": 6336
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.610481486736078e-07,
+      "loss": 0.6769,
+      "step": 6337
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.592418695791427e-07,
+      "loss": 0.7316,
+      "step": 6338
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.574374019987875e-07,
+      "loss": 0.6769,
+      "step": 6339
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.556347462901493e-07,
+      "loss": 0.7233,
+      "step": 6340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.538339028104792e-07,
+      "loss": 0.7893,
+      "step": 6341
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.520348719166671e-07,
+      "loss": 0.7899,
+      "step": 6342
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.502376539652446e-07,
+      "loss": 0.6609,
+      "step": 6343
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.484422493123868e-07,
+      "loss": 0.3054,
+      "step": 6344
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.466486583139022e-07,
+      "loss": 0.3274,
+      "step": 6345
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.448568813252489e-07,
+      "loss": 0.3168,
+      "step": 6346
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.430669187015184e-07,
+      "loss": 0.742,
+      "step": 6347
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.412787707974457e-07,
+      "loss": 0.7984,
+      "step": 6348
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.394924379674052e-07,
+      "loss": 0.711,
+      "step": 6349
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.377079205654126e-07,
+      "loss": 0.7487,
+      "step": 6350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.359252189451239e-07,
+      "loss": 0.7792,
+      "step": 6351
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.341443334598332e-07,
+      "loss": 0.826,
+      "step": 6352
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.323652644624791e-07,
+      "loss": 0.762,
+      "step": 6353
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.305880123056332e-07,
+      "loss": 0.7476,
+      "step": 6354
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.288125773415146e-07,
+      "loss": 0.7878,
+      "step": 6355
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.270389599219752e-07,
+      "loss": 0.7554,
+      "step": 6356
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.25267160398513e-07,
+      "loss": 0.7811,
+      "step": 6357
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.234971791222624e-07,
+      "loss": 0.7847,
+      "step": 6358
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.217290164439973e-07,
+      "loss": 0.3467,
+      "step": 6359
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.199626727141341e-07,
+      "loss": 0.7248,
+      "step": 6360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.181981482827228e-07,
+      "loss": 0.7174,
+      "step": 6361
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.164354434994603e-07,
+      "loss": 0.8336,
+      "step": 6362
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.146745587136761e-07,
+      "loss": 0.6874,
+      "step": 6363
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.129154942743433e-07,
+      "loss": 0.8238,
+      "step": 6364
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.111582505300753e-07,
+      "loss": 0.6617,
+      "step": 6365
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.094028278291189e-07,
+      "loss": 0.7324,
+      "step": 6366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.076492265193658e-07,
+      "loss": 0.6839,
+      "step": 6367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.058974469483427e-07,
+      "loss": 0.3433,
+      "step": 6368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.041474894632185e-07,
+      "loss": 0.8003,
+      "step": 6369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.023993544107988e-07,
+      "loss": 0.7579,
+      "step": 6370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.00653042137528e-07,
+      "loss": 0.7224,
+      "step": 6371
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.989085529894924e-07,
+      "loss": 0.801,
+      "step": 6372
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.971658873124122e-07,
+      "loss": 0.736,
+      "step": 6373
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.95425045451651e-07,
+      "loss": 0.7361,
+      "step": 6374
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.936860277522051e-07,
+      "loss": 0.7633,
+      "step": 6375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.919488345587167e-07,
+      "loss": 0.7578,
+      "step": 6376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.902134662154592e-07,
+      "loss": 0.7191,
+      "step": 6377
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.884799230663487e-07,
+      "loss": 0.8889,
+      "step": 6378
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.867482054549424e-07,
+      "loss": 0.7555,
+      "step": 6379
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.85018313724425e-07,
+      "loss": 0.7537,
+      "step": 6380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.832902482176308e-07,
+      "loss": 0.6911,
+      "step": 6381
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.815640092770249e-07,
+      "loss": 0.7685,
+      "step": 6382
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.798395972447148e-07,
+      "loss": 0.7751,
+      "step": 6383
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.78117012462446e-07,
+      "loss": 0.8089,
+      "step": 6384
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.763962552715954e-07,
+      "loss": 0.7925,
+      "step": 6385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.746773260131868e-07,
+      "loss": 0.7976,
+      "step": 6386
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.729602250278734e-07,
+      "loss": 0.7713,
+      "step": 6387
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.712449526559552e-07,
+      "loss": 0.3801,
+      "step": 6388
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.69531509237359e-07,
+      "loss": 0.7453,
+      "step": 6389
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.678198951116567e-07,
+      "loss": 0.747,
+      "step": 6390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.661101106180591e-07,
+      "loss": 0.8315,
+      "step": 6391
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.644021560954062e-07,
+      "loss": 0.7566,
+      "step": 6392
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.626960318821841e-07,
+      "loss": 0.6975,
+      "step": 6393
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.609917383165089e-07,
+      "loss": 0.7606,
+      "step": 6394
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.592892757361403e-07,
+      "loss": 0.8163,
+      "step": 6395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.575886444784697e-07,
+      "loss": 0.7493,
+      "step": 6396
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.558898448805285e-07,
+      "loss": 0.7985,
+      "step": 6397
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.541928772789864e-07,
+      "loss": 0.7771,
+      "step": 6398
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.524977420101442e-07,
+      "loss": 0.8116,
+      "step": 6399
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.50804439409949e-07,
+      "loss": 0.7805,
+      "step": 6400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.491129698139743e-07,
+      "loss": 0.7283,
+      "step": 6401
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.474233335574388e-07,
+      "loss": 0.7447,
+      "step": 6402
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.457355309751913e-07,
+      "loss": 0.7446,
+      "step": 6403
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.440495624017219e-07,
+      "loss": 0.7895,
+      "step": 6404
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.423654281711568e-07,
+      "loss": 0.7919,
+      "step": 6405
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.406831286172545e-07,
+      "loss": 0.7942,
+      "step": 6406
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.390026640734172e-07,
+      "loss": 0.6955,
+      "step": 6407
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.37324034872674e-07,
+      "loss": 0.7952,
+      "step": 6408
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.356472413477001e-07,
+      "loss": 0.8357,
+      "step": 6409
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.339722838307983e-07,
+      "loss": 0.78,
+      "step": 6410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.322991626539122e-07,
+      "loss": 0.679,
+      "step": 6411
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.306278781486243e-07,
+      "loss": 0.7573,
+      "step": 6412
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.28958430646145e-07,
+      "loss": 0.6946,
+      "step": 6413
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.272908204773287e-07,
+      "loss": 0.8003,
+      "step": 6414
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.256250479726589e-07,
+      "loss": 0.6899,
+      "step": 6415
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.239611134622615e-07,
+      "loss": 0.7228,
+      "step": 6416
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.222990172758914e-07,
+      "loss": 0.712,
+      "step": 6417
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.206387597429454e-07,
+      "loss": 0.8225,
+      "step": 6418
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.189803411924511e-07,
+      "loss": 0.6803,
+      "step": 6419
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.173237619530748e-07,
+      "loss": 0.7325,
+      "step": 6420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.156690223531182e-07,
+      "loss": 0.7915,
+      "step": 6421
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.140161227205133e-07,
+      "loss": 0.7469,
+      "step": 6422
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.123650633828339e-07,
+      "loss": 0.7536,
+      "step": 6423
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.107158446672891e-07,
+      "loss": 0.7937,
+      "step": 6424
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.090684669007176e-07,
+      "loss": 0.8044,
+      "step": 6425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.074229304095981e-07,
+      "loss": 0.7547,
+      "step": 6426
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.05779235520041e-07,
+      "loss": 0.7199,
+      "step": 6427
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.041373825577969e-07,
+      "loss": 0.7193,
+      "step": 6428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.024973718482441e-07,
+      "loss": 0.7984,
+      "step": 6429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.008592037164008e-07,
+      "loss": 0.7336,
+      "step": 6430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.992228784869215e-07,
+      "loss": 0.7942,
+      "step": 6431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.97588396484089e-07,
+      "loss": 0.7913,
+      "step": 6432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.959557580318288e-07,
+      "loss": 0.7753,
+      "step": 6433
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.943249634536919e-07,
+      "loss": 0.7778,
+      "step": 6434
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.926960130728744e-07,
+      "loss": 0.7742,
+      "step": 6435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.910689072121979e-07,
+      "loss": 0.7694,
+      "step": 6436
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.894436461941212e-07,
+      "loss": 0.7771,
+      "step": 6437
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.878202303407433e-07,
+      "loss": 0.6782,
+      "step": 6438
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.861986599737869e-07,
+      "loss": 0.7844,
+      "step": 6439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.845789354146182e-07,
+      "loss": 0.7094,
+      "step": 6440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.829610569842316e-07,
+      "loss": 0.7272,
+      "step": 6441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.813450250032604e-07,
+      "loss": 0.8153,
+      "step": 6442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.797308397919677e-07,
+      "loss": 0.7742,
+      "step": 6443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.781185016702519e-07,
+      "loss": 0.721,
+      "step": 6444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.765080109576494e-07,
+      "loss": 0.8147,
+      "step": 6445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.748993679733229e-07,
+      "loss": 0.6957,
+      "step": 6446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.732925730360762e-07,
+      "loss": 0.7572,
+      "step": 6447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.716876264643413e-07,
+      "loss": 0.7489,
+      "step": 6448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.700845285761881e-07,
+      "loss": 0.8095,
+      "step": 6449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.68483279689317e-07,
+      "loss": 0.6262,
+      "step": 6450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.668838801210642e-07,
+      "loss": 0.7429,
+      "step": 6451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.652863301883971e-07,
+      "loss": 0.8007,
+      "step": 6452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.636906302079194e-07,
+      "loss": 0.7196,
+      "step": 6453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.620967804958667e-07,
+      "loss": 0.6842,
+      "step": 6454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.605047813681065e-07,
+      "loss": 0.7494,
+      "step": 6455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.589146331401431e-07,
+      "loss": 0.6885,
+      "step": 6456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.573263361271099e-07,
+      "loss": 0.7206,
+      "step": 6457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.557398906437762e-07,
+      "loss": 0.8414,
+      "step": 6458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.541552970045428e-07,
+      "loss": 0.7952,
+      "step": 6459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.525725555234453e-07,
+      "loss": 0.7318,
+      "step": 6460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.509916665141514e-07,
+      "loss": 0.7994,
+      "step": 6461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.494126302899606e-07,
+      "loss": 0.3761,
+      "step": 6462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.478354471638049e-07,
+      "loss": 0.7309,
+      "step": 6463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.46260117448253e-07,
+      "loss": 0.7377,
+      "step": 6464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.446866414555008e-07,
+      "loss": 0.7937,
+      "step": 6465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.431150194973823e-07,
+      "loss": 0.7783,
+      "step": 6466
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.415452518853571e-07,
+      "loss": 0.7737,
+      "step": 6467
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.399773389305275e-07,
+      "loss": 0.7494,
+      "step": 6468
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.38411280943616e-07,
+      "loss": 0.7041,
+      "step": 6469
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.368470782349878e-07,
+      "loss": 0.6905,
+      "step": 6470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.352847311146359e-07,
+      "loss": 0.777,
+      "step": 6471
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.33724239892185e-07,
+      "loss": 0.8036,
+      "step": 6472
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.321656048768943e-07,
+      "loss": 0.7817,
+      "step": 6473
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.306088263776511e-07,
+      "loss": 0.7199,
+      "step": 6474
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.29053904702982e-07,
+      "loss": 0.8352,
+      "step": 6475
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.275008401610361e-07,
+      "loss": 0.6941,
+      "step": 6476
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.259496330596038e-07,
+      "loss": 0.7945,
+      "step": 6477
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.244002837060991e-07,
+      "loss": 0.7355,
+      "step": 6478
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.228527924075744e-07,
+      "loss": 0.7639,
+      "step": 6479
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.213071594707132e-07,
+      "loss": 0.8299,
+      "step": 6480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.197633852018248e-07,
+      "loss": 0.7902,
+      "step": 6481
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.182214699068567e-07,
+      "loss": 0.7134,
+      "step": 6482
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.166814138913835e-07,
+      "loss": 0.7855,
+      "step": 6483
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.151432174606154e-07,
+      "loss": 0.6839,
+      "step": 6484
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.136068809193896e-07,
+      "loss": 0.7319,
+      "step": 6485
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.120724045721793e-07,
+      "loss": 0.7385,
+      "step": 6486
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.105397887230857e-07,
+      "loss": 0.7801,
+      "step": 6487
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.090090336758426e-07,
+      "loss": 0.7716,
+      "step": 6488
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.074801397338159e-07,
+      "loss": 0.7254,
+      "step": 6489
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.05953107199998e-07,
+      "loss": 0.7057,
+      "step": 6490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.044279363770211e-07,
+      "loss": 0.8427,
+      "step": 6491
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.029046275671391e-07,
+      "loss": 0.7643,
+      "step": 6492
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.013831810722425e-07,
+      "loss": 0.7359,
+      "step": 6493
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.998635971938526e-07,
+      "loss": 0.8075,
+      "step": 6494
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.983458762331174e-07,
+      "loss": 0.7688,
+      "step": 6495
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.968300184908227e-07,
+      "loss": 0.788,
+      "step": 6496
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.953160242673772e-07,
+      "loss": 0.7441,
+      "step": 6497
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.938038938628276e-07,
+      "loss": 0.7816,
+      "step": 6498
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.922936275768432e-07,
+      "loss": 0.7677,
+      "step": 6499
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.90785225708731e-07,
+      "loss": 0.756,
+      "step": 6500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.892786885574275e-07,
+      "loss": 0.7611,
+      "step": 6501
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.877740164214941e-07,
+      "loss": 0.7958,
+      "step": 6502
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.862712095991307e-07,
+      "loss": 0.7253,
+      "step": 6503
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.847702683881606e-07,
+      "loss": 0.7282,
+      "step": 6504
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.832711930860402e-07,
+      "loss": 0.736,
+      "step": 6505
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.817739839898573e-07,
+      "loss": 0.767,
+      "step": 6506
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.802786413963268e-07,
+      "loss": 0.7756,
+      "step": 6507
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.787851656017974e-07,
+      "loss": 0.8303,
+      "step": 6508
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.772935569022442e-07,
+      "loss": 0.3431,
+      "step": 6509
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.758038155932755e-07,
+      "loss": 0.7399,
+      "step": 6510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.743159419701261e-07,
+      "loss": 0.7349,
+      "step": 6511
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.728299363276646e-07,
+      "loss": 0.7778,
+      "step": 6512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.713457989603877e-07,
+      "loss": 0.3222,
+      "step": 6513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.698635301624189e-07,
+      "loss": 0.7911,
+      "step": 6514
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.683831302275167e-07,
+      "loss": 0.7786,
+      "step": 6515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.66904599449064e-07,
+      "loss": 0.8276,
+      "step": 6516
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.654279381200789e-07,
+      "loss": 0.7693,
+      "step": 6517
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.639531465332027e-07,
+      "loss": 0.8148,
+      "step": 6518
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.624802249807104e-07,
+      "loss": 0.8241,
+      "step": 6519
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.610091737545076e-07,
+      "loss": 0.7384,
+      "step": 6520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.595399931461243e-07,
+      "loss": 0.7347,
+      "step": 6521
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.580726834467242e-07,
+      "loss": 0.3429,
+      "step": 6522
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.566072449470972e-07,
+      "loss": 0.3303,
+      "step": 6523
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.551436779376651e-07,
+      "loss": 0.7796,
+      "step": 6524
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.536819827084749e-07,
+      "loss": 0.7737,
+      "step": 6525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.522221595492072e-07,
+      "loss": 0.8072,
+      "step": 6526
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.507642087491704e-07,
+      "loss": 0.6819,
+      "step": 6527
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.493081305972992e-07,
+      "loss": 0.8101,
+      "step": 6528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.478539253821602e-07,
+      "loss": 0.8542,
+      "step": 6529
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.464015933919454e-07,
+      "loss": 0.7206,
+      "step": 6530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.44951134914481e-07,
+      "loss": 0.7294,
+      "step": 6531
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.435025502372149e-07,
+      "loss": 0.8067,
+      "step": 6532
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.420558396472297e-07,
+      "loss": 0.7183,
+      "step": 6533
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.406110034312362e-07,
+      "loss": 0.7247,
+      "step": 6534
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.391680418755673e-07,
+      "loss": 0.7475,
+      "step": 6535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.377269552661924e-07,
+      "loss": 0.7945,
+      "step": 6536
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.362877438887038e-07,
+      "loss": 0.77,
+      "step": 6537
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.348504080283267e-07,
+      "loss": 0.7514,
+      "step": 6538
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.334149479699081e-07,
+      "loss": 0.7203,
+      "step": 6539
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.319813639979299e-07,
+      "loss": 0.7184,
+      "step": 6540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.305496563965006e-07,
+      "loss": 0.7575,
+      "step": 6541
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.291198254493535e-07,
+      "loss": 0.8085,
+      "step": 6542
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.276918714398549e-07,
+      "loss": 0.7186,
+      "step": 6543
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.26265794650993e-07,
+      "loss": 0.8092,
+      "step": 6544
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.248415953653907e-07,
+      "loss": 0.7974,
+      "step": 6545
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.234192738652955e-07,
+      "loss": 0.3377,
+      "step": 6546
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.219988304325785e-07,
+      "loss": 0.8155,
+      "step": 6547
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.20580265348748e-07,
+      "loss": 0.8427,
+      "step": 6548
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.191635788949323e-07,
+      "loss": 0.7394,
+      "step": 6549
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.177487713518925e-07,
+      "loss": 0.6837,
+      "step": 6550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.163358430000109e-07,
+      "loss": 0.7785,
+      "step": 6551
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.149247941193048e-07,
+      "loss": 0.7673,
+      "step": 6552
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.135156249894158e-07,
+      "loss": 0.8087,
+      "step": 6553
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.121083358896106e-07,
+      "loss": 0.7469,
+      "step": 6554
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.107029270987885e-07,
+      "loss": 0.7021,
+      "step": 6555
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.092993988954698e-07,
+      "loss": 0.703,
+      "step": 6556
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.078977515578087e-07,
+      "loss": 0.6842,
+      "step": 6557
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.06497985363581e-07,
+      "loss": 0.7604,
+      "step": 6558
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.051001005901923e-07,
+      "loss": 0.7164,
+      "step": 6559
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.037040975146778e-07,
+      "loss": 0.7983,
+      "step": 6560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.023099764136941e-07,
+      "loss": 0.6609,
+      "step": 6561
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.009177375635299e-07,
+      "loss": 0.7957,
+      "step": 6562
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.995273812400969e-07,
+      "loss": 0.7148,
+      "step": 6563
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.981389077189392e-07,
+      "loss": 0.7528,
+      "step": 6564
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.967523172752209e-07,
+      "loss": 0.7794,
+      "step": 6565
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.953676101837368e-07,
+      "loss": 0.7764,
+      "step": 6566
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.939847867189107e-07,
+      "loss": 0.7958,
+      "step": 6567
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.926038471547856e-07,
+      "loss": 0.7448,
+      "step": 6568
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.912247917650414e-07,
+      "loss": 0.8018,
+      "step": 6569
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.898476208229741e-07,
+      "loss": 0.7604,
+      "step": 6570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.884723346015152e-07,
+      "loss": 0.8086,
+      "step": 6571
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.870989333732156e-07,
+      "loss": 0.767,
+      "step": 6572
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.857274174102566e-07,
+      "loss": 0.7144,
+      "step": 6573
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.843577869844474e-07,
+      "loss": 0.366,
+      "step": 6574
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.829900423672185e-07,
+      "loss": 0.7418,
+      "step": 6575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.816241838296309e-07,
+      "loss": 0.8108,
+      "step": 6576
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.802602116423671e-07,
+      "loss": 0.7523,
+      "step": 6577
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.788981260757442e-07,
+      "loss": 0.7643,
+      "step": 6578
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.77537927399695e-07,
+      "loss": 0.8568,
+      "step": 6579
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7617961588378724e-07,
+      "loss": 0.7423,
+      "step": 6580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.748231917972079e-07,
+      "loss": 0.6952,
+      "step": 6581
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.734686554087753e-07,
+      "loss": 0.7563,
+      "step": 6582
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7211600698693017e-07,
+      "loss": 0.752,
+      "step": 6583
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.707652467997392e-07,
+      "loss": 0.7723,
+      "step": 6584
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6941637511489945e-07,
+      "loss": 0.73,
+      "step": 6585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.680693921997248e-07,
+      "loss": 0.7551,
+      "step": 6586
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6672429832116505e-07,
+      "loss": 0.6987,
+      "step": 6587
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.653810937457892e-07,
+      "loss": 0.3578,
+      "step": 6588
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.64039778739791e-07,
+      "loss": 0.7714,
+      "step": 6589
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.627003535689945e-07,
+      "loss": 0.7508,
+      "step": 6590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.61362818498845e-07,
+      "loss": 0.8015,
+      "step": 6591
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.600271737944162e-07,
+      "loss": 0.7614,
+      "step": 6592
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.586934197204074e-07,
+      "loss": 0.7588,
+      "step": 6593
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.573615565411382e-07,
+      "loss": 0.6855,
+      "step": 6594
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5603158452056095e-07,
+      "loss": 0.6944,
+      "step": 6595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5470350392224695e-07,
+      "loss": 0.7708,
+      "step": 6596
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5337731500939566e-07,
+      "loss": 0.7294,
+      "step": 6597
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5205301804483014e-07,
+      "loss": 0.6896,
+      "step": 6598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5073061329099945e-07,
+      "loss": 0.8333,
+      "step": 6599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.494101010099805e-07,
+      "loss": 0.8318,
+      "step": 6600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.480914814634685e-07,
+      "loss": 0.7495,
+      "step": 6601
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4677475491278876e-07,
+      "loss": 0.6939,
+      "step": 6602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.454599216188904e-07,
+      "loss": 0.7929,
+      "step": 6603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4414698184234607e-07,
+      "loss": 0.8014,
+      "step": 6604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4283593584335427e-07,
+      "loss": 0.7511,
+      "step": 6605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.415267838817383e-07,
+      "loss": 0.7472,
+      "step": 6606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.402195262169451e-07,
+      "loss": 0.7238,
+      "step": 6607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.389141631080462e-07,
+      "loss": 0.8012,
+      "step": 6608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.376106948137404e-07,
+      "loss": 0.7503,
+      "step": 6609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.363091215923465e-07,
+      "loss": 0.6706,
+      "step": 6610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3500944370181266e-07,
+      "loss": 0.7662,
+      "step": 6611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.337116613997061e-07,
+      "loss": 0.8008,
+      "step": 6612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.324157749432234e-07,
+      "loss": 0.707,
+      "step": 6613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.311217845891813e-07,
+      "loss": 0.741,
+      "step": 6614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.298296905940236e-07,
+      "loss": 0.3146,
+      "step": 6615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.285394932138176e-07,
+      "loss": 0.7679,
+      "step": 6616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.272511927042533e-07,
+      "loss": 0.736,
+      "step": 6617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2596478932064754e-07,
+      "loss": 0.7897,
+      "step": 6618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2468028331793756e-07,
+      "loss": 0.3564,
+      "step": 6619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.233976749506885e-07,
+      "loss": 0.8707,
+      "step": 6620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2211696447308603e-07,
+      "loss": 0.7795,
+      "step": 6621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2083815213894154e-07,
+      "loss": 0.7445,
+      "step": 6622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1956123820169117e-07,
+      "loss": 0.7746,
+      "step": 6623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1828622291439137e-07,
+      "loss": 0.7888,
+      "step": 6624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.170131065297267e-07,
+      "loss": 0.8183,
+      "step": 6625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.157418893000009e-07,
+      "loss": 0.7273,
+      "step": 6626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.144725714771458e-07,
+      "loss": 0.802,
+      "step": 6627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1320515331271236e-07,
+      "loss": 0.7578,
+      "step": 6628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.119396350578797e-07,
+      "loss": 0.6902,
+      "step": 6629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1067601696344717e-07,
+      "loss": 0.7585,
+      "step": 6630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0941429927983776e-07,
+      "loss": 0.7964,
+      "step": 6631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0815448225709707e-07,
+      "loss": 0.7502,
+      "step": 6632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.068965661448998e-07,
+      "loss": 0.7804,
+      "step": 6633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.056405511925354e-07,
+      "loss": 0.7691,
+      "step": 6634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.043864376489237e-07,
+      "loss": 0.7547,
+      "step": 6635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.031342257626014e-07,
+      "loss": 0.7797,
+      "step": 6636
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0188391578173556e-07,
+      "loss": 0.792,
+      "step": 6637
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.006355079541102e-07,
+      "loss": 0.7015,
+      "step": 6638
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.993890025271352e-07,
+      "loss": 0.7896,
+      "step": 6639
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.981443997478407e-07,
+      "loss": 0.7444,
+      "step": 6640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9690169986288494e-07,
+      "loss": 0.7178,
+      "step": 6641
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.956609031185454e-07,
+      "loss": 0.7324,
+      "step": 6642
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9442200976072097e-07,
+      "loss": 0.7153,
+      "step": 6643
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9318502003493855e-07,
+      "loss": 0.7516,
+      "step": 6644
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.91949934186342e-07,
+      "loss": 0.7822,
+      "step": 6645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.907167524597022e-07,
+      "loss": 0.6841,
+      "step": 6646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8948547509940813e-07,
+      "loss": 0.8069,
+      "step": 6647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8825610234947685e-07,
+      "loss": 0.7401,
+      "step": 6648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8702863445354566e-07,
+      "loss": 0.8024,
+      "step": 6649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.858030716548711e-07,
+      "loss": 0.724,
+      "step": 6650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.845794141963377e-07,
+      "loss": 0.7302,
+      "step": 6651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8335766232044824e-07,
+      "loss": 0.7921,
+      "step": 6652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.821378162693312e-07,
+      "loss": 0.3484,
+      "step": 6653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8091987628473216e-07,
+      "loss": 0.8147,
+      "step": 6654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.797038426080246e-07,
+      "loss": 0.7666,
+      "step": 6655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7848971548020253e-07,
+      "loss": 0.7959,
+      "step": 6656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7727749514187894e-07,
+      "loss": 0.8461,
+      "step": 6657
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.76067181833295e-07,
+      "loss": 0.7431,
+      "step": 6658
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7485877579430675e-07,
+      "loss": 0.7755,
+      "step": 6659
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.736522772643991e-07,
+      "loss": 0.7382,
+      "step": 6660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7244768648267204e-07,
+      "loss": 0.72,
+      "step": 6661
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7124500368785457e-07,
+      "loss": 0.7758,
+      "step": 6662
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7004422911829375e-07,
+      "loss": 0.7298,
+      "step": 6663
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6884536301195707e-07,
+      "loss": 0.751,
+      "step": 6664
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.676484056064389e-07,
+      "loss": 0.8058,
+      "step": 6665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6645335713894724e-07,
+      "loss": 0.7976,
+      "step": 6666
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.652602178463216e-07,
+      "loss": 0.7655,
+      "step": 6667
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6406898796501387e-07,
+      "loss": 0.8069,
+      "step": 6668
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.628796677311042e-07,
+      "loss": 0.6899,
+      "step": 6669
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.616922573802928e-07,
+      "loss": 0.7606,
+      "step": 6670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.605067571478993e-07,
+      "loss": 0.711,
+      "step": 6671
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5932316726886575e-07,
+      "loss": 0.7616,
+      "step": 6672
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.581414879777545e-07,
+      "loss": 0.7456,
+      "step": 6673
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.569617195087527e-07,
+      "loss": 0.3677,
+      "step": 6674
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.557838620956666e-07,
+      "loss": 0.7713,
+      "step": 6675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.546079159719218e-07,
+      "loss": 0.7111,
+      "step": 6676
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5343388137057067e-07,
+      "loss": 0.7705,
+      "step": 6677
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5226175852427933e-07,
+      "loss": 0.7664,
+      "step": 6678
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5109154766534205e-07,
+      "loss": 0.7689,
+      "step": 6679
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4992324902566767e-07,
+      "loss": 0.7678,
+      "step": 6680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.487568628367921e-07,
+      "loss": 0.7012,
+      "step": 6681
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475923893298705e-07,
+      "loss": 0.8277,
+      "step": 6682
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4642982873567487e-07,
+      "loss": 0.7522,
+      "step": 6683
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.452691812846043e-07,
+      "loss": 0.8261,
+      "step": 6684
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.441104472066725e-07,
+      "loss": 0.7428,
+      "step": 6685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4295362673152143e-07,
+      "loss": 0.7146,
+      "step": 6686
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4179872008840543e-07,
+      "loss": 0.7765,
+      "step": 6687
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.406457275062058e-07,
+      "loss": 0.7445,
+      "step": 6688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3949464921342325e-07,
+      "loss": 0.7449,
+      "step": 6689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3834548543817516e-07,
+      "loss": 0.7693,
+      "step": 6690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3719823640820713e-07,
+      "loss": 0.7547,
+      "step": 6691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.360529023508763e-07,
+      "loss": 0.7444,
+      "step": 6692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3490948349316874e-07,
+      "loss": 0.342,
+      "step": 6693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.337679800616833e-07,
+      "loss": 0.7809,
+      "step": 6694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3262839228264565e-07,
+      "loss": 0.6998,
+      "step": 6695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3149072038189955e-07,
+      "loss": 0.6894,
+      "step": 6696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.303549645849058e-07,
+      "loss": 0.7529,
+      "step": 6697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.29221125116751e-07,
+      "loss": 0.8138,
+      "step": 6698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2808920220213757e-07,
+      "loss": 0.7773,
+      "step": 6699
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2695919606539174e-07,
+      "loss": 0.8105,
+      "step": 6700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.258311069304554e-07,
+      "loss": 0.7342,
+      "step": 6701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.247049350208942e-07,
+      "loss": 0.7755,
+      "step": 6702
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.23580680559894e-07,
+      "loss": 0.8308,
+      "step": 6703
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.224583437702566e-07,
+      "loss": 0.6679,
+      "step": 6704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2133792487441083e-07,
+      "loss": 0.7622,
+      "step": 6705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2021942409439676e-07,
+      "loss": 0.3254,
+      "step": 6706
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1910284165188156e-07,
+      "loss": 0.7907,
+      "step": 6707
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1798817776814703e-07,
+      "loss": 0.7232,
+      "step": 6708
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1687543266409883e-07,
+      "loss": 0.7477,
+      "step": 6709
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.157646065602615e-07,
+      "loss": 0.7683,
+      "step": 6710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.146556996767758e-07,
+      "loss": 0.7349,
+      "step": 6711
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.135487122334069e-07,
+      "loss": 0.7195,
+      "step": 6712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.12443644449536e-07,
+      "loss": 0.7778,
+      "step": 6713
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1134049654416576e-07,
+      "loss": 0.6496,
+      "step": 6714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1023926873591794e-07,
+      "loss": 0.3484,
+      "step": 6715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.091399612430335e-07,
+      "loss": 0.7477,
+      "step": 6716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0804257428337367e-07,
+      "loss": 0.7419,
+      "step": 6717
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0694710807441796e-07,
+      "loss": 0.7716,
+      "step": 6718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0585356283326597e-07,
+      "loss": 0.6874,
+      "step": 6719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.047619387766354e-07,
+      "loss": 0.7159,
+      "step": 6720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.036722361208644e-07,
+      "loss": 0.7525,
+      "step": 6721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.025844550819113e-07,
+      "loss": 0.7307,
+      "step": 6722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0149859587535136e-07,
+      "loss": 0.7488,
+      "step": 6723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0041465871638143e-07,
+      "loss": 0.8216,
+      "step": 6724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.99332643819813e-07,
+      "loss": 0.7167,
+      "step": 6725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9825255140008336e-07,
+      "loss": 0.8074,
+      "step": 6726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9717438167124133e-07,
+      "loss": 0.7196,
+      "step": 6727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.960981348469605e-07,
+      "loss": 0.7863,
+      "step": 6728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9502381114053236e-07,
+      "loss": 0.769,
+      "step": 6729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9395141076486333e-07,
+      "loss": 0.8,
+      "step": 6730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.928809339324845e-07,
+      "loss": 0.6829,
+      "step": 6731
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.918123808555406e-07,
+      "loss": 0.7759,
+      "step": 6732
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.907457517458001e-07,
+      "loss": 0.7151,
+      "step": 6733
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.896810468146427e-07,
+      "loss": 0.78,
+      "step": 6734
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.886182662730752e-07,
+      "loss": 0.7635,
+      "step": 6735
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.875574103317202e-07,
+      "loss": 0.7362,
+      "step": 6736
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8649847920081406e-07,
+      "loss": 0.6862,
+      "step": 6737
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8544147309021885e-07,
+      "loss": 0.8168,
+      "step": 6738
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.843863922094092e-07,
+      "loss": 0.7243,
+      "step": 6739
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.833332367674846e-07,
+      "loss": 0.7072,
+      "step": 6740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8228200697315576e-07,
+      "loss": 0.7967,
+      "step": 6741
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8123270303475725e-07,
+      "loss": 0.7787,
+      "step": 6742
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.801853251602382e-07,
+      "loss": 0.766,
+      "step": 6743
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7913987355716823e-07,
+      "loss": 0.6845,
+      "step": 6744
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7809634843273703e-07,
+      "loss": 0.8094,
+      "step": 6745
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7705474999374814e-07,
+      "loss": 0.3674,
+      "step": 6746
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7601507844662644e-07,
+      "loss": 0.6986,
+      "step": 6747
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.749773339974116e-07,
+      "loss": 0.6995,
+      "step": 6748
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7394151685176783e-07,
+      "loss": 0.8001,
+      "step": 6749
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.729076272149689e-07,
+      "loss": 0.8115,
+      "step": 6750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.71875665291913e-07,
+      "loss": 0.7758,
+      "step": 6751
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7084563128711436e-07,
+      "loss": 0.7273,
+      "step": 6752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.698175254047042e-07,
+      "loss": 0.6607,
+      "step": 6753
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6879134784843386e-07,
+      "loss": 0.7992,
+      "step": 6754
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6776709882166963e-07,
+      "loss": 0.7836,
+      "step": 6755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.667447785273969e-07,
+      "loss": 0.7735,
+      "step": 6756
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.657243871682191e-07,
+      "loss": 0.7669,
+      "step": 6757
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.647059249463579e-07,
+      "loss": 0.7708,
+      "step": 6758
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6368939206365296e-07,
+      "loss": 0.3264,
+      "step": 6759
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6267478872155747e-07,
+      "loss": 0.6829,
+      "step": 6760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.616621151211474e-07,
+      "loss": 0.708,
+      "step": 6761
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6065137146311536e-07,
+      "loss": 0.778,
+      "step": 6762
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5964255794776795e-07,
+      "loss": 0.7386,
+      "step": 6763
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5863567477503406e-07,
+      "loss": 0.7706,
+      "step": 6764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5763072214445515e-07,
+      "loss": 0.6143,
+      "step": 6765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5662770025519426e-07,
+      "loss": 0.7628,
+      "step": 6766
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5562660930603e-07,
+      "loss": 0.7202,
+      "step": 6767
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.546274494953582e-07,
+      "loss": 0.7569,
+      "step": 6768
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5363022102119164e-07,
+      "loss": 0.7513,
+      "step": 6769
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.52634924081161e-07,
+      "loss": 0.7798,
+      "step": 6770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.516415588725163e-07,
+      "loss": 0.7653,
+      "step": 6771
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.50650125592119e-07,
+      "loss": 0.7204,
+      "step": 6772
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.496606244364541e-07,
+      "loss": 0.7405,
+      "step": 6773
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4867305560161815e-07,
+      "loss": 0.7154,
+      "step": 6774
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4768741928333117e-07,
+      "loss": 0.3213,
+      "step": 6775
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.467037156769214e-07,
+      "loss": 0.7788,
+      "step": 6776
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.457219449773429e-07,
+      "loss": 0.831,
+      "step": 6777
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4474210737916337e-07,
+      "loss": 0.8019,
+      "step": 6778
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.43764203076563e-07,
+      "loss": 0.7556,
+      "step": 6779
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.427882322633468e-07,
+      "loss": 0.7875,
+      "step": 6780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.418141951329289e-07,
+      "loss": 0.6741,
+      "step": 6781
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4084209187834717e-07,
+      "loss": 0.698,
+      "step": 6782
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.398719226922508e-07,
+      "loss": 0.7772,
+      "step": 6783
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.389036877669071e-07,
+      "loss": 0.7941,
+      "step": 6784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3793738729420368e-07,
+      "loss": 0.8125,
+      "step": 6785
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3697302146563962e-07,
+      "loss": 0.7394,
+      "step": 6786
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.360105904723331e-07,
+      "loss": 0.7728,
+      "step": 6787
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3505009450501936e-07,
+      "loss": 0.7013,
+      "step": 6788
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3409153375404837e-07,
+      "loss": 0.7044,
+      "step": 6789
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3313490840938702e-07,
+      "loss": 0.7611,
+      "step": 6790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3218021866062035e-07,
+      "loss": 0.683,
+      "step": 6791
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3122746469695035e-07,
+      "loss": 0.7124,
+      "step": 6792
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.302766467071893e-07,
+      "loss": 0.7048,
+      "step": 6793
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2932776487977426e-07,
+      "loss": 0.7503,
+      "step": 6794
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2838081940275148e-07,
+      "loss": 0.7723,
+      "step": 6795
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.274358104637886e-07,
+      "loss": 0.8227,
+      "step": 6796
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2649273825016694e-07,
+      "loss": 0.8293,
+      "step": 6797
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2555160294878253e-07,
+      "loss": 0.7795,
+      "step": 6798
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2461240474615176e-07,
+      "loss": 0.8362,
+      "step": 6799
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.236751438284035e-07,
+      "loss": 0.8237,
+      "step": 6800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.227398203812847e-07,
+      "loss": 0.732,
+      "step": 6801
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2180643459015716e-07,
+      "loss": 0.7724,
+      "step": 6802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.208749866399995e-07,
+      "loss": 0.7406,
+      "step": 6803
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1994547671540634e-07,
+      "loss": 0.6876,
+      "step": 6804
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1901790500058584e-07,
+      "loss": 0.7748,
+      "step": 6805
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1809227167936653e-07,
+      "loss": 0.665,
+      "step": 6806
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1716857693518833e-07,
+      "loss": 0.7996,
+      "step": 6807
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1624682095111148e-07,
+      "loss": 0.7393,
+      "step": 6808
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1532700390980544e-07,
+      "loss": 0.7328,
+      "step": 6809
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.144091259935621e-07,
+      "loss": 0.6643,
+      "step": 6810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1349318738428716e-07,
+      "loss": 0.7492,
+      "step": 6811
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1257918826349978e-07,
+      "loss": 0.7653,
+      "step": 6812
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.116671288123362e-07,
+      "loss": 0.7721,
+      "step": 6813
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.107570092115485e-07,
+      "loss": 0.7597,
+      "step": 6814
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0984882964150467e-07,
+      "loss": 0.8164,
+      "step": 6815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.089425902821851e-07,
+      "loss": 0.7612,
+      "step": 6816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0803829131319064e-07,
+      "loss": 0.7099,
+      "step": 6817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0713593291373568e-07,
+      "loss": 0.8115,
+      "step": 6818
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.062355152626472e-07,
+      "loss": 0.7308,
+      "step": 6819
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0533703853837238e-07,
+      "loss": 0.8046,
+      "step": 6820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0444050291896777e-07,
+      "loss": 0.7426,
+      "step": 6821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0354590858211233e-07,
+      "loss": 0.7946,
+      "step": 6822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0265325570509308e-07,
+      "loss": 0.8361,
+      "step": 6823
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0176254446481859e-07,
+      "loss": 0.7582,
+      "step": 6824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.008737750378098e-07,
+      "loss": 0.7808,
+      "step": 6825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9998694760020143e-07,
+      "loss": 0.7753,
+      "step": 6826
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.991020623277462e-07,
+      "loss": 0.8175,
+      "step": 6827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.982191193958083e-07,
+      "loss": 0.7632,
+      "step": 6828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9733811897937215e-07,
+      "loss": 0.8223,
+      "step": 6829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.964590612530326e-07,
+      "loss": 0.7264,
+      "step": 6830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9558194639100026e-07,
+      "loss": 0.7376,
+      "step": 6831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9470677456710497e-07,
+      "loss": 0.7826,
+      "step": 6832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9383354595478465e-07,
+      "loss": 0.8042,
+      "step": 6833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9296226072709755e-07,
+      "loss": 0.7514,
+      "step": 6834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9209291905671335e-07,
+      "loss": 0.6912,
+      "step": 6835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9122552111591975e-07,
+      "loss": 0.7397,
+      "step": 6836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9036006707661704e-07,
+      "loss": 0.7403,
+      "step": 6837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8949655711032022e-07,
+      "loss": 0.7483,
+      "step": 6838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8863499138816022e-07,
+      "loss": 0.716,
+      "step": 6839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8777537008088153e-07,
+      "loss": 0.8121,
+      "step": 6840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8691769335884459e-07,
+      "loss": 0.7044,
+      "step": 6841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8606196139202115e-07,
+      "loss": 0.7152,
+      "step": 6842
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8520817435000227e-07,
+      "loss": 0.7929,
+      "step": 6843
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8435633240199257e-07,
+      "loss": 0.7464,
+      "step": 6844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8350643571680705e-07,
+      "loss": 0.7813,
+      "step": 6845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8265848446287982e-07,
+      "loss": 0.7337,
+      "step": 6846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8181247880825647e-07,
+      "loss": 0.7273,
+      "step": 6847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.809684189206007e-07,
+      "loss": 0.7266,
+      "step": 6848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.801263049671853e-07,
+      "loss": 0.7714,
+      "step": 6849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.792861371149013e-07,
+      "loss": 0.7359,
+      "step": 6850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7844791553025432e-07,
+      "loss": 0.7949,
+      "step": 6851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7761164037936153e-07,
+      "loss": 0.3345,
+      "step": 6852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7677731182795698e-07,
+      "loss": 0.7448,
+      "step": 6853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7594493004138623e-07,
+      "loss": 0.7154,
+      "step": 6854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7511449518461177e-07,
+      "loss": 0.7624,
+      "step": 6855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7428600742220858e-07,
+      "loss": 0.7076,
+      "step": 6856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7345946691836534e-07,
+      "loss": 0.7364,
+      "step": 6857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7263487383688883e-07,
+      "loss": 0.7756,
+      "step": 6858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7181222834119272e-07,
+      "loss": 0.2964,
+      "step": 6859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7099153059431327e-07,
+      "loss": 0.7291,
+      "step": 6860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7017278075889154e-07,
+      "loss": 0.779,
+      "step": 6861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6935597899719103e-07,
+      "loss": 0.7484,
+      "step": 6862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6854112547108225e-07,
+      "loss": 0.7455,
+      "step": 6863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6772822034205494e-07,
+      "loss": 0.7308,
+      "step": 6864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.669172637712102e-07,
+      "loss": 0.7755,
+      "step": 6865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6610825591926283e-07,
+      "loss": 0.7707,
+      "step": 6866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6530119694654235e-07,
+      "loss": 0.7675,
+      "step": 6867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.644960870129908e-07,
+      "loss": 0.7422,
+      "step": 6868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6369292627816613e-07,
+      "loss": 0.7762,
+      "step": 6869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6289171490123657e-07,
+      "loss": 0.7538,
+      "step": 6870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.620924530409873e-07,
+      "loss": 0.7192,
+      "step": 6871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.612951408558161e-07,
+      "loss": 0.7759,
+      "step": 6872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6049977850373434e-07,
+      "loss": 0.7857,
+      "step": 6873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.597063661423659e-07,
+      "loss": 0.8028,
+      "step": 6874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5891490392894949e-07,
+      "loss": 0.7313,
+      "step": 6875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.581253920203385e-07,
+      "loss": 0.7285,
+      "step": 6876
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5733783057299558e-07,
+      "loss": 0.7575,
+      "step": 6877
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5655221974300138e-07,
+      "loss": 0.7802,
+      "step": 6878
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.55768559686047e-07,
+      "loss": 0.7361,
+      "step": 6879
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5498685055743923e-07,
+      "loss": 0.8091,
+      "step": 6880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5420709251209753e-07,
+      "loss": 0.7183,
+      "step": 6881
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5342928570455162e-07,
+      "loss": 0.7623,
+      "step": 6882
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.526534302889493e-07,
+      "loss": 0.7498,
+      "step": 6883
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5187952641904867e-07,
+      "loss": 0.7881,
+      "step": 6884
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5110757424822266e-07,
+      "loss": 0.7505,
+      "step": 6885
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5033757392945547e-07,
+      "loss": 0.7332,
+      "step": 6886
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4956952561534731e-07,
+      "loss": 0.8353,
+      "step": 6887
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4880342945810867e-07,
+      "loss": 0.6854,
+      "step": 6888
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.480392856095636e-07,
+      "loss": 0.7889,
+      "step": 6889
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4727709422115098e-07,
+      "loss": 0.8047,
+      "step": 6890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4651685544392337e-07,
+      "loss": 0.7914,
+      "step": 6891
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.457585694285424e-07,
+      "loss": 0.8046,
+      "step": 6892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4500223632528564e-07,
+      "loss": 0.7455,
+      "step": 6893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4424785628404325e-07,
+      "loss": 0.7342,
+      "step": 6894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4349542945431893e-07,
+      "loss": 0.784,
+      "step": 6895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4274495598522676e-07,
+      "loss": 0.7494,
+      "step": 6896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4199643602549772e-07,
+      "loss": 0.8071,
+      "step": 6897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.412498697234721e-07,
+      "loss": 0.681,
+      "step": 6898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4050525722710373e-07,
+      "loss": 0.7587,
+      "step": 6899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3976259868396124e-07,
+      "loss": 0.6461,
+      "step": 6900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.390218942412225e-07,
+      "loss": 0.7206,
+      "step": 6901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3828314404568333e-07,
+      "loss": 0.7264,
+      "step": 6902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.375463482437456e-07,
+      "loss": 0.7209,
+      "step": 6903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.368115069814313e-07,
+      "loss": 0.7727,
+      "step": 6904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3607862040436737e-07,
+      "loss": 0.7075,
+      "step": 6905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3534768865779867e-07,
+      "loss": 0.7778,
+      "step": 6906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.346187118865816e-07,
+      "loss": 0.7347,
+      "step": 6907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.338916902351839e-07,
+      "loss": 0.7385,
+      "step": 6908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3316662384768807e-07,
+      "loss": 0.7766,
+      "step": 6909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3244351286778588e-07,
+      "loss": 0.6908,
+      "step": 6910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3172235743878492e-07,
+      "loss": 0.8114,
+      "step": 6911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3100315770360304e-07,
+      "loss": 0.7793,
+      "step": 6912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3028591380476964e-07,
+      "loss": 0.6982,
+      "step": 6913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2957062588443091e-07,
+      "loss": 0.6711,
+      "step": 6914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.288572940843391e-07,
+      "loss": 0.3323,
+      "step": 6915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2814591854586555e-07,
+      "loss": 0.7666,
+      "step": 6916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2743649940998858e-07,
+      "loss": 0.7593,
+      "step": 6917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2672903681730132e-07,
+      "loss": 0.3607,
+      "step": 6918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2602353090800824e-07,
+      "loss": 0.7363,
+      "step": 6919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2531998182192638e-07,
+      "loss": 0.711,
+      "step": 6920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.246183896984865e-07,
+      "loss": 0.7841,
+      "step": 6921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2391875467672842e-07,
+      "loss": 0.7076,
+      "step": 6922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2322107689530573e-07,
+      "loss": 0.7399,
+      "step": 6923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2252535649248443e-07,
+      "loss": 0.7805,
+      "step": 6924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2183159360614316e-07,
+      "loss": 0.7185,
+      "step": 6925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.211397883737697e-07,
+      "loss": 0.752,
+      "step": 6926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2044994093246775e-07,
+      "loss": 0.6875,
+      "step": 6927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1976205141895124e-07,
+      "loss": 0.7451,
+      "step": 6928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1907611996954449e-07,
+      "loss": 0.6984,
+      "step": 6929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1839214672018651e-07,
+      "loss": 0.7416,
+      "step": 6930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1771013180642554e-07,
+      "loss": 0.7324,
+      "step": 6931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.170300753634257e-07,
+      "loss": 0.3516,
+      "step": 6932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1635197752595917e-07,
+      "loss": 0.7343,
+      "step": 6933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1567583842840958e-07,
+      "loss": 0.8083,
+      "step": 6934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1500165820477748e-07,
+      "loss": 0.7423,
+      "step": 6935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1432943698866939e-07,
+      "loss": 0.7778,
+      "step": 6936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1365917491330647e-07,
+      "loss": 0.76,
+      "step": 6937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1299087211152137e-07,
+      "loss": 0.7051,
+      "step": 6938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1232452871575817e-07,
+      "loss": 0.7375,
+      "step": 6939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1166014485807341e-07,
+      "loss": 0.6984,
+      "step": 6940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1099772067013292e-07,
+      "loss": 0.7714,
+      "step": 6941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1033725628321722e-07,
+      "loss": 0.355,
+      "step": 6942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0967875182821608e-07,
+      "loss": 0.795,
+      "step": 6943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0902220743563396e-07,
+      "loss": 0.712,
+      "step": 6944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0836762323558237e-07,
+      "loss": 0.8,
+      "step": 6945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.077149993577864e-07,
+      "loss": 0.7353,
+      "step": 6946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0706433593158594e-07,
+      "loss": 0.367,
+      "step": 6947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0641563308592561e-07,
+      "loss": 0.7218,
+      "step": 6948
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0576889094936927e-07,
+      "loss": 0.6357,
+      "step": 6949
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0512410965008435e-07,
+      "loss": 0.6838,
+      "step": 6950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0448128931585533e-07,
+      "loss": 0.7129,
+      "step": 6951
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0384043007407696e-07,
+      "loss": 0.8525,
+      "step": 6952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.032015320517532e-07,
+      "loss": 0.7397,
+      "step": 6953
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0256459537550168e-07,
+      "loss": 0.7608,
+      "step": 6954
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0192962017154917e-07,
+      "loss": 0.7085,
+      "step": 6955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0129660656573726e-07,
+      "loss": 0.7749,
+      "step": 6956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0066555468351446e-07,
+      "loss": 0.7161,
+      "step": 6957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0003646464994299e-07,
+      "loss": 0.7786,
+      "step": 6958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.940933658969642e-08,
+      "loss": 0.692,
+      "step": 6959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.878417062705758e-08,
+      "loss": 0.7103,
+      "step": 6960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.816096688592403e-08,
+      "loss": 0.7898,
+      "step": 6961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.753972548980029e-08,
+      "loss": 0.845,
+      "step": 6962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.692044656180565e-08,
+      "loss": 0.7014,
+      "step": 6963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.630313022466642e-08,
+      "loss": 0.681,
+      "step": 6964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.568777660072249e-08,
+      "loss": 0.7473,
+      "step": 6965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.507438581192629e-08,
+      "loss": 0.7116,
+      "step": 6966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.44629579798384e-08,
+      "loss": 0.3496,
+      "step": 6967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.385349322563297e-08,
+      "loss": 0.7208,
+      "step": 6968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.324599167009119e-08,
+      "loss": 0.7539,
+      "step": 6969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.264045343360894e-08,
+      "loss": 0.7666,
+      "step": 6970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.203687863619137e-08,
+      "loss": 0.7666,
+      "step": 6971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.143526739745612e-08,
+      "loss": 0.729,
+      "step": 6972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.083561983662892e-08,
+      "loss": 0.7816,
+      "step": 6973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.023793607254805e-08,
+      "loss": 0.7055,
+      "step": 6974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.96422162236632e-08,
+      "loss": 0.687,
+      "step": 6975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.904846040803216e-08,
+      "loss": 0.7855,
+      "step": 6976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.84566687433286e-08,
+      "loss": 0.7985,
+      "step": 6977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.786684134683088e-08,
+      "loss": 0.7666,
+      "step": 6978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.727897833543109e-08,
+      "loss": 0.713,
+      "step": 6979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.66930798256338e-08,
+      "loss": 0.7186,
+      "step": 6980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.610914593355058e-08,
+      "loss": 0.7233,
+      "step": 6981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.552717677490552e-08,
+      "loss": 0.7422,
+      "step": 6982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.494717246503414e-08,
+      "loss": 0.7755,
+      "step": 6983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.436913311888118e-08,
+      "loss": 0.7511,
+      "step": 6984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.379305885100164e-08,
+      "loss": 0.7504,
+      "step": 6985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.3218949775562e-08,
+      "loss": 0.7468,
+      "step": 6986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.264680600634012e-08,
+      "loss": 0.7158,
+      "step": 6987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.207662765672308e-08,
+      "loss": 0.7835,
+      "step": 6988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.150841483970828e-08,
+      "loss": 0.8417,
+      "step": 6989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.094216766790342e-08,
+      "loss": 0.7444,
+      "step": 6990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.037788625352872e-08,
+      "loss": 0.3364,
+      "step": 6991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.98155707084125e-08,
+      "loss": 0.7212,
+      "step": 6992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.925522114399453e-08,
+      "loss": 0.7485,
+      "step": 6993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.869683767132485e-08,
+      "loss": 0.6969,
+      "step": 6994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.814042040106384e-08,
+      "loss": 0.7857,
+      "step": 6995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.758596944348218e-08,
+      "loss": 0.7263,
+      "step": 6996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.703348490846086e-08,
+      "loss": 0.8082,
+      "step": 6997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.648296690549228e-08,
+      "loss": 0.6983,
+      "step": 6998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.593441554367586e-08,
+      "loss": 0.6903,
+      "step": 6999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.538783093172464e-08,
+      "loss": 0.7541,
+      "step": 7000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.484321317796084e-08,
+      "loss": 0.7814,
+      "step": 7001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.430056239031592e-08,
+      "loss": 0.7501,
+      "step": 7002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.375987867633383e-08,
+      "loss": 0.7738,
+      "step": 7003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.322116214316443e-08,
+      "loss": 0.7509,
+      "step": 7004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.268441289757455e-08,
+      "loss": 0.8204,
+      "step": 7005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.214963104593353e-08,
+      "loss": 0.6759,
+      "step": 7006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.161681669422548e-08,
+      "loss": 0.7602,
+      "step": 7007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.108596994804373e-08,
+      "loss": 0.7052,
+      "step": 7008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.055709091259189e-08,
+      "loss": 0.7879,
+      "step": 7009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.003017969268167e-08,
+      "loss": 0.7139,
+      "step": 7010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.950523639273732e-08,
+      "loss": 0.3103,
+      "step": 7011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.898226111679229e-08,
+      "loss": 0.7005,
+      "step": 7012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.846125396848924e-08,
+      "loss": 0.7754,
+      "step": 7013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.794221505108112e-08,
+      "loss": 0.7378,
+      "step": 7014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.74251444674312e-08,
+      "loss": 0.7652,
+      "step": 7015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.691004232001197e-08,
+      "loss": 0.751,
+      "step": 7016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.639690871090842e-08,
+      "loss": 0.7588,
+      "step": 7017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.588574374181034e-08,
+      "loss": 0.6745,
+      "step": 7018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.537654751402111e-08,
+      "loss": 0.7162,
+      "step": 7019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.486932012845448e-08,
+      "loss": 0.7026,
+      "step": 7020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.436406168563225e-08,
+      "loss": 0.7324,
+      "step": 7021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.386077228568543e-08,
+      "loss": 0.7435,
+      "step": 7022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.335945202835647e-08,
+      "loss": 0.6888,
+      "step": 7023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.286010101299811e-08,
+      "loss": 0.7247,
+      "step": 7024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.236271933857008e-08,
+      "loss": 0.8429,
+      "step": 7025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.186730710364353e-08,
+      "loss": 0.7379,
+      "step": 7026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.137386440639992e-08,
+      "loss": 0.8132,
+      "step": 7027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.08823913446288e-08,
+      "loss": 0.7975,
+      "step": 7028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.039288801573118e-08,
+      "loss": 0.6315,
+      "step": 7029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.990535451671608e-08,
+      "loss": 0.7073,
+      "step": 7030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.941979094420292e-08,
+      "loss": 0.7786,
+      "step": 7031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.893619739442025e-08,
+      "loss": 0.741,
+      "step": 7032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8454573963208085e-08,
+      "loss": 0.8041,
+      "step": 7033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.79749207460123e-08,
+      "loss": 0.6366,
+      "step": 7034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7497237837892404e-08,
+      "loss": 0.3273,
+      "step": 7035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7021525333514905e-08,
+      "loss": 0.7206,
+      "step": 7036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.65477833271566e-08,
+      "loss": 0.7665,
+      "step": 7037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.60760119127024e-08,
+      "loss": 0.7804,
+      "step": 7038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5606211183649726e-08,
+      "loss": 0.8045,
+      "step": 7039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.513838123310189e-08,
+      "loss": 0.7875,
+      "step": 7040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.467252215377472e-08,
+      "loss": 0.7392,
+      "step": 7041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.420863403799104e-08,
+      "loss": 0.7611,
+      "step": 7042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.374671697768507e-08,
+      "loss": 0.6955,
+      "step": 7043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.328677106439806e-08,
+      "loss": 0.7987,
+      "step": 7044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2828796389283734e-08,
+      "loss": 0.6688,
+      "step": 7045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.237279304310283e-08,
+      "loss": 0.7933,
+      "step": 7046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.19187611162264e-08,
+      "loss": 0.7485,
+      "step": 7047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.146670069863357e-08,
+      "loss": 0.3751,
+      "step": 7048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1016611879913805e-08,
+      "loss": 0.8035,
+      "step": 7049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.056849474926573e-08,
+      "loss": 0.743,
+      "step": 7050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0122349395497205e-08,
+      "loss": 0.7422,
+      "step": 7051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.967817590702417e-08,
+      "loss": 0.7802,
+      "step": 7052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.923597437187621e-08,
+      "loss": 0.7866,
+      "step": 7053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.879574487768435e-08,
+      "loss": 0.7774,
+      "step": 7054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8357487511696555e-08,
+      "loss": 0.6984,
+      "step": 7055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.792120236076559e-08,
+      "loss": 0.7961,
+      "step": 7056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.748688951135338e-08,
+      "loss": 0.7009,
+      "step": 7057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.705454904953333e-08,
+      "loss": 0.7864,
+      "step": 7058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.662418106098576e-08,
+      "loss": 0.738,
+      "step": 7059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6195785631001356e-08,
+      "loss": 0.8159,
+      "step": 7060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.576936284447886e-08,
+      "loss": 0.6792,
+      "step": 7061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.534491278592845e-08,
+      "loss": 0.722,
+      "step": 7062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.492243553946507e-08,
+      "loss": 0.6888,
+      "step": 7063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4501931188817294e-08,
+      "loss": 0.8005,
+      "step": 7064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.408339981731957e-08,
+      "loss": 0.6782,
+      "step": 7065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.366684150791667e-08,
+      "loss": 0.7675,
+      "step": 7066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3252256343161434e-08,
+      "loss": 0.6818,
+      "step": 7067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.283964440521593e-08,
+      "loss": 0.8172,
+      "step": 7068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2429005775853626e-08,
+      "loss": 0.7891,
+      "step": 7069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2020340536453875e-08,
+      "loss": 0.8235,
+      "step": 7070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.161364876800522e-08,
+      "loss": 0.7244,
+      "step": 7071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.120893055110653e-08,
+      "loss": 0.712,
+      "step": 7072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.080618596596364e-08,
+      "loss": 0.737,
+      "step": 7073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.040541509239382e-08,
+      "loss": 0.7634,
+      "step": 7074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0006618009820195e-08,
+      "loss": 0.7255,
+      "step": 7075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9609794797278446e-08,
+      "loss": 0.7141,
+      "step": 7076
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.92149455334101e-08,
+      "loss": 0.8279,
+      "step": 7077
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8822070296464785e-08,
+      "loss": 0.6521,
+      "step": 7078
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.843116916430467e-08,
+      "loss": 0.7405,
+      "step": 7079
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8042242214396673e-08,
+      "loss": 0.7565,
+      "step": 7080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.765528952382025e-08,
+      "loss": 0.7412,
+      "step": 7081
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.727031116925961e-08,
+      "loss": 0.8134,
+      "step": 7082
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6887307227010395e-08,
+      "loss": 0.7335,
+      "step": 7083
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.650627777297744e-08,
+      "loss": 0.7225,
+      "step": 7084
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6127222882671456e-08,
+      "loss": 0.7107,
+      "step": 7085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.575014263121457e-08,
+      "loss": 0.7438,
+      "step": 7086
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5375037093335897e-08,
+      "loss": 0.7291,
+      "step": 7087
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.500190634337375e-08,
+      "loss": 0.8065,
+      "step": 7088
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4630750455275644e-08,
+      "loss": 0.7216,
+      "step": 7089
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.426156950259718e-08,
+      "loss": 0.6736,
+      "step": 7090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3894363558502065e-08,
+      "loss": 0.758,
+      "step": 7091
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.352913269576319e-08,
+      "loss": 0.8311,
+      "step": 7092
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.316587698676155e-08,
+      "loss": 0.7839,
+      "step": 7093
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.280459650348844e-08,
+      "loss": 0.6901,
+      "step": 7094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.244529131754215e-08,
+      "loss": 0.7749,
+      "step": 7095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.208796150012794e-08,
+      "loss": 0.747,
+      "step": 7096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.173260712206361e-08,
+      "loss": 0.7895,
+      "step": 7097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.137922825377171e-08,
+      "loss": 0.7906,
+      "step": 7098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1027824965286226e-08,
+      "loss": 0.783,
+      "step": 7099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0678397326247e-08,
+      "loss": 0.7044,
+      "step": 7100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0330945405903086e-08,
+      "loss": 0.386,
+      "step": 7101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9985469273113857e-08,
+      "loss": 0.705,
+      "step": 7102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9641968996345637e-08,
+      "loss": 0.7447,
+      "step": 7103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9300444643672877e-08,
+      "loss": 0.6475,
+      "step": 7104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8960896282778095e-08,
+      "loss": 0.6967,
+      "step": 7105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8623323980953023e-08,
+      "loss": 0.7007,
+      "step": 7106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8287727805099695e-08,
+      "loss": 0.7714,
+      "step": 7107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.795410782172381e-08,
+      "loss": 0.7774,
+      "step": 7108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.762246409694358e-08,
+      "loss": 0.8039,
+      "step": 7109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7292796696484212e-08,
+      "loss": 0.7786,
+      "step": 7110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6965105685679005e-08,
+      "loss": 0.785,
+      "step": 7111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.663939112946934e-08,
+      "loss": 0.7645,
+      "step": 7112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.631565309240469e-08,
+      "loss": 0.6722,
+      "step": 7113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5993891638644852e-08,
+      "loss": 0.7753,
+      "step": 7114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.567410683195659e-08,
+      "loss": 0.7435,
+      "step": 7115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5356298735712547e-08,
+      "loss": 0.7864,
+      "step": 7116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5040467412899006e-08,
+      "loss": 0.7346,
+      "step": 7117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4726612926104786e-08,
+      "loss": 0.8141,
+      "step": 7118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4414735337530138e-08,
+      "loss": 0.7986,
+      "step": 7119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4104834708983395e-08,
+      "loss": 0.7991,
+      "step": 7120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3796911101880983e-08,
+      "loss": 0.7542,
+      "step": 7121
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.349096457724631e-08,
+      "loss": 0.6801,
+      "step": 7122
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.318699519571199e-08,
+      "loss": 0.7428,
+      "step": 7123
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2885003017519835e-08,
+      "loss": 0.7139,
+      "step": 7124
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2584988102516414e-08,
+      "loss": 0.7189,
+      "step": 7125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2286950510159722e-08,
+      "loss": 0.7112,
+      "step": 7126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1990890299515843e-08,
+      "loss": 0.7051,
+      "step": 7127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1696807529256735e-08,
+      "loss": 0.7346,
+      "step": 7128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1404702257663558e-08,
+      "loss": 0.7208,
+      "step": 7129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1114574542626664e-08,
+      "loss": 0.7812,
+      "step": 7130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.082642444164229e-08,
+      "loss": 0.8056,
+      "step": 7131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0540252011816975e-08,
+      "loss": 0.7802,
+      "step": 7132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.025605730986313e-08,
+      "loss": 0.7244,
+      "step": 7133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9973840392103482e-08,
+      "loss": 0.7675,
+      "step": 7134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9693601314467737e-08,
+      "loss": 0.7789,
+      "step": 7135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.941534013249369e-08,
+      "loss": 0.7541,
+      "step": 7136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9139056901327225e-08,
+      "loss": 0.7898,
+      "step": 7137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8864751675720104e-08,
+      "loss": 0.7557,
+      "step": 7138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859242451003773e-08,
+      "loss": 0.7528,
+      "step": 7139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8322075458245825e-08,
+      "loss": 0.7648,
+      "step": 7140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8053704573924858e-08,
+      "loss": 0.7133,
+      "step": 7141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7787311910261173e-08,
+      "loss": 0.3491,
+      "step": 7142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7522897520045878e-08,
+      "loss": 0.7963,
+      "step": 7143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.72604614556815e-08,
+      "loss": 0.6832,
+      "step": 7144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.700000376917754e-08,
+      "loss": 0.7596,
+      "step": 7145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6741524512152717e-08,
+      "loss": 0.7683,
+      "step": 7146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6485023735831606e-08,
+      "loss": 0.7564,
+      "step": 7147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.623050149104577e-08,
+      "loss": 0.8042,
+      "step": 7148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.59779578282393e-08,
+      "loss": 0.7732,
+      "step": 7149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572739279745994e-08,
+      "loss": 0.7569,
+      "step": 7150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5478806448364635e-08,
+      "loss": 0.8252,
+      "step": 7151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5232198830217316e-08,
+      "loss": 0.7556,
+      "step": 7152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4987569991892214e-08,
+      "loss": 0.8115,
+      "step": 7153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4744919981868333e-08,
+      "loss": 0.7672,
+      "step": 7154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4504248848236091e-08,
+      "loss": 0.7573,
+      "step": 7155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.426555663868956e-08,
+      "loss": 0.7699,
+      "step": 7156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.402884340053312e-08,
+      "loss": 0.6883,
+      "step": 7157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3794109180680359e-08,
+      "loss": 0.6969,
+      "step": 7158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.356135402564851e-08,
+      "loss": 0.7418,
+      "step": 7159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3330577981566229e-08,
+      "loss": 0.6836,
+      "step": 7160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3101781094168043e-08,
+      "loss": 0.7785,
+      "step": 7161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.287496340879657e-08,
+      "loss": 0.7265,
+      "step": 7162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2650124970404743e-08,
+      "loss": 0.7563,
+      "step": 7163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2427265823548029e-08,
+      "loss": 0.7627,
+      "step": 7164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2206386012395544e-08,
+      "loss": 0.7221,
+      "step": 7165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1987485580718939e-08,
+      "loss": 0.688,
+      "step": 7166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1770564571900178e-08,
+      "loss": 0.3199,
+      "step": 7167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1555623028929318e-08,
+      "loss": 0.7694,
+      "step": 7168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1342660994403399e-08,
+      "loss": 0.7658,
+      "step": 7169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1131678510527544e-08,
+      "loss": 0.7219,
+      "step": 7170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0922675619112755e-08,
+      "loss": 0.7677,
+      "step": 7171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0715652361581452e-08,
+      "loss": 0.8164,
+      "step": 7172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0510608778958598e-08,
+      "loss": 0.734,
+      "step": 7173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0307544911881684e-08,
+      "loss": 0.744,
+      "step": 7174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0106460800594075e-08,
+      "loss": 0.7848,
+      "step": 7175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.907356484946119e-09,
+      "loss": 0.7854,
+      "step": 7176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.710232004395137e-09,
+      "loss": 0.6973,
+      "step": 7177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.51508739800877e-09,
+      "loss": 0.7251,
+      "step": 7178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.32192270446053e-09,
+      "loss": 0.7566,
+      "step": 7179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.1307379620309e-09,
+      "loss": 0.7563,
+      "step": 7180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.941533208609577e-09,
+      "loss": 0.7295,
+      "step": 7181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.75430848169323e-09,
+      "loss": 0.7594,
+      "step": 7182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.569063818385515e-09,
+      "loss": 0.8374,
+      "step": 7183
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.385799255398175e-09,
+      "loss": 0.8035,
+      "step": 7184
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.204514829051047e-09,
+      "loss": 0.7298,
+      "step": 7185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.025210575270948e-09,
+      "loss": 0.8117,
+      "step": 7186
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.847886529591675e-09,
+      "loss": 0.7279,
+      "step": 7187
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.672542727156229e-09,
+      "loss": 0.7444,
+      "step": 7188
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.49917920271459e-09,
+      "loss": 0.8041,
+      "step": 7189
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.3277959906215005e-09,
+      "loss": 0.8243,
+      "step": 7190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.158393124845342e-09,
+      "loss": 0.726,
+      "step": 7191
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.9909706389548195e-09,
+      "loss": 0.8014,
+      "step": 7192
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.825528566131168e-09,
+      "loss": 0.7456,
+      "step": 7193
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.662066939162603e-09,
+      "loss": 0.3295,
+      "step": 7194
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.500585790442104e-09,
+      "loss": 0.7313,
+      "step": 7195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.341085151974069e-09,
+      "loss": 0.7554,
+      "step": 7196
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.183565055365437e-09,
+      "loss": 0.8118,
+      "step": 7197
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.02802553183679e-09,
+      "loss": 0.7336,
+      "step": 7198
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.874466612210139e-09,
+      "loss": 0.7416,
+      "step": 7199
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.722888326920028e-09,
+      "loss": 0.33,
+      "step": 7200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.573290706004653e-09,
+      "loss": 0.3513,
+      "step": 7201
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.425673779112517e-09,
+      "loss": 0.8421,
+      "step": 7202
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.280037575496888e-09,
+      "loss": 0.7983,
+      "step": 7203
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.136382124020234e-09,
+      "loss": 0.7704,
+      "step": 7204
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.994707453153114e-09,
+      "loss": 0.7449,
+      "step": 7205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.855013590971958e-09,
+      "loss": 0.7729,
+      "step": 7206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.717300565161287e-09,
+      "loss": 0.7585,
+      "step": 7207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.581568403012604e-09,
+      "loss": 0.7739,
+      "step": 7208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.447817131426613e-09,
+      "loss": 0.7246,
+      "step": 7209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.316046776908778e-09,
+      "loss": 0.7691,
+      "step": 7210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.186257365574875e-09,
+      "loss": 0.7837,
+      "step": 7211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.058448923144331e-09,
+      "loss": 0.3498,
+      "step": 7212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.932621474947995e-09,
+      "loss": 0.7107,
+      "step": 7213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.808775045921476e-09,
+      "loss": 0.3725,
+      "step": 7214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6869096606095835e-09,
+      "loss": 0.6974,
+      "step": 7215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5670253431618894e-09,
+      "loss": 0.7271,
+      "step": 7216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4491221173393873e-09,
+      "loss": 0.2976,
+      "step": 7217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.33320000650561e-09,
+      "loss": 0.8034,
+      "step": 7218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2192590336355133e-09,
+      "loss": 0.7894,
+      "step": 7219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.107299221309923e-09,
+      "loss": 0.7935,
+      "step": 7220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.997320591716646e-09,
+      "loss": 0.7112,
+      "step": 7221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8893231666504706e-09,
+      "loss": 0.7763,
+      "step": 7222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7833069675153868e-09,
+      "loss": 0.7807,
+      "step": 7223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.679272015322365e-09,
+      "loss": 0.7518,
+      "step": 7224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5772183306871366e-09,
+      "loss": 0.8205,
+      "step": 7225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.477145933835745e-09,
+      "loss": 0.7295,
+      "step": 7226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.379054844600104e-09,
+      "loss": 0.7479,
+      "step": 7227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2829450824202182e-09,
+      "loss": 0.7562,
+      "step": 7228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1888166663430744e-09,
+      "loss": 0.7521,
+      "step": 7229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0966696150237496e-09,
+      "loss": 0.7006,
+      "step": 7230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0065039467220827e-09,
+      "loss": 0.7151,
+      "step": 7231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9183196793082225e-09,
+      "loss": 0.7183,
+      "step": 7232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8321168302593007e-09,
+      "loss": 0.66,
+      "step": 7233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7478954166583183e-09,
+      "loss": 0.6753,
+      "step": 7234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6656554551952586e-09,
+      "loss": 0.7653,
+      "step": 7235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5853969621693056e-09,
+      "loss": 0.76,
+      "step": 7236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5071199534866242e-09,
+      "loss": 0.7463,
+      "step": 7237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.430824444659251e-09,
+      "loss": 0.7372,
+      "step": 7238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3565104508084237e-09,
+      "loss": 0.7286,
+      "step": 7239
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2841779866601401e-09,
+      "loss": 0.7513,
+      "step": 7240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2138270665507101e-09,
+      "loss": 0.8005,
+      "step": 7241
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1454577044212045e-09,
+      "loss": 0.3142,
+      "step": 7242
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.079069913820785e-09,
+      "loss": 0.7127,
+      "step": 7243
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0146637079078148e-09,
+      "loss": 0.7788,
+      "step": 7244
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.522390994454177e-10,
+      "loss": 0.7446,
+      "step": 7245
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.917961008048093e-10,
+      "loss": 0.3382,
+      "step": 7246
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.333347239641853e-10,
+      "loss": 0.7933,
+      "step": 7247
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.768549805109438e-10,
+      "loss": 0.3381,
+      "step": 7248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.223568816361326e-10,
+      "loss": 0.7997,
+      "step": 7249
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.698404381411117e-10,
+      "loss": 0.702,
+      "step": 7250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.193056604342219e-10,
+      "loss": 0.7034,
+      "step": 7251
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.707525585296747e-10,
+      "loss": 0.7359,
+      "step": 7252
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.241811420497734e-10,
+      "loss": 0.7722,
+      "step": 7253
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.795914202238017e-10,
+      "loss": 0.7821,
+      "step": 7254
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.36983401889135e-10,
+      "loss": 0.7927,
+      "step": 7255
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.963570954901297e-10,
+      "loss": 0.7899,
+      "step": 7256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.577125090770128e-10,
+      "loss": 0.7493,
+      "step": 7257
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2104965030921307e-10,
+      "loss": 0.734,
+      "step": 7258
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8636852645202996e-10,
+      "loss": 0.823,
+      "step": 7259
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5366914437885413e-10,
+      "loss": 0.3355,
+      "step": 7260
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2295151057005749e-10,
+      "loss": 0.8065,
+      "step": 7261
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.942156311129928e-10,
+      "loss": 0.694,
+      "step": 7262
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.674615117031042e-10,
+      "loss": 0.7413,
+      "step": 7263
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4268915764170666e-10,
+      "loss": 0.7375,
+      "step": 7264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1989857383931658e-10,
+      "loss": 0.8101,
+      "step": 7265
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.908976481121102e-11,
+      "loss": 0.7619,
+      "step": 7266
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.026273468297874e-11,
+      "loss": 0.7585,
+      "step": 7267
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.341748718385887e-11,
+      "loss": 0.7087,
+      "step": 7268
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.855402565340228e-11,
+      "loss": 0.7251,
+      "step": 7269
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.567235303592043e-11,
+      "loss": 0.779,
+      "step": 7270
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.477247188714671e-11,
+      "loss": 0.6913,
+      "step": 7271
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.585438436424447e-11,
+      "loss": 0.3235,
+      "step": 7272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.918092235798982e-12,
+      "loss": 0.7199,
+      "step": 7273
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.963596876266351e-12,
+      "loss": 0.7272,
+      "step": 7274
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.908992681939566e-13,
+      "loss": 0.711,
+      "step": 7275
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.8177,
+      "step": 7276
+    },
+    {
+      "epoch": 1.0,
+      "step": 7276,
+      "total_flos": 1.020416037322752e+16,
+      "train_loss": 0.7892695052232829,
+      "train_runtime": 97609.8479,
+      "train_samples_per_second": 8.051,
+      "train_steps_per_second": 0.075
+    }
+  ],
+  "max_steps": 7276,
+  "num_train_epochs": 1,
+  "total_flos": 1.020416037322752e+16,
+  "trial_name": null,
+  "trial_params": null
+}