diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,31213 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 5198,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.615384615384616e-08,
+      "loss": 1.0063,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769231e-07,
+      "loss": 1.291,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.884615384615385e-07,
+      "loss": 1.333,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8461538461538463e-07,
+      "loss": 1.3145,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.807692307692307e-07,
+      "loss": 1.3564,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.76923076923077e-07,
+      "loss": 1.3223,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.730769230769231e-07,
+      "loss": 1.3398,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307693e-07,
+      "loss": 1.3447,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.653846153846154e-07,
+      "loss": 1.2949,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.615384615384615e-07,
+      "loss": 1.3428,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0576923076923078e-06,
+      "loss": 1.2754,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 1.3135,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2499999999999999e-06,
+      "loss": 1.2988,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3461538461538462e-06,
+      "loss": 1.292,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4423076923076924e-06,
+      "loss": 1.2002,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615385e-06,
+      "loss": 1.2148,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6346153846153848e-06,
+      "loss": 1.2598,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7307692307692308e-06,
+      "loss": 1.165,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.826923076923077e-06,
+      "loss": 1.1719,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.923076923076923e-06,
+      "loss": 1.2227,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0192307692307692e-06,
+      "loss": 1.1006,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1153846153846155e-06,
+      "loss": 1.1816,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.211538461538462e-06,
+      "loss": 1.1416,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 1.0566,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.403846153846154e-06,
+      "loss": 1.165,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4999999999999998e-06,
+      "loss": 1.124,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.596153846153846e-06,
+      "loss": 1.1172,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 1.1201,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7884615384615386e-06,
+      "loss": 1.1167,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.884615384615385e-06,
+      "loss": 1.0679,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9807692307692307e-06,
+      "loss": 1.0669,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.076923076923077e-06,
+      "loss": 1.1089,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1730769230769233e-06,
+      "loss": 1.0352,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2692307692307696e-06,
+      "loss": 1.0566,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3653846153846154e-06,
+      "loss": 1.0347,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 1.0107,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5576923076923075e-06,
+      "loss": 1.043,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.653846153846154e-06,
+      "loss": 0.3071,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.75e-06,
+      "loss": 1.0635,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153846e-06,
+      "loss": 1.0605,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.942307692307692e-06,
+      "loss": 1.0337,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0384615384615385e-06,
+      "loss": 1.0293,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.134615384615385e-06,
+      "loss": 1.083,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 0.9819,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3269230769230765e-06,
+      "loss": 1.0469,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.423076923076924e-06,
+      "loss": 1.0518,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.519230769230769e-06,
+      "loss": 1.0059,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 0.9897,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.711538461538462e-06,
+      "loss": 0.9854,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.807692307692308e-06,
+      "loss": 0.9883,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.903846153846154e-06,
+      "loss": 0.9746,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999999999999996e-06,
+      "loss": 1.0181,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.096153846153847e-06,
+      "loss": 1.0161,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.192307692307692e-06,
+      "loss": 0.999,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.288461538461539e-06,
+      "loss": 0.9854,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 1.0195,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.480769230769231e-06,
+      "loss": 0.3168,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.576923076923077e-06,
+      "loss": 0.98,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.673076923076923e-06,
+      "loss": 1.0186,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.76923076923077e-06,
+      "loss": 1.0361,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.865384615384615e-06,
+      "loss": 1.0024,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9615384615384615e-06,
+      "loss": 0.9653,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.057692307692308e-06,
+      "loss": 0.9805,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.153846153846154e-06,
+      "loss": 0.9268,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.25e-06,
+      "loss": 1.0083,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.3461538461538466e-06,
+      "loss": 0.9912,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.442307692307692e-06,
+      "loss": 0.9707,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 0.9019,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.6346153846153846e-06,
+      "loss": 0.9849,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.730769230769231e-06,
+      "loss": 0.9746,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.826923076923077e-06,
+      "loss": 0.9688,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 0.9526,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.01923076923077e-06,
+      "loss": 0.9175,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.115384615384615e-06,
+      "loss": 0.9932,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.211538461538462e-06,
+      "loss": 0.9492,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 0.938,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.403846153846155e-06,
+      "loss": 0.9434,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.5e-06,
+      "loss": 0.9106,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.596153846153846e-06,
+      "loss": 0.9814,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.692307692307692e-06,
+      "loss": 1.0254,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.78846153846154e-06,
+      "loss": 0.9629,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.884615384615384e-06,
+      "loss": 1.0107,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.98076923076923e-06,
+      "loss": 1.0332,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 1.0327,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.173076923076923e-06,
+      "loss": 0.9702,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.26923076923077e-06,
+      "loss": 0.9512,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.365384615384616e-06,
+      "loss": 0.9453,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 0.9985,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.557692307692307e-06,
+      "loss": 0.9683,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.653846153846153e-06,
+      "loss": 0.2966,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 0.9106,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 0.9785,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.942307692307692e-06,
+      "loss": 0.9883,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.038461538461538e-06,
+      "loss": 0.9922,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.134615384615384e-06,
+      "loss": 0.9409,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.3081,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.326923076923077e-06,
+      "loss": 0.9521,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.423076923076923e-06,
+      "loss": 0.959,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.51923076923077e-06,
+      "loss": 0.9824,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 0.8892,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.711538461538462e-06,
+      "loss": 0.2925,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.807692307692308e-06,
+      "loss": 0.9414,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.903846153846155e-06,
+      "loss": 1.0498,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999999999999999e-06,
+      "loss": 0.9736,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0096153846153847e-05,
+      "loss": 0.9639,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0192307692307693e-05,
+      "loss": 0.2644,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.028846153846154e-05,
+      "loss": 0.9438,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0384615384615384e-05,
+      "loss": 1.0166,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.048076923076923e-05,
+      "loss": 0.9434,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0576923076923078e-05,
+      "loss": 0.9702,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0673076923076923e-05,
+      "loss": 0.9712,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 0.9424,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0865384615384616e-05,
+      "loss": 1.0293,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0961538461538462e-05,
+      "loss": 0.9346,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1057692307692308e-05,
+      "loss": 0.9653,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1153846153846154e-05,
+      "loss": 0.9595,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.125e-05,
+      "loss": 0.2949,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1346153846153845e-05,
+      "loss": 0.9805,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1442307692307692e-05,
+      "loss": 0.9536,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.153846153846154e-05,
+      "loss": 1.0566,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1634615384615386e-05,
+      "loss": 0.938,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.173076923076923e-05,
+      "loss": 0.9912,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1826923076923077e-05,
+      "loss": 1.0098,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1923076923076923e-05,
+      "loss": 0.9102,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2019230769230771e-05,
+      "loss": 0.9438,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2115384615384615e-05,
+      "loss": 0.9658,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2211538461538462e-05,
+      "loss": 0.9492,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2307692307692308e-05,
+      "loss": 1.0029,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2403846153846153e-05,
+      "loss": 0.855,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.25e-05,
+      "loss": 0.9351,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2596153846153847e-05,
+      "loss": 0.9536,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2692307692307693e-05,
+      "loss": 0.9688,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2788461538461538e-05,
+      "loss": 0.9136,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2884615384615384e-05,
+      "loss": 0.9058,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2980769230769232e-05,
+      "loss": 0.9385,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 0.9253,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3173076923076923e-05,
+      "loss": 0.897,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3269230769230769e-05,
+      "loss": 0.9868,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3365384615384617e-05,
+      "loss": 0.9966,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3461538461538462e-05,
+      "loss": 0.9243,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3557692307692308e-05,
+      "loss": 0.9976,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3653846153846154e-05,
+      "loss": 0.9658,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.375e-05,
+      "loss": 0.9644,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.9927,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3942307692307693e-05,
+      "loss": 0.9668,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.403846153846154e-05,
+      "loss": 0.959,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4134615384615384e-05,
+      "loss": 0.9517,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.423076923076923e-05,
+      "loss": 0.9165,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4326923076923078e-05,
+      "loss": 0.9326,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4423076923076924e-05,
+      "loss": 0.9375,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4519230769230769e-05,
+      "loss": 0.9307,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 0.9834,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4711538461538462e-05,
+      "loss": 0.291,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.480769230769231e-05,
+      "loss": 0.9824,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4903846153846154e-05,
+      "loss": 0.9019,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5e-05,
+      "loss": 0.8936,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999998544120884e-05,
+      "loss": 1.0107,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999994176484101e-05,
+      "loss": 0.9561,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999986897091346e-05,
+      "loss": 0.9785,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999976705945446e-05,
+      "loss": 0.9312,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999963603050354e-05,
+      "loss": 0.938,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999947588411161e-05,
+      "loss": 0.9585,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999928662034086e-05,
+      "loss": 0.9775,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999906823926473e-05,
+      "loss": 0.9653,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999882074096802e-05,
+      "loss": 0.3306,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.499985441255468e-05,
+      "loss": 0.9497,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999823839310847e-05,
+      "loss": 0.9575,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999790354377175e-05,
+      "loss": 0.9692,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999753957766664e-05,
+      "loss": 0.9321,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999714649493437e-05,
+      "loss": 0.9458,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999672429572764e-05,
+      "loss": 0.9214,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999627298021031e-05,
+      "loss": 0.8359,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999579254855761e-05,
+      "loss": 0.9771,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999528300095609e-05,
+      "loss": 0.9355,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999474433760351e-05,
+      "loss": 0.9644,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999417655870907e-05,
+      "loss": 0.959,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999357966449314e-05,
+      "loss": 0.9263,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999295365518749e-05,
+      "loss": 0.9277,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999229853103515e-05,
+      "loss": 0.8921,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999161429229045e-05,
+      "loss": 0.9683,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999090093921903e-05,
+      "loss": 0.2848,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4999015847209788e-05,
+      "loss": 0.9282,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4998938689121522e-05,
+      "loss": 1.001,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4998858619687061e-05,
+      "loss": 0.9229,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4998775638937489e-05,
+      "loss": 0.9272,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4998689746905024e-05,
+      "loss": 0.9858,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4998600943623012e-05,
+      "loss": 0.9639,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4998509229125931e-05,
+      "loss": 0.3303,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4998414603449382e-05,
+      "loss": 0.9199,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.499831706663011e-05,
+      "loss": 0.8594,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4998216618705977e-05,
+      "loss": 0.2904,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.499811325971598e-05,
+      "loss": 0.9277,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4998006989700249e-05,
+      "loss": 0.9316,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4997897808700042e-05,
+      "loss": 1.0181,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4997785716757743e-05,
+      "loss": 0.9595,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4997670713916873e-05,
+      "loss": 0.9077,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4997552800222082e-05,
+      "loss": 0.9761,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4997431975719142e-05,
+      "loss": 0.9834,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4997308240454967e-05,
+      "loss": 0.9634,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4997181594477595e-05,
+      "loss": 0.9087,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4997052037836189e-05,
+      "loss": 0.9355,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4996919570581052e-05,
+      "loss": 0.9873,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4996784192763612e-05,
+      "loss": 0.9863,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4996645904436427e-05,
+      "loss": 0.9648,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4996504705653183e-05,
+      "loss": 0.9077,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.49963605964687e-05,
+      "loss": 0.3115,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4996213576938927e-05,
+      "loss": 0.9307,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4996063647120941e-05,
+      "loss": 0.9341,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.499591080707295e-05,
+      "loss": 0.9307,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4995755056854293e-05,
+      "loss": 0.9785,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4995596396525437e-05,
+      "loss": 0.9717,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4995434826147976e-05,
+      "loss": 0.9502,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4995270345784643e-05,
+      "loss": 0.8804,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.499510295549929e-05,
+      "loss": 0.9106,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4994932655356906e-05,
+      "loss": 0.9365,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4994759445423609e-05,
+      "loss": 0.9463,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4994583325766641e-05,
+      "loss": 0.9634,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.499440429645438e-05,
+      "loss": 0.2771,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4994222357556332e-05,
+      "loss": 0.3052,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4994037509143133e-05,
+      "loss": 0.9961,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4993849751286543e-05,
+      "loss": 0.2812,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.499365908405946e-05,
+      "loss": 0.9399,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4993465507535907e-05,
+      "loss": 0.9351,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4993269021791033e-05,
+      "loss": 1.0107,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4993069626901126e-05,
+      "loss": 0.9692,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4992867322943596e-05,
+      "loss": 0.9277,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4992662109996983e-05,
+      "loss": 0.2766,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4992453988140961e-05,
+      "loss": 0.8965,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4992242957456328e-05,
+      "loss": 0.9717,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4992029018025014e-05,
+      "loss": 0.9116,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4991812169930075e-05,
+      "loss": 0.9062,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4991592413255703e-05,
+      "loss": 0.916,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.499136974808721e-05,
+      "loss": 0.9199,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4991144174511047e-05,
+      "loss": 0.939,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.499091569261479e-05,
+      "loss": 0.9214,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4990684302487139e-05,
+      "loss": 0.9312,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.499045000421793e-05,
+      "loss": 0.9819,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4990212797898127e-05,
+      "loss": 0.9629,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.498997268361982e-05,
+      "loss": 0.9692,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4989729661476229e-05,
+      "loss": 0.9399,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4989483731561704e-05,
+      "loss": 0.9805,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4989234893971725e-05,
+      "loss": 0.957,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4988983148802898e-05,
+      "loss": 0.8647,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4988728496152958e-05,
+      "loss": 0.9004,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4988470936120773e-05,
+      "loss": 0.9287,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4988210468806335e-05,
+      "loss": 0.9409,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4987947094310765e-05,
+      "loss": 0.9893,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4987680812736316e-05,
+      "loss": 0.9258,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4987411624186368e-05,
+      "loss": 0.9355,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4987139528765428e-05,
+      "loss": 0.9087,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4986864526579133e-05,
+      "loss": 0.8945,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4986586617734249e-05,
+      "loss": 0.3225,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4986305802338669e-05,
+      "loss": 0.937,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4986022080501414e-05,
+      "loss": 0.9424,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4985735452332638e-05,
+      "loss": 0.9214,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4985445917943617e-05,
+      "loss": 0.9604,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.498515347744676e-05,
+      "loss": 0.9004,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.49848581309556e-05,
+      "loss": 0.8608,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4984559878584803e-05,
+      "loss": 0.9385,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4984258720450162e-05,
+      "loss": 0.9741,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4983954656668593e-05,
+      "loss": 0.936,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4983647687358147e-05,
+      "loss": 0.9438,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4983337812637998e-05,
+      "loss": 0.8062,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4983025032628453e-05,
+      "loss": 0.9531,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4982709347450944e-05,
+      "loss": 0.9341,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4982390757228024e-05,
+      "loss": 0.938,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.498206926208339e-05,
+      "loss": 0.9312,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4981744862141851e-05,
+      "loss": 0.9346,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4981417557529354e-05,
+      "loss": 1.0054,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4981087348372968e-05,
+      "loss": 0.8589,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.498075423480089e-05,
+      "loss": 0.9204,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.498041821694245e-05,
+      "loss": 0.9458,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.49800792949281e-05,
+      "loss": 0.9629,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4979737468889423e-05,
+      "loss": 0.9419,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4979392738959121e-05,
+      "loss": 0.9404,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4979045105271038e-05,
+      "loss": 0.9595,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4978694567960133e-05,
+      "loss": 0.9233,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4978341127162497e-05,
+      "loss": 0.9658,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.497798478301535e-05,
+      "loss": 0.9507,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4977625535657036e-05,
+      "loss": 0.9561,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4977263385227025e-05,
+      "loss": 0.9541,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.497689833186592e-05,
+      "loss": 0.9253,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4976530375715445e-05,
+      "loss": 0.9268,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4976159516918453e-05,
+      "loss": 0.9136,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4975785755618924e-05,
+      "loss": 0.9712,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4975409091961968e-05,
+      "loss": 0.8633,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4975029526093814e-05,
+      "loss": 0.918,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4974647058161828e-05,
+      "loss": 0.8394,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.497426168831449e-05,
+      "loss": 0.8857,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4973873416701423e-05,
+      "loss": 0.9409,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4973482243473362e-05,
+      "loss": 0.9248,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4973088168782172e-05,
+      "loss": 0.9609,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.497269119278085e-05,
+      "loss": 0.9092,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4972291315623515e-05,
+      "loss": 0.9102,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4971888537465414e-05,
+      "loss": 0.958,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4971482858462918e-05,
+      "loss": 0.8994,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4971074278773522e-05,
+      "loss": 0.9585,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4970662798555857e-05,
+      "loss": 0.3242,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4970248417969672e-05,
+      "loss": 0.9229,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4969831137175842e-05,
+      "loss": 0.9438,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4969410956336371e-05,
+      "loss": 0.896,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4968987875614386e-05,
+      "loss": 0.9619,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4968561895174145e-05,
+      "loss": 0.9873,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4968133015181026e-05,
+      "loss": 0.9028,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4967701235801533e-05,
+      "loss": 0.8828,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4967266557203301e-05,
+      "loss": 0.939,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4966828979555086e-05,
+      "loss": 0.8721,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.496638850302677e-05,
+      "loss": 0.9678,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4965945127789364e-05,
+      "loss": 0.8828,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4965498854014998e-05,
+      "loss": 0.9194,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4965049681876932e-05,
+      "loss": 0.9312,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.496459761154955e-05,
+      "loss": 0.9111,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4964142643208363e-05,
+      "loss": 0.8564,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.496368477703e-05,
+      "loss": 0.9614,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.496322401319223e-05,
+      "loss": 0.9868,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.496276035187393e-05,
+      "loss": 0.8672,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4962293793255108e-05,
+      "loss": 0.9336,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4961824337516905e-05,
+      "loss": 0.9438,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4961351984841575e-05,
+      "loss": 0.9331,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4960876735412501e-05,
+      "loss": 0.9189,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4960398589414193e-05,
+      "loss": 0.9102,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4959917547032285e-05,
+      "loss": 0.8916,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.495943360845353e-05,
+      "loss": 0.2745,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4958946773865815e-05,
+      "loss": 1.0063,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4958457043458145e-05,
+      "loss": 0.9434,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4957964417420645e-05,
+      "loss": 0.9814,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4957468895944575e-05,
+      "loss": 0.8682,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.495697047922231e-05,
+      "loss": 0.979,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4956469167447355e-05,
+      "loss": 0.9141,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4955964960814334e-05,
+      "loss": 0.9136,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4955457859519e-05,
+      "loss": 0.8813,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4954947863758224e-05,
+      "loss": 0.9448,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.495443497373001e-05,
+      "loss": 0.9233,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4953919189633473e-05,
+      "loss": 0.9131,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4953400511668861e-05,
+      "loss": 0.8608,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4952878940037541e-05,
+      "loss": 0.9331,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4952354474942009e-05,
+      "loss": 0.9351,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4951827116585874e-05,
+      "loss": 0.9116,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4951296865173882e-05,
+      "loss": 0.9351,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4950763720911891e-05,
+      "loss": 0.9512,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4950227684006884e-05,
+      "loss": 0.9194,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4949688754666973e-05,
+      "loss": 0.9468,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4949146933101386e-05,
+      "loss": 0.8682,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4948602219520477e-05,
+      "loss": 0.9712,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4948054614135727e-05,
+      "loss": 0.9053,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.494750411715973e-05,
+      "loss": 0.9092,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4946950728806209e-05,
+      "loss": 0.9263,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4946394449290007e-05,
+      "loss": 0.9609,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4945835278827096e-05,
+      "loss": 0.916,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4945273217634562e-05,
+      "loss": 0.9424,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4944708265930617e-05,
+      "loss": 0.875,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4944140423934592e-05,
+      "loss": 0.9155,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4943569691866947e-05,
+      "loss": 0.3118,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4942996069949256e-05,
+      "loss": 0.8989,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4942419558404222e-05,
+      "loss": 0.9722,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4941840157455667e-05,
+      "loss": 0.9009,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4941257867328532e-05,
+      "loss": 0.9067,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4940672688248882e-05,
+      "loss": 0.9092,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4940084620443906e-05,
+      "loss": 0.9736,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4939493664141909e-05,
+      "loss": 0.9766,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4938899819572323e-05,
+      "loss": 0.9224,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4938303086965698e-05,
+      "loss": 0.9238,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4937703466553705e-05,
+      "loss": 0.8887,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4937100958569143e-05,
+      "loss": 0.9199,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4936495563245918e-05,
+      "loss": 0.9043,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4935887280819072e-05,
+      "loss": 0.9043,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4935276111524754e-05,
+      "loss": 0.936,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.493466205560025e-05,
+      "loss": 0.9463,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4934045113283953e-05,
+      "loss": 0.9062,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.493342528481538e-05,
+      "loss": 0.9194,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4932802570435171e-05,
+      "loss": 0.9526,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4932176970385085e-05,
+      "loss": 0.936,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4931548484908001e-05,
+      "loss": 0.9233,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4930917114247922e-05,
+      "loss": 0.8984,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4930282858649963e-05,
+      "loss": 0.937,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4929645718360367e-05,
+      "loss": 0.897,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4929005693626493e-05,
+      "loss": 0.8906,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.492836278469682e-05,
+      "loss": 0.9663,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4927716991820947e-05,
+      "loss": 0.8838,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4927068315249595e-05,
+      "loss": 0.918,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4926416755234605e-05,
+      "loss": 0.8984,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4925762312028926e-05,
+      "loss": 0.9258,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4925104985886645e-05,
+      "loss": 0.9014,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4924444777062951e-05,
+      "loss": 0.9048,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4923781685814167e-05,
+      "loss": 0.8911,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4923115712397723e-05,
+      "loss": 0.9087,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4922446857072174e-05,
+      "loss": 0.9355,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4921775120097193e-05,
+      "loss": 0.9341,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4921100501733571e-05,
+      "loss": 0.8955,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.492042300224322e-05,
+      "loss": 0.936,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4919742621889163e-05,
+      "loss": 0.9331,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4919059360935554e-05,
+      "loss": 0.938,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4918373219647653e-05,
+      "loss": 0.9624,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4917684198291847e-05,
+      "loss": 0.9277,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4916992297135637e-05,
+      "loss": 0.9541,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4916297516447645e-05,
+      "loss": 0.9253,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4915599856497604e-05,
+      "loss": 0.9507,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4914899317556371e-05,
+      "loss": 0.8848,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4914195899895921e-05,
+      "loss": 0.9595,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4913489603789348e-05,
+      "loss": 0.9385,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4912780429510852e-05,
+      "loss": 0.8706,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4912068377335766e-05,
+      "loss": 0.8936,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4911353447540533e-05,
+      "loss": 0.9136,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.491063564040271e-05,
+      "loss": 0.9448,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4909914956200974e-05,
+      "loss": 0.9355,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4909191395215122e-05,
+      "loss": 0.9463,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4908464957726065e-05,
+      "loss": 0.9619,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4907735644015832e-05,
+      "loss": 0.8989,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4907003454367563e-05,
+      "loss": 0.8911,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4906268389065523e-05,
+      "loss": 0.2871,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.490553044839509e-05,
+      "loss": 0.2974,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4904789632642757e-05,
+      "loss": 0.897,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4904045942096136e-05,
+      "loss": 0.9263,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4903299377043947e-05,
+      "loss": 0.9141,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.490254993777604e-05,
+      "loss": 0.9233,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.490179762458337e-05,
+      "loss": 0.8931,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.490104243775801e-05,
+      "loss": 0.9497,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.490028437759315e-05,
+      "loss": 0.9565,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4899523444383096e-05,
+      "loss": 0.8955,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4898759638423266e-05,
+      "loss": 0.9443,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.48979929600102e-05,
+      "loss": 0.9346,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4897223409441546e-05,
+      "loss": 0.9526,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.489645098701607e-05,
+      "loss": 0.9199,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4895675693033654e-05,
+      "loss": 0.9556,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4894897527795293e-05,
+      "loss": 0.9106,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4894116491603099e-05,
+      "loss": 0.8354,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4893332584760294e-05,
+      "loss": 0.9399,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4892545807571222e-05,
+      "loss": 0.8945,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4891756160341334e-05,
+      "loss": 0.9253,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4890963643377198e-05,
+      "loss": 0.9131,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4890168256986499e-05,
+      "loss": 0.9712,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4889370001478028e-05,
+      "loss": 0.9497,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4888568877161703e-05,
+      "loss": 0.875,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4887764884348543e-05,
+      "loss": 0.8638,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4886958023350686e-05,
+      "loss": 0.3611,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4886148294481385e-05,
+      "loss": 0.3271,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4885335698055002e-05,
+      "loss": 0.8921,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4884520234387018e-05,
+      "loss": 0.9185,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.488370190379402e-05,
+      "loss": 0.9341,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4882880706593717e-05,
+      "loss": 0.9048,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4882056643104923e-05,
+      "loss": 0.8828,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4881229713647569e-05,
+      "loss": 0.9727,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4880399918542695e-05,
+      "loss": 0.897,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.487956725811246e-05,
+      "loss": 0.9204,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.487873173268013e-05,
+      "loss": 0.9121,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4877893342570083e-05,
+      "loss": 0.9282,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4877052088107812e-05,
+      "loss": 0.9268,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4876207969619923e-05,
+      "loss": 0.939,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4875360987434127e-05,
+      "loss": 0.9307,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4874511141879257e-05,
+      "loss": 0.9287,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4873658433285247e-05,
+      "loss": 0.8599,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4872802861983153e-05,
+      "loss": 0.9731,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4871944428305133e-05,
+      "loss": 0.9258,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4871083132584464e-05,
+      "loss": 0.8608,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.487021897515553e-05,
+      "loss": 0.8877,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4869351956353824e-05,
+      "loss": 0.9019,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4868482076515955e-05,
+      "loss": 0.877,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.486760933597964e-05,
+      "loss": 0.9478,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4866733735083706e-05,
+      "loss": 0.9365,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4865855274168093e-05,
+      "loss": 0.8984,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4864973953573846e-05,
+      "loss": 0.8838,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4864089773643128e-05,
+      "loss": 0.8882,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4863202734719208e-05,
+      "loss": 0.9111,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4862312837146463e-05,
+      "loss": 0.9277,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4861420081270385e-05,
+      "loss": 0.8926,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4860524467437568e-05,
+      "loss": 0.916,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4859625995995723e-05,
+      "loss": 0.9771,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4858724667293665e-05,
+      "loss": 0.895,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4857820481681325e-05,
+      "loss": 0.8638,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4856913439509737e-05,
+      "loss": 0.8418,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4856003541131045e-05,
+      "loss": 0.835,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4855090786898506e-05,
+      "loss": 0.9756,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4854175177166476e-05,
+      "loss": 0.9155,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4853256712290434e-05,
+      "loss": 0.9497,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4852335392626955e-05,
+      "loss": 0.9292,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4851411218533728e-05,
+      "loss": 0.9751,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4850484190369551e-05,
+      "loss": 0.9116,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4849554308494326e-05,
+      "loss": 0.9194,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4848621573269066e-05,
+      "loss": 0.8677,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.484768598505589e-05,
+      "loss": 0.9072,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4846747544218025e-05,
+      "loss": 0.8599,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.484580625111981e-05,
+      "loss": 0.9058,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4844862106126684e-05,
+      "loss": 0.855,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4843915109605197e-05,
+      "loss": 0.9199,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4842965261923007e-05,
+      "loss": 0.9644,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4842012563448875e-05,
+      "loss": 0.8726,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4841057014552673e-05,
+      "loss": 0.9175,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.484009861560538e-05,
+      "loss": 0.8618,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4839137366979077e-05,
+      "loss": 0.9487,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4838173269046952e-05,
+      "loss": 0.4392,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4837206322183305e-05,
+      "loss": 0.9219,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4836236526763538e-05,
+      "loss": 0.9268,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4835263883164153e-05,
+      "loss": 0.9248,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.483428839176277e-05,
+      "loss": 0.9341,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4833310052938105e-05,
+      "loss": 0.896,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4832328867069988e-05,
+      "loss": 0.876,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4831344834539344e-05,
+      "loss": 0.3862,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4830357955728208e-05,
+      "loss": 0.8896,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4829368231019722e-05,
+      "loss": 0.8882,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4828375660798132e-05,
+      "loss": 0.9229,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4827380245448788e-05,
+      "loss": 0.9453,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4826381985358144e-05,
+      "loss": 0.9541,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4825380880913756e-05,
+      "loss": 0.9448,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4824376932504292e-05,
+      "loss": 0.8984,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4823370140519514e-05,
+      "loss": 0.8774,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.48223605053503e-05,
+      "loss": 0.8706,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4821348027388622e-05,
+      "loss": 0.9932,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4820332707027554e-05,
+      "loss": 0.8672,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4819314544661286e-05,
+      "loss": 0.8921,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4818293540685099e-05,
+      "loss": 0.9067,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4817269695495383e-05,
+      "loss": 0.9404,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4816243009489632e-05,
+      "loss": 0.9087,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.481521348306644e-05,
+      "loss": 0.3298,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4814181116625501e-05,
+      "loss": 0.9551,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.481314591056762e-05,
+      "loss": 0.9229,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4812107865294694e-05,
+      "loss": 0.9141,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4811066981209735e-05,
+      "loss": 0.8779,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4810023258716846e-05,
+      "loss": 0.8662,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4808976698221235e-05,
+      "loss": 0.897,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4807927300129215e-05,
+      "loss": 0.9072,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4806875064848197e-05,
+      "loss": 0.9077,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4805819992786697e-05,
+      "loss": 0.9287,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.480476208435433e-05,
+      "loss": 0.3418,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4803701339961808e-05,
+      "loss": 0.9297,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4802637760020957e-05,
+      "loss": 0.9097,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4801571344944686e-05,
+      "loss": 0.9263,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.480050209514702e-05,
+      "loss": 0.9429,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4799430011043077e-05,
+      "loss": 0.8794,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4798355093049077e-05,
+      "loss": 0.8809,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.479727734158234e-05,
+      "loss": 0.9595,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4796196757061287e-05,
+      "loss": 0.9048,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4795113339905437e-05,
+      "loss": 0.8574,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.479402709053541e-05,
+      "loss": 0.9209,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4792938009372925e-05,
+      "loss": 0.9268,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4791846096840803e-05,
+      "loss": 0.9526,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.479075135336296e-05,
+      "loss": 0.8916,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4789653779364415e-05,
+      "loss": 0.8662,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4788553375271282e-05,
+      "loss": 0.9819,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4787450141510776e-05,
+      "loss": 0.873,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4786344078511211e-05,
+      "loss": 0.8984,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4785235186701997e-05,
+      "loss": 0.9126,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4784123466513647e-05,
+      "loss": 0.9268,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4783008918377767e-05,
+      "loss": 0.9058,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4781891542727066e-05,
+      "loss": 0.8696,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4780771339995344e-05,
+      "loss": 0.9097,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4779648310617501e-05,
+      "loss": 0.9019,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4778522455029541e-05,
+      "loss": 0.8887,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4777393773668555e-05,
+      "loss": 0.8921,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.477626226697274e-05,
+      "loss": 0.9365,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4775127935381385e-05,
+      "loss": 0.9307,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4773990779334871e-05,
+      "loss": 0.9463,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4772850799274689e-05,
+      "loss": 0.9219,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4771707995643415e-05,
+      "loss": 0.9009,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4770562368884722e-05,
+      "loss": 0.895,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4769413919443388e-05,
+      "loss": 0.9448,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4768262647765277e-05,
+      "loss": 0.9482,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4767108554297351e-05,
+      "loss": 0.9424,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4765951639487671e-05,
+      "loss": 0.8682,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.476479190378539e-05,
+      "loss": 0.8809,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4763629347640758e-05,
+      "loss": 0.8853,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4762463971505118e-05,
+      "loss": 0.937,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4761295775830914e-05,
+      "loss": 0.877,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4760124761071673e-05,
+      "loss": 0.8945,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4758950927682028e-05,
+      "loss": 0.8779,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4757774276117698e-05,
+      "loss": 0.8638,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4756594806835503e-05,
+      "loss": 0.8862,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4755412520293353e-05,
+      "loss": 0.8848,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.475422741695025e-05,
+      "loss": 0.938,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4753039497266292e-05,
+      "loss": 0.9189,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4751848761702672e-05,
+      "loss": 0.9404,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4750655210721676e-05,
+      "loss": 0.8413,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4749458844786679e-05,
+      "loss": 0.8511,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4748259664362154e-05,
+      "loss": 0.9199,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.474705766991366e-05,
+      "loss": 0.9082,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4745852861907855e-05,
+      "loss": 0.8486,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.474464524081249e-05,
+      "loss": 0.9072,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4743434807096397e-05,
+      "loss": 0.3149,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4742221561229516e-05,
+      "loss": 0.3611,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4741005503682869e-05,
+      "loss": 0.8911,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.473978663492857e-05,
+      "loss": 0.894,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4738564955439827e-05,
+      "loss": 0.936,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4737340465690933e-05,
+      "loss": 0.8882,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4736113166157285e-05,
+      "loss": 0.9272,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4734883057315357e-05,
+      "loss": 0.9414,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4733650139642725e-05,
+      "loss": 0.9424,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4732414413618045e-05,
+      "loss": 0.8887,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4731175879721069e-05,
+      "loss": 0.8989,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4729934538432641e-05,
+      "loss": 0.9038,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4728690390234692e-05,
+      "loss": 0.9048,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4727443435610242e-05,
+      "loss": 0.8628,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4726193675043401e-05,
+      "loss": 0.9185,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4724941109019372e-05,
+      "loss": 0.3491,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4723685738024442e-05,
+      "loss": 0.9048,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4722427562545989e-05,
+      "loss": 0.8921,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.472116658307248e-05,
+      "loss": 0.8638,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4719902800093473e-05,
+      "loss": 0.8696,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4718636214099609e-05,
+      "loss": 0.9053,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4717366825582623e-05,
+      "loss": 0.8701,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4716094635035334e-05,
+      "loss": 0.8843,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4714819642951648e-05,
+      "loss": 0.8794,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4713541849826566e-05,
+      "loss": 0.9453,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4712261256156168e-05,
+      "loss": 0.8379,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4710977862437626e-05,
+      "loss": 0.8862,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4709691669169194e-05,
+      "loss": 0.896,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.470840267685022e-05,
+      "loss": 0.9038,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4707110885981132e-05,
+      "loss": 0.8818,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4705816297063453e-05,
+      "loss": 0.9346,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4704518910599784e-05,
+      "loss": 0.895,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4703218727093813e-05,
+      "loss": 0.323,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.470191574705032e-05,
+      "loss": 0.3123,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4700609970975164e-05,
+      "loss": 0.853,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4699301399375294e-05,
+      "loss": 0.8711,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4697990032758742e-05,
+      "loss": 0.873,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4696675871634622e-05,
+      "loss": 0.9209,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4695358916513142e-05,
+      "loss": 0.8745,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4694039167905586e-05,
+      "loss": 0.9033,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.469271662632433e-05,
+      "loss": 0.8721,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4691391292282823e-05,
+      "loss": 0.897,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4690063166295613e-05,
+      "loss": 0.8813,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4688732248878319e-05,
+      "loss": 0.9067,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.468739854054765e-05,
+      "loss": 0.8921,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4686062041821399e-05,
+      "loss": 0.9224,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.468472275321844e-05,
+      "loss": 0.8955,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.468338067525873e-05,
+      "loss": 0.9609,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4682035808463314e-05,
+      "loss": 0.9258,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.468068815335431e-05,
+      "loss": 0.8906,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4679337710454927e-05,
+      "loss": 0.8447,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4677984480289452e-05,
+      "loss": 0.9385,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4676628463383258e-05,
+      "loss": 0.8799,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4675269660262794e-05,
+      "loss": 0.8779,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4673908071455598e-05,
+      "loss": 0.9106,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4672543697490283e-05,
+      "loss": 0.9165,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4671176538896546e-05,
+      "loss": 0.9292,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4669806596205168e-05,
+      "loss": 0.9028,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4668433869948003e-05,
+      "loss": 0.9097,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4667058360657995e-05,
+      "loss": 0.876,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4665680068869162e-05,
+      "loss": 0.8906,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4664298995116603e-05,
+      "loss": 0.874,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4662915139936501e-05,
+      "loss": 0.8335,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4661528503866119e-05,
+      "loss": 0.8838,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4660139087443789e-05,
+      "loss": 0.9146,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4658746891208936e-05,
+      "loss": 0.8794,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4657351915702055e-05,
+      "loss": 0.9146,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4655954161464728e-05,
+      "loss": 0.8936,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4654553629039608e-05,
+      "loss": 0.8374,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.465315031897043e-05,
+      "loss": 0.8667,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4651744231802007e-05,
+      "loss": 0.8491,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4650335368080232e-05,
+      "loss": 0.9131,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4648923728352075e-05,
+      "loss": 0.854,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4647509313165583e-05,
+      "loss": 0.8257,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4646092123069879e-05,
+      "loss": 0.8896,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4644672158615163e-05,
+      "loss": 0.9189,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4643249420352719e-05,
+      "loss": 0.8916,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4641823908834901e-05,
+      "loss": 0.958,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4640395624615142e-05,
+      "loss": 0.8921,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4638964568247948e-05,
+      "loss": 0.9585,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4637530740288909e-05,
+      "loss": 0.3516,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4636094141294683e-05,
+      "loss": 0.8809,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4634654771823008e-05,
+      "loss": 0.8389,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4633212632432698e-05,
+      "loss": 0.8833,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4631767723683641e-05,
+      "loss": 0.873,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.46303200461368e-05,
+      "loss": 0.8916,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4628869600354214e-05,
+      "loss": 0.8594,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4627416386898992e-05,
+      "loss": 0.9243,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4625960406335324e-05,
+      "loss": 0.8608,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4624501659228475e-05,
+      "loss": 0.8931,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4623040146144778e-05,
+      "loss": 0.9009,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4621575867651639e-05,
+      "loss": 0.9321,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4620108824317549e-05,
+      "loss": 0.8623,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4618639016712061e-05,
+      "loss": 0.9224,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4617166445405804e-05,
+      "loss": 0.8408,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.461569111097048e-05,
+      "loss": 0.8569,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4614213013978868e-05,
+      "loss": 0.9087,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4612732155004813e-05,
+      "loss": 0.3562,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4611248534623238e-05,
+      "loss": 0.8237,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4609762153410135e-05,
+      "loss": 0.8779,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4608273011942566e-05,
+      "loss": 0.3604,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.460678111079867e-05,
+      "loss": 0.9097,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4605286450557654e-05,
+      "loss": 0.8647,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4603789031799794e-05,
+      "loss": 0.3345,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4602288855106441e-05,
+      "loss": 0.9351,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4600785921060018e-05,
+      "loss": 0.9351,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.459928023024401e-05,
+      "loss": 0.9355,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4597771783242984e-05,
+      "loss": 0.9292,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.459626058064257e-05,
+      "loss": 0.9023,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4594746623029462e-05,
+      "loss": 0.8896,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4593229910991437e-05,
+      "loss": 0.9048,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4591710445117334e-05,
+      "loss": 0.9087,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4590188225997064e-05,
+      "loss": 0.9165,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.45886632542216e-05,
+      "loss": 0.8984,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.458713553038299e-05,
+      "loss": 0.9263,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4585605055074352e-05,
+      "loss": 0.3213,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4584071828889866e-05,
+      "loss": 0.8765,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4582535852424785e-05,
+      "loss": 0.8569,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4580997126275424e-05,
+      "loss": 0.875,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4579455651039176e-05,
+      "loss": 0.8643,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4577911427314491e-05,
+      "loss": 0.8696,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4576364455700891e-05,
+      "loss": 0.9072,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4574814736798963e-05,
+      "loss": 0.8916,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.457326227121036e-05,
+      "loss": 0.8813,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4571707059537802e-05,
+      "loss": 0.8833,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4570149102385079e-05,
+      "loss": 0.9082,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4568588400357045e-05,
+      "loss": 0.3101,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4567024954059613e-05,
+      "loss": 0.9155,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4565458764099767e-05,
+      "loss": 0.9067,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4563889831085561e-05,
+      "loss": 0.9448,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4562318155626104e-05,
+      "loss": 0.8594,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4560743738331576e-05,
+      "loss": 0.8779,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4559166579813222e-05,
+      "loss": 0.8604,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4557586680683347e-05,
+      "loss": 0.8486,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4556004041555323e-05,
+      "loss": 0.9141,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4554418663043583e-05,
+      "loss": 0.8828,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4552830545763629e-05,
+      "loss": 0.9253,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4551239690332019e-05,
+      "loss": 0.9146,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4549646097366381e-05,
+      "loss": 0.8916,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4548049767485402e-05,
+      "loss": 0.9775,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4546450701308833e-05,
+      "loss": 0.8989,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4544848899457483e-05,
+      "loss": 0.9419,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4543244362553229e-05,
+      "loss": 0.9429,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4541637091219009e-05,
+      "loss": 0.8535,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4540027086078819e-05,
+      "loss": 0.9038,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.453841434775772e-05,
+      "loss": 0.9019,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.453679887688183e-05,
+      "loss": 0.9351,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4535180674078331e-05,
+      "loss": 0.8652,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4533559739975468e-05,
+      "loss": 0.8877,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.453193607520254e-05,
+      "loss": 0.8994,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4530309680389913e-05,
+      "loss": 0.8833,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4528680556169005e-05,
+      "loss": 0.916,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.45270487031723e-05,
+      "loss": 0.9131,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4525414122033343e-05,
+      "loss": 0.8779,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4523776813386731e-05,
+      "loss": 0.2992,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4522136777868123e-05,
+      "loss": 0.896,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4520494016114239e-05,
+      "loss": 0.8901,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4518848528762857e-05,
+      "loss": 0.9092,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4517200316452807e-05,
+      "loss": 0.8696,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4515549379823988e-05,
+      "loss": 0.9136,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4513895719517344e-05,
+      "loss": 0.8364,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.451223933617489e-05,
+      "loss": 0.9097,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4510580230439682e-05,
+      "loss": 0.8608,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4508918402955854e-05,
+      "loss": 0.8911,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4507253854368572e-05,
+      "loss": 0.8716,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4505586585324079e-05,
+      "loss": 0.8652,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4503916596469665e-05,
+      "loss": 0.8896,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4502243888453674e-05,
+      "loss": 0.9014,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4500568461925513e-05,
+      "loss": 0.9014,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4498890317535639e-05,
+      "loss": 0.8779,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4497209455935566e-05,
+      "loss": 0.9023,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4495525877777857e-05,
+      "loss": 0.8877,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4493839583716142e-05,
+      "loss": 0.3245,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4492150574405095e-05,
+      "loss": 0.3042,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4490458850500449e-05,
+      "loss": 0.9062,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4488764412658988e-05,
+      "loss": 0.9253,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4487067261538553e-05,
+      "loss": 0.9028,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4485367397798031e-05,
+      "loss": 0.8569,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4483664822097376e-05,
+      "loss": 0.8804,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4481959535097581e-05,
+      "loss": 0.8823,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.44802515374607e-05,
+      "loss": 0.8945,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4478540829849835e-05,
+      "loss": 0.9609,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.447682741292914e-05,
+      "loss": 0.9463,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4475111287363827e-05,
+      "loss": 0.8994,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4473392453820149e-05,
+      "loss": 0.9048,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.447167091296542e-05,
+      "loss": 0.8662,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4469946665468e-05,
+      "loss": 0.874,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4468219711997305e-05,
+      "loss": 0.9268,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4466490053223793e-05,
+      "loss": 0.8838,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4464757689818978e-05,
+      "loss": 0.9033,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4463022622455428e-05,
+      "loss": 0.8442,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4461284851806748e-05,
+      "loss": 0.894,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4459544378547608e-05,
+      "loss": 0.8945,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4457801203353715e-05,
+      "loss": 0.8809,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.445605532690183e-05,
+      "loss": 0.9609,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4454306749869762e-05,
+      "loss": 0.874,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4452555472936372e-05,
+      "loss": 0.875,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4450801496781563e-05,
+      "loss": 0.897,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4449044822086291e-05,
+      "loss": 0.9229,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4447285449532556e-05,
+      "loss": 0.8608,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4445523379803404e-05,
+      "loss": 0.8667,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4443758613582941e-05,
+      "loss": 0.9043,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4441991151556297e-05,
+      "loss": 0.8408,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4440220994409673e-05,
+      "loss": 0.9214,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4438448142830297e-05,
+      "loss": 0.8545,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4436672597506457e-05,
+      "loss": 0.897,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4434894359127477e-05,
+      "loss": 0.835,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.443311342838373e-05,
+      "loss": 0.9062,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4431329805966636e-05,
+      "loss": 0.8745,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.442954349256866e-05,
+      "loss": 0.8643,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.442775448888331e-05,
+      "loss": 0.9219,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4425962795605135e-05,
+      "loss": 0.7808,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4424168413429737e-05,
+      "loss": 0.8398,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4422371343053755e-05,
+      "loss": 0.3484,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4420571585174873e-05,
+      "loss": 0.9165,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4418769140491817e-05,
+      "loss": 0.3667,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4416964009704364e-05,
+      "loss": 0.9238,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4415156193513322e-05,
+      "loss": 0.3071,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4413345692620551e-05,
+      "loss": 0.8418,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4411532507728947e-05,
+      "loss": 0.9009,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4409716639542454e-05,
+      "loss": 0.9409,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4407898088766054e-05,
+      "loss": 0.8882,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4406076856105768e-05,
+      "loss": 0.9214,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4404252942268664e-05,
+      "loss": 0.8452,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4402426347962846e-05,
+      "loss": 0.9102,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4400597073897468e-05,
+      "loss": 0.8628,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4398765120782705e-05,
+      "loss": 0.8315,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4396930489329793e-05,
+      "loss": 0.8579,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4395093180250996e-05,
+      "loss": 0.8867,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4393253194259622e-05,
+      "loss": 0.8398,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4391410532070015e-05,
+      "loss": 0.8125,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4389565194397564e-05,
+      "loss": 0.9424,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4387717181958689e-05,
+      "loss": 0.8779,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4385866495470851e-05,
+      "loss": 0.8687,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4384013135652554e-05,
+      "loss": 0.8887,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4382157103223335e-05,
+      "loss": 0.8584,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4380298398903764e-05,
+      "loss": 0.8804,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4378437023415464e-05,
+      "loss": 0.8335,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4376572977481079e-05,
+      "loss": 0.9082,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4374706261824296e-05,
+      "loss": 0.9053,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.437283687716984e-05,
+      "loss": 0.8491,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4370964824243468e-05,
+      "loss": 0.9199,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4369090103771979e-05,
+      "loss": 0.8833,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4367212716483201e-05,
+      "loss": 0.8989,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4365332663106003e-05,
+      "loss": 0.8721,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4363449944370283e-05,
+      "loss": 0.8647,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4361564561006978e-05,
+      "loss": 0.8359,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4359676513748061e-05,
+      "loss": 0.9102,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4357785803326536e-05,
+      "loss": 0.9297,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.435589243047644e-05,
+      "loss": 0.9062,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.435399639593285e-05,
+      "loss": 0.9258,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4352097700431867e-05,
+      "loss": 0.8672,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4350196344710633e-05,
+      "loss": 0.897,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.434829232950732e-05,
+      "loss": 0.8491,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4346385655561129e-05,
+      "loss": 0.8677,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.43444763236123e-05,
+      "loss": 0.8584,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4342564334402098e-05,
+      "loss": 0.832,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4340649688672826e-05,
+      "loss": 0.9009,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4338732387167816e-05,
+      "loss": 0.9473,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4336812430631426e-05,
+      "loss": 0.8608,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4334889819809055e-05,
+      "loss": 0.8975,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4332964555447122e-05,
+      "loss": 0.9004,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4331036638293084e-05,
+      "loss": 0.3311,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4329106069095421e-05,
+      "loss": 0.9155,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4327172848603652e-05,
+      "loss": 0.8735,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4325236977568315e-05,
+      "loss": 0.9229,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4323298456740984e-05,
+      "loss": 0.8677,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4321357286874258e-05,
+      "loss": 0.8784,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.431941346872177e-05,
+      "loss": 0.8896,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.431746700303817e-05,
+      "loss": 0.8926,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.431551789057915e-05,
+      "loss": 0.3149,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4313566132101417e-05,
+      "loss": 0.3173,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4311611728362719e-05,
+      "loss": 0.9351,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.430965468012181e-05,
+      "loss": 0.8994,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4307694988138498e-05,
+      "loss": 0.8667,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.430573265317359e-05,
+      "loss": 0.9209,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4303767675988941e-05,
+      "loss": 0.8882,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.430180005734742e-05,
+      "loss": 0.9185,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4299829798012924e-05,
+      "loss": 0.7866,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4297856898750372e-05,
+      "loss": 0.8638,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4295881360325718e-05,
+      "loss": 0.8569,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.429390318350593e-05,
+      "loss": 0.873,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4291922369059006e-05,
+      "loss": 0.9346,
+      "step": 859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4289938917753963e-05,
+      "loss": 0.8779,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.428795283036085e-05,
+      "loss": 0.8584,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4285964107650734e-05,
+      "loss": 0.8892,
+      "step": 862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.42839727503957e-05,
+      "loss": 0.9585,
+      "step": 863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4281978759368865e-05,
+      "loss": 0.8452,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4279982135344369e-05,
+      "loss": 0.9116,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4277982879097364e-05,
+      "loss": 0.8677,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4275980991404033e-05,
+      "loss": 0.2898,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4273976473041577e-05,
+      "loss": 0.9028,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.427196932478822e-05,
+      "loss": 0.9038,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4269959547423206e-05,
+      "loss": 0.8701,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4267947141726796e-05,
+      "loss": 0.8599,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4265932108480281e-05,
+      "loss": 0.8726,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4263914448465964e-05,
+      "loss": 0.8423,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4261894162467167e-05,
+      "loss": 0.8496,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4259871251268239e-05,
+      "loss": 0.8804,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4257845715654543e-05,
+      "loss": 0.856,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4255817556412462e-05,
+      "loss": 0.8438,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4253786774329393e-05,
+      "loss": 0.8652,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4251753370193758e-05,
+      "loss": 0.8799,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4249717344794998e-05,
+      "loss": 0.8901,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4247678698923562e-05,
+      "loss": 0.8657,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4245637433370927e-05,
+      "loss": 0.3167,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4243593548929582e-05,
+      "loss": 0.8315,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4241547046393032e-05,
+      "loss": 0.8843,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4239497926555802e-05,
+      "loss": 0.8921,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4237446190213428e-05,
+      "loss": 0.9072,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4235391838162466e-05,
+      "loss": 0.9253,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4233334871200484e-05,
+      "loss": 0.8452,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.423127529012607e-05,
+      "loss": 0.9092,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4229213095738826e-05,
+      "loss": 0.834,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4227148288839363e-05,
+      "loss": 0.8809,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4225080870229313e-05,
+      "loss": 0.897,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4223010840711318e-05,
+      "loss": 0.9043,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4220938201089032e-05,
+      "loss": 0.9443,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.421886295216713e-05,
+      "loss": 0.8052,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4216785094751292e-05,
+      "loss": 0.854,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4214704629648215e-05,
+      "loss": 0.8506,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4212621557665607e-05,
+      "loss": 0.3302,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4210535879612185e-05,
+      "loss": 0.8618,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4208447596297686e-05,
+      "loss": 0.853,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4206356708532853e-05,
+      "loss": 0.9019,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.420426321712944e-05,
+      "loss": 0.8999,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4202167122900209e-05,
+      "loss": 0.8853,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4200068426658938e-05,
+      "loss": 0.8721,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4197967129220416e-05,
+      "loss": 0.8389,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4195863231400434e-05,
+      "loss": 0.9248,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4193756734015802e-05,
+      "loss": 0.9087,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4191647637884334e-05,
+      "loss": 0.8306,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.418953594382485e-05,
+      "loss": 0.3416,
+      "step": 909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4187421652657187e-05,
+      "loss": 0.8926,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4185304765202184e-05,
+      "loss": 0.8804,
+      "step": 911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4183185282281687e-05,
+      "loss": 0.8823,
+      "step": 912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4181063204718557e-05,
+      "loss": 0.8936,
+      "step": 913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4178938533336655e-05,
+      "loss": 0.8652,
+      "step": 914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4176811268960851e-05,
+      "loss": 0.8394,
+      "step": 915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4174681412417023e-05,
+      "loss": 0.8999,
+      "step": 916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4172548964532055e-05,
+      "loss": 0.8696,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4170413926133837e-05,
+      "loss": 0.8496,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4168276298051264e-05,
+      "loss": 0.8994,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4166136081114236e-05,
+      "loss": 0.8667,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.416399327615366e-05,
+      "loss": 0.8818,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4161847884001445e-05,
+      "loss": 0.8115,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4159699905490508e-05,
+      "loss": 0.8208,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4157549341454767e-05,
+      "loss": 0.8379,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4155396192729144e-05,
+      "loss": 0.8452,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.415324046014957e-05,
+      "loss": 0.9019,
+      "step": 926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.415108214455297e-05,
+      "loss": 0.3547,
+      "step": 927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.414892124677728e-05,
+      "loss": 0.9062,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.414675776766143e-05,
+      "loss": 0.8613,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4144591708045363e-05,
+      "loss": 0.9258,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4142423068770012e-05,
+      "loss": 0.8716,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4140251850677322e-05,
+      "loss": 0.8672,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4138078054610233e-05,
+      "loss": 0.9155,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4135901681412687e-05,
+      "loss": 0.9141,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4133722731929628e-05,
+      "loss": 0.8174,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4131541207006998e-05,
+      "loss": 0.8486,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.412935710749174e-05,
+      "loss": 0.8794,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.41271704342318e-05,
+      "loss": 0.8413,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4124981188076115e-05,
+      "loss": 0.8232,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4122789369874629e-05,
+      "loss": 0.8579,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.412059498047828e-05,
+      "loss": 0.8594,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4118398020739005e-05,
+      "loss": 0.8652,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4116198491509743e-05,
+      "loss": 0.855,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.411399639364442e-05,
+      "loss": 0.8608,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4111791727997974e-05,
+      "loss": 0.9351,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4109584495426328e-05,
+      "loss": 0.8828,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4107374696786406e-05,
+      "loss": 0.3455,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4105162332936125e-05,
+      "loss": 0.8779,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.410294740473441e-05,
+      "loss": 0.8872,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4100729913041162e-05,
+      "loss": 0.8867,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4098509858717293e-05,
+      "loss": 0.8706,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4096287242624702e-05,
+      "loss": 0.8882,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4094062065626289e-05,
+      "loss": 0.9038,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4091834328585941e-05,
+      "loss": 0.8286,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4089604032368543e-05,
+      "loss": 0.8354,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4087371177839972e-05,
+      "loss": 0.8022,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4085135765867103e-05,
+      "loss": 0.8442,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4082897797317794e-05,
+      "loss": 0.8784,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4080657273060904e-05,
+      "loss": 0.8516,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4078414193966283e-05,
+      "loss": 0.8613,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4076168560904772e-05,
+      "loss": 0.8696,
+      "step": 961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.40739203747482e-05,
+      "loss": 0.8696,
+      "step": 962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.407166963636939e-05,
+      "loss": 0.8906,
+      "step": 963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.406941634664216e-05,
+      "loss": 0.8843,
+      "step": 964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4067160506441313e-05,
+      "loss": 0.9287,
+      "step": 965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4064902116642642e-05,
+      "loss": 0.9136,
+      "step": 966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4062641178122935e-05,
+      "loss": 0.8687,
+      "step": 967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4060377691759964e-05,
+      "loss": 0.8906,
+      "step": 968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4058111658432491e-05,
+      "loss": 0.7979,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.405584307902027e-05,
+      "loss": 0.9092,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4053571954404042e-05,
+      "loss": 0.916,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4051298285465536e-05,
+      "loss": 0.8003,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4049022073087467e-05,
+      "loss": 0.9033,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4046743318153539e-05,
+      "loss": 0.3267,
+      "step": 974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4044462021548446e-05,
+      "loss": 0.8672,
+      "step": 975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.404217818415786e-05,
+      "loss": 0.8589,
+      "step": 976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4039891806868452e-05,
+      "loss": 0.98,
+      "step": 977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.403760289056787e-05,
+      "loss": 0.8789,
+      "step": 978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.403531143614475e-05,
+      "loss": 0.853,
+      "step": 979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.403301744448871e-05,
+      "loss": 0.855,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4030720916490361e-05,
+      "loss": 0.8242,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4028421853041294e-05,
+      "loss": 0.8594,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4026120255034083e-05,
+      "loss": 0.8618,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4023816123362288e-05,
+      "loss": 0.8472,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4021509458920453e-05,
+      "loss": 0.8877,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4019200262604103e-05,
+      "loss": 0.8945,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4016888535309749e-05,
+      "loss": 0.9238,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4014574277934881e-05,
+      "loss": 0.9146,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4012257491377977e-05,
+      "loss": 0.8579,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4009938176538492e-05,
+      "loss": 0.8906,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4007616334316862e-05,
+      "loss": 0.8745,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4005291965614505e-05,
+      "loss": 0.9424,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4002965071333824e-05,
+      "loss": 0.8389,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.4000635652378201e-05,
+      "loss": 0.8818,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3998303709651992e-05,
+      "loss": 0.7852,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3995969244060543e-05,
+      "loss": 0.9321,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3993632256510167e-05,
+      "loss": 0.8389,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3991292747908169e-05,
+      "loss": 0.875,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3988950719162822e-05,
+      "loss": 0.8354,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3986606171183387e-05,
+      "loss": 0.8267,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3984259104880096e-05,
+      "loss": 0.8472,
+      "step": 1001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3981909521164158e-05,
+      "loss": 0.8394,
+      "step": 1002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.397955742094777e-05,
+      "loss": 0.3228,
+      "step": 1003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.397720280514409e-05,
+      "loss": 0.9014,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3974845674667265e-05,
+      "loss": 0.8208,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3972486030432414e-05,
+      "loss": 0.894,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3970123873355633e-05,
+      "loss": 0.9209,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3967759204353988e-05,
+      "loss": 0.8774,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3965392024345531e-05,
+      "loss": 0.8823,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.396302233424928e-05,
+      "loss": 0.9263,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.396065013498523e-05,
+      "loss": 0.8584,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3958275427474352e-05,
+      "loss": 0.8862,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3955898212638586e-05,
+      "loss": 0.8911,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3953518491400852e-05,
+      "loss": 0.3694,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3951136264685037e-05,
+      "loss": 0.8149,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3948751533416009e-05,
+      "loss": 0.916,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3946364298519595e-05,
+      "loss": 0.8682,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3943974560922607e-05,
+      "loss": 0.9102,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3941582321552822e-05,
+      "loss": 0.8667,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.393918758133899e-05,
+      "loss": 0.8809,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3936790341210827e-05,
+      "loss": 0.8901,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3934390602099033e-05,
+      "loss": 0.9214,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3931988364935264e-05,
+      "loss": 0.8643,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3929583630652152e-05,
+      "loss": 0.9092,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3927176400183296e-05,
+      "loss": 0.8833,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3924766674463267e-05,
+      "loss": 0.9355,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3922354454427604e-05,
+      "loss": 0.8467,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3919939741012815e-05,
+      "loss": 0.8892,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3917522535156374e-05,
+      "loss": 0.9507,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3915102837796721e-05,
+      "loss": 0.8965,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3912680649873271e-05,
+      "loss": 0.896,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3910255972326396e-05,
+      "loss": 0.8765,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.390782880609744e-05,
+      "loss": 0.9146,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3905399152128714e-05,
+      "loss": 0.9326,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3902967011363493e-05,
+      "loss": 0.938,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3900532384746017e-05,
+      "loss": 0.853,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3898095273221492e-05,
+      "loss": 0.8662,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3895655677736091e-05,
+      "loss": 0.8237,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3893213599236945e-05,
+      "loss": 0.8989,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3890769038672156e-05,
+      "loss": 0.8862,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3888321996990786e-05,
+      "loss": 0.8848,
+      "step": 1041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.388587247514286e-05,
+      "loss": 0.9692,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3883420474079366e-05,
+      "loss": 0.8838,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3880965994752258e-05,
+      "loss": 0.8491,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3878509038114447e-05,
+      "loss": 0.8398,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3876049605119809e-05,
+      "loss": 0.8213,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3873587696723181e-05,
+      "loss": 0.9268,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3871123313880359e-05,
+      "loss": 0.3103,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3868656457548102e-05,
+      "loss": 0.8086,
+      "step": 1049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3866187128684131e-05,
+      "loss": 0.854,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3863715328247122e-05,
+      "loss": 0.8584,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3861241057196712e-05,
+      "loss": 0.9141,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.38587643164935e-05,
+      "loss": 0.915,
+      "step": 1053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3856285107099043e-05,
+      "loss": 0.8564,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3853803429975851e-05,
+      "loss": 0.3242,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3851319286087403e-05,
+      "loss": 0.9058,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3848832676398124e-05,
+      "loss": 0.8286,
+      "step": 1057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3846343601873405e-05,
+      "loss": 0.8916,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3843852063479587e-05,
+      "loss": 0.9038,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3841358062183975e-05,
+      "loss": 0.8691,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3838861598954821e-05,
+      "loss": 0.8994,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3836362674761341e-05,
+      "loss": 0.896,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3833861290573704e-05,
+      "loss": 0.9199,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3831357447363035e-05,
+      "loss": 0.3247,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3828851146101408e-05,
+      "loss": 0.3308,
+      "step": 1065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3826342387761856e-05,
+      "loss": 0.875,
+      "step": 1066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3823831173318368e-05,
+      "loss": 0.9038,
+      "step": 1067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3821317503745883e-05,
+      "loss": 0.856,
+      "step": 1068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3818801380020293e-05,
+      "loss": 0.8613,
+      "step": 1069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3816282803118445e-05,
+      "loss": 0.8428,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3813761774018136e-05,
+      "loss": 0.8589,
+      "step": 1071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3811238293698119e-05,
+      "loss": 0.8672,
+      "step": 1072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3808712363138093e-05,
+      "loss": 0.8081,
+      "step": 1073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3806183983318712e-05,
+      "loss": 0.8965,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.380365315522158e-05,
+      "loss": 0.855,
+      "step": 1075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3801119879829254e-05,
+      "loss": 0.8936,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3798584158125237e-05,
+      "loss": 0.8223,
+      "step": 1077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3796045991093981e-05,
+      "loss": 0.9292,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3793505379720892e-05,
+      "loss": 0.8447,
+      "step": 1079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3790962324992325e-05,
+      "loss": 0.9102,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3788416827895577e-05,
+      "loss": 0.8877,
+      "step": 1081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3785868889418902e-05,
+      "loss": 0.9092,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3783318510551494e-05,
+      "loss": 0.8843,
+      "step": 1083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3780765692283498e-05,
+      "loss": 0.8662,
+      "step": 1084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.377821043560601e-05,
+      "loss": 0.8438,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3775652741511065e-05,
+      "loss": 0.8618,
+      "step": 1086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3773092610991648e-05,
+      "loss": 0.8682,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3770530045041689e-05,
+      "loss": 0.9087,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3767965044656067e-05,
+      "loss": 0.8726,
+      "step": 1089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3765397610830604e-05,
+      "loss": 0.7998,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3762827744562063e-05,
+      "loss": 0.8716,
+      "step": 1091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3760255446848155e-05,
+      "loss": 0.8511,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3757680718687538e-05,
+      "loss": 0.8438,
+      "step": 1093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3755103561079806e-05,
+      "loss": 0.9028,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3752523975025503e-05,
+      "loss": 0.8486,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3749941961526112e-05,
+      "loss": 0.8516,
+      "step": 1096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.374735752158406e-05,
+      "loss": 0.8149,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3744770656202713e-05,
+      "loss": 0.8359,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3742181366386385e-05,
+      "loss": 0.8892,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3739589653140327e-05,
+      "loss": 0.9214,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3736995517470729e-05,
+      "loss": 0.8779,
+      "step": 1101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3734398960384726e-05,
+      "loss": 0.9351,
+      "step": 1102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.373179998289039e-05,
+      "loss": 0.8691,
+      "step": 1103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3729198585996734e-05,
+      "loss": 0.8506,
+      "step": 1104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3726594770713707e-05,
+      "loss": 0.9072,
+      "step": 1105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3723988538052204e-05,
+      "loss": 0.8804,
+      "step": 1106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3721379889024053e-05,
+      "loss": 0.895,
+      "step": 1107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3718768824642018e-05,
+      "loss": 0.8687,
+      "step": 1108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.371615534591981e-05,
+      "loss": 0.9336,
+      "step": 1109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3713539453872064e-05,
+      "loss": 0.8857,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3710921149514368e-05,
+      "loss": 0.8325,
+      "step": 1111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.370830043386323e-05,
+      "loss": 0.8652,
+      "step": 1112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3705677307936108e-05,
+      "loss": 0.3555,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3703051772751385e-05,
+      "loss": 0.895,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.370042382932839e-05,
+      "loss": 0.8633,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3697793478687376e-05,
+      "loss": 0.8569,
+      "step": 1116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3695160721849538e-05,
+      "loss": 0.3208,
+      "step": 1117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3692525559836999e-05,
+      "loss": 0.3328,
+      "step": 1118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3689887993672824e-05,
+      "loss": 0.8545,
+      "step": 1119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3687248024381008e-05,
+      "loss": 0.3387,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3684605652986474e-05,
+      "loss": 0.855,
+      "step": 1121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3681960880515083e-05,
+      "loss": 0.9185,
+      "step": 1122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3679313707993628e-05,
+      "loss": 0.8677,
+      "step": 1123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.367666413644983e-05,
+      "loss": 0.7773,
+      "step": 1124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3674012166912348e-05,
+      "loss": 0.8413,
+      "step": 1125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3671357800410763e-05,
+      "loss": 0.8887,
+      "step": 1126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3668701037975593e-05,
+      "loss": 0.8599,
+      "step": 1127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3666041880638288e-05,
+      "loss": 0.8594,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3663380329431222e-05,
+      "loss": 0.8174,
+      "step": 1129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3660716385387699e-05,
+      "loss": 0.8872,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3658050049541958e-05,
+      "loss": 0.4019,
+      "step": 1131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3655381322929157e-05,
+      "loss": 0.8535,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3652710206585393e-05,
+      "loss": 0.3396,
+      "step": 1133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.365003670154768e-05,
+      "loss": 0.9194,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.364736080885397e-05,
+      "loss": 0.7974,
+      "step": 1135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3644682529543132e-05,
+      "loss": 0.8784,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.364200186465497e-05,
+      "loss": 0.8706,
+      "step": 1137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.363931881523021e-05,
+      "loss": 0.8306,
+      "step": 1138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.36366333823105e-05,
+      "loss": 0.834,
+      "step": 1139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3633945566938423e-05,
+      "loss": 0.8931,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3631255370157477e-05,
+      "loss": 0.8779,
+      "step": 1141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.362856279301209e-05,
+      "loss": 0.873,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3625867836547617e-05,
+      "loss": 0.8716,
+      "step": 1143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.362317050181033e-05,
+      "loss": 0.3298,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3620470789847426e-05,
+      "loss": 0.8354,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3617768701707028e-05,
+      "loss": 0.895,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3615064238438178e-05,
+      "loss": 0.8813,
+      "step": 1147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3612357401090842e-05,
+      "loss": 0.8599,
+      "step": 1148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3609648190715913e-05,
+      "loss": 0.8359,
+      "step": 1149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3606936608365189e-05,
+      "loss": 0.3477,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3604222655091408e-05,
+      "loss": 0.8892,
+      "step": 1151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.360150633194822e-05,
+      "loss": 0.8125,
+      "step": 1152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.359878763999019e-05,
+      "loss": 0.3486,
+      "step": 1153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3596066580272813e-05,
+      "loss": 0.9204,
+      "step": 1154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3593343153852495e-05,
+      "loss": 0.9082,
+      "step": 1155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3590617361786566e-05,
+      "loss": 0.8794,
+      "step": 1156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3587889205133268e-05,
+      "loss": 0.8721,
+      "step": 1157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3585158684951768e-05,
+      "loss": 0.9321,
+      "step": 1158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3582425802302151e-05,
+      "loss": 0.8494,
+      "step": 1159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3579690558245413e-05,
+      "loss": 0.8511,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3576952953843471e-05,
+      "loss": 0.8979,
+      "step": 1161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3574212990159154e-05,
+      "loss": 0.313,
+      "step": 1162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3571470668256215e-05,
+      "loss": 0.3096,
+      "step": 1163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3568725989199316e-05,
+      "loss": 0.9004,
+      "step": 1164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3565978954054036e-05,
+      "loss": 0.8716,
+      "step": 1165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3563229563886867e-05,
+      "loss": 0.8794,
+      "step": 1166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3560477819765217e-05,
+      "loss": 0.8853,
+      "step": 1167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3557723722757412e-05,
+      "loss": 0.8286,
+      "step": 1168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.3554967273932684e-05,
+      "loss": 0.8921,
+      "step": 1169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3552208474361177e-05,
+      "loss": 0.3418,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3549447325113962e-05,
+      "loss": 0.8965,
+      "step": 1171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3546683827263006e-05,
+      "loss": 0.9248,
+      "step": 1172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3543917981881193e-05,
+      "loss": 0.8281,
+      "step": 1173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3541149790042321e-05,
+      "loss": 0.8193,
+      "step": 1174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.35383792528211e-05,
+      "loss": 0.8306,
+      "step": 1175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3535606371293143e-05,
+      "loss": 0.3284,
+      "step": 1176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3532831146534982e-05,
+      "loss": 0.8032,
+      "step": 1177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3530053579624054e-05,
+      "loss": 0.8687,
+      "step": 1178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3527273671638703e-05,
+      "loss": 0.8628,
+      "step": 1179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.352449142365819e-05,
+      "loss": 0.8916,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3521706836762677e-05,
+      "loss": 0.8286,
+      "step": 1181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3518919912033235e-05,
+      "loss": 0.8931,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3516130650551847e-05,
+      "loss": 0.9136,
+      "step": 1183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3513339053401399e-05,
+      "loss": 0.8579,
+      "step": 1184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3510545121665685e-05,
+      "loss": 0.3279,
+      "step": 1185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3507748856429405e-05,
+      "loss": 0.833,
+      "step": 1186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3504950258778168e-05,
+      "loss": 0.916,
+      "step": 1187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3502149329798481e-05,
+      "loss": 0.8315,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3499346070577766e-05,
+      "loss": 0.8896,
+      "step": 1189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3496540482204342e-05,
+      "loss": 0.855,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3493732565767438e-05,
+      "loss": 0.894,
+      "step": 1191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3490922322357179e-05,
+      "loss": 0.8682,
+      "step": 1192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.34881097530646e-05,
+      "loss": 0.834,
+      "step": 1193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3485294858981641e-05,
+      "loss": 0.8472,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3482477641201138e-05,
+      "loss": 0.8774,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3479658100816832e-05,
+      "loss": 0.8511,
+      "step": 1196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3476836238923364e-05,
+      "loss": 0.8638,
+      "step": 1197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.347401205661628e-05,
+      "loss": 0.8506,
+      "step": 1198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3471185554992024e-05,
+      "loss": 0.8423,
+      "step": 1199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3468356735147944e-05,
+      "loss": 0.873,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.346552559818228e-05,
+      "loss": 0.8755,
+      "step": 1201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3462692145194182e-05,
+      "loss": 0.8013,
+      "step": 1202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.345985637728369e-05,
+      "loss": 0.334,
+      "step": 1203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.345701829555175e-05,
+      "loss": 0.8838,
+      "step": 1204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3454177901100198e-05,
+      "loss": 0.8418,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.345133519503178e-05,
+      "loss": 0.9102,
+      "step": 1206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3448490178450128e-05,
+      "loss": 0.8838,
+      "step": 1207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3445642852459776e-05,
+      "loss": 0.8911,
+      "step": 1208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3442793218166155e-05,
+      "loss": 0.8662,
+      "step": 1209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.343994127667559e-05,
+      "loss": 0.8857,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3437087029095302e-05,
+      "loss": 0.8701,
+      "step": 1211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.343423047653341e-05,
+      "loss": 0.8496,
+      "step": 1212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3431371620098926e-05,
+      "loss": 0.9302,
+      "step": 1213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3428510460901754e-05,
+      "loss": 0.8428,
+      "step": 1214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3425647000052702e-05,
+      "loss": 0.8174,
+      "step": 1215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3422781238663452e-05,
+      "loss": 0.835,
+      "step": 1216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.34199131778466e-05,
+      "loss": 0.3243,
+      "step": 1217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3417042818715625e-05,
+      "loss": 0.939,
+      "step": 1218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3414170162384897e-05,
+      "loss": 0.8076,
+      "step": 1219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3411295209969682e-05,
+      "loss": 0.8511,
+      "step": 1220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3408417962586134e-05,
+      "loss": 0.8765,
+      "step": 1221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.34055384213513e-05,
+      "loss": 0.8452,
+      "step": 1222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3402656587383118e-05,
+      "loss": 0.8354,
+      "step": 1223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3399772461800415e-05,
+      "loss": 0.8384,
+      "step": 1224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3396886045722905e-05,
+      "loss": 0.8423,
+      "step": 1225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3393997340271196e-05,
+      "loss": 0.3333,
+      "step": 1226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3391106346566786e-05,
+      "loss": 0.8379,
+      "step": 1227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3388213065732054e-05,
+      "loss": 0.3298,
+      "step": 1228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3385317498890273e-05,
+      "loss": 0.835,
+      "step": 1229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.33824196471656e-05,
+      "loss": 0.9277,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3379519511683084e-05,
+      "loss": 0.9185,
+      "step": 1231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3376617093568656e-05,
+      "loss": 0.8418,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3373712393949135e-05,
+      "loss": 0.8813,
+      "step": 1233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3370805413952223e-05,
+      "loss": 0.8652,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3367896154706512e-05,
+      "loss": 0.8735,
+      "step": 1235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3364984617341477e-05,
+      "loss": 0.8579,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3362070802987475e-05,
+      "loss": 0.791,
+      "step": 1237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3359154712775752e-05,
+      "loss": 0.8491,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3356236347838431e-05,
+      "loss": 0.9019,
+      "step": 1239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3353315709308525e-05,
+      "loss": 0.8809,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3350392798319925e-05,
+      "loss": 0.8877,
+      "step": 1241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3347467616007405e-05,
+      "loss": 0.8145,
+      "step": 1242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.334454016350662e-05,
+      "loss": 0.3218,
+      "step": 1243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3341610441954114e-05,
+      "loss": 0.834,
+      "step": 1244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.33386784524873e-05,
+      "loss": 0.8438,
+      "step": 1245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.333574419624448e-05,
+      "loss": 0.8984,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3332807674364833e-05,
+      "loss": 0.8618,
+      "step": 1247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3329868887988417e-05,
+      "loss": 0.8369,
+      "step": 1248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.332692783825617e-05,
+      "loss": 0.8467,
+      "step": 1249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3323984526309908e-05,
+      "loss": 0.897,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3321038953292327e-05,
+      "loss": 0.8384,
+      "step": 1251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3318091120347003e-05,
+      "loss": 0.8691,
+      "step": 1252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3315141028618381e-05,
+      "loss": 0.9229,
+      "step": 1253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3312188679251792e-05,
+      "loss": 0.8267,
+      "step": 1254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3309234073393437e-05,
+      "loss": 0.8521,
+      "step": 1255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3306277212190397e-05,
+      "loss": 0.3115,
+      "step": 1256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3303318096790629e-05,
+      "loss": 0.8477,
+      "step": 1257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3300356728342959e-05,
+      "loss": 0.8701,
+      "step": 1258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3297393107997097e-05,
+      "loss": 0.8418,
+      "step": 1259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3294427236903621e-05,
+      "loss": 0.8486,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3291459116213982e-05,
+      "loss": 0.8706,
+      "step": 1261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3288488747080507e-05,
+      "loss": 0.8906,
+      "step": 1262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.32855161306564e-05,
+      "loss": 0.8853,
+      "step": 1263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3282541268095727e-05,
+      "loss": 0.8564,
+      "step": 1264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3279564160553437e-05,
+      "loss": 0.8691,
+      "step": 1265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3276584809185344e-05,
+      "loss": 0.8511,
+      "step": 1266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3273603215148133e-05,
+      "loss": 0.8765,
+      "step": 1267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3270619379599364e-05,
+      "loss": 0.9326,
+      "step": 1268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3267633303697461e-05,
+      "loss": 0.8652,
+      "step": 1269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3264644988601726e-05,
+      "loss": 0.7993,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.3261654435472326e-05,
+      "loss": 0.9053,
+      "step": 1271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.325866164547029e-05,
+      "loss": 0.3365,
+      "step": 1272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.325566661975753e-05,
+      "loss": 0.8691,
+      "step": 1273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3252669359496815e-05,
+      "loss": 0.835,
+      "step": 1274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3249669865851785e-05,
+      "loss": 0.8457,
+      "step": 1275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3246668139986944e-05,
+      "loss": 0.8779,
+      "step": 1276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3243664183067669e-05,
+      "loss": 0.8325,
+      "step": 1277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3240657996260196e-05,
+      "loss": 0.8174,
+      "step": 1278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3237649580731632e-05,
+      "loss": 0.8628,
+      "step": 1279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3234638937649949e-05,
+      "loss": 0.8525,
+      "step": 1280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.323162606818398e-05,
+      "loss": 0.8408,
+      "step": 1281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.322861097350343e-05,
+      "loss": 0.8218,
+      "step": 1282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3225593654778854e-05,
+      "loss": 0.876,
+      "step": 1283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3222574113181682e-05,
+      "loss": 0.8691,
+      "step": 1284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3219552349884208e-05,
+      "loss": 0.8496,
+      "step": 1285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3216528366059582e-05,
+      "loss": 0.8867,
+      "step": 1286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3213502162881817e-05,
+      "loss": 0.8672,
+      "step": 1287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3210473741525793e-05,
+      "loss": 0.873,
+      "step": 1288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3207443103167243e-05,
+      "loss": 0.8857,
+      "step": 1289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3204410248982767e-05,
+      "loss": 0.3362,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3201375180149824e-05,
+      "loss": 0.8452,
+      "step": 1291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3198337897846731e-05,
+      "loss": 0.8599,
+      "step": 1292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.319529840325267e-05,
+      "loss": 0.8813,
+      "step": 1293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3192256697547674e-05,
+      "loss": 0.9214,
+      "step": 1294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3189212781912634e-05,
+      "loss": 0.3398,
+      "step": 1295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3186166657529308e-05,
+      "loss": 0.8564,
+      "step": 1296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3183118325580309e-05,
+      "loss": 0.9058,
+      "step": 1297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3180067787249095e-05,
+      "loss": 0.9204,
+      "step": 1298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3177015043719997e-05,
+      "loss": 0.3325,
+      "step": 1299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3173960096178194e-05,
+      "loss": 0.8848,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3170902945809725e-05,
+      "loss": 0.8164,
+      "step": 1301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3167843593801472e-05,
+      "loss": 0.344,
+      "step": 1302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.316478204134119e-05,
+      "loss": 0.8755,
+      "step": 1303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3161718289617475e-05,
+      "loss": 0.833,
+      "step": 1304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.315865233981978e-05,
+      "loss": 0.8896,
+      "step": 1305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3155584193138414e-05,
+      "loss": 0.833,
+      "step": 1306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3152513850764536e-05,
+      "loss": 0.8804,
+      "step": 1307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3149441313890161e-05,
+      "loss": 0.8481,
+      "step": 1308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.314636658370815e-05,
+      "loss": 0.3333,
+      "step": 1309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3143289661412224e-05,
+      "loss": 0.9238,
+      "step": 1310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3140210548196946e-05,
+      "loss": 0.8394,
+      "step": 1311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3137129245257734e-05,
+      "loss": 0.8765,
+      "step": 1312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3134045753790859e-05,
+      "loss": 0.853,
+      "step": 1313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3130960074993433e-05,
+      "loss": 0.8311,
+      "step": 1314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3127872210063429e-05,
+      "loss": 0.8574,
+      "step": 1315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3124782160199659e-05,
+      "loss": 0.8682,
+      "step": 1316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3121689926601787e-05,
+      "loss": 0.8506,
+      "step": 1317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3118595510470328e-05,
+      "loss": 0.8906,
+      "step": 1318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3115498913006632e-05,
+      "loss": 0.8633,
+      "step": 1319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3112400135412916e-05,
+      "loss": 0.9102,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3109299178892226e-05,
+      "loss": 0.8662,
+      "step": 1321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.310619604464846e-05,
+      "loss": 0.8398,
+      "step": 1322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3103090733886362e-05,
+      "loss": 0.8359,
+      "step": 1323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3099983247811522e-05,
+      "loss": 0.8662,
+      "step": 1324
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.3096873587630372e-05,
+      "loss": 0.324,
+      "step": 1325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3093761754550188e-05,
+      "loss": 0.8208,
+      "step": 1326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3090647749779092e-05,
+      "loss": 0.8242,
+      "step": 1327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3087531574526049e-05,
+      "loss": 0.8301,
+      "step": 1328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3084413230000865e-05,
+      "loss": 0.9072,
+      "step": 1329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3081292717414187e-05,
+      "loss": 0.8477,
+      "step": 1330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3078170037977507e-05,
+      "loss": 0.8291,
+      "step": 1331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3075045192903155e-05,
+      "loss": 0.9346,
+      "step": 1332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3071918183404307e-05,
+      "loss": 0.8281,
+      "step": 1333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3068789010694973e-05,
+      "loss": 0.8491,
+      "step": 1334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3065657675990005e-05,
+      "loss": 0.8857,
+      "step": 1335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3062524180505096e-05,
+      "loss": 0.856,
+      "step": 1336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3059388525456778e-05,
+      "loss": 0.8828,
+      "step": 1337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3056250712062417e-05,
+      "loss": 0.9028,
+      "step": 1338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3053110741540223e-05,
+      "loss": 0.8613,
+      "step": 1339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3049968615109242e-05,
+      "loss": 0.8608,
+      "step": 1340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.304682433398935e-05,
+      "loss": 0.8418,
+      "step": 1341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3043677899401273e-05,
+      "loss": 0.8975,
+      "step": 1342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3040529312566558e-05,
+      "loss": 0.8511,
+      "step": 1343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3037378574707599e-05,
+      "loss": 0.8389,
+      "step": 1344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3034225687047619e-05,
+      "loss": 0.8677,
+      "step": 1345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.303107065081068e-05,
+      "loss": 0.8306,
+      "step": 1346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3027913467221671e-05,
+      "loss": 0.897,
+      "step": 1347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3024754137506324e-05,
+      "loss": 0.8662,
+      "step": 1348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3021592662891199e-05,
+      "loss": 0.8467,
+      "step": 1349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3018429044603688e-05,
+      "loss": 0.8237,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3015263283872014e-05,
+      "loss": 0.8843,
+      "step": 1351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3012095381925242e-05,
+      "loss": 0.8452,
+      "step": 1352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3008925339993252e-05,
+      "loss": 0.3137,
+      "step": 1353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3005753159306768e-05,
+      "loss": 0.834,
+      "step": 1354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3002578841097338e-05,
+      "loss": 0.8628,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2999402386597343e-05,
+      "loss": 0.9233,
+      "step": 1356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2996223797039993e-05,
+      "loss": 0.9404,
+      "step": 1357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2993043073659323e-05,
+      "loss": 0.9141,
+      "step": 1358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2989860217690201e-05,
+      "loss": 0.8701,
+      "step": 1359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2986675230368324e-05,
+      "loss": 0.8853,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2983488112930208e-05,
+      "loss": 0.8765,
+      "step": 1361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2980298866613204e-05,
+      "loss": 0.8276,
+      "step": 1362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2977107492655488e-05,
+      "loss": 0.873,
+      "step": 1363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2973913992296062e-05,
+      "loss": 0.8716,
+      "step": 1364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2970718366774751e-05,
+      "loss": 0.833,
+      "step": 1365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2967520617332209e-05,
+      "loss": 0.8711,
+      "step": 1366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.296432074520991e-05,
+      "loss": 0.8892,
+      "step": 1367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2961118751650156e-05,
+      "loss": 0.8159,
+      "step": 1368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.295791463789607e-05,
+      "loss": 0.9028,
+      "step": 1369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2954708405191602e-05,
+      "loss": 0.8462,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2951500054781517e-05,
+      "loss": 0.8569,
+      "step": 1371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2948289587911412e-05,
+      "loss": 0.896,
+      "step": 1372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2945077005827698e-05,
+      "loss": 0.8813,
+      "step": 1373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2941862309777613e-05,
+      "loss": 0.8145,
+      "step": 1374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2938645501009209e-05,
+      "loss": 0.3428,
+      "step": 1375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2935426580771363e-05,
+      "loss": 0.9121,
+      "step": 1376
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2932205550313772e-05,
+      "loss": 0.3359,
+      "step": 1377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.292898241088695e-05,
+      "loss": 0.8579,
+      "step": 1378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.292575716374223e-05,
+      "loss": 0.8579,
+      "step": 1379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2922529810131763e-05,
+      "loss": 0.9004,
+      "step": 1380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2919300351308524e-05,
+      "loss": 0.9097,
+      "step": 1381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2916068788526295e-05,
+      "loss": 0.875,
+      "step": 1382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2912835123039683e-05,
+      "loss": 0.7979,
+      "step": 1383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2909599356104104e-05,
+      "loss": 0.8428,
+      "step": 1384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2906361488975799e-05,
+      "loss": 0.8916,
+      "step": 1385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2903121522911818e-05,
+      "loss": 0.834,
+      "step": 1386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2899879459170026e-05,
+      "loss": 0.8569,
+      "step": 1387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2896635299009105e-05,
+      "loss": 0.8667,
+      "step": 1388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.289338904368855e-05,
+      "loss": 0.8638,
+      "step": 1389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2890140694468666e-05,
+      "loss": 0.8013,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2886890252610578e-05,
+      "loss": 0.3325,
+      "step": 1391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2883637719376218e-05,
+      "loss": 0.8418,
+      "step": 1392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2880383096028332e-05,
+      "loss": 0.8154,
+      "step": 1393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2877126383830475e-05,
+      "loss": 0.8711,
+      "step": 1394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2873867584047015e-05,
+      "loss": 0.3616,
+      "step": 1395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2870606697943132e-05,
+      "loss": 0.8516,
+      "step": 1396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2867343726784813e-05,
+      "loss": 0.9185,
+      "step": 1397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2864078671838859e-05,
+      "loss": 0.3125,
+      "step": 1398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2860811534372872e-05,
+      "loss": 0.854,
+      "step": 1399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.285754231565527e-05,
+      "loss": 0.8613,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2854271016955273e-05,
+      "loss": 0.8394,
+      "step": 1401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2850997639542918e-05,
+      "loss": 0.8589,
+      "step": 1402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2847722184689035e-05,
+      "loss": 0.8535,
+      "step": 1403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2844444653665274e-05,
+      "loss": 0.8105,
+      "step": 1404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2841165047744084e-05,
+      "loss": 0.8613,
+      "step": 1405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2837883368198722e-05,
+      "loss": 0.8433,
+      "step": 1406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2834599616303246e-05,
+      "loss": 0.8428,
+      "step": 1407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2831313793332525e-05,
+      "loss": 0.8457,
+      "step": 1408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2828025900562224e-05,
+      "loss": 0.8472,
+      "step": 1409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2824735939268823e-05,
+      "loss": 0.8628,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2821443910729591e-05,
+      "loss": 0.845,
+      "step": 1411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2818149816222613e-05,
+      "loss": 0.8726,
+      "step": 1412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2814853657026765e-05,
+      "loss": 0.8179,
+      "step": 1413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2811555434421731e-05,
+      "loss": 0.8696,
+      "step": 1414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2808255149687997e-05,
+      "loss": 0.3389,
+      "step": 1415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2804952804106846e-05,
+      "loss": 0.8809,
+      "step": 1416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.280164839896036e-05,
+      "loss": 0.8584,
+      "step": 1417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2798341935531422e-05,
+      "loss": 0.8086,
+      "step": 1418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2795033415103721e-05,
+      "loss": 0.8525,
+      "step": 1419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2791722838961733e-05,
+      "loss": 0.8511,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.278841020839074e-05,
+      "loss": 0.834,
+      "step": 1421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2785095524676821e-05,
+      "loss": 0.2988,
+      "step": 1422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2781778789106847e-05,
+      "loss": 0.8291,
+      "step": 1423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2778460002968488e-05,
+      "loss": 0.8418,
+      "step": 1424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2775139167550216e-05,
+      "loss": 0.8452,
+      "step": 1425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.277181628414129e-05,
+      "loss": 0.3479,
+      "step": 1426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2768491354031766e-05,
+      "loss": 0.349,
+      "step": 1427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2765164378512501e-05,
+      "loss": 0.8608,
+      "step": 1428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.276183535887514e-05,
+      "loss": 0.8496,
+      "step": 1429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2758504296412122e-05,
+      "loss": 0.8735,
+      "step": 1430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.275517119241668e-05,
+      "loss": 0.8633,
+      "step": 1431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2751836048182843e-05,
+      "loss": 0.9263,
+      "step": 1432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.274849886500542e-05,
+      "loss": 0.8755,
+      "step": 1433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2745159644180031e-05,
+      "loss": 0.8618,
+      "step": 1434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2741818387003072e-05,
+      "loss": 0.8257,
+      "step": 1435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2738475094771733e-05,
+      "loss": 0.8193,
+      "step": 1436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2735129768783995e-05,
+      "loss": 0.8608,
+      "step": 1437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2731782410338631e-05,
+      "loss": 0.9424,
+      "step": 1438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.27284330207352e-05,
+      "loss": 0.3435,
+      "step": 1439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.272508160127405e-05,
+      "loss": 0.8901,
+      "step": 1440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.272172815325632e-05,
+      "loss": 0.3669,
+      "step": 1441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2718372677983928e-05,
+      "loss": 0.8691,
+      "step": 1442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2715015176759588e-05,
+      "loss": 0.9116,
+      "step": 1443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2711655650886802e-05,
+      "loss": 0.8242,
+      "step": 1444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2708294101669848e-05,
+      "loss": 0.8818,
+      "step": 1445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2704930530413796e-05,
+      "loss": 0.9292,
+      "step": 1446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2701564938424501e-05,
+      "loss": 0.8701,
+      "step": 1447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2698197327008604e-05,
+      "loss": 0.8984,
+      "step": 1448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2694827697473525e-05,
+      "loss": 0.8892,
+      "step": 1449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2691456051127466e-05,
+      "loss": 0.8765,
+      "step": 1450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2688082389279426e-05,
+      "loss": 0.876,
+      "step": 1451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.268470671323917e-05,
+      "loss": 0.8896,
+      "step": 1452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2681329024317255e-05,
+      "loss": 0.8608,
+      "step": 1453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.267794932382501e-05,
+      "loss": 0.8838,
+      "step": 1454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.267456761307456e-05,
+      "loss": 0.8062,
+      "step": 1455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2671183893378795e-05,
+      "loss": 0.9175,
+      "step": 1456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2667798166051392e-05,
+      "loss": 0.8804,
+      "step": 1457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2664410432406812e-05,
+      "loss": 0.8345,
+      "step": 1458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2661020693760283e-05,
+      "loss": 0.7759,
+      "step": 1459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2657628951427824e-05,
+      "loss": 0.8691,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2654235206726221e-05,
+      "loss": 0.8276,
+      "step": 1461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2650839460973049e-05,
+      "loss": 0.79,
+      "step": 1462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2647441715486648e-05,
+      "loss": 0.876,
+      "step": 1463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.264404197158614e-05,
+      "loss": 0.8418,
+      "step": 1464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2640640230591427e-05,
+      "loss": 0.8516,
+      "step": 1465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2637236493823176e-05,
+      "loss": 0.9209,
+      "step": 1466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2633830762602839e-05,
+      "loss": 0.8882,
+      "step": 1467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2630423038252636e-05,
+      "loss": 0.328,
+      "step": 1468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2627013322095566e-05,
+      "loss": 0.8418,
+      "step": 1469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2623601615455393e-05,
+      "loss": 0.3406,
+      "step": 1470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2620187919656665e-05,
+      "loss": 0.8213,
+      "step": 1471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2616772236024693e-05,
+      "loss": 0.9312,
+      "step": 1472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2613354565885562e-05,
+      "loss": 0.8223,
+      "step": 1473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2609934910566131e-05,
+      "loss": 0.8174,
+      "step": 1474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2606513271394028e-05,
+      "loss": 0.8247,
+      "step": 1475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.260308964969765e-05,
+      "loss": 0.8711,
+      "step": 1476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2599664046806167e-05,
+      "loss": 0.8525,
+      "step": 1477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2596236464049512e-05,
+      "loss": 0.8823,
+      "step": 1478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2592806902758397e-05,
+      "loss": 0.8579,
+      "step": 1479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2589375364264288e-05,
+      "loss": 0.3452,
+      "step": 1480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2585941849899432e-05,
+      "loss": 0.8633,
+      "step": 1481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2582506360996835e-05,
+      "loss": 0.8589,
+      "step": 1482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2579068898890273e-05,
+      "loss": 0.853,
+      "step": 1483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2575629464914287e-05,
+      "loss": 0.8657,
+      "step": 1484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2572188060404185e-05,
+      "loss": 0.8643,
+      "step": 1485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2568744686696037e-05,
+      "loss": 0.8232,
+      "step": 1486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2565299345126678e-05,
+      "loss": 0.8555,
+      "step": 1487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.256185203703371e-05,
+      "loss": 0.8164,
+      "step": 1488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2558402763755497e-05,
+      "loss": 0.8306,
+      "step": 1489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2554951526631166e-05,
+      "loss": 0.8877,
+      "step": 1490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2551498327000604e-05,
+      "loss": 0.8374,
+      "step": 1491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2548043166204465e-05,
+      "loss": 0.8765,
+      "step": 1492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2544586045584159e-05,
+      "loss": 0.8989,
+      "step": 1493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2541126966481859e-05,
+      "loss": 0.8696,
+      "step": 1494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.25376659302405e-05,
+      "loss": 0.8115,
+      "step": 1495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2534202938203775e-05,
+      "loss": 0.8931,
+      "step": 1496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2530737991716137e-05,
+      "loss": 0.8511,
+      "step": 1497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.25272710921228e-05,
+      "loss": 0.8496,
+      "step": 1498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2523802240769724e-05,
+      "loss": 0.3376,
+      "step": 1499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2520331439003649e-05,
+      "loss": 0.811,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2516858688172054e-05,
+      "loss": 0.8672,
+      "step": 1501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.251338398962318e-05,
+      "loss": 0.8438,
+      "step": 1502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2509907344706024e-05,
+      "loss": 0.8379,
+      "step": 1503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2506428754770342e-05,
+      "loss": 0.8477,
+      "step": 1504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2502948221166642e-05,
+      "loss": 0.8613,
+      "step": 1505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2499465745246183e-05,
+      "loss": 0.8853,
+      "step": 1506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2495981328360989e-05,
+      "loss": 0.7725,
+      "step": 1507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2492494971863823e-05,
+      "loss": 0.894,
+      "step": 1508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2489006677108214e-05,
+      "loss": 0.8384,
+      "step": 1509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2485516445448435e-05,
+      "loss": 0.7764,
+      "step": 1510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2482024278239517e-05,
+      "loss": 0.8623,
+      "step": 1511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2478530176837238e-05,
+      "loss": 0.8394,
+      "step": 1512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2475034142598127e-05,
+      "loss": 0.8125,
+      "step": 1513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2471536176879467e-05,
+      "loss": 0.8652,
+      "step": 1514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.246803628103929e-05,
+      "loss": 0.8823,
+      "step": 1515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2464534456436374e-05,
+      "loss": 0.9004,
+      "step": 1516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2461030704430246e-05,
+      "loss": 0.8784,
+      "step": 1517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2457525026381186e-05,
+      "loss": 0.9194,
+      "step": 1518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2454017423650218e-05,
+      "loss": 0.8867,
+      "step": 1519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2450507897599116e-05,
+      "loss": 0.8501,
+      "step": 1520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2446996449590396e-05,
+      "loss": 0.9009,
+      "step": 1521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2443483080987324e-05,
+      "loss": 0.8936,
+      "step": 1522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.243996779315391e-05,
+      "loss": 0.8809,
+      "step": 1523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2436450587454913e-05,
+      "loss": 0.8848,
+      "step": 1524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2432931465255829e-05,
+      "loss": 0.8462,
+      "step": 1525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2429410427922904e-05,
+      "loss": 0.8164,
+      "step": 1526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2425887476823129e-05,
+      "loss": 0.8589,
+      "step": 1527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2422362613324229e-05,
+      "loss": 0.8555,
+      "step": 1528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2418835838794682e-05,
+      "loss": 0.9019,
+      "step": 1529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2415307154603703e-05,
+      "loss": 0.8027,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2411776562121247e-05,
+      "loss": 0.8608,
+      "step": 1531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2408244062718012e-05,
+      "loss": 0.8174,
+      "step": 1532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2404709657765435e-05,
+      "loss": 0.8823,
+      "step": 1533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2401173348635697e-05,
+      "loss": 0.8672,
+      "step": 1534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.239763513670171e-05,
+      "loss": 0.3347,
+      "step": 1535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2394095023337133e-05,
+      "loss": 0.8921,
+      "step": 1536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2390553009916361e-05,
+      "loss": 0.8979,
+      "step": 1537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2387009097814523e-05,
+      "loss": 0.8555,
+      "step": 1538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2383463288407488e-05,
+      "loss": 0.8438,
+      "step": 1539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.237991558307186e-05,
+      "loss": 0.812,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2376365983184987e-05,
+      "loss": 0.333,
+      "step": 1541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2372814490124938e-05,
+      "loss": 0.8521,
+      "step": 1542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2369261105270529e-05,
+      "loss": 0.876,
+      "step": 1543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2365705830001305e-05,
+      "loss": 0.3228,
+      "step": 1544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2362148665697546e-05,
+      "loss": 0.8311,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2358589613740269e-05,
+      "loss": 0.896,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2355028675511214e-05,
+      "loss": 0.8823,
+      "step": 1547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2351465852392866e-05,
+      "loss": 0.8242,
+      "step": 1548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2347901145768433e-05,
+      "loss": 0.8833,
+      "step": 1549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2344334557021856e-05,
+      "loss": 0.8325,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.234076608753781e-05,
+      "loss": 0.8076,
+      "step": 1551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2337195738701696e-05,
+      "loss": 0.8784,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2333623511899645e-05,
+      "loss": 0.8066,
+      "step": 1553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2330049408518523e-05,
+      "loss": 0.8545,
+      "step": 1554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2326473429945915e-05,
+      "loss": 0.8726,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2322895577570142e-05,
+      "loss": 0.8994,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2319315852780249e-05,
+      "loss": 0.8438,
+      "step": 1557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2315734256966007e-05,
+      "loss": 0.8394,
+      "step": 1558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2312150791517917e-05,
+      "loss": 0.7974,
+      "step": 1559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2308565457827202e-05,
+      "loss": 0.8726,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2304978257285812e-05,
+      "loss": 0.8662,
+      "step": 1561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2301389191286423e-05,
+      "loss": 0.8716,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2297798261222434e-05,
+      "loss": 0.8726,
+      "step": 1563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2294205468487965e-05,
+      "loss": 0.8706,
+      "step": 1564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2290610814477866e-05,
+      "loss": 0.8818,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.22870143005877e-05,
+      "loss": 0.8926,
+      "step": 1566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2283415928213763e-05,
+      "loss": 0.9053,
+      "step": 1567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2279815698753067e-05,
+      "loss": 0.8804,
+      "step": 1568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.227621361360334e-05,
+      "loss": 0.8149,
+      "step": 1569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2272609674163038e-05,
+      "loss": 0.9092,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2269003881831335e-05,
+      "loss": 0.7664,
+      "step": 1571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2265396238008125e-05,
+      "loss": 0.8921,
+      "step": 1572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2261786744094018e-05,
+      "loss": 0.8916,
+      "step": 1573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2258175401490341e-05,
+      "loss": 0.8237,
+      "step": 1574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2254562211599146e-05,
+      "loss": 0.8281,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2250947175823198e-05,
+      "loss": 0.8379,
+      "step": 1576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2247330295565973e-05,
+      "loss": 0.8809,
+      "step": 1577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2243711572231676e-05,
+      "loss": 0.8867,
+      "step": 1578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2240091007225212e-05,
+      "loss": 0.8203,
+      "step": 1579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2236468601952214e-05,
+      "loss": 0.8677,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2232844357819024e-05,
+      "loss": 0.9072,
+      "step": 1581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2229218276232698e-05,
+      "loss": 0.8613,
+      "step": 1582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2225590358601005e-05,
+      "loss": 0.8979,
+      "step": 1583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2221960606332426e-05,
+      "loss": 0.8418,
+      "step": 1584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2218329020836159e-05,
+      "loss": 0.3652,
+      "step": 1585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2214695603522111e-05,
+      "loss": 0.915,
+      "step": 1586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2211060355800897e-05,
+      "loss": 0.8594,
+      "step": 1587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2207423279083847e-05,
+      "loss": 0.333,
+      "step": 1588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2203784374782997e-05,
+      "loss": 0.9004,
+      "step": 1589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2200143644311098e-05,
+      "loss": 0.3472,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2196501089081605e-05,
+      "loss": 0.8218,
+      "step": 1591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2192856710508684e-05,
+      "loss": 0.8633,
+      "step": 1592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2189210510007207e-05,
+      "loss": 0.8872,
+      "step": 1593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2185562488992757e-05,
+      "loss": 0.8467,
+      "step": 1594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2181912648881623e-05,
+      "loss": 0.873,
+      "step": 1595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.217826099109079e-05,
+      "loss": 0.8467,
+      "step": 1596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2174607517037969e-05,
+      "loss": 0.9663,
+      "step": 1597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2170952228141553e-05,
+      "loss": 0.8682,
+      "step": 1598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.216729512582066e-05,
+      "loss": 0.3213,
+      "step": 1599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2163636211495096e-05,
+      "loss": 0.877,
+      "step": 1600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2159975486585385e-05,
+      "loss": 0.8828,
+      "step": 1601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.215631295251274e-05,
+      "loss": 0.8369,
+      "step": 1602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2152648610699085e-05,
+      "loss": 0.8628,
+      "step": 1603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2148982462567045e-05,
+      "loss": 0.8071,
+      "step": 1604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2145314509539944e-05,
+      "loss": 0.8213,
+      "step": 1605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2141644753041809e-05,
+      "loss": 0.8174,
+      "step": 1606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2137973194497361e-05,
+      "loss": 0.7876,
+      "step": 1607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2134299835332033e-05,
+      "loss": 0.8408,
+      "step": 1608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.213062467697194e-05,
+      "loss": 0.833,
+      "step": 1609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2126947720843914e-05,
+      "loss": 0.8179,
+      "step": 1610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2123268968375472e-05,
+      "loss": 0.8696,
+      "step": 1611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2119588420994832e-05,
+      "loss": 0.9097,
+      "step": 1612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.211590608013091e-05,
+      "loss": 0.8818,
+      "step": 1613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2112221947213316e-05,
+      "loss": 0.8281,
+      "step": 1614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2108536023672361e-05,
+      "loss": 0.8843,
+      "step": 1615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2104848310939043e-05,
+      "loss": 0.853,
+      "step": 1616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2101158810445063e-05,
+      "loss": 0.9092,
+      "step": 1617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2097467523622809e-05,
+      "loss": 0.8872,
+      "step": 1618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2093774451905367e-05,
+      "loss": 0.8652,
+      "step": 1619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2090079596726514e-05,
+      "loss": 0.8472,
+      "step": 1620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2086382959520723e-05,
+      "loss": 0.7798,
+      "step": 1621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2082684541723152e-05,
+      "loss": 0.8081,
+      "step": 1622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2078984344769654e-05,
+      "loss": 0.8359,
+      "step": 1623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2075282370096776e-05,
+      "loss": 0.8848,
+      "step": 1624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.207157861914175e-05,
+      "loss": 0.854,
+      "step": 1625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2067873093342498e-05,
+      "loss": 0.9028,
+      "step": 1626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2064165794137638e-05,
+      "loss": 0.8452,
+      "step": 1627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2060456722966467e-05,
+      "loss": 0.8496,
+      "step": 1628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2056745881268973e-05,
+      "loss": 0.8394,
+      "step": 1629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2053033270485835e-05,
+      "loss": 0.8315,
+      "step": 1630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.204931889205842e-05,
+      "loss": 0.8047,
+      "step": 1631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2045602747428768e-05,
+      "loss": 0.8975,
+      "step": 1632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2041884838039623e-05,
+      "loss": 0.8286,
+      "step": 1633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.20381651653344e-05,
+      "loss": 0.8662,
+      "step": 1634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2034443730757206e-05,
+      "loss": 0.8252,
+      "step": 1635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2030720535752831e-05,
+      "loss": 0.9341,
+      "step": 1636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2026995581766748e-05,
+      "loss": 0.8599,
+      "step": 1637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.202326887024511e-05,
+      "loss": 0.8887,
+      "step": 1638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2019540402634757e-05,
+      "loss": 0.8916,
+      "step": 1639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2015810180383208e-05,
+      "loss": 0.9155,
+      "step": 1640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.201207820493866e-05,
+      "loss": 0.8408,
+      "step": 1641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2008344477749998e-05,
+      "loss": 0.3198,
+      "step": 1642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2004609000266786e-05,
+      "loss": 0.8101,
+      "step": 1643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.200087177393926e-05,
+      "loss": 0.8013,
+      "step": 1644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.199713280021834e-05,
+      "loss": 0.9258,
+      "step": 1645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.199339208055563e-05,
+      "loss": 0.8467,
+      "step": 1646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.19896496164034e-05,
+      "loss": 0.8291,
+      "step": 1647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1985905409214605e-05,
+      "loss": 0.8433,
+      "step": 1648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1982159460442876e-05,
+      "loss": 0.8872,
+      "step": 1649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.197841177154252e-05,
+      "loss": 0.8735,
+      "step": 1650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1974662343968518e-05,
+      "loss": 0.8276,
+      "step": 1651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1970911179176528e-05,
+      "loss": 0.8652,
+      "step": 1652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.196715827862288e-05,
+      "loss": 0.8691,
+      "step": 1653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1963403643764577e-05,
+      "loss": 0.8472,
+      "step": 1654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1959647276059301e-05,
+      "loss": 0.8418,
+      "step": 1655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1955889176965403e-05,
+      "loss": 0.8179,
+      "step": 1656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1952129347941907e-05,
+      "loss": 0.8716,
+      "step": 1657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1948367790448506e-05,
+      "loss": 0.8604,
+      "step": 1658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1944604505945566e-05,
+      "loss": 0.812,
+      "step": 1659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1940839495894125e-05,
+      "loss": 0.897,
+      "step": 1660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1937072761755889e-05,
+      "loss": 0.8423,
+      "step": 1661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1933304304993234e-05,
+      "loss": 0.8813,
+      "step": 1662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1929534127069206e-05,
+      "loss": 0.8193,
+      "step": 1663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1925762229447515e-05,
+      "loss": 0.8022,
+      "step": 1664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1921988613592543e-05,
+      "loss": 0.8789,
+      "step": 1665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1918213280969339e-05,
+      "loss": 0.8418,
+      "step": 1666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1914436233043615e-05,
+      "loss": 0.3501,
+      "step": 1667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.191065747128175e-05,
+      "loss": 0.8442,
+      "step": 1668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1906876997150793e-05,
+      "loss": 0.8418,
+      "step": 1669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1903094812118452e-05,
+      "loss": 0.8389,
+      "step": 1670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1899310917653102e-05,
+      "loss": 0.897,
+      "step": 1671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1895525315223784e-05,
+      "loss": 0.8022,
+      "step": 1672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1891738006300193e-05,
+      "loss": 0.8457,
+      "step": 1673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1887948992352698e-05,
+      "loss": 0.8726,
+      "step": 1674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1884158274852319e-05,
+      "loss": 0.8701,
+      "step": 1675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1880365855270749e-05,
+      "loss": 0.8379,
+      "step": 1676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1876571735080333e-05,
+      "loss": 0.8921,
+      "step": 1677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1872775915754083e-05,
+      "loss": 0.8672,
+      "step": 1678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.186897839876566e-05,
+      "loss": 0.9092,
+      "step": 1679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1865179185589395e-05,
+      "loss": 0.8647,
+      "step": 1680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1861378277700273e-05,
+      "loss": 0.8555,
+      "step": 1681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1857575676573936e-05,
+      "loss": 0.8423,
+      "step": 1682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1853771383686686e-05,
+      "loss": 0.3403,
+      "step": 1683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.184996540051548e-05,
+      "loss": 0.8745,
+      "step": 1684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1846157728537931e-05,
+      "loss": 0.8081,
+      "step": 1685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1842348369232314e-05,
+      "loss": 0.9111,
+      "step": 1686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1838537324077545e-05,
+      "loss": 0.8418,
+      "step": 1687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1834724594553204e-05,
+      "loss": 0.8433,
+      "step": 1688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.183091018213953e-05,
+      "loss": 0.8198,
+      "step": 1689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1827094088317404e-05,
+      "loss": 0.3376,
+      "step": 1690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1823276314568367e-05,
+      "loss": 0.9038,
+      "step": 1691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.181945686237461e-05,
+      "loss": 0.8794,
+      "step": 1692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1815635733218977e-05,
+      "loss": 0.8638,
+      "step": 1693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.181181292858496e-05,
+      "loss": 0.8682,
+      "step": 1694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1807988449956704e-05,
+      "loss": 0.8232,
+      "step": 1695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1804162298819003e-05,
+      "loss": 0.8584,
+      "step": 1696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.18003344766573e-05,
+      "loss": 0.8896,
+      "step": 1697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.179650498495769e-05,
+      "loss": 0.8677,
+      "step": 1698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1792673825206912e-05,
+      "loss": 0.7808,
+      "step": 1699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1788840998892354e-05,
+      "loss": 0.9199,
+      "step": 1700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1785006507502048e-05,
+      "loss": 0.8706,
+      "step": 1701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1781170352524682e-05,
+      "loss": 0.8188,
+      "step": 1702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1777332535449578e-05,
+      "loss": 0.3328,
+      "step": 1703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1773493057766713e-05,
+      "loss": 0.3469,
+      "step": 1704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1769651920966702e-05,
+      "loss": 0.8286,
+      "step": 1705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1765809126540805e-05,
+      "loss": 0.8867,
+      "step": 1706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.176196467598093e-05,
+      "loss": 0.8232,
+      "step": 1707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1758118570779626e-05,
+      "loss": 0.8325,
+      "step": 1708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1754270812430079e-05,
+      "loss": 0.8491,
+      "step": 1709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1750421402426123e-05,
+      "loss": 0.8027,
+      "step": 1710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1746570342262234e-05,
+      "loss": 0.7993,
+      "step": 1711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1742717633433523e-05,
+      "loss": 0.8193,
+      "step": 1712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.173886327743575e-05,
+      "loss": 0.8008,
+      "step": 1713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.17350072757653e-05,
+      "loss": 0.855,
+      "step": 1714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.173114962991921e-05,
+      "loss": 0.874,
+      "step": 1715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1727290341395154e-05,
+      "loss": 0.8604,
+      "step": 1716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1723429411691432e-05,
+      "loss": 0.8574,
+      "step": 1717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1719566842307e-05,
+      "loss": 0.8438,
+      "step": 1718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1715702634741436e-05,
+      "loss": 0.8843,
+      "step": 1719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1711836790494957e-05,
+      "loss": 0.8516,
+      "step": 1720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.170796931106842e-05,
+      "loss": 0.8623,
+      "step": 1721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.170410019796331e-05,
+      "loss": 0.8652,
+      "step": 1722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1700229452681752e-05,
+      "loss": 0.8638,
+      "step": 1723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1696357076726502e-05,
+      "loss": 0.8501,
+      "step": 1724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.169248307160095e-05,
+      "loss": 0.8315,
+      "step": 1725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1688607438809118e-05,
+      "loss": 0.8179,
+      "step": 1726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1684730179855661e-05,
+      "loss": 0.897,
+      "step": 1727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1680851296245865e-05,
+      "loss": 0.8389,
+      "step": 1728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1676970789485642e-05,
+      "loss": 0.8442,
+      "step": 1729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1673088661081542e-05,
+      "loss": 0.8857,
+      "step": 1730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1669204912540742e-05,
+      "loss": 0.8525,
+      "step": 1731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1665319545371044e-05,
+      "loss": 0.894,
+      "step": 1732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.166143256108088e-05,
+      "loss": 0.8135,
+      "step": 1733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1657543961179316e-05,
+      "loss": 0.8335,
+      "step": 1734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1653653747176038e-05,
+      "loss": 0.8179,
+      "step": 1735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.164976192058136e-05,
+      "loss": 0.7842,
+      "step": 1736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1645868482906225e-05,
+      "loss": 0.8525,
+      "step": 1737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1641973435662198e-05,
+      "loss": 0.8433,
+      "step": 1738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.163807678036147e-05,
+      "loss": 0.7979,
+      "step": 1739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1634178518516858e-05,
+      "loss": 0.8647,
+      "step": 1740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1630278651641802e-05,
+      "loss": 0.8926,
+      "step": 1741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1626377181250361e-05,
+      "loss": 0.8354,
+      "step": 1742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1622474108857227e-05,
+      "loss": 0.8589,
+      "step": 1743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.16185694359777e-05,
+      "loss": 0.8682,
+      "step": 1744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1614663164127712e-05,
+      "loss": 0.8921,
+      "step": 1745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1610755294823811e-05,
+      "loss": 0.8789,
+      "step": 1746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.160684582958317e-05,
+      "loss": 0.8413,
+      "step": 1747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1602934769923574e-05,
+      "loss": 0.8423,
+      "step": 1748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1599022117363432e-05,
+      "loss": 0.7925,
+      "step": 1749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1595107873421768e-05,
+      "loss": 0.8228,
+      "step": 1750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1591192039618233e-05,
+      "loss": 0.8496,
+      "step": 1751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1587274617473084e-05,
+      "loss": 0.8203,
+      "step": 1752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1583355608507194e-05,
+      "loss": 0.8833,
+      "step": 1753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1579435014242068e-05,
+      "loss": 0.834,
+      "step": 1754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1575512836199812e-05,
+      "loss": 0.3374,
+      "step": 1755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1571589075903144e-05,
+      "loss": 0.3267,
+      "step": 1756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.156766373487541e-05,
+      "loss": 0.8853,
+      "step": 1757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.156373681464056e-05,
+      "loss": 0.8398,
+      "step": 1758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1559808316723159e-05,
+      "loss": 0.876,
+      "step": 1759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1555878242648389e-05,
+      "loss": 0.3237,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1551946593942035e-05,
+      "loss": 0.8433,
+      "step": 1761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1548013372130502e-05,
+      "loss": 0.8149,
+      "step": 1762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1544078578740799e-05,
+      "loss": 0.8438,
+      "step": 1763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.154014221530055e-05,
+      "loss": 0.8442,
+      "step": 1764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1536204283337988e-05,
+      "loss": 0.8706,
+      "step": 1765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1532264784381953e-05,
+      "loss": 0.8564,
+      "step": 1766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1528323719961894e-05,
+      "loss": 0.8232,
+      "step": 1767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1524381091607868e-05,
+      "loss": 0.894,
+      "step": 1768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1520436900850535e-05,
+      "loss": 0.8608,
+      "step": 1769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1516491149221172e-05,
+      "loss": 0.8423,
+      "step": 1770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1512543838251653e-05,
+      "loss": 0.8735,
+      "step": 1771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1508594969474459e-05,
+      "loss": 0.8652,
+      "step": 1772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.150464454442268e-05,
+      "loss": 0.8545,
+      "step": 1773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1500692564630004e-05,
+      "loss": 0.8452,
+      "step": 1774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1496739031630729e-05,
+      "loss": 0.8706,
+      "step": 1775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1492783946959745e-05,
+      "loss": 0.8252,
+      "step": 1776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1488827312152559e-05,
+      "loss": 0.8794,
+      "step": 1777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1484869128745274e-05,
+      "loss": 0.8218,
+      "step": 1778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1480909398274585e-05,
+      "loss": 0.8325,
+      "step": 1779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1476948122277805e-05,
+      "loss": 0.8394,
+      "step": 1780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1472985302292834e-05,
+      "loss": 0.8389,
+      "step": 1781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1469020939858173e-05,
+      "loss": 0.3376,
+      "step": 1782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1465055036512924e-05,
+      "loss": 0.8672,
+      "step": 1783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1461087593796793e-05,
+      "loss": 0.812,
+      "step": 1784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1457118613250072e-05,
+      "loss": 0.8223,
+      "step": 1785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1453148096413657e-05,
+      "loss": 0.8765,
+      "step": 1786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1449176044829039e-05,
+      "loss": 0.8096,
+      "step": 1787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.144520246003831e-05,
+      "loss": 0.8794,
+      "step": 1788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1441227343584145e-05,
+      "loss": 0.8594,
+      "step": 1789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1437250697009826e-05,
+      "loss": 0.8442,
+      "step": 1790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1433272521859224e-05,
+      "loss": 0.8462,
+      "step": 1791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1429292819676803e-05,
+      "loss": 0.8735,
+      "step": 1792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.1425311592007618e-05,
+      "loss": 0.811,
+      "step": 1793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.142132884039732e-05,
+      "loss": 0.9175,
+      "step": 1794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1417344566392152e-05,
+      "loss": 0.813,
+      "step": 1795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1413358771538947e-05,
+      "loss": 0.8657,
+      "step": 1796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1409371457385124e-05,
+      "loss": 0.7993,
+      "step": 1797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1405382625478698e-05,
+      "loss": 0.8975,
+      "step": 1798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.140139227736827e-05,
+      "loss": 0.8691,
+      "step": 1799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1397400414603033e-05,
+      "loss": 0.8257,
+      "step": 1800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1393407038732761e-05,
+      "loss": 0.8374,
+      "step": 1801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1389412151307822e-05,
+      "loss": 0.8618,
+      "step": 1802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1385415753879173e-05,
+      "loss": 0.7832,
+      "step": 1803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1381417847998347e-05,
+      "loss": 0.8848,
+      "step": 1804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1377418435217471e-05,
+      "loss": 0.7603,
+      "step": 1805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1373417517089259e-05,
+      "loss": 0.8271,
+      "step": 1806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1369415095166997e-05,
+      "loss": 0.8394,
+      "step": 1807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1365411171004568e-05,
+      "loss": 0.8613,
+      "step": 1808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.136140574615643e-05,
+      "loss": 0.875,
+      "step": 1809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1357398822177635e-05,
+      "loss": 0.8491,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.13533904006238e-05,
+      "loss": 0.3298,
+      "step": 1811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1349380483051135e-05,
+      "loss": 0.8438,
+      "step": 1812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.134536907101643e-05,
+      "loss": 0.3403,
+      "step": 1813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1341356166077051e-05,
+      "loss": 0.8452,
+      "step": 1814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1337341769790945e-05,
+      "loss": 0.8213,
+      "step": 1815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1333325883716644e-05,
+      "loss": 0.8423,
+      "step": 1816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1329308509413245e-05,
+      "loss": 0.8657,
+      "step": 1817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1325289648440438e-05,
+      "loss": 0.8062,
+      "step": 1818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1321269302358483e-05,
+      "loss": 0.8931,
+      "step": 1819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1317247472728213e-05,
+      "loss": 0.7505,
+      "step": 1820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.131322416111104e-05,
+      "loss": 0.8647,
+      "step": 1821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1309199369068956e-05,
+      "loss": 0.3188,
+      "step": 1822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1305173098164523e-05,
+      "loss": 0.7861,
+      "step": 1823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1301145349960879e-05,
+      "loss": 0.8081,
+      "step": 1824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.129711612602173e-05,
+      "loss": 0.8296,
+      "step": 1825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1293085427911364e-05,
+      "loss": 0.8267,
+      "step": 1826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1289053257194634e-05,
+      "loss": 0.873,
+      "step": 1827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1285019615436968e-05,
+      "loss": 0.3318,
+      "step": 1828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1280984504204367e-05,
+      "loss": 0.8779,
+      "step": 1829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1276947925063398e-05,
+      "loss": 0.9062,
+      "step": 1830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1272909879581201e-05,
+      "loss": 0.8164,
+      "step": 1831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1268870369325482e-05,
+      "loss": 0.8301,
+      "step": 1832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.126482939586452e-05,
+      "loss": 0.8169,
+      "step": 1833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1260786960767159e-05,
+      "loss": 0.8379,
+      "step": 1834
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1256743065602815e-05,
+      "loss": 0.8896,
+      "step": 1835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1252697711941462e-05,
+      "loss": 0.812,
+      "step": 1836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1248650901353649e-05,
+      "loss": 0.8447,
+      "step": 1837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1244602635410486e-05,
+      "loss": 0.3215,
+      "step": 1838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1240552915683649e-05,
+      "loss": 0.8813,
+      "step": 1839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.123650174374538e-05,
+      "loss": 0.8237,
+      "step": 1840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1232449121168482e-05,
+      "loss": 0.8813,
+      "step": 1841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1228395049526323e-05,
+      "loss": 0.8726,
+      "step": 1842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1224339530392835e-05,
+      "loss": 0.8486,
+      "step": 1843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1220282565342506e-05,
+      "loss": 0.7617,
+      "step": 1844
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.1216224155950394e-05,
+      "loss": 0.8921,
+      "step": 1845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1212164303792111e-05,
+      "loss": 0.8794,
+      "step": 1846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.120810301044383e-05,
+      "loss": 0.874,
+      "step": 1847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1204040277482289e-05,
+      "loss": 0.8291,
+      "step": 1848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1199976106484776e-05,
+      "loss": 0.8262,
+      "step": 1849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1195910499029147e-05,
+      "loss": 0.855,
+      "step": 1850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1191843456693807e-05,
+      "loss": 0.3506,
+      "step": 1851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1187774981057724e-05,
+      "loss": 0.8481,
+      "step": 1852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.118370507370042e-05,
+      "loss": 0.8179,
+      "step": 1853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1179633736201968e-05,
+      "loss": 0.8325,
+      "step": 1854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.117556097014301e-05,
+      "loss": 0.7969,
+      "step": 1855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1171486777104733e-05,
+      "loss": 0.8442,
+      "step": 1856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1167411158668872e-05,
+      "loss": 0.7705,
+      "step": 1857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1163334116417728e-05,
+      "loss": 0.8228,
+      "step": 1858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1159255651934148e-05,
+      "loss": 0.8179,
+      "step": 1859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.115517576680153e-05,
+      "loss": 0.7979,
+      "step": 1860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.115109446260383e-05,
+      "loss": 0.834,
+      "step": 1861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1147011740925547e-05,
+      "loss": 0.8545,
+      "step": 1862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1142927603351738e-05,
+      "loss": 0.8242,
+      "step": 1863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1138842051468003e-05,
+      "loss": 0.873,
+      "step": 1864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1134755086860494e-05,
+      "loss": 0.7783,
+      "step": 1865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1130666711115911e-05,
+      "loss": 0.8594,
+      "step": 1866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1126576925821505e-05,
+      "loss": 0.8306,
+      "step": 1867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1122485732565069e-05,
+      "loss": 0.8237,
+      "step": 1868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1118393132934948e-05,
+      "loss": 0.8525,
+      "step": 1869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1114299128520023e-05,
+      "loss": 0.8096,
+      "step": 1870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1110203720909737e-05,
+      "loss": 0.8789,
+      "step": 1871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1106106911694062e-05,
+      "loss": 0.8628,
+      "step": 1872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1102008702463524e-05,
+      "loss": 0.8223,
+      "step": 1873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1097909094809183e-05,
+      "loss": 0.8672,
+      "step": 1874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.109380809032265e-05,
+      "loss": 0.8271,
+      "step": 1875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1089705690596084e-05,
+      "loss": 0.793,
+      "step": 1876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1085601897222169e-05,
+      "loss": 0.8828,
+      "step": 1877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1081496711794142e-05,
+      "loss": 0.8271,
+      "step": 1878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1077390135905777e-05,
+      "loss": 0.8623,
+      "step": 1879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1073282171151386e-05,
+      "loss": 0.8843,
+      "step": 1880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1069172819125826e-05,
+      "loss": 0.8008,
+      "step": 1881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1065062081424489e-05,
+      "loss": 0.8521,
+      "step": 1882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.10609499596433e-05,
+      "loss": 0.8237,
+      "step": 1883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1056836455378733e-05,
+      "loss": 0.8774,
+      "step": 1884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1052721570227787e-05,
+      "loss": 0.8228,
+      "step": 1885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1048605305788005e-05,
+      "loss": 0.875,
+      "step": 1886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1044487663657458e-05,
+      "loss": 0.8745,
+      "step": 1887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1040368645434762e-05,
+      "loss": 0.8457,
+      "step": 1888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1036248252719059e-05,
+      "loss": 0.8408,
+      "step": 1889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1032126487110027e-05,
+      "loss": 0.8496,
+      "step": 1890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.102800335020788e-05,
+      "loss": 0.854,
+      "step": 1891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.102387884361336e-05,
+      "loss": 0.3528,
+      "step": 1892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1019752968927737e-05,
+      "loss": 0.8013,
+      "step": 1893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1015625727752826e-05,
+      "loss": 0.8188,
+      "step": 1894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.1011497121690962e-05,
+      "loss": 0.8599,
+      "step": 1895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.100736715234501e-05,
+      "loss": 0.8472,
+      "step": 1896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.100323582131837e-05,
+      "loss": 0.8101,
+      "step": 1897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.099910313021496e-05,
+      "loss": 0.8228,
+      "step": 1898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.099496908063924e-05,
+      "loss": 0.8408,
+      "step": 1899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0990833674196186e-05,
+      "loss": 0.8931,
+      "step": 1900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0986696912491306e-05,
+      "loss": 0.8433,
+      "step": 1901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0982558797130637e-05,
+      "loss": 0.8604,
+      "step": 1902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0978419329720733e-05,
+      "loss": 0.8555,
+      "step": 1903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.097427851186868e-05,
+      "loss": 0.8203,
+      "step": 1904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0970136345182086e-05,
+      "loss": 0.8818,
+      "step": 1905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0965992831269083e-05,
+      "loss": 0.8535,
+      "step": 1906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0961847971738324e-05,
+      "loss": 0.8345,
+      "step": 1907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0957701768198987e-05,
+      "loss": 0.8682,
+      "step": 1908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0953554222260773e-05,
+      "loss": 0.3467,
+      "step": 1909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.09494053355339e-05,
+      "loss": 0.8799,
+      "step": 1910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.094525510962911e-05,
+      "loss": 0.8652,
+      "step": 1911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0941103546157662e-05,
+      "loss": 0.3209,
+      "step": 1912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0936950646731336e-05,
+      "loss": 0.7827,
+      "step": 1913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0932796412962433e-05,
+      "loss": 0.8867,
+      "step": 1914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0928640846463768e-05,
+      "loss": 0.854,
+      "step": 1915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0924483948848671e-05,
+      "loss": 0.8848,
+      "step": 1916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0920325721731002e-05,
+      "loss": 0.8062,
+      "step": 1917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.091616616672512e-05,
+      "loss": 0.8628,
+      "step": 1918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0912005285445909e-05,
+      "loss": 0.8237,
+      "step": 1919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0907843079508769e-05,
+      "loss": 0.832,
+      "step": 1920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0903679550529609e-05,
+      "loss": 0.8271,
+      "step": 1921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0899514700124857e-05,
+      "loss": 0.8462,
+      "step": 1922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0895348529911447e-05,
+      "loss": 0.8296,
+      "step": 1923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0891181041506835e-05,
+      "loss": 0.8965,
+      "step": 1924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.088701223652898e-05,
+      "loss": 0.8413,
+      "step": 1925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0882842116596355e-05,
+      "loss": 0.8481,
+      "step": 1926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0878670683327948e-05,
+      "loss": 0.7925,
+      "step": 1927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0874497938343249e-05,
+      "loss": 0.9048,
+      "step": 1928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0870323883262262e-05,
+      "loss": 0.8291,
+      "step": 1929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0866148519705505e-05,
+      "loss": 0.8965,
+      "step": 1930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.086197184929399e-05,
+      "loss": 0.8179,
+      "step": 1931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0857793873649244e-05,
+      "loss": 0.8662,
+      "step": 1932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0853614594393306e-05,
+      "loss": 0.8003,
+      "step": 1933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0849434013148714e-05,
+      "loss": 0.8696,
+      "step": 1934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0845252131538516e-05,
+      "loss": 0.8525,
+      "step": 1935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0841068951186256e-05,
+      "loss": 0.3296,
+      "step": 1936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0836884473715999e-05,
+      "loss": 0.9062,
+      "step": 1937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0832698700752292e-05,
+      "loss": 0.8604,
+      "step": 1938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0828511633920201e-05,
+      "loss": 0.8901,
+      "step": 1939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0824323274845292e-05,
+      "loss": 0.8428,
+      "step": 1940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0820133625153626e-05,
+      "loss": 0.8213,
+      "step": 1941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0815942686471771e-05,
+      "loss": 0.8169,
+      "step": 1942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0811750460426797e-05,
+      "loss": 0.8389,
+      "step": 1943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0807556948646266e-05,
+      "loss": 0.8462,
+      "step": 1944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0803362152758242e-05,
+      "loss": 0.8315,
+      "step": 1945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0799166074391296e-05,
+      "loss": 0.8379,
+      "step": 1946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0794968715174483e-05,
+      "loss": 0.8887,
+      "step": 1947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0790770076737368e-05,
+      "loss": 0.8384,
+      "step": 1948
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.0786570160710003e-05,
+      "loss": 0.8096,
+      "step": 1949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0782368968722946e-05,
+      "loss": 0.8818,
+      "step": 1950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0778166502407237e-05,
+      "loss": 0.3264,
+      "step": 1951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.077396276339442e-05,
+      "loss": 0.8789,
+      "step": 1952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0769757753316533e-05,
+      "loss": 0.8418,
+      "step": 1953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0765551473806106e-05,
+      "loss": 0.8618,
+      "step": 1954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.076134392649616e-05,
+      "loss": 0.7988,
+      "step": 1955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.075713511302021e-05,
+      "loss": 0.8677,
+      "step": 1956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0752925035012264e-05,
+      "loss": 0.8379,
+      "step": 1957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0748713694106817e-05,
+      "loss": 0.3201,
+      "step": 1958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0744501091938858e-05,
+      "loss": 0.9111,
+      "step": 1959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0740287230143863e-05,
+      "loss": 0.8345,
+      "step": 1960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.07360721103578e-05,
+      "loss": 0.8291,
+      "step": 1961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.073185573421712e-05,
+      "loss": 0.8486,
+      "step": 1962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0727638103358771e-05,
+      "loss": 0.7852,
+      "step": 1963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.072341921942018e-05,
+      "loss": 0.8545,
+      "step": 1964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.071919908403926e-05,
+      "loss": 0.8472,
+      "step": 1965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0714977698854413e-05,
+      "loss": 0.8613,
+      "step": 1966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0710755065504533e-05,
+      "loss": 0.8291,
+      "step": 1967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0706531185628984e-05,
+      "loss": 0.3274,
+      "step": 1968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0702306060867627e-05,
+      "loss": 0.7988,
+      "step": 1969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0698079692860799e-05,
+      "loss": 0.8208,
+      "step": 1970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0693852083249317e-05,
+      "loss": 0.8442,
+      "step": 1971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0689623233674492e-05,
+      "loss": 0.3101,
+      "step": 1972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0685393145778106e-05,
+      "loss": 0.8589,
+      "step": 1973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0681161821202424e-05,
+      "loss": 0.8384,
+      "step": 1974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.067692926159019e-05,
+      "loss": 0.8604,
+      "step": 1975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0672695468584634e-05,
+      "loss": 0.7827,
+      "step": 1976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.066846044382946e-05,
+      "loss": 0.8506,
+      "step": 1977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0664224188968843e-05,
+      "loss": 0.8896,
+      "step": 1978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0659986705647452e-05,
+      "loss": 0.7856,
+      "step": 1979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0655747995510422e-05,
+      "loss": 0.8877,
+      "step": 1980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0651508060203361e-05,
+      "loss": 0.877,
+      "step": 1981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0647266901372363e-05,
+      "loss": 0.8086,
+      "step": 1982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0643024520663992e-05,
+      "loss": 0.8579,
+      "step": 1983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0638780919725282e-05,
+      "loss": 0.8696,
+      "step": 1984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0634536100203748e-05,
+      "loss": 0.8154,
+      "step": 1985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0630290063747374e-05,
+      "loss": 0.875,
+      "step": 1986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0626042812004621e-05,
+      "loss": 0.8765,
+      "step": 1987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0621794346624412e-05,
+      "loss": 0.9238,
+      "step": 1988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0617544669256154e-05,
+      "loss": 0.7856,
+      "step": 1989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0613293781549714e-05,
+      "loss": 0.8618,
+      "step": 1990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0609041685155436e-05,
+      "loss": 0.8262,
+      "step": 1991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0604788381724125e-05,
+      "loss": 0.7622,
+      "step": 1992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0600533872907064e-05,
+      "loss": 0.353,
+      "step": 1993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0596278160355999e-05,
+      "loss": 0.8594,
+      "step": 1994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0592021245723147e-05,
+      "loss": 0.8491,
+      "step": 1995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0587763130661182e-05,
+      "loss": 0.8447,
+      "step": 1996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0583503816823259e-05,
+      "loss": 0.8501,
+      "step": 1997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0579243305862985e-05,
+      "loss": 0.8569,
+      "step": 1998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0574981599434437e-05,
+      "loss": 0.3318,
+      "step": 1999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0570718699192158e-05,
+      "loss": 0.8579,
+      "step": 2000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0566454606791151e-05,
+      "loss": 0.8613,
+      "step": 2001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0562189323886883e-05,
+      "loss": 0.8569,
+      "step": 2002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.055792285213529e-05,
+      "loss": 0.298,
+      "step": 2003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0553655193192756e-05,
+      "loss": 0.854,
+      "step": 2004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0549386348716135e-05,
+      "loss": 0.8647,
+      "step": 2005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.054511632036274e-05,
+      "loss": 0.3506,
+      "step": 2006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0540845109790343e-05,
+      "loss": 0.8213,
+      "step": 2007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0536572718657174e-05,
+      "loss": 0.8345,
+      "step": 2008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0532299148621927e-05,
+      "loss": 0.8774,
+      "step": 2009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0528024401343743e-05,
+      "loss": 0.343,
+      "step": 2010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0523748478482229e-05,
+      "loss": 0.834,
+      "step": 2011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0519471381697444e-05,
+      "loss": 0.7512,
+      "step": 2012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0515193112649906e-05,
+      "loss": 0.8711,
+      "step": 2013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0510913673000584e-05,
+      "loss": 0.8442,
+      "step": 2014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0506633064410907e-05,
+      "loss": 0.8496,
+      "step": 2015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0502351288542755e-05,
+      "loss": 0.8232,
+      "step": 2016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0498068347058455e-05,
+      "loss": 0.8301,
+      "step": 2017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0493784241620797e-05,
+      "loss": 0.876,
+      "step": 2018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0489498973893019e-05,
+      "loss": 0.8755,
+      "step": 2019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0485212545538805e-05,
+      "loss": 0.8403,
+      "step": 2020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.04809249582223e-05,
+      "loss": 0.7866,
+      "step": 2021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0476636213608088e-05,
+      "loss": 0.8662,
+      "step": 2022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0472346313361213e-05,
+      "loss": 0.8096,
+      "step": 2023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0468055259147153e-05,
+      "loss": 0.8184,
+      "step": 2024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0463763052631851e-05,
+      "loss": 0.8374,
+      "step": 2025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0459469695481686e-05,
+      "loss": 0.7827,
+      "step": 2026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0455175189363488e-05,
+      "loss": 0.8628,
+      "step": 2027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.045087953594453e-05,
+      "loss": 0.8379,
+      "step": 2028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0446582736892538e-05,
+      "loss": 0.8188,
+      "step": 2029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0442284793875671e-05,
+      "loss": 0.8325,
+      "step": 2030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0437985708562543e-05,
+      "loss": 0.8442,
+      "step": 2031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0433685482622203e-05,
+      "loss": 0.8906,
+      "step": 2032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0429384117724149e-05,
+      "loss": 0.854,
+      "step": 2033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.042508161553832e-05,
+      "loss": 0.8838,
+      "step": 2034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0420777977735095e-05,
+      "loss": 0.7593,
+      "step": 2035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0416473205985294e-05,
+      "loss": 0.8301,
+      "step": 2036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0412167301960174e-05,
+      "loss": 0.9053,
+      "step": 2037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0407860267331443e-05,
+      "loss": 0.8311,
+      "step": 2038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0403552103771235e-05,
+      "loss": 0.8281,
+      "step": 2039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0399242812952125e-05,
+      "loss": 0.3547,
+      "step": 2040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0394932396547133e-05,
+      "loss": 0.8726,
+      "step": 2041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0390620856229707e-05,
+      "loss": 0.2799,
+      "step": 2042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.038630819367374e-05,
+      "loss": 0.8867,
+      "step": 2043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0381994410553554e-05,
+      "loss": 0.8486,
+      "step": 2044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0377679508543907e-05,
+      "loss": 0.853,
+      "step": 2045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0373363489319992e-05,
+      "loss": 0.8369,
+      "step": 2046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.036904635455744e-05,
+      "loss": 0.7988,
+      "step": 2047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0364728105932308e-05,
+      "loss": 0.8604,
+      "step": 2048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0360408745121089e-05,
+      "loss": 0.8081,
+      "step": 2049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0356088273800711e-05,
+      "loss": 0.9009,
+      "step": 2050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0351766693648525e-05,
+      "loss": 0.8354,
+      "step": 2051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0347444006342319e-05,
+      "loss": 0.876,
+      "step": 2052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.034312021356031e-05,
+      "loss": 0.835,
+      "step": 2053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0338795316981143e-05,
+      "loss": 0.3486,
+      "step": 2054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.033446931828389e-05,
+      "loss": 0.8325,
+      "step": 2055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0330142219148057e-05,
+      "loss": 0.8403,
+      "step": 2056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0325814021253568e-05,
+      "loss": 0.8413,
+      "step": 2057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.032148472628078e-05,
+      "loss": 0.8066,
+      "step": 2058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0317154335910472e-05,
+      "loss": 0.3148,
+      "step": 2059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.031282285182386e-05,
+      "loss": 0.8442,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0308490275702564e-05,
+      "loss": 0.8374,
+      "step": 2061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0304156609228644e-05,
+      "loss": 0.8628,
+      "step": 2062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.029982185408458e-05,
+      "loss": 0.8574,
+      "step": 2063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.029548601195327e-05,
+      "loss": 0.8257,
+      "step": 2064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.029114908451804e-05,
+      "loss": 0.8384,
+      "step": 2065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0286811073462632e-05,
+      "loss": 0.8242,
+      "step": 2066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0282471980471214e-05,
+      "loss": 0.8433,
+      "step": 2067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0278131807228367e-05,
+      "loss": 0.8291,
+      "step": 2068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0273790555419103e-05,
+      "loss": 0.8267,
+      "step": 2069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0269448226728838e-05,
+      "loss": 0.8052,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0265104822843418e-05,
+      "loss": 0.3242,
+      "step": 2071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0260760345449098e-05,
+      "loss": 0.3347,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0256414796232557e-05,
+      "loss": 0.8594,
+      "step": 2073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0252068176880888e-05,
+      "loss": 0.7861,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0247720489081594e-05,
+      "loss": 0.8081,
+      "step": 2075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0243371734522601e-05,
+      "loss": 0.8384,
+      "step": 2076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0239021914892242e-05,
+      "loss": 0.8193,
+      "step": 2077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.023467103187927e-05,
+      "loss": 0.8101,
+      "step": 2078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0230319087172844e-05,
+      "loss": 0.8442,
+      "step": 2079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.022596608246254e-05,
+      "loss": 0.8647,
+      "step": 2080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0221612019438349e-05,
+      "loss": 0.832,
+      "step": 2081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0217256899790661e-05,
+      "loss": 0.7803,
+      "step": 2082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0212900725210289e-05,
+      "loss": 0.8477,
+      "step": 2083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0208543497388446e-05,
+      "loss": 0.8008,
+      "step": 2084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0204185218016759e-05,
+      "loss": 0.8398,
+      "step": 2085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0199825888787264e-05,
+      "loss": 0.8394,
+      "step": 2086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.01954655113924e-05,
+      "loss": 0.8418,
+      "step": 2087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0191104087525018e-05,
+      "loss": 0.8672,
+      "step": 2088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.018674161887837e-05,
+      "loss": 0.8633,
+      "step": 2089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0182378107146121e-05,
+      "loss": 0.8345,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0178013554022333e-05,
+      "loss": 0.8315,
+      "step": 2091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0173647961201473e-05,
+      "loss": 0.8584,
+      "step": 2092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0169281330378422e-05,
+      "loss": 0.8833,
+      "step": 2093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0164913663248452e-05,
+      "loss": 0.8716,
+      "step": 2094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0160544961507242e-05,
+      "loss": 0.8081,
+      "step": 2095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0156175226850873e-05,
+      "loss": 0.8579,
+      "step": 2096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0151804460975826e-05,
+      "loss": 0.8496,
+      "step": 2097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.014743266557898e-05,
+      "loss": 0.8003,
+      "step": 2098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0143059842357621e-05,
+      "loss": 0.8564,
+      "step": 2099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.013868599300943e-05,
+      "loss": 0.8843,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0134311119232483e-05,
+      "loss": 0.8555,
+      "step": 2101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0129935222725259e-05,
+      "loss": 0.8042,
+      "step": 2102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0125558305186628e-05,
+      "loss": 0.7983,
+      "step": 2103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0121180368315863e-05,
+      "loss": 0.8022,
+      "step": 2104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0116801413812628e-05,
+      "loss": 0.9102,
+      "step": 2105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0112421443376986e-05,
+      "loss": 0.9028,
+      "step": 2106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0108040458709391e-05,
+      "loss": 0.8184,
+      "step": 2107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0103658461510693e-05,
+      "loss": 0.8301,
+      "step": 2108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0099275453482134e-05,
+      "loss": 0.7959,
+      "step": 2109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0094891436325349e-05,
+      "loss": 0.8794,
+      "step": 2110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0090506411742362e-05,
+      "loss": 0.8271,
+      "step": 2111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0086120381435592e-05,
+      "loss": 0.8091,
+      "step": 2112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0081733347107848e-05,
+      "loss": 0.8472,
+      "step": 2113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0077345310462328e-05,
+      "loss": 0.3472,
+      "step": 2114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0072956273202616e-05,
+      "loss": 0.8735,
+      "step": 2115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0068566237032691e-05,
+      "loss": 0.8232,
+      "step": 2116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0064175203656913e-05,
+      "loss": 0.3457,
+      "step": 2117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0059783174780033e-05,
+      "loss": 0.8691,
+      "step": 2118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.005539015210719e-05,
+      "loss": 0.8369,
+      "step": 2119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0050996137343905e-05,
+      "loss": 0.7852,
+      "step": 2120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0046601132196087e-05,
+      "loss": 0.8101,
+      "step": 2121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0042205138370027e-05,
+      "loss": 0.8647,
+      "step": 2122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0037808157572402e-05,
+      "loss": 0.8804,
+      "step": 2123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.003341019151027e-05,
+      "loss": 0.8013,
+      "step": 2124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0029011241891073e-05,
+      "loss": 0.8467,
+      "step": 2125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0024611310422638e-05,
+      "loss": 0.8096,
+      "step": 2126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0020210398813164e-05,
+      "loss": 0.3325,
+      "step": 2127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0015808508771243e-05,
+      "loss": 0.8823,
+      "step": 2128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0011405642005835e-05,
+      "loss": 0.7993,
+      "step": 2129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0007001800226288e-05,
+      "loss": 0.8857,
+      "step": 2130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0002596985142322e-05,
+      "loss": 0.9014,
+      "step": 2131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.998191198464042e-06,
+      "loss": 0.8594,
+      "step": 2132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.99378444190192e-06,
+      "loss": 0.8345,
+      "step": 2133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.989376717166817e-06,
+      "loss": 0.8784,
+      "step": 2134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.98496802596996e-06,
+      "loss": 0.834,
+      "step": 2135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.980558370022956e-06,
+      "loss": 0.8545,
+      "step": 2136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.976147751037783e-06,
+      "loss": 0.8359,
+      "step": 2137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.971736170726797e-06,
+      "loss": 0.8652,
+      "step": 2138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.967323630802725e-06,
+      "loss": 0.8862,
+      "step": 2139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.962910132978667e-06,
+      "loss": 0.8389,
+      "step": 2140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.958495678968096e-06,
+      "loss": 0.7649,
+      "step": 2141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.954080270484856e-06,
+      "loss": 0.8657,
+      "step": 2142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.949663909243154e-06,
+      "loss": 0.8521,
+      "step": 2143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.94524659695758e-06,
+      "loss": 0.8608,
+      "step": 2144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.940828335343085e-06,
+      "loss": 0.7622,
+      "step": 2145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.93640912611499e-06,
+      "loss": 0.8301,
+      "step": 2146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.931988970988985e-06,
+      "loss": 0.8281,
+      "step": 2147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.927567871681124e-06,
+      "loss": 0.8618,
+      "step": 2148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.923145829907833e-06,
+      "loss": 0.8955,
+      "step": 2149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.918722847385898e-06,
+      "loss": 0.8301,
+      "step": 2150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.914298925832475e-06,
+      "loss": 0.8545,
+      "step": 2151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.909874066965083e-06,
+      "loss": 0.8032,
+      "step": 2152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.905448272501603e-06,
+      "loss": 0.8716,
+      "step": 2153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.90102154416028e-06,
+      "loss": 0.8374,
+      "step": 2154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.89659388365973e-06,
+      "loss": 0.8545,
+      "step": 2155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.892165292718912e-06,
+      "loss": 0.8047,
+      "step": 2156
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.887735773057161e-06,
+      "loss": 0.8267,
+      "step": 2157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.883305326394176e-06,
+      "loss": 0.8789,
+      "step": 2158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.87887395445e-06,
+      "loss": 0.8252,
+      "step": 2159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.87444165894505e-06,
+      "loss": 0.7861,
+      "step": 2160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.870008441600092e-06,
+      "loss": 0.853,
+      "step": 2161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.865574304136256e-06,
+      "loss": 0.9131,
+      "step": 2162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.861139248275028e-06,
+      "loss": 0.8652,
+      "step": 2163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.856703275738244e-06,
+      "loss": 0.8252,
+      "step": 2164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.852266388248106e-06,
+      "loss": 0.8076,
+      "step": 2165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.847828587527166e-06,
+      "loss": 0.8213,
+      "step": 2166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.84338987529833e-06,
+      "loss": 0.769,
+      "step": 2167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.838950253284859e-06,
+      "loss": 0.8335,
+      "step": 2168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.834509723210366e-06,
+      "loss": 0.8438,
+      "step": 2169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.830068286798819e-06,
+      "loss": 0.8486,
+      "step": 2170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.825625945774537e-06,
+      "loss": 0.8115,
+      "step": 2171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.82118270186219e-06,
+      "loss": 0.8032,
+      "step": 2172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.816738556786795e-06,
+      "loss": 0.8354,
+      "step": 2173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.812293512273727e-06,
+      "loss": 0.8428,
+      "step": 2174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.807847570048704e-06,
+      "loss": 0.8462,
+      "step": 2175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.803400731837793e-06,
+      "loss": 0.8037,
+      "step": 2176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.798952999367404e-06,
+      "loss": 0.7935,
+      "step": 2177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.794504374364312e-06,
+      "loss": 0.854,
+      "step": 2178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.790054858555616e-06,
+      "loss": 0.9136,
+      "step": 2179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.785604453668776e-06,
+      "loss": 0.8066,
+      "step": 2180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.781153161431592e-06,
+      "loss": 0.8726,
+      "step": 2181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.776700983572208e-06,
+      "loss": 0.8003,
+      "step": 2182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.77224792181911e-06,
+      "loss": 0.8379,
+      "step": 2183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.767793977901136e-06,
+      "loss": 0.856,
+      "step": 2184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.763339153547456e-06,
+      "loss": 0.8618,
+      "step": 2185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.758883450487591e-06,
+      "loss": 0.8213,
+      "step": 2186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.754426870451391e-06,
+      "loss": 0.8784,
+      "step": 2187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.74996941516906e-06,
+      "loss": 0.8604,
+      "step": 2188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.745511086371133e-06,
+      "loss": 0.8804,
+      "step": 2189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.741051885788486e-06,
+      "loss": 0.8223,
+      "step": 2190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.736591815152334e-06,
+      "loss": 0.835,
+      "step": 2191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.732130876194235e-06,
+      "loss": 0.8037,
+      "step": 2192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.727669070646073e-06,
+      "loss": 0.8315,
+      "step": 2193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.723206400240078e-06,
+      "loss": 0.8013,
+      "step": 2194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.71874286670881e-06,
+      "loss": 0.8198,
+      "step": 2195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.714278471785167e-06,
+      "loss": 0.8682,
+      "step": 2196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.709813217202382e-06,
+      "loss": 0.8438,
+      "step": 2197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.705347104694021e-06,
+      "loss": 0.8672,
+      "step": 2198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.700880135993982e-06,
+      "loss": 0.3362,
+      "step": 2199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.696412312836495e-06,
+      "loss": 0.8335,
+      "step": 2200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.691943636956124e-06,
+      "loss": 0.3523,
+      "step": 2201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.687474110087761e-06,
+      "loss": 0.7842,
+      "step": 2202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.68300373396663e-06,
+      "loss": 0.8433,
+      "step": 2203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.678532510328289e-06,
+      "loss": 0.7876,
+      "step": 2204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.674060440908618e-06,
+      "loss": 0.8071,
+      "step": 2205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.669587527443829e-06,
+      "loss": 0.8677,
+      "step": 2206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.665113771670461e-06,
+      "loss": 0.8247,
+      "step": 2207
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.66063917532538e-06,
+      "loss": 0.325,
+      "step": 2208
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.656163740145778e-06,
+      "loss": 0.8184,
+      "step": 2209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.651687467869169e-06,
+      "loss": 0.8296,
+      "step": 2210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.647210360233406e-06,
+      "loss": 0.8301,
+      "step": 2211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.642732418976649e-06,
+      "loss": 0.3267,
+      "step": 2212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.638253645837388e-06,
+      "loss": 0.8447,
+      "step": 2213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.633774042554442e-06,
+      "loss": 0.7725,
+      "step": 2214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.629293610866945e-06,
+      "loss": 0.8394,
+      "step": 2215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.624812352514351e-06,
+      "loss": 0.8184,
+      "step": 2216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.620330269236446e-06,
+      "loss": 0.7949,
+      "step": 2217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.615847362773324e-06,
+      "loss": 0.8071,
+      "step": 2218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.611363634865405e-06,
+      "loss": 0.877,
+      "step": 2219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.606879087253427e-06,
+      "loss": 0.9019,
+      "step": 2220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.602393721678445e-06,
+      "loss": 0.8394,
+      "step": 2221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.597907539881834e-06,
+      "loss": 0.8608,
+      "step": 2222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.59342054360528e-06,
+      "loss": 0.334,
+      "step": 2223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.588932734590794e-06,
+      "loss": 0.8301,
+      "step": 2224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.584444114580696e-06,
+      "loss": 0.8853,
+      "step": 2225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.579954685317624e-06,
+      "loss": 0.8203,
+      "step": 2226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.575464448544529e-06,
+      "loss": 0.8271,
+      "step": 2227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.570973406004674e-06,
+      "loss": 0.7983,
+      "step": 2228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.566481559441635e-06,
+      "loss": 0.8462,
+      "step": 2229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.561988910599305e-06,
+      "loss": 0.749,
+      "step": 2230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.557495461221885e-06,
+      "loss": 0.8047,
+      "step": 2231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.553001213053883e-06,
+      "loss": 0.8242,
+      "step": 2232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.548506167840127e-06,
+      "loss": 0.8325,
+      "step": 2233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.544010327325741e-06,
+      "loss": 0.8184,
+      "step": 2234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.539513693256172e-06,
+      "loss": 0.8296,
+      "step": 2235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.53501626737716e-06,
+      "loss": 0.8569,
+      "step": 2236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.530518051434767e-06,
+      "loss": 0.8901,
+      "step": 2237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.526019047175355e-06,
+      "loss": 0.8506,
+      "step": 2238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.52151925634559e-06,
+      "loss": 0.8784,
+      "step": 2239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.517018680692447e-06,
+      "loss": 0.8208,
+      "step": 2240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.512517321963204e-06,
+      "loss": 0.8374,
+      "step": 2241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.508015181905444e-06,
+      "loss": 0.8394,
+      "step": 2242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.503512262267051e-06,
+      "loss": 0.7847,
+      "step": 2243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.499008564796216e-06,
+      "loss": 0.8198,
+      "step": 2244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.494504091241428e-06,
+      "loss": 0.8364,
+      "step": 2245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.489998843351481e-06,
+      "loss": 0.8135,
+      "step": 2246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.485492822875463e-06,
+      "loss": 0.792,
+      "step": 2247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.48098603156277e-06,
+      "loss": 0.8311,
+      "step": 2248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.47647847116309e-06,
+      "loss": 0.8677,
+      "step": 2249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.471970143426417e-06,
+      "loss": 0.7827,
+      "step": 2250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.467461050103038e-06,
+      "loss": 0.895,
+      "step": 2251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.462951192943538e-06,
+      "loss": 0.8193,
+      "step": 2252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.458440573698797e-06,
+      "loss": 0.3274,
+      "step": 2253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.453929194119994e-06,
+      "loss": 0.7949,
+      "step": 2254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.449417055958603e-06,
+      "loss": 0.8638,
+      "step": 2255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.444904160966388e-06,
+      "loss": 0.833,
+      "step": 2256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.440390510895415e-06,
+      "loss": 0.8354,
+      "step": 2257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.435876107498035e-06,
+      "loss": 0.8872,
+      "step": 2258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.4313609525269e-06,
+      "loss": 0.8184,
+      "step": 2259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.42684504773494e-06,
+      "loss": 0.874,
+      "step": 2260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.42232839487539e-06,
+      "loss": 0.8315,
+      "step": 2261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.41781099570177e-06,
+      "loss": 0.874,
+      "step": 2262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.41329285196789e-06,
+      "loss": 0.8511,
+      "step": 2263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.408773965427846e-06,
+      "loss": 0.3192,
+      "step": 2264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.40425433783603e-06,
+      "loss": 0.8501,
+      "step": 2265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.399733970947114e-06,
+      "loss": 0.8647,
+      "step": 2266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.395212866516063e-06,
+      "loss": 0.7891,
+      "step": 2267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.390691026298118e-06,
+      "loss": 0.8325,
+      "step": 2268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.38616845204882e-06,
+      "loss": 0.8804,
+      "step": 2269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.381645145523989e-06,
+      "loss": 0.8125,
+      "step": 2270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.37712110847972e-06,
+      "loss": 0.8477,
+      "step": 2271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.372596342672413e-06,
+      "loss": 0.8677,
+      "step": 2272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.368070849858728e-06,
+      "loss": 0.8569,
+      "step": 2273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.363544631795617e-06,
+      "loss": 0.8887,
+      "step": 2274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.359017690240316e-06,
+      "loss": 0.7969,
+      "step": 2275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.354490026950341e-06,
+      "loss": 0.8286,
+      "step": 2276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.349961643683486e-06,
+      "loss": 0.8086,
+      "step": 2277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.345432542197822e-06,
+      "loss": 0.8213,
+      "step": 2278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.340902724251711e-06,
+      "loss": 0.8145,
+      "step": 2279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.336372191603775e-06,
+      "loss": 0.8672,
+      "step": 2280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.331840946012923e-06,
+      "loss": 0.853,
+      "step": 2281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.327308989238347e-06,
+      "loss": 0.8477,
+      "step": 2282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.322776323039505e-06,
+      "loss": 0.8999,
+      "step": 2283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.318242949176137e-06,
+      "loss": 0.8623,
+      "step": 2284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.313708869408246e-06,
+      "loss": 0.7993,
+      "step": 2285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.30917408549613e-06,
+      "loss": 0.8306,
+      "step": 2286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.304638599200339e-06,
+      "loss": 0.9038,
+      "step": 2287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.300102412281708e-06,
+      "loss": 0.832,
+      "step": 2288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.295565526501344e-06,
+      "loss": 0.8564,
+      "step": 2289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.291027943620618e-06,
+      "loss": 0.8369,
+      "step": 2290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.286489665401175e-06,
+      "loss": 0.8374,
+      "step": 2291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.281950693604937e-06,
+      "loss": 0.8564,
+      "step": 2292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.277411029994086e-06,
+      "loss": 0.7959,
+      "step": 2293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.272870676331072e-06,
+      "loss": 0.8594,
+      "step": 2294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.268329634378619e-06,
+      "loss": 0.8213,
+      "step": 2295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.263787905899718e-06,
+      "loss": 0.7764,
+      "step": 2296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.259245492657624e-06,
+      "loss": 0.8403,
+      "step": 2297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.254702396415855e-06,
+      "loss": 0.8027,
+      "step": 2298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.250158618938203e-06,
+      "loss": 0.3257,
+      "step": 2299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.245614161988714e-06,
+      "loss": 0.8867,
+      "step": 2300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.2410690273317e-06,
+      "loss": 0.8506,
+      "step": 2301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.236523216731748e-06,
+      "loss": 0.8804,
+      "step": 2302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.231976731953689e-06,
+      "loss": 0.8877,
+      "step": 2303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.227429574762631e-06,
+      "loss": 0.897,
+      "step": 2304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.222881746923935e-06,
+      "loss": 0.8096,
+      "step": 2305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.218333250203227e-06,
+      "loss": 0.8579,
+      "step": 2306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.213784086366382e-06,
+      "loss": 0.8545,
+      "step": 2307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.209234257179551e-06,
+      "loss": 0.8525,
+      "step": 2308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.20468376440913e-06,
+      "loss": 0.8091,
+      "step": 2309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.200132609821778e-06,
+      "loss": 0.8232,
+      "step": 2310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.195580795184406e-06,
+      "loss": 0.8101,
+      "step": 2311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.191028322264194e-06,
+      "loss": 0.7808,
+      "step": 2312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.186475192828562e-06,
+      "loss": 0.8203,
+      "step": 2313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.18192140864519e-06,
+      "loss": 0.8633,
+      "step": 2314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.177366971482021e-06,
+      "loss": 0.834,
+      "step": 2315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.172811883107238e-06,
+      "loss": 0.3401,
+      "step": 2316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.168256145289284e-06,
+      "loss": 0.8511,
+      "step": 2317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.163699759796857e-06,
+      "loss": 0.854,
+      "step": 2318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.159142728398902e-06,
+      "loss": 0.8711,
+      "step": 2319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.15458505286461e-06,
+      "loss": 0.8438,
+      "step": 2320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.150026734963432e-06,
+      "loss": 0.8589,
+      "step": 2321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.145467776465068e-06,
+      "loss": 0.8452,
+      "step": 2322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.140908179139453e-06,
+      "loss": 0.8091,
+      "step": 2323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.136347944756784e-06,
+      "loss": 0.8271,
+      "step": 2324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.131787075087506e-06,
+      "loss": 0.9087,
+      "step": 2325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.127225571902302e-06,
+      "loss": 0.8628,
+      "step": 2326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.122663436972099e-06,
+      "loss": 0.8711,
+      "step": 2327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.118100672068082e-06,
+      "loss": 0.7881,
+      "step": 2328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.113537278961671e-06,
+      "loss": 0.79,
+      "step": 2329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.108973259424533e-06,
+      "loss": 0.8203,
+      "step": 2330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.104408615228577e-06,
+      "loss": 0.8105,
+      "step": 2331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.099843348145957e-06,
+      "loss": 0.8677,
+      "step": 2332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.095277459949063e-06,
+      "loss": 0.7871,
+      "step": 2333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.09071095241053e-06,
+      "loss": 0.8149,
+      "step": 2334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.086143827303241e-06,
+      "loss": 0.8179,
+      "step": 2335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.081576086400302e-06,
+      "loss": 0.8179,
+      "step": 2336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.07700773147507e-06,
+      "loss": 0.2904,
+      "step": 2337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.072438764301142e-06,
+      "loss": 0.7944,
+      "step": 2338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.067869186652345e-06,
+      "loss": 0.748,
+      "step": 2339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.063299000302744e-06,
+      "loss": 0.8281,
+      "step": 2340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.058728207026645e-06,
+      "loss": 0.7915,
+      "step": 2341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.054156808598589e-06,
+      "loss": 0.873,
+      "step": 2342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.049584806793346e-06,
+      "loss": 0.334,
+      "step": 2343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.04501220338593e-06,
+      "loss": 0.8555,
+      "step": 2344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.04043900015158e-06,
+      "loss": 0.8799,
+      "step": 2345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.03586519886577e-06,
+      "loss": 0.8398,
+      "step": 2346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.03129080130421e-06,
+      "loss": 0.8584,
+      "step": 2347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.026715809242837e-06,
+      "loss": 0.8652,
+      "step": 2348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.022140224457821e-06,
+      "loss": 0.7871,
+      "step": 2349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.017564048725558e-06,
+      "loss": 0.8701,
+      "step": 2350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.012987283822683e-06,
+      "loss": 0.8047,
+      "step": 2351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.00840993152605e-06,
+      "loss": 0.8394,
+      "step": 2352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.003831993612744e-06,
+      "loss": 0.8174,
+      "step": 2353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.999253471860081e-06,
+      "loss": 0.8335,
+      "step": 2354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.9946743680456e-06,
+      "loss": 0.8018,
+      "step": 2355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.990094683947064e-06,
+      "loss": 0.8618,
+      "step": 2356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.985514421342466e-06,
+      "loss": 0.8267,
+      "step": 2357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.980933582010025e-06,
+      "loss": 0.8398,
+      "step": 2358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.976352167728178e-06,
+      "loss": 0.7666,
+      "step": 2359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.971770180275583e-06,
+      "loss": 0.8433,
+      "step": 2360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.967187621431133e-06,
+      "loss": 0.8364,
+      "step": 2361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.96260449297393e-06,
+      "loss": 0.8843,
+      "step": 2362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.958020796683304e-06,
+      "loss": 0.8813,
+      "step": 2363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.953436534338803e-06,
+      "loss": 0.3484,
+      "step": 2364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.948851707720198e-06,
+      "loss": 0.7969,
+      "step": 2365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.944266318607473e-06,
+      "loss": 0.8545,
+      "step": 2366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.939680368780836e-06,
+      "loss": 0.8472,
+      "step": 2367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.93509386002071e-06,
+      "loss": 0.7739,
+      "step": 2368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.930506794107733e-06,
+      "loss": 0.8037,
+      "step": 2369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.92591917282277e-06,
+      "loss": 0.8271,
+      "step": 2370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.921330997946887e-06,
+      "loss": 0.8242,
+      "step": 2371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.916742271261373e-06,
+      "loss": 0.8018,
+      "step": 2372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.912152994547726e-06,
+      "loss": 0.791,
+      "step": 2373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.907563169587668e-06,
+      "loss": 0.792,
+      "step": 2374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.902972798163123e-06,
+      "loss": 0.8613,
+      "step": 2375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.89838188205623e-06,
+      "loss": 0.8032,
+      "step": 2376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.893790423049344e-06,
+      "loss": 0.8076,
+      "step": 2377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.889198422925029e-06,
+      "loss": 0.8223,
+      "step": 2378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.88460588346605e-06,
+      "loss": 0.3218,
+      "step": 2379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.880012806455396e-06,
+      "loss": 0.8623,
+      "step": 2380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.875419193676255e-06,
+      "loss": 0.8247,
+      "step": 2381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.870825046912025e-06,
+      "loss": 0.8247,
+      "step": 2382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.866230367946313e-06,
+      "loss": 0.8579,
+      "step": 2383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.861635158562934e-06,
+      "loss": 0.3131,
+      "step": 2384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.857039420545904e-06,
+      "loss": 0.8164,
+      "step": 2385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.852443155679444e-06,
+      "loss": 0.8423,
+      "step": 2386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.847846365747986e-06,
+      "loss": 0.814,
+      "step": 2387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.84324905253616e-06,
+      "loss": 0.7808,
+      "step": 2388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.838651217828803e-06,
+      "loss": 0.8315,
+      "step": 2389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.834052863410954e-06,
+      "loss": 0.7998,
+      "step": 2390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.82945399106785e-06,
+      "loss": 0.8398,
+      "step": 2391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.82485460258493e-06,
+      "loss": 0.7964,
+      "step": 2392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.820254699747841e-06,
+      "loss": 0.8491,
+      "step": 2393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.815654284342416e-06,
+      "loss": 0.8003,
+      "step": 2394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.8110533581547e-06,
+      "loss": 0.791,
+      "step": 2395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.806451922970929e-06,
+      "loss": 0.8037,
+      "step": 2396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.80184998057754e-06,
+      "loss": 0.304,
+      "step": 2397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.797247532761165e-06,
+      "loss": 0.8633,
+      "step": 2398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.792644581308632e-06,
+      "loss": 0.8066,
+      "step": 2399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.788041128006965e-06,
+      "loss": 0.8564,
+      "step": 2400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.783437174643383e-06,
+      "loss": 0.8413,
+      "step": 2401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.778832723005298e-06,
+      "loss": 0.7896,
+      "step": 2402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.774227774880321e-06,
+      "loss": 0.8726,
+      "step": 2403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.769622332056248e-06,
+      "loss": 0.7881,
+      "step": 2404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.765016396321073e-06,
+      "loss": 0.8179,
+      "step": 2405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.760409969462976e-06,
+      "loss": 0.8447,
+      "step": 2406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.755803053270332e-06,
+      "loss": 0.8247,
+      "step": 2407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.751195649531703e-06,
+      "loss": 0.8511,
+      "step": 2408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.746587760035845e-06,
+      "loss": 0.8125,
+      "step": 2409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.741979386571696e-06,
+      "loss": 0.8896,
+      "step": 2410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.737370530928385e-06,
+      "loss": 0.7842,
+      "step": 2411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.732761194895233e-06,
+      "loss": 0.325,
+      "step": 2412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.728151380261738e-06,
+      "loss": 0.8418,
+      "step": 2413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.723541088817592e-06,
+      "loss": 0.8428,
+      "step": 2414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.718930322352663e-06,
+      "loss": 0.7681,
+      "step": 2415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.71431908265702e-06,
+      "loss": 0.8477,
+      "step": 2416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.709707371520897e-06,
+      "loss": 0.8774,
+      "step": 2417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.70509519073472e-06,
+      "loss": 0.8257,
+      "step": 2418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.700482542089097e-06,
+      "loss": 0.8105,
+      "step": 2419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.69586942737482e-06,
+      "loss": 0.8452,
+      "step": 2420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.691255848382851e-06,
+      "loss": 0.8511,
+      "step": 2421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.68664180690435e-06,
+      "loss": 0.8389,
+      "step": 2422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.682027304730641e-06,
+      "loss": 0.8145,
+      "step": 2423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.677412343653234e-06,
+      "loss": 0.7979,
+      "step": 2424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.672796925463817e-06,
+      "loss": 0.8442,
+      "step": 2425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.668181051954252e-06,
+      "loss": 0.8359,
+      "step": 2426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.66356472491658e-06,
+      "loss": 0.8047,
+      "step": 2427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.65894794614302e-06,
+      "loss": 0.8599,
+      "step": 2428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.654330717425963e-06,
+      "loss": 0.8594,
+      "step": 2429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.649713040557977e-06,
+      "loss": 0.3276,
+      "step": 2430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.645094917331803e-06,
+      "loss": 0.8423,
+      "step": 2431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.640476349540356e-06,
+      "loss": 0.8408,
+      "step": 2432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.635857338976721e-06,
+      "loss": 0.8721,
+      "step": 2433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.631237887434158e-06,
+      "loss": 0.8091,
+      "step": 2434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.6266179967061e-06,
+      "loss": 0.8765,
+      "step": 2435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.621997668586141e-06,
+      "loss": 0.812,
+      "step": 2436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.61737690486806e-06,
+      "loss": 0.832,
+      "step": 2437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.612755707345788e-06,
+      "loss": 0.8169,
+      "step": 2438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.60813407781344e-06,
+      "loss": 0.8037,
+      "step": 2439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.603512018065288e-06,
+      "loss": 0.8179,
+      "step": 2440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.598889529895775e-06,
+      "loss": 0.3271,
+      "step": 2441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.594266615099511e-06,
+      "loss": 0.8193,
+      "step": 2442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.58964327547127e-06,
+      "loss": 0.8081,
+      "step": 2443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.585019512805993e-06,
+      "loss": 0.7783,
+      "step": 2444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.580395328898783e-06,
+      "loss": 0.8501,
+      "step": 2445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.575770725544907e-06,
+      "loss": 0.8599,
+      "step": 2446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.571145704539796e-06,
+      "loss": 0.8066,
+      "step": 2447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.566520267679038e-06,
+      "loss": 0.3655,
+      "step": 2448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.561894416758393e-06,
+      "loss": 0.8525,
+      "step": 2449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.557268153573774e-06,
+      "loss": 0.8564,
+      "step": 2450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.552641479921255e-06,
+      "loss": 0.8525,
+      "step": 2451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.54801439759707e-06,
+      "loss": 0.7773,
+      "step": 2452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.543386908397613e-06,
+      "loss": 0.8892,
+      "step": 2453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.53875901411943e-06,
+      "loss": 0.7871,
+      "step": 2454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.534130716559235e-06,
+      "loss": 0.8003,
+      "step": 2455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.529502017513888e-06,
+      "loss": 0.8252,
+      "step": 2456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.524872918780411e-06,
+      "loss": 0.8818,
+      "step": 2457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.520243422155982e-06,
+      "loss": 0.79,
+      "step": 2458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.515613529437926e-06,
+      "loss": 0.8599,
+      "step": 2459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.51098324242373e-06,
+      "loss": 0.3162,
+      "step": 2460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.506352562911028e-06,
+      "loss": 0.856,
+      "step": 2461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.501721492697615e-06,
+      "loss": 0.8022,
+      "step": 2462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.497090033581427e-06,
+      "loss": 0.7959,
+      "step": 2463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.492458187360556e-06,
+      "loss": 0.7725,
+      "step": 2464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.487825955833245e-06,
+      "loss": 0.8311,
+      "step": 2465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.48319334079789e-06,
+      "loss": 0.7939,
+      "step": 2466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.478560344053024e-06,
+      "loss": 0.8691,
+      "step": 2467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.473926967397338e-06,
+      "loss": 0.8159,
+      "step": 2468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.469293212629674e-06,
+      "loss": 0.8013,
+      "step": 2469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.464659081549012e-06,
+      "loss": 0.8306,
+      "step": 2470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.460024575954475e-06,
+      "loss": 0.8569,
+      "step": 2471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.455389697645348e-06,
+      "loss": 0.8198,
+      "step": 2472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.450754448421044e-06,
+      "loss": 0.8867,
+      "step": 2473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.446118830081124e-06,
+      "loss": 0.856,
+      "step": 2474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.441482844425302e-06,
+      "loss": 0.7905,
+      "step": 2475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.436846493253424e-06,
+      "loss": 0.8364,
+      "step": 2476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.432209778365479e-06,
+      "loss": 0.8262,
+      "step": 2477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.427572701561602e-06,
+      "loss": 0.8052,
+      "step": 2478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.422935264642066e-06,
+      "loss": 0.8267,
+      "step": 2479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.418297469407281e-06,
+      "loss": 0.7891,
+      "step": 2480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.4136593176578e-06,
+      "loss": 0.7974,
+      "step": 2481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.409020811194314e-06,
+      "loss": 0.8584,
+      "step": 2482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.404381951817653e-06,
+      "loss": 0.769,
+      "step": 2483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.399742741328779e-06,
+      "loss": 0.7363,
+      "step": 2484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.395103181528792e-06,
+      "loss": 0.7651,
+      "step": 2485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.39046327421893e-06,
+      "loss": 0.9053,
+      "step": 2486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.385823021200565e-06,
+      "loss": 0.7847,
+      "step": 2487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.381182424275206e-06,
+      "loss": 0.8569,
+      "step": 2488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.37654148524449e-06,
+      "loss": 0.8804,
+      "step": 2489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.371900205910188e-06,
+      "loss": 0.8071,
+      "step": 2490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.367258588074206e-06,
+      "loss": 0.8198,
+      "step": 2491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.362616633538578e-06,
+      "loss": 0.8418,
+      "step": 2492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.357974344105474e-06,
+      "loss": 0.8076,
+      "step": 2493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.353331721577188e-06,
+      "loss": 0.8228,
+      "step": 2494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.348688767756146e-06,
+      "loss": 0.8345,
+      "step": 2495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.344045484444904e-06,
+      "loss": 0.8711,
+      "step": 2496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.339401873446143e-06,
+      "loss": 0.8262,
+      "step": 2497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.334757936562676e-06,
+      "loss": 0.7935,
+      "step": 2498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.330113675597433e-06,
+      "loss": 0.8823,
+      "step": 2499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.32546909235348e-06,
+      "loss": 0.7935,
+      "step": 2500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.320824188634004e-06,
+      "loss": 0.8496,
+      "step": 2501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.316178966242316e-06,
+      "loss": 0.8076,
+      "step": 2502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.311533426981853e-06,
+      "loss": 0.8398,
+      "step": 2503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.30688757265617e-06,
+      "loss": 0.7432,
+      "step": 2504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.302241405068952e-06,
+      "loss": 0.8169,
+      "step": 2505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.297594926023994e-06,
+      "loss": 0.8232,
+      "step": 2506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.292948137325226e-06,
+      "loss": 0.3303,
+      "step": 2507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.28830104077669e-06,
+      "loss": 0.8379,
+      "step": 2508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.283653638182547e-06,
+      "loss": 0.7842,
+      "step": 2509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.279005931347081e-06,
+      "loss": 0.8892,
+      "step": 2510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.27435792207469e-06,
+      "loss": 0.8379,
+      "step": 2511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.26970961216989e-06,
+      "loss": 0.8906,
+      "step": 2512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.265061003437317e-06,
+      "loss": 0.8687,
+      "step": 2513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.260412097681722e-06,
+      "loss": 0.8525,
+      "step": 2514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.255762896707968e-06,
+      "loss": 0.8262,
+      "step": 2515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.251113402321038e-06,
+      "loss": 0.8584,
+      "step": 2516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.246463616326021e-06,
+      "loss": 0.8115,
+      "step": 2517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.241813540528128e-06,
+      "loss": 0.7695,
+      "step": 2518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.237163176732674e-06,
+      "loss": 0.8389,
+      "step": 2519
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.232512526745098e-06,
+      "loss": 0.8833,
+      "step": 2520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.227861592370936e-06,
+      "loss": 0.8433,
+      "step": 2521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.223210375415843e-06,
+      "loss": 0.7119,
+      "step": 2522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.218558877685582e-06,
+      "loss": 0.8403,
+      "step": 2523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.213907100986023e-06,
+      "loss": 0.7437,
+      "step": 2524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.209255047123148e-06,
+      "loss": 0.8228,
+      "step": 2525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.20460271790304e-06,
+      "loss": 0.8135,
+      "step": 2526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.1999501151319e-06,
+      "loss": 0.8608,
+      "step": 2527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.195297240616025e-06,
+      "loss": 0.8289,
+      "step": 2528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.190644096161822e-06,
+      "loss": 0.8818,
+      "step": 2529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.1859906835758e-06,
+      "loss": 0.8672,
+      "step": 2530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.181337004664577e-06,
+      "loss": 0.8223,
+      "step": 2531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.176683061234866e-06,
+      "loss": 0.8521,
+      "step": 2532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.172028855093495e-06,
+      "loss": 0.8379,
+      "step": 2533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.167374388047382e-06,
+      "loss": 0.8452,
+      "step": 2534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.162719661903555e-06,
+      "loss": 0.8442,
+      "step": 2535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.158064678469136e-06,
+      "loss": 0.8633,
+      "step": 2536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.153409439551351e-06,
+      "loss": 0.7954,
+      "step": 2537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.148753946957524e-06,
+      "loss": 0.8218,
+      "step": 2538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.144098202495076e-06,
+      "loss": 0.8584,
+      "step": 2539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.139442207971532e-06,
+      "loss": 0.8086,
+      "step": 2540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.134785965194505e-06,
+      "loss": 0.7925,
+      "step": 2541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.13012947597171e-06,
+      "loss": 0.8706,
+      "step": 2542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.125472742110956e-06,
+      "loss": 0.8247,
+      "step": 2543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.120815765420148e-06,
+      "loss": 0.8364,
+      "step": 2544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.116158547707284e-06,
+      "loss": 0.8667,
+      "step": 2545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.111501090780458e-06,
+      "loss": 0.8364,
+      "step": 2546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.106843396447855e-06,
+      "loss": 0.8467,
+      "step": 2547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.102185466517747e-06,
+      "loss": 0.8296,
+      "step": 2548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.09752730279851e-06,
+      "loss": 0.8472,
+      "step": 2549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.092868907098602e-06,
+      "loss": 0.7852,
+      "step": 2550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.08821028122657e-06,
+      "loss": 0.833,
+      "step": 2551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.083551426991051e-06,
+      "loss": 0.8384,
+      "step": 2552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.07889234620078e-06,
+      "loss": 0.8604,
+      "step": 2553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.074233040664566e-06,
+      "loss": 0.7422,
+      "step": 2554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.069573512191315e-06,
+      "loss": 0.8545,
+      "step": 2555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.064913762590017e-06,
+      "loss": 0.8408,
+      "step": 2556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.060253793669744e-06,
+      "loss": 0.8374,
+      "step": 2557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.055593607239657e-06,
+      "loss": 0.3219,
+      "step": 2558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.050933205109004e-06,
+      "loss": 0.8447,
+      "step": 2559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.046272589087111e-06,
+      "loss": 0.8076,
+      "step": 2560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.041611760983391e-06,
+      "loss": 0.8423,
+      "step": 2561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.036950722607337e-06,
+      "loss": 0.8086,
+      "step": 2562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.032289475768524e-06,
+      "loss": 0.8291,
+      "step": 2563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.02762802227661e-06,
+      "loss": 0.7529,
+      "step": 2564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.022966363941329e-06,
+      "loss": 0.8252,
+      "step": 2565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.018304502572502e-06,
+      "loss": 0.7837,
+      "step": 2566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.01364243998002e-06,
+      "loss": 0.8179,
+      "step": 2567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.008980177973858e-06,
+      "loss": 0.8857,
+      "step": 2568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.004317718364066e-06,
+      "loss": 0.8271,
+      "step": 2569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.999655062960771e-06,
+      "loss": 0.8843,
+      "step": 2570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.994992213574178e-06,
+      "loss": 0.8296,
+      "step": 2571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.990329172014563e-06,
+      "loss": 0.8345,
+      "step": 2572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.985665940092281e-06,
+      "loss": 0.8247,
+      "step": 2573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.98100251961776e-06,
+      "loss": 0.8174,
+      "step": 2574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.976338912401501e-06,
+      "loss": 0.7866,
+      "step": 2575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.971675120254072e-06,
+      "loss": 0.7905,
+      "step": 2576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.967011144986125e-06,
+      "loss": 0.7837,
+      "step": 2577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.962346988408368e-06,
+      "loss": 0.8833,
+      "step": 2578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.957682652331593e-06,
+      "loss": 0.8174,
+      "step": 2579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.953018138566654e-06,
+      "loss": 0.812,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.948353448924475e-06,
+      "loss": 0.769,
+      "step": 2581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.943688585216051e-06,
+      "loss": 0.7935,
+      "step": 2582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.93902354925244e-06,
+      "loss": 0.8193,
+      "step": 2583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.934358342844773e-06,
+      "loss": 0.7861,
+      "step": 2584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.929692967804241e-06,
+      "loss": 0.8511,
+      "step": 2585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.925027425942102e-06,
+      "loss": 0.8438,
+      "step": 2586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.920361719069684e-06,
+      "loss": 0.8218,
+      "step": 2587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.915695848998374e-06,
+      "loss": 0.8081,
+      "step": 2588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.911029817539623e-06,
+      "loss": 0.835,
+      "step": 2589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.906363626504945e-06,
+      "loss": 0.7891,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.901697277705912e-06,
+      "loss": 0.7793,
+      "step": 2591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.897030772954166e-06,
+      "loss": 0.814,
+      "step": 2592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.892364114061405e-06,
+      "loss": 0.8882,
+      "step": 2593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.887697302839386e-06,
+      "loss": 0.7881,
+      "step": 2594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.883030341099925e-06,
+      "loss": 0.8579,
+      "step": 2595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.878363230654898e-06,
+      "loss": 0.8623,
+      "step": 2596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.873695973316237e-06,
+      "loss": 0.8755,
+      "step": 2597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.869028570895931e-06,
+      "loss": 0.8125,
+      "step": 2598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.864361025206029e-06,
+      "loss": 0.8135,
+      "step": 2599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.859693338058631e-06,
+      "loss": 0.7803,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.855025511265896e-06,
+      "loss": 0.833,
+      "step": 2601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.850357546640032e-06,
+      "loss": 0.8438,
+      "step": 2602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.845689445993306e-06,
+      "loss": 0.3213,
+      "step": 2603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.841021211138033e-06,
+      "loss": 0.8809,
+      "step": 2604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.836352843886586e-06,
+      "loss": 0.8057,
+      "step": 2605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.831684346051385e-06,
+      "loss": 0.8042,
+      "step": 2606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.827015719444897e-06,
+      "loss": 0.8589,
+      "step": 2607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.822346965879653e-06,
+      "loss": 0.8457,
+      "step": 2608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.817678087168213e-06,
+      "loss": 0.8271,
+      "step": 2609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.813009085123202e-06,
+      "loss": 0.9023,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.808339961557288e-06,
+      "loss": 0.7358,
+      "step": 2611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.803670718283185e-06,
+      "loss": 0.7988,
+      "step": 2612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.799001357113653e-06,
+      "loss": 0.811,
+      "step": 2613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.794331879861499e-06,
+      "loss": 0.7769,
+      "step": 2614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.789662288339577e-06,
+      "loss": 0.8262,
+      "step": 2615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.784992584360779e-06,
+      "loss": 0.3169,
+      "step": 2616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.780322769738046e-06,
+      "loss": 0.854,
+      "step": 2617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.775652846284366e-06,
+      "loss": 0.7964,
+      "step": 2618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.770982815812758e-06,
+      "loss": 0.877,
+      "step": 2619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.76631268013629e-06,
+      "loss": 0.8979,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.761642441068072e-06,
+      "loss": 0.7866,
+      "step": 2621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.756972100421249e-06,
+      "loss": 0.7637,
+      "step": 2622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.752301660009007e-06,
+      "loss": 0.8057,
+      "step": 2623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.747631121644575e-06,
+      "loss": 0.8506,
+      "step": 2624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.742960487141215e-06,
+      "loss": 0.8247,
+      "step": 2625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.73828975831223e-06,
+      "loss": 0.8247,
+      "step": 2626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.733618936970954e-06,
+      "loss": 0.8281,
+      "step": 2627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.728948024930764e-06,
+      "loss": 0.3306,
+      "step": 2628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.72427702400507e-06,
+      "loss": 0.7866,
+      "step": 2629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.719605936007309e-06,
+      "loss": 0.7974,
+      "step": 2630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.714934762750964e-06,
+      "loss": 0.2939,
+      "step": 2631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.710263506049545e-06,
+      "loss": 0.7852,
+      "step": 2632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.705592167716592e-06,
+      "loss": 0.8657,
+      "step": 2633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.700920749565682e-06,
+      "loss": 0.813,
+      "step": 2634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.696249253410418e-06,
+      "loss": 0.7793,
+      "step": 2635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.691577681064439e-06,
+      "loss": 0.8091,
+      "step": 2636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.686906034341403e-06,
+      "loss": 0.7881,
+      "step": 2637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.682234315055015e-06,
+      "loss": 0.8701,
+      "step": 2638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.677562525018987e-06,
+      "loss": 0.8223,
+      "step": 2639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.672890666047074e-06,
+      "loss": 0.8267,
+      "step": 2640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.668218739953054e-06,
+      "loss": 0.7808,
+      "step": 2641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.663546748550727e-06,
+      "loss": 0.8994,
+      "step": 2642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.658874693653916e-06,
+      "loss": 0.8311,
+      "step": 2643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.654202577076484e-06,
+      "loss": 0.8809,
+      "step": 2644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.6495304006323e-06,
+      "loss": 0.8027,
+      "step": 2645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.644858166135267e-06,
+      "loss": 0.8091,
+      "step": 2646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.640185875399306e-06,
+      "loss": 0.7676,
+      "step": 2647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.635513530238362e-06,
+      "loss": 0.8379,
+      "step": 2648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.630841132466399e-06,
+      "loss": 0.7891,
+      "step": 2649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.626168683897403e-06,
+      "loss": 0.8062,
+      "step": 2650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.6214961863453795e-06,
+      "loss": 0.7988,
+      "step": 2651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.616823641624354e-06,
+      "loss": 0.8003,
+      "step": 2652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.612151051548368e-06,
+      "loss": 0.8149,
+      "step": 2653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.607478417931484e-06,
+      "loss": 0.8374,
+      "step": 2654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.6028057425877775e-06,
+      "loss": 0.8188,
+      "step": 2655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.59813302733134e-06,
+      "loss": 0.3003,
+      "step": 2656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.593460273976283e-06,
+      "loss": 0.7837,
+      "step": 2657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.58878748433673e-06,
+      "loss": 0.8452,
+      "step": 2658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.5841146602268195e-06,
+      "loss": 0.8145,
+      "step": 2659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.579441803460699e-06,
+      "loss": 0.314,
+      "step": 2660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.574768915852538e-06,
+      "loss": 0.8398,
+      "step": 2661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.5700959992165086e-06,
+      "loss": 0.7983,
+      "step": 2662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.565423055366798e-06,
+      "loss": 0.3462,
+      "step": 2663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.560750086117605e-06,
+      "loss": 0.8047,
+      "step": 2664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.556077093283138e-06,
+      "loss": 0.7759,
+      "step": 2665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.551404078677611e-06,
+      "loss": 0.8511,
+      "step": 2666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.546731044115251e-06,
+      "loss": 0.8037,
+      "step": 2667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.542057991410294e-06,
+      "loss": 0.875,
+      "step": 2668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.537384922376972e-06,
+      "loss": 0.79,
+      "step": 2669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.532711838829538e-06,
+      "loss": 0.8413,
+      "step": 2670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.528038742582243e-06,
+      "loss": 0.811,
+      "step": 2671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.5233656354493415e-06,
+      "loss": 0.8364,
+      "step": 2672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.518692519245094e-06,
+      "loss": 0.3306,
+      "step": 2673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.514019395783771e-06,
+      "loss": 0.8218,
+      "step": 2674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.509346266879634e-06,
+      "loss": 0.8311,
+      "step": 2675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.504673134346952e-06,
+      "loss": 0.8262,
+      "step": 2676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.5e-06,
+      "loss": 0.8701,
+      "step": 2677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.4953268656530476e-06,
+      "loss": 0.8096,
+      "step": 2678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.490653733120369e-06,
+      "loss": 0.8057,
+      "step": 2679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.485980604216231e-06,
+      "loss": 0.7715,
+      "step": 2680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.481307480754905e-06,
+      "loss": 0.8115,
+      "step": 2681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.4766343645506615e-06,
+      "loss": 0.8945,
+      "step": 2682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.471961257417759e-06,
+      "loss": 0.7695,
+      "step": 2683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.467288161170461e-06,
+      "loss": 0.9102,
+      "step": 2684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.462615077623029e-06,
+      "loss": 0.8232,
+      "step": 2685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.457942008589708e-06,
+      "loss": 0.8066,
+      "step": 2686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.453268955884748e-06,
+      "loss": 0.8613,
+      "step": 2687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.448595921322391e-06,
+      "loss": 0.7886,
+      "step": 2688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.443922906716863e-06,
+      "loss": 0.8267,
+      "step": 2689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.439249913882397e-06,
+      "loss": 0.8071,
+      "step": 2690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.434576944633203e-06,
+      "loss": 0.8628,
+      "step": 2691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.429904000783492e-06,
+      "loss": 0.812,
+      "step": 2692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.425231084147464e-06,
+      "loss": 0.8262,
+      "step": 2693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.420558196539303e-06,
+      "loss": 0.8267,
+      "step": 2694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.4158853397731826e-06,
+      "loss": 0.7893,
+      "step": 2695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.411212515663273e-06,
+      "loss": 0.8428,
+      "step": 2696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.406539726023718e-06,
+      "loss": 0.7656,
+      "step": 2697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.401866972668661e-06,
+      "loss": 0.7422,
+      "step": 2698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.397194257412226e-06,
+      "loss": 0.8491,
+      "step": 2699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.392521582068517e-06,
+      "loss": 0.8291,
+      "step": 2700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.387848948451631e-06,
+      "loss": 0.8682,
+      "step": 2701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.383176358375648e-06,
+      "loss": 0.8608,
+      "step": 2702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.378503813654622e-06,
+      "loss": 0.8296,
+      "step": 2703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.373831316102597e-06,
+      "loss": 0.7974,
+      "step": 2704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.369158867533603e-06,
+      "loss": 0.8296,
+      "step": 2705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.364486469761639e-06,
+      "loss": 0.811,
+      "step": 2706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.359814124600695e-06,
+      "loss": 0.8486,
+      "step": 2707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.355141833864736e-06,
+      "loss": 0.8408,
+      "step": 2708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.350469599367701e-06,
+      "loss": 0.7871,
+      "step": 2709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.345797422923516e-06,
+      "loss": 0.8188,
+      "step": 2710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.341125306346084e-06,
+      "loss": 0.8169,
+      "step": 2711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.336453251449276e-06,
+      "loss": 0.833,
+      "step": 2712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.331781260046947e-06,
+      "loss": 0.8257,
+      "step": 2713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.327109333952928e-06,
+      "loss": 0.7954,
+      "step": 2714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.322437474981014e-06,
+      "loss": 0.8115,
+      "step": 2715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.31776568494499e-06,
+      "loss": 0.8096,
+      "step": 2716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.313093965658597e-06,
+      "loss": 0.7842,
+      "step": 2717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.308422318935564e-06,
+      "loss": 0.8389,
+      "step": 2718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.303750746589584e-06,
+      "loss": 0.8125,
+      "step": 2719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.299079250434319e-06,
+      "loss": 0.8281,
+      "step": 2720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.294407832283409e-06,
+      "loss": 0.7769,
+      "step": 2721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.289736493950458e-06,
+      "loss": 0.8594,
+      "step": 2722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.285065237249037e-06,
+      "loss": 0.8052,
+      "step": 2723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.280394063992691e-06,
+      "loss": 0.8032,
+      "step": 2724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.275722975994934e-06,
+      "loss": 0.7769,
+      "step": 2725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.271051975069236e-06,
+      "loss": 0.791,
+      "step": 2726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.2663810630290465e-06,
+      "loss": 0.8428,
+      "step": 2727
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.261710241687774e-06,
+      "loss": 0.8296,
+      "step": 2728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.257039512858786e-06,
+      "loss": 0.7939,
+      "step": 2729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.252368878355425e-06,
+      "loss": 0.875,
+      "step": 2730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.247698339990995e-06,
+      "loss": 0.8052,
+      "step": 2731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.2430278995787535e-06,
+      "loss": 0.8613,
+      "step": 2732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.23835755893193e-06,
+      "loss": 0.8091,
+      "step": 2733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.233687319863712e-06,
+      "loss": 0.8433,
+      "step": 2734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.229017184187244e-06,
+      "loss": 0.8052,
+      "step": 2735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.224347153715634e-06,
+      "loss": 0.8042,
+      "step": 2736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.219677230261954e-06,
+      "loss": 0.8027,
+      "step": 2737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.215007415639222e-06,
+      "loss": 0.771,
+      "step": 2738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.210337711660424e-06,
+      "loss": 0.7944,
+      "step": 2739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.205668120138503e-06,
+      "loss": 0.8071,
+      "step": 2740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.200998642886347e-06,
+      "loss": 0.8271,
+      "step": 2741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.1963292817168176e-06,
+      "loss": 0.7959,
+      "step": 2742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.191660038442713e-06,
+      "loss": 0.3696,
+      "step": 2743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.1869909148767975e-06,
+      "loss": 0.8345,
+      "step": 2744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.1823219128317894e-06,
+      "loss": 0.8726,
+      "step": 2745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.17765303412035e-06,
+      "loss": 0.8027,
+      "step": 2746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.172984280555103e-06,
+      "loss": 0.8159,
+      "step": 2747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.16831565394862e-06,
+      "loss": 0.8428,
+      "step": 2748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.163647156113415e-06,
+      "loss": 0.8394,
+      "step": 2749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.158978788861967e-06,
+      "loss": 0.8511,
+      "step": 2750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.154310554006697e-06,
+      "loss": 0.7671,
+      "step": 2751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.149642453359969e-06,
+      "loss": 0.8379,
+      "step": 2752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.1449744887341065e-06,
+      "loss": 0.8057,
+      "step": 2753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.140306661941372e-06,
+      "loss": 0.7866,
+      "step": 2754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.135638974793973e-06,
+      "loss": 0.8604,
+      "step": 2755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.1309714291040694e-06,
+      "loss": 0.8691,
+      "step": 2756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.126304026683766e-06,
+      "loss": 0.8652,
+      "step": 2757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.121636769345103e-06,
+      "loss": 0.7837,
+      "step": 2758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.1169696589000755e-06,
+      "loss": 0.7954,
+      "step": 2759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.1123026971606165e-06,
+      "loss": 0.8262,
+      "step": 2760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.107635885938595e-06,
+      "loss": 0.792,
+      "step": 2761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.102969227045833e-06,
+      "loss": 0.3276,
+      "step": 2762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.09830272229409e-06,
+      "loss": 0.8423,
+      "step": 2763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.0936363734950576e-06,
+      "loss": 0.8652,
+      "step": 2764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.08897018246038e-06,
+      "loss": 0.7402,
+      "step": 2765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.084304151001627e-06,
+      "loss": 0.8296,
+      "step": 2766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.0796382809303165e-06,
+      "loss": 0.7993,
+      "step": 2767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.0749725740579005e-06,
+      "loss": 0.874,
+      "step": 2768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.070307032195762e-06,
+      "loss": 0.8154,
+      "step": 2769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.0656416571552285e-06,
+      "loss": 0.8369,
+      "step": 2770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.060976450747562e-06,
+      "loss": 0.8145,
+      "step": 2771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.056311414783951e-06,
+      "loss": 0.8311,
+      "step": 2772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.051646551075526e-06,
+      "loss": 0.8022,
+      "step": 2773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.0469818614333495e-06,
+      "loss": 0.8623,
+      "step": 2774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.042317347668409e-06,
+      "loss": 0.8794,
+      "step": 2775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.0376530115916325e-06,
+      "loss": 0.3279,
+      "step": 2776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.032988855013879e-06,
+      "loss": 0.7832,
+      "step": 2777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.028324879745929e-06,
+      "loss": 0.7827,
+      "step": 2778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.023661087598501e-06,
+      "loss": 0.8398,
+      "step": 2779
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.0189974803822425e-06,
+      "loss": 0.8315,
+      "step": 2780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.01433405990772e-06,
+      "loss": 0.8491,
+      "step": 2781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.0096708279854374e-06,
+      "loss": 0.8267,
+      "step": 2782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.005007786425825e-06,
+      "loss": 0.8271,
+      "step": 2783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.00034493703923e-06,
+      "loss": 0.8037,
+      "step": 2784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.9956822816359364e-06,
+      "loss": 0.834,
+      "step": 2785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.991019822026144e-06,
+      "loss": 0.8281,
+      "step": 2786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.9863575600199814e-06,
+      "loss": 0.7905,
+      "step": 2787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.981695497427499e-06,
+      "loss": 0.812,
+      "step": 2788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.977033636058672e-06,
+      "loss": 0.8447,
+      "step": 2789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.972371977723391e-06,
+      "loss": 0.8457,
+      "step": 2790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.967710524231479e-06,
+      "loss": 0.8032,
+      "step": 2791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.963049277392665e-06,
+      "loss": 0.8608,
+      "step": 2792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.958388239016611e-06,
+      "loss": 0.8042,
+      "step": 2793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.953727410912892e-06,
+      "loss": 0.8281,
+      "step": 2794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.949066794890998e-06,
+      "loss": 0.8423,
+      "step": 2795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.944406392760343e-06,
+      "loss": 0.7905,
+      "step": 2796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.939746206330259e-06,
+      "loss": 0.8223,
+      "step": 2797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.935086237409985e-06,
+      "loss": 0.8481,
+      "step": 2798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.930426487808685e-06,
+      "loss": 0.8267,
+      "step": 2799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.925766959335436e-06,
+      "loss": 0.813,
+      "step": 2800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.9211076537992216e-06,
+      "loss": 0.7773,
+      "step": 2801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.916448573008948e-06,
+      "loss": 0.8433,
+      "step": 2802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.911789718773433e-06,
+      "loss": 0.8184,
+      "step": 2803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.9071310929014e-06,
+      "loss": 0.8184,
+      "step": 2804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.90247269720149e-06,
+      "loss": 0.8359,
+      "step": 2805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.897814533482253e-06,
+      "loss": 0.7974,
+      "step": 2806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.893156603552148e-06,
+      "loss": 0.3295,
+      "step": 2807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.888498909219542e-06,
+      "loss": 0.7847,
+      "step": 2808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.883841452292717e-06,
+      "loss": 0.7852,
+      "step": 2809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.879184234579852e-06,
+      "loss": 0.8252,
+      "step": 2810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.874527257889045e-06,
+      "loss": 0.8447,
+      "step": 2811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.869870524028291e-06,
+      "loss": 0.8071,
+      "step": 2812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.865214034805496e-06,
+      "loss": 0.7988,
+      "step": 2813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.860557792028467e-06,
+      "loss": 0.7559,
+      "step": 2814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.8559017975049235e-06,
+      "loss": 0.7998,
+      "step": 2815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.851246053042477e-06,
+      "loss": 0.8755,
+      "step": 2816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.846590560448652e-06,
+      "loss": 0.8184,
+      "step": 2817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.841935321530865e-06,
+      "loss": 0.8354,
+      "step": 2818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.837280338096447e-06,
+      "loss": 0.791,
+      "step": 2819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.832625611952621e-06,
+      "loss": 0.8062,
+      "step": 2820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.827971144906507e-06,
+      "loss": 0.8438,
+      "step": 2821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.823316938765134e-06,
+      "loss": 0.8311,
+      "step": 2822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.818662995335427e-06,
+      "loss": 0.8965,
+      "step": 2823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.8140093164242015e-06,
+      "loss": 0.875,
+      "step": 2824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.809355903838179e-06,
+      "loss": 0.8604,
+      "step": 2825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.804702759383976e-06,
+      "loss": 0.312,
+      "step": 2826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.800049884868101e-06,
+      "loss": 0.7837,
+      "step": 2827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.7953972820969585e-06,
+      "loss": 0.7603,
+      "step": 2828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.790744952876855e-06,
+      "loss": 0.834,
+      "step": 2829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.786092899013978e-06,
+      "loss": 0.749,
+      "step": 2830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.781441122314419e-06,
+      "loss": 0.8008,
+      "step": 2831
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.776789624584158e-06,
+      "loss": 0.79,
+      "step": 2832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.772138407629066e-06,
+      "loss": 0.813,
+      "step": 2833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.7674874732549025e-06,
+      "loss": 0.8677,
+      "step": 2834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.762836823267326e-06,
+      "loss": 0.8062,
+      "step": 2835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.758186459471874e-06,
+      "loss": 0.8018,
+      "step": 2836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.75353638367398e-06,
+      "loss": 0.8213,
+      "step": 2837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.748886597678964e-06,
+      "loss": 0.8301,
+      "step": 2838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.7442371032920326e-06,
+      "loss": 0.8159,
+      "step": 2839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.7395879023182804e-06,
+      "loss": 0.7844,
+      "step": 2840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.734938996562685e-06,
+      "loss": 0.8433,
+      "step": 2841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.730290387830111e-06,
+      "loss": 0.8774,
+      "step": 2842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.725642077925314e-06,
+      "loss": 0.8428,
+      "step": 2843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.720994068652921e-06,
+      "loss": 0.8247,
+      "step": 2844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.716346361817454e-06,
+      "loss": 0.813,
+      "step": 2845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.711698959223311e-06,
+      "loss": 0.8081,
+      "step": 2846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.707051862674775e-06,
+      "loss": 0.8164,
+      "step": 2847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.702405073976006e-06,
+      "loss": 0.8032,
+      "step": 2848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.697758594931052e-06,
+      "loss": 0.8101,
+      "step": 2849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.693112427343831e-06,
+      "loss": 0.8081,
+      "step": 2850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.688466573018149e-06,
+      "loss": 0.8237,
+      "step": 2851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.683821033757684e-06,
+      "loss": 0.812,
+      "step": 2852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.679175811365997e-06,
+      "loss": 0.8369,
+      "step": 2853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.67453090764652e-06,
+      "loss": 0.8208,
+      "step": 2854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.66988632440257e-06,
+      "loss": 0.7803,
+      "step": 2855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.665242063437326e-06,
+      "loss": 0.8101,
+      "step": 2856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.660598126553857e-06,
+      "loss": 0.8164,
+      "step": 2857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.655954515555097e-06,
+      "loss": 0.834,
+      "step": 2858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.651311232243855e-06,
+      "loss": 0.7832,
+      "step": 2859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.646668278422812e-06,
+      "loss": 0.3428,
+      "step": 2860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.6420256558945275e-06,
+      "loss": 0.8833,
+      "step": 2861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.637383366461422e-06,
+      "loss": 0.7832,
+      "step": 2862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.632741411925794e-06,
+      "loss": 0.8315,
+      "step": 2863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.6280997940898135e-06,
+      "loss": 0.8003,
+      "step": 2864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.623458514755512e-06,
+      "loss": 0.8262,
+      "step": 2865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.618817575724797e-06,
+      "loss": 0.7715,
+      "step": 2866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.614176978799435e-06,
+      "loss": 0.8579,
+      "step": 2867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.60953672578107e-06,
+      "loss": 0.8462,
+      "step": 2868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.604896818471211e-06,
+      "loss": 0.8281,
+      "step": 2869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.6002572586712235e-06,
+      "loss": 0.8135,
+      "step": 2870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.595618048182349e-06,
+      "loss": 0.8105,
+      "step": 2871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.5909791888056854e-06,
+      "loss": 0.8057,
+      "step": 2872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.586340682342201e-06,
+      "loss": 0.8623,
+      "step": 2873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.5817025305927194e-06,
+      "loss": 0.7852,
+      "step": 2874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.577064735357937e-06,
+      "loss": 0.8286,
+      "step": 2875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.5724272984383985e-06,
+      "loss": 0.8574,
+      "step": 2876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.567790221634521e-06,
+      "loss": 0.8115,
+      "step": 2877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.563153506746577e-06,
+      "loss": 0.8066,
+      "step": 2878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.558517155574698e-06,
+      "loss": 0.8081,
+      "step": 2879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.553881169918875e-06,
+      "loss": 0.8022,
+      "step": 2880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.549245551578959e-06,
+      "loss": 0.8447,
+      "step": 2881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.544610302354654e-06,
+      "loss": 0.8081,
+      "step": 2882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.539975424045523e-06,
+      "loss": 0.8013,
+      "step": 2883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.535340918450991e-06,
+      "loss": 0.8418,
+      "step": 2884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.530706787370326e-06,
+      "loss": 0.8643,
+      "step": 2885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.526073032602659e-06,
+      "loss": 0.7578,
+      "step": 2886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.521439655946979e-06,
+      "loss": 0.8135,
+      "step": 2887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.516806659202112e-06,
+      "loss": 0.8506,
+      "step": 2888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.512174044166752e-06,
+      "loss": 0.8394,
+      "step": 2889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.507541812639445e-06,
+      "loss": 0.8252,
+      "step": 2890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.502909966418575e-06,
+      "loss": 0.8213,
+      "step": 2891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.498278507302387e-06,
+      "loss": 0.8242,
+      "step": 2892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.493647437088972e-06,
+      "loss": 0.8286,
+      "step": 2893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.489016757576271e-06,
+      "loss": 0.8755,
+      "step": 2894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.484386470562076e-06,
+      "loss": 0.7544,
+      "step": 2895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.479756577844021e-06,
+      "loss": 0.856,
+      "step": 2896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.4751270812195886e-06,
+      "loss": 0.8096,
+      "step": 2897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.470497982486114e-06,
+      "loss": 0.8096,
+      "step": 2898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.465869283440767e-06,
+      "loss": 0.7729,
+      "step": 2899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.4612409858805695e-06,
+      "loss": 0.8262,
+      "step": 2900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.456613091602391e-06,
+      "loss": 0.8267,
+      "step": 2901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.4519856024029305e-06,
+      "loss": 0.8179,
+      "step": 2902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.447358520078743e-06,
+      "loss": 0.7637,
+      "step": 2903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.442731846426227e-06,
+      "loss": 0.8496,
+      "step": 2904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.438105583241607e-06,
+      "loss": 0.3066,
+      "step": 2905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.433479732320962e-06,
+      "loss": 0.8052,
+      "step": 2906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.428854295460208e-06,
+      "loss": 0.8179,
+      "step": 2907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.424229274455095e-06,
+      "loss": 0.8633,
+      "step": 2908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.419604671101216e-06,
+      "loss": 0.8462,
+      "step": 2909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.414980487194008e-06,
+      "loss": 0.8931,
+      "step": 2910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.4103567245287305e-06,
+      "loss": 0.8042,
+      "step": 2911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.405733384900489e-06,
+      "loss": 0.8755,
+      "step": 2912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.401110470104226e-06,
+      "loss": 0.8652,
+      "step": 2913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.396487981934712e-06,
+      "loss": 0.8159,
+      "step": 2914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.391865922186562e-06,
+      "loss": 0.7788,
+      "step": 2915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.387244292654212e-06,
+      "loss": 0.8296,
+      "step": 2916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.3826230951319424e-06,
+      "loss": 0.8003,
+      "step": 2917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.37800233141386e-06,
+      "loss": 0.8213,
+      "step": 2918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.373382003293903e-06,
+      "loss": 0.8018,
+      "step": 2919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.368762112565842e-06,
+      "loss": 0.8296,
+      "step": 2920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.364142661023282e-06,
+      "loss": 0.8379,
+      "step": 2921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.3595236504596465e-06,
+      "loss": 0.7734,
+      "step": 2922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.354905082668197e-06,
+      "loss": 0.8149,
+      "step": 2923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.350286959442025e-06,
+      "loss": 0.8188,
+      "step": 2924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.3456692825740396e-06,
+      "loss": 0.8682,
+      "step": 2925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.341052053856981e-06,
+      "loss": 0.8589,
+      "step": 2926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.336435275083423e-06,
+      "loss": 0.8242,
+      "step": 2927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.33181894804575e-06,
+      "loss": 0.8325,
+      "step": 2928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.327203074536183e-06,
+      "loss": 0.7852,
+      "step": 2929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.322587656346767e-06,
+      "loss": 0.8359,
+      "step": 2930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.31797269526936e-06,
+      "loss": 0.8159,
+      "step": 2931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.313358193095649e-06,
+      "loss": 0.8179,
+      "step": 2932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.308744151617149e-06,
+      "loss": 0.8643,
+      "step": 2933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.304130572625182e-06,
+      "loss": 0.7905,
+      "step": 2934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.299517457910902e-06,
+      "loss": 0.8169,
+      "step": 2935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.294904809265282e-06,
+      "loss": 0.834,
+      "step": 2936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.290292628479105e-06,
+      "loss": 0.8735,
+      "step": 2937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.28568091734298e-06,
+      "loss": 0.8218,
+      "step": 2938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.281069677647336e-06,
+      "loss": 0.8579,
+      "step": 2939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.27645891118241e-06,
+      "loss": 0.8623,
+      "step": 2940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.271848619738265e-06,
+      "loss": 0.8022,
+      "step": 2941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.267238805104769e-06,
+      "loss": 0.8179,
+      "step": 2942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.262629469071615e-06,
+      "loss": 0.9106,
+      "step": 2943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.258020613428307e-06,
+      "loss": 0.7725,
+      "step": 2944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.2534122399641574e-06,
+      "loss": 0.8535,
+      "step": 2945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.248804350468297e-06,
+      "loss": 0.3158,
+      "step": 2946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.24419694672967e-06,
+      "loss": 0.8535,
+      "step": 2947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.239590030537025e-06,
+      "loss": 0.8276,
+      "step": 2948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.234983603678926e-06,
+      "loss": 0.8354,
+      "step": 2949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.2303776679437515e-06,
+      "loss": 0.8188,
+      "step": 2950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.225772225119679e-06,
+      "loss": 0.8491,
+      "step": 2951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.221167276994701e-06,
+      "loss": 0.3225,
+      "step": 2952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.21656282535662e-06,
+      "loss": 0.7817,
+      "step": 2953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.211958871993037e-06,
+      "loss": 0.7964,
+      "step": 2954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.2073554186913685e-06,
+      "loss": 0.8345,
+      "step": 2955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.202752467238836e-06,
+      "loss": 0.8027,
+      "step": 2956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.198150019422461e-06,
+      "loss": 0.8618,
+      "step": 2957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.19354807702907e-06,
+      "loss": 0.8472,
+      "step": 2958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.188946641845302e-06,
+      "loss": 0.853,
+      "step": 2959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.184345715657585e-06,
+      "loss": 0.8218,
+      "step": 2960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.17974530025216e-06,
+      "loss": 0.8201,
+      "step": 2961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.1751453974150705e-06,
+      "loss": 0.7866,
+      "step": 2962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.170546008932153e-06,
+      "loss": 0.8706,
+      "step": 2963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.165947136589046e-06,
+      "loss": 0.8135,
+      "step": 2964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.161348782171198e-06,
+      "loss": 0.8037,
+      "step": 2965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.15675094746384e-06,
+      "loss": 0.8359,
+      "step": 2966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.152153634252017e-06,
+      "loss": 0.7136,
+      "step": 2967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.147556844320558e-06,
+      "loss": 0.7534,
+      "step": 2968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.1429605794540975e-06,
+      "loss": 0.7939,
+      "step": 2969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.1383648414370675e-06,
+      "loss": 0.8164,
+      "step": 2970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.133769632053687e-06,
+      "loss": 0.8203,
+      "step": 2971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.129174953087975e-06,
+      "loss": 0.8228,
+      "step": 2972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.124580806323748e-06,
+      "loss": 0.7915,
+      "step": 2973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.119987193544605e-06,
+      "loss": 0.7944,
+      "step": 2974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.1153941165339494e-06,
+      "loss": 0.7822,
+      "step": 2975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.110801577074975e-06,
+      "loss": 0.8091,
+      "step": 2976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.106209576950657e-06,
+      "loss": 0.8569,
+      "step": 2977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.101618117943771e-06,
+      "loss": 0.7944,
+      "step": 2978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.097027201836881e-06,
+      "loss": 0.79,
+      "step": 2979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.092436830412333e-06,
+      "loss": 0.7974,
+      "step": 2980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.087847005452273e-06,
+      "loss": 0.7861,
+      "step": 2981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.08325772873863e-06,
+      "loss": 0.7935,
+      "step": 2982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.078669002053116e-06,
+      "loss": 0.8032,
+      "step": 2983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.074080827177231e-06,
+      "loss": 0.7876,
+      "step": 2984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.069493205892266e-06,
+      "loss": 0.7803,
+      "step": 2985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.064906139979291e-06,
+      "loss": 0.772,
+      "step": 2986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.060319631219165e-06,
+      "loss": 0.8315,
+      "step": 2987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.055733681392529e-06,
+      "loss": 0.7832,
+      "step": 2988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.051148292279803e-06,
+      "loss": 0.3364,
+      "step": 2989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.046563465661199e-06,
+      "loss": 0.8076,
+      "step": 2990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.041979203316699e-06,
+      "loss": 0.8003,
+      "step": 2991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.037395507026071e-06,
+      "loss": 0.8096,
+      "step": 2992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.032812378568871e-06,
+      "loss": 0.7974,
+      "step": 2993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.028229819724417e-06,
+      "loss": 0.8555,
+      "step": 2994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.023647832271824e-06,
+      "loss": 0.7739,
+      "step": 2995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.019066417989976e-06,
+      "loss": 0.7705,
+      "step": 2996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.014485578657534e-06,
+      "loss": 0.853,
+      "step": 2997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.009905316052935e-06,
+      "loss": 0.7651,
+      "step": 2998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.005325631954403e-06,
+      "loss": 0.8423,
+      "step": 2999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.00074652813992e-06,
+      "loss": 0.8423,
+      "step": 3000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.9961680063872555e-06,
+      "loss": 0.8354,
+      "step": 3001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.9915900684739525e-06,
+      "loss": 0.8223,
+      "step": 3002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.98701271617732e-06,
+      "loss": 0.8154,
+      "step": 3003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.982435951274441e-06,
+      "loss": 0.8149,
+      "step": 3004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.977859775542183e-06,
+      "loss": 0.7847,
+      "step": 3005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.973284190757164e-06,
+      "loss": 0.8438,
+      "step": 3006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.968709198695788e-06,
+      "loss": 0.8057,
+      "step": 3007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.964134801134231e-06,
+      "loss": 0.3301,
+      "step": 3008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.95956099984842e-06,
+      "loss": 0.7666,
+      "step": 3009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.954987796614069e-06,
+      "loss": 0.8462,
+      "step": 3010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.950415193206655e-06,
+      "loss": 0.8203,
+      "step": 3011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.945843191401412e-06,
+      "loss": 0.7339,
+      "step": 3012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.941271792973355e-06,
+      "loss": 0.7534,
+      "step": 3013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.936700999697259e-06,
+      "loss": 0.8711,
+      "step": 3014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.932130813347657e-06,
+      "loss": 0.7607,
+      "step": 3015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.9275612356988605e-06,
+      "loss": 0.8271,
+      "step": 3016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.92299226852493e-06,
+      "loss": 0.7744,
+      "step": 3017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.918423913599699e-06,
+      "loss": 0.791,
+      "step": 3018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.913856172696763e-06,
+      "loss": 0.8037,
+      "step": 3019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.909289047589469e-06,
+      "loss": 0.7964,
+      "step": 3020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.904722540050938e-06,
+      "loss": 0.7827,
+      "step": 3021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.900156651854045e-06,
+      "loss": 0.8218,
+      "step": 3022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.895591384771423e-06,
+      "loss": 0.8105,
+      "step": 3023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.891026740575467e-06,
+      "loss": 0.8325,
+      "step": 3024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.886462721038331e-06,
+      "loss": 0.834,
+      "step": 3025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.881899327931919e-06,
+      "loss": 0.874,
+      "step": 3026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.8773365630279015e-06,
+      "loss": 0.791,
+      "step": 3027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.872774428097703e-06,
+      "loss": 0.8018,
+      "step": 3028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.868212924912494e-06,
+      "loss": 0.8398,
+      "step": 3029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.863652055243215e-06,
+      "loss": 0.7812,
+      "step": 3030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.85909182086055e-06,
+      "loss": 0.8115,
+      "step": 3031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.854532223534935e-06,
+      "loss": 0.8218,
+      "step": 3032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.849973265036566e-06,
+      "loss": 0.7939,
+      "step": 3033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.8454149471353915e-06,
+      "loss": 0.7896,
+      "step": 3034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.8408572716011e-06,
+      "loss": 0.7979,
+      "step": 3035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.836300240203143e-06,
+      "loss": 0.3208,
+      "step": 3036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.831743854710716e-06,
+      "loss": 0.8472,
+      "step": 3037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.827188116892764e-06,
+      "loss": 0.8213,
+      "step": 3038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.822633028517983e-06,
+      "loss": 0.8325,
+      "step": 3039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.8180785913548104e-06,
+      "loss": 0.8076,
+      "step": 3040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.81352480717144e-06,
+      "loss": 0.8193,
+      "step": 3041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.8089716777358085e-06,
+      "loss": 0.7881,
+      "step": 3042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.804419204815594e-06,
+      "loss": 0.8325,
+      "step": 3043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.799867390178224e-06,
+      "loss": 0.8208,
+      "step": 3044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.795316235590873e-06,
+      "loss": 0.8477,
+      "step": 3045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.790765742820451e-06,
+      "loss": 0.7881,
+      "step": 3046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.786215913633617e-06,
+      "loss": 0.8213,
+      "step": 3047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.781666749796776e-06,
+      "loss": 0.8164,
+      "step": 3048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.777118253076066e-06,
+      "loss": 0.8052,
+      "step": 3049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.772570425237369e-06,
+      "loss": 0.3257,
+      "step": 3050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.768023268046312e-06,
+      "loss": 0.8228,
+      "step": 3051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.763476783268255e-06,
+      "loss": 0.8022,
+      "step": 3052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.758930972668299e-06,
+      "loss": 0.7798,
+      "step": 3053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.75438583801129e-06,
+      "loss": 0.7837,
+      "step": 3054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.749841381061799e-06,
+      "loss": 0.8164,
+      "step": 3055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.745297603584145e-06,
+      "loss": 0.812,
+      "step": 3056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.74075450734238e-06,
+      "loss": 0.7954,
+      "step": 3057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.736212094100283e-06,
+      "loss": 0.7891,
+      "step": 3058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.7316703656213805e-06,
+      "loss": 0.7856,
+      "step": 3059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.727129323668931e-06,
+      "loss": 0.8379,
+      "step": 3060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.722588970005917e-06,
+      "loss": 0.8755,
+      "step": 3061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.718049306395063e-06,
+      "loss": 0.323,
+      "step": 3062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.713510334598825e-06,
+      "loss": 0.3271,
+      "step": 3063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.708972056379384e-06,
+      "loss": 0.8159,
+      "step": 3064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.70443447349866e-06,
+      "loss": 0.834,
+      "step": 3065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.699897587718292e-06,
+      "loss": 0.8149,
+      "step": 3066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.69536140079966e-06,
+      "loss": 0.835,
+      "step": 3067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.690825914503872e-06,
+      "loss": 0.8174,
+      "step": 3068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.686291130591754e-06,
+      "loss": 0.835,
+      "step": 3069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.681757050823865e-06,
+      "loss": 0.8374,
+      "step": 3070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.677223676960496e-06,
+      "loss": 0.8242,
+      "step": 3071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.672691010761652e-06,
+      "loss": 0.8184,
+      "step": 3072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.668159053987074e-06,
+      "loss": 0.8394,
+      "step": 3073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.663627808396228e-06,
+      "loss": 0.8389,
+      "step": 3074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.65909727574829e-06,
+      "loss": 0.7676,
+      "step": 3075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.654567457802176e-06,
+      "loss": 0.8311,
+      "step": 3076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.650038356316517e-06,
+      "loss": 0.3369,
+      "step": 3077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.64550997304966e-06,
+      "loss": 0.7891,
+      "step": 3078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.640982309759684e-06,
+      "loss": 0.7817,
+      "step": 3079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6364553682043865e-06,
+      "loss": 0.8599,
+      "step": 3080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.631929150141274e-06,
+      "loss": 0.8379,
+      "step": 3081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.627403657327588e-06,
+      "loss": 0.3385,
+      "step": 3082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.622878891520279e-06,
+      "loss": 0.8169,
+      "step": 3083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.618354854476012e-06,
+      "loss": 0.8257,
+      "step": 3084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.613831547951178e-06,
+      "loss": 0.7852,
+      "step": 3085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.609308973701884e-06,
+      "loss": 0.7964,
+      "step": 3086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.60478713348394e-06,
+      "loss": 0.8281,
+      "step": 3087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.600266029052886e-06,
+      "loss": 0.8247,
+      "step": 3088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.595745662163972e-06,
+      "loss": 0.3491,
+      "step": 3089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.591226034572153e-06,
+      "loss": 0.3052,
+      "step": 3090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.586707148032114e-06,
+      "loss": 0.8589,
+      "step": 3091
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.582189004298231e-06,
+      "loss": 0.3523,
+      "step": 3092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.57767160512461e-06,
+      "loss": 0.8286,
+      "step": 3093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.573154952265062e-06,
+      "loss": 0.8262,
+      "step": 3094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.568639047473102e-06,
+      "loss": 0.7998,
+      "step": 3095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.564123892501964e-06,
+      "loss": 0.8193,
+      "step": 3096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.559609489104587e-06,
+      "loss": 0.814,
+      "step": 3097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.555095839033612e-06,
+      "loss": 0.8359,
+      "step": 3098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.5505829440413985e-06,
+      "loss": 0.8462,
+      "step": 3099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.546070805880009e-06,
+      "loss": 0.8291,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.541559426301205e-06,
+      "loss": 0.875,
+      "step": 3101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.537048807056464e-06,
+      "loss": 0.7871,
+      "step": 3102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.532538949896965e-06,
+      "loss": 0.7993,
+      "step": 3103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.528029856573585e-06,
+      "loss": 0.8184,
+      "step": 3104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.52352152883691e-06,
+      "loss": 0.8672,
+      "step": 3105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.519013968437233e-06,
+      "loss": 0.8179,
+      "step": 3106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.514507177124538e-06,
+      "loss": 0.8779,
+      "step": 3107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.510001156648521e-06,
+      "loss": 0.79,
+      "step": 3108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.505495908758574e-06,
+      "loss": 0.894,
+      "step": 3109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.500991435203785e-06,
+      "loss": 0.7607,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.496487737732948e-06,
+      "loss": 0.8037,
+      "step": 3111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.491984818094558e-06,
+      "loss": 0.8179,
+      "step": 3112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.487482678036796e-06,
+      "loss": 0.812,
+      "step": 3113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.482981319307556e-06,
+      "loss": 0.8086,
+      "step": 3114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.478480743654411e-06,
+      "loss": 0.7983,
+      "step": 3115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.473980952824647e-06,
+      "loss": 0.8174,
+      "step": 3116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.469481948565236e-06,
+      "loss": 0.8257,
+      "step": 3117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.464983732622843e-06,
+      "loss": 0.8247,
+      "step": 3118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.460486306743831e-06,
+      "loss": 0.8057,
+      "step": 3119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.455989672674261e-06,
+      "loss": 0.7681,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.451493832159875e-06,
+      "loss": 0.873,
+      "step": 3121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.4469987869461165e-06,
+      "loss": 0.7954,
+      "step": 3122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.442504538778118e-06,
+      "loss": 0.8584,
+      "step": 3123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.438011089400695e-06,
+      "loss": 0.8159,
+      "step": 3124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.433518440558364e-06,
+      "loss": 0.8442,
+      "step": 3125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.429026593995329e-06,
+      "loss": 0.7793,
+      "step": 3126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.424535551455472e-06,
+      "loss": 0.8296,
+      "step": 3127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.420045314682376e-06,
+      "loss": 0.3164,
+      "step": 3128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.415555885419305e-06,
+      "loss": 0.7988,
+      "step": 3129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.411067265409207e-06,
+      "loss": 0.8379,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.40657945639472e-06,
+      "loss": 0.7715,
+      "step": 3131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.40209246011817e-06,
+      "loss": 0.8179,
+      "step": 3132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.397606278321555e-06,
+      "loss": 0.8438,
+      "step": 3133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.3931209127465745e-06,
+      "loss": 0.8003,
+      "step": 3134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.388636365134595e-06,
+      "loss": 0.8306,
+      "step": 3135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.384152637226677e-06,
+      "loss": 0.8774,
+      "step": 3136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.379669730763555e-06,
+      "loss": 0.8286,
+      "step": 3137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.375187647485649e-06,
+      "loss": 0.8462,
+      "step": 3138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.370706389133057e-06,
+      "loss": 0.8457,
+      "step": 3139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.36622595744556e-06,
+      "loss": 0.7798,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.361746354162612e-06,
+      "loss": 0.7837,
+      "step": 3141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.357267581023353e-06,
+      "loss": 0.7837,
+      "step": 3142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.352789639766598e-06,
+      "loss": 0.7935,
+      "step": 3143
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.34831253213083e-06,
+      "loss": 0.8276,
+      "step": 3144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.343836259854224e-06,
+      "loss": 0.813,
+      "step": 3145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.339360824674624e-06,
+      "loss": 0.8047,
+      "step": 3146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.334886228329541e-06,
+      "loss": 0.8203,
+      "step": 3147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.330412472556171e-06,
+      "loss": 0.854,
+      "step": 3148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.325939559091384e-06,
+      "loss": 0.7261,
+      "step": 3149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.321467489671712e-06,
+      "loss": 0.8672,
+      "step": 3150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.316996266033369e-06,
+      "loss": 0.8071,
+      "step": 3151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.312525889912242e-06,
+      "loss": 0.8452,
+      "step": 3152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.308056363043878e-06,
+      "loss": 0.8423,
+      "step": 3153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.3035876871635056e-06,
+      "loss": 0.8394,
+      "step": 3154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.299119864006019e-06,
+      "loss": 0.8125,
+      "step": 3155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.294652895305979e-06,
+      "loss": 0.7964,
+      "step": 3156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.2901867827976165e-06,
+      "loss": 0.8345,
+      "step": 3157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.285721528214835e-06,
+      "loss": 0.8203,
+      "step": 3158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.2812571332911904e-06,
+      "loss": 0.8247,
+      "step": 3159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.2767935997599234e-06,
+      "loss": 0.8579,
+      "step": 3160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.272330929353927e-06,
+      "loss": 0.7759,
+      "step": 3161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.267869123805766e-06,
+      "loss": 0.7939,
+      "step": 3162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.263408184847664e-06,
+      "loss": 0.8472,
+      "step": 3163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.2589481142115156e-06,
+      "loss": 0.7832,
+      "step": 3164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.254488913628868e-06,
+      "loss": 0.3252,
+      "step": 3165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.250030584830942e-06,
+      "loss": 0.8022,
+      "step": 3166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.2455731295486085e-06,
+      "loss": 0.8477,
+      "step": 3167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.24111654951241e-06,
+      "loss": 0.8374,
+      "step": 3168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.236660846452545e-06,
+      "loss": 0.3296,
+      "step": 3169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.2322060220988645e-06,
+      "loss": 0.8037,
+      "step": 3170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.22775207818089e-06,
+      "loss": 0.8159,
+      "step": 3171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.223299016427796e-06,
+      "loss": 0.7886,
+      "step": 3172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.21884683856841e-06,
+      "loss": 0.855,
+      "step": 3173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.214395546331226e-06,
+      "loss": 0.3237,
+      "step": 3174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.209945141444387e-06,
+      "loss": 0.8618,
+      "step": 3175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.205495625635691e-06,
+      "loss": 0.7451,
+      "step": 3176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.201047000632595e-06,
+      "loss": 0.8438,
+      "step": 3177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.196599268162211e-06,
+      "loss": 0.8086,
+      "step": 3178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.192152429951298e-06,
+      "loss": 0.7397,
+      "step": 3179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.187706487726272e-06,
+      "loss": 0.793,
+      "step": 3180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.183261443213204e-06,
+      "loss": 0.7471,
+      "step": 3181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.1788172981378116e-06,
+      "loss": 0.8091,
+      "step": 3182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.1743740542254625e-06,
+      "loss": 0.7808,
+      "step": 3183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.1699317132011834e-06,
+      "loss": 0.813,
+      "step": 3184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.165490276789635e-06,
+      "loss": 0.8403,
+      "step": 3185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.161049746715143e-06,
+      "loss": 0.8491,
+      "step": 3186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.156610124701672e-06,
+      "loss": 0.8252,
+      "step": 3187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.152171412472836e-06,
+      "loss": 0.7803,
+      "step": 3188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.1477336117518975e-06,
+      "loss": 0.814,
+      "step": 3189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.143296724261758e-06,
+      "loss": 0.8345,
+      "step": 3190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.1388607517249735e-06,
+      "loss": 0.8564,
+      "step": 3191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.134425695863745e-06,
+      "loss": 0.792,
+      "step": 3192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.129991558399908e-06,
+      "loss": 0.8076,
+      "step": 3193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.1255583410549505e-06,
+      "loss": 0.3076,
+      "step": 3194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.1211260455500025e-06,
+      "loss": 0.8291,
+      "step": 3195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.116694673605826e-06,
+      "loss": 0.8667,
+      "step": 3196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.112264226942837e-06,
+      "loss": 0.8457,
+      "step": 3197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.1078347072810915e-06,
+      "loss": 0.8281,
+      "step": 3198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.1034061163402725e-06,
+      "loss": 0.8374,
+      "step": 3199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.098978455839719e-06,
+      "loss": 0.3423,
+      "step": 3200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.094551727498399e-06,
+      "loss": 0.8267,
+      "step": 3201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.0901259330349195e-06,
+      "loss": 0.8149,
+      "step": 3202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.085701074167526e-06,
+      "loss": 0.832,
+      "step": 3203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.081277152614104e-06,
+      "loss": 0.8364,
+      "step": 3204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.076854170092168e-06,
+      "loss": 0.8093,
+      "step": 3205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.072432128318877e-06,
+      "loss": 0.8164,
+      "step": 3206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.068011029011018e-06,
+      "loss": 0.7983,
+      "step": 3207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.063590873885012e-06,
+      "loss": 0.8398,
+      "step": 3208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.059171664656915e-06,
+      "loss": 0.7349,
+      "step": 3209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.054753403042422e-06,
+      "loss": 0.7959,
+      "step": 3210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.050336090756847e-06,
+      "loss": 0.7661,
+      "step": 3211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.045919729515145e-06,
+      "loss": 0.7964,
+      "step": 3212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.041504321031904e-06,
+      "loss": 0.7896,
+      "step": 3213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.037089867021333e-06,
+      "loss": 0.8105,
+      "step": 3214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.032676369197277e-06,
+      "loss": 0.854,
+      "step": 3215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.028263829273205e-06,
+      "loss": 0.8257,
+      "step": 3216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.023852248962218e-06,
+      "loss": 0.8315,
+      "step": 3217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.019441629977047e-06,
+      "loss": 0.7725,
+      "step": 3218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.015031974030041e-06,
+      "loss": 0.8096,
+      "step": 3219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.010623282833184e-06,
+      "loss": 0.793,
+      "step": 3220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.00621555809808e-06,
+      "loss": 0.8003,
+      "step": 3221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.001808801535961e-06,
+      "loss": 0.7979,
+      "step": 3222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9974030148576775e-06,
+      "loss": 0.749,
+      "step": 3223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.992998199773714e-06,
+      "loss": 0.7866,
+      "step": 3224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.988594357994165e-06,
+      "loss": 0.8228,
+      "step": 3225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9841914912287576e-06,
+      "loss": 0.3193,
+      "step": 3226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.979789601186836e-06,
+      "loss": 0.8193,
+      "step": 3227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9753886895773645e-06,
+      "loss": 0.8315,
+      "step": 3228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.970988758108926e-06,
+      "loss": 0.8223,
+      "step": 3229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.966589808489733e-06,
+      "loss": 0.8042,
+      "step": 3230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9621918424275995e-06,
+      "loss": 0.7905,
+      "step": 3231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.957794861629973e-06,
+      "loss": 0.8257,
+      "step": 3232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.953398867803914e-06,
+      "loss": 0.8032,
+      "step": 3233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.949003862656096e-06,
+      "loss": 0.8672,
+      "step": 3234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.94460984789281e-06,
+      "loss": 0.8804,
+      "step": 3235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.940216825219969e-06,
+      "loss": 0.8271,
+      "step": 3236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.935824796343089e-06,
+      "loss": 0.8486,
+      "step": 3237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.931433762967309e-06,
+      "loss": 0.8389,
+      "step": 3238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.927043726797385e-06,
+      "loss": 0.7847,
+      "step": 3239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.922654689537674e-06,
+      "loss": 0.8081,
+      "step": 3240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.918266652892153e-06,
+      "loss": 0.8267,
+      "step": 3241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9138796185644094e-06,
+      "loss": 0.8276,
+      "step": 3242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.909493588257639e-06,
+      "loss": 0.8159,
+      "step": 3243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.905108563674655e-06,
+      "loss": 0.8081,
+      "step": 3244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.900724546517866e-06,
+      "loss": 0.833,
+      "step": 3245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.896341538489307e-06,
+      "loss": 0.855,
+      "step": 3246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.89195954129061e-06,
+      "loss": 0.7759,
+      "step": 3247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.887578556623015e-06,
+      "loss": 0.7822,
+      "step": 3248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.883198586187372e-06,
+      "loss": 0.8428,
+      "step": 3249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8788196316841396e-06,
+      "loss": 0.8032,
+      "step": 3250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.874441694813374e-06,
+      "loss": 0.7505,
+      "step": 3251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8700647772747425e-06,
+      "loss": 0.8037,
+      "step": 3252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.865688880767518e-06,
+      "loss": 0.8569,
+      "step": 3253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.86131400699057e-06,
+      "loss": 0.2983,
+      "step": 3254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.856940157642377e-06,
+      "loss": 0.8521,
+      "step": 3255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8525673344210205e-06,
+      "loss": 0.7842,
+      "step": 3256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8481955390241765e-06,
+      "loss": 0.8291,
+      "step": 3257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.843824773149127e-06,
+      "loss": 0.7803,
+      "step": 3258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8394550384927595e-06,
+      "loss": 0.813,
+      "step": 3259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.83508633675155e-06,
+      "loss": 0.8398,
+      "step": 3260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.830718669621577e-06,
+      "loss": 0.8286,
+      "step": 3261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8263520387985275e-06,
+      "loss": 0.834,
+      "step": 3262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.821986445977669e-06,
+      "loss": 0.813,
+      "step": 3263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.817621892853882e-06,
+      "loss": 0.8711,
+      "step": 3264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.81325838112163e-06,
+      "loss": 0.7251,
+      "step": 3265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.808895912474984e-06,
+      "loss": 0.877,
+      "step": 3266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.804534488607601e-06,
+      "loss": 0.7974,
+      "step": 3267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.800174111212738e-06,
+      "loss": 0.7817,
+      "step": 3268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.795814781983241e-06,
+      "loss": 0.3528,
+      "step": 3269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.791456502611556e-06,
+      "loss": 0.8154,
+      "step": 3270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.7870992747897125e-06,
+      "loss": 0.7842,
+      "step": 3271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.782743100209339e-06,
+      "loss": 0.813,
+      "step": 3272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.778387980561652e-06,
+      "loss": 0.8389,
+      "step": 3273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.774033917537459e-06,
+      "loss": 0.7896,
+      "step": 3274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.769680912827156e-06,
+      "loss": 0.791,
+      "step": 3275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.765328968120733e-06,
+      "loss": 0.8105,
+      "step": 3276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.7609780851077585e-06,
+      "loss": 0.8325,
+      "step": 3277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.7566282654773995e-06,
+      "loss": 0.8296,
+      "step": 3278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.752279510918407e-06,
+      "loss": 0.7886,
+      "step": 3279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.747931823119115e-06,
+      "loss": 0.8936,
+      "step": 3280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.743585203767442e-06,
+      "loss": 0.7646,
+      "step": 3281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.739239654550904e-06,
+      "loss": 0.8359,
+      "step": 3282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.734895177156585e-06,
+      "loss": 0.8613,
+      "step": 3283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.730551773271162e-06,
+      "loss": 0.7969,
+      "step": 3284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.726209444580899e-06,
+      "loss": 0.8413,
+      "step": 3285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.721868192771633e-06,
+      "loss": 0.8525,
+      "step": 3286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.717528019528788e-06,
+      "loss": 0.7861,
+      "step": 3287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.71318892653737e-06,
+      "loss": 0.7734,
+      "step": 3288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.708850915481962e-06,
+      "loss": 0.813,
+      "step": 3289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.7045139880467324e-06,
+      "loss": 0.8252,
+      "step": 3290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.700178145915422e-06,
+      "loss": 0.8203,
+      "step": 3291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.695843390771357e-06,
+      "loss": 0.8452,
+      "step": 3292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.691509724297439e-06,
+      "loss": 0.7412,
+      "step": 3293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.687177148176143e-06,
+      "loss": 0.8066,
+      "step": 3294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.682845664089525e-06,
+      "loss": 0.8174,
+      "step": 3295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.678515273719223e-06,
+      "loss": 0.8315,
+      "step": 3296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.674185978746434e-06,
+      "loss": 0.8188,
+      "step": 3297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.669857780851944e-06,
+      "loss": 0.7871,
+      "step": 3298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.66553068171611e-06,
+      "loss": 0.8442,
+      "step": 3299
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.661204683018859e-06,
+      "loss": 0.8384,
+      "step": 3300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.65687978643969e-06,
+      "loss": 0.8447,
+      "step": 3301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.652555993657683e-06,
+      "loss": 0.7959,
+      "step": 3302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.648233306351476e-06,
+      "loss": 0.8281,
+      "step": 3303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.643911726199289e-06,
+      "loss": 0.8154,
+      "step": 3304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.639591254878911e-06,
+      "loss": 0.7988,
+      "step": 3305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.635271894067693e-06,
+      "loss": 0.8022,
+      "step": 3306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.630953645442561e-06,
+      "loss": 0.7808,
+      "step": 3307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.626636510680009e-06,
+      "loss": 0.7983,
+      "step": 3308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.622320491456095e-06,
+      "loss": 0.8301,
+      "step": 3309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.618005589446446e-06,
+      "loss": 0.8193,
+      "step": 3310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.613691806326261e-06,
+      "loss": 0.8394,
+      "step": 3311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.609379143770293e-06,
+      "loss": 0.8447,
+      "step": 3312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.605067603452868e-06,
+      "loss": 0.8101,
+      "step": 3313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.600757187047878e-06,
+      "loss": 0.8359,
+      "step": 3314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.596447896228767e-06,
+      "loss": 0.3269,
+      "step": 3315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.59213973266856e-06,
+      "loss": 0.8438,
+      "step": 3316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.587832698039825e-06,
+      "loss": 0.7935,
+      "step": 3317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.583526794014707e-06,
+      "loss": 0.8066,
+      "step": 3318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.579222022264906e-06,
+      "loss": 0.8169,
+      "step": 3319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.574918384461681e-06,
+      "loss": 0.792,
+      "step": 3320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.5706158822758495e-06,
+      "loss": 0.7778,
+      "step": 3321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.5663145173777986e-06,
+      "loss": 0.8857,
+      "step": 3322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.5620142914374595e-06,
+      "loss": 0.7334,
+      "step": 3323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.557715206124329e-06,
+      "loss": 0.7964,
+      "step": 3324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.553417263107465e-06,
+      "loss": 0.8237,
+      "step": 3325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.54912046405547e-06,
+      "loss": 0.833,
+      "step": 3326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.544824810636513e-06,
+      "loss": 0.772,
+      "step": 3327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.540530304518316e-06,
+      "loss": 0.8271,
+      "step": 3328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.5362369473681495e-06,
+      "loss": 0.8838,
+      "step": 3329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.531944740852847e-06,
+      "loss": 0.7744,
+      "step": 3330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.5276536866387905e-06,
+      "loss": 0.8691,
+      "step": 3331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.523363786391913e-06,
+      "loss": 0.7461,
+      "step": 3332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.519075041777701e-06,
+      "loss": 0.8076,
+      "step": 3333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.514787454461196e-06,
+      "loss": 0.7891,
+      "step": 3334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.510501026106982e-06,
+      "loss": 0.8022,
+      "step": 3335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.506215758379202e-06,
+      "loss": 0.7876,
+      "step": 3336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.501931652941546e-06,
+      "loss": 0.7559,
+      "step": 3337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.497648711457246e-06,
+      "loss": 0.8101,
+      "step": 3338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4933669355890945e-06,
+      "loss": 0.7554,
+      "step": 3339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.489086326999417e-06,
+      "loss": 0.3247,
+      "step": 3340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.484806887350095e-06,
+      "loss": 0.7822,
+      "step": 3341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.480528618302559e-06,
+      "loss": 0.7969,
+      "step": 3342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.476251521517773e-06,
+      "loss": 0.8447,
+      "step": 3343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.471975598656257e-06,
+      "loss": 0.7979,
+      "step": 3344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.467700851378075e-06,
+      "loss": 0.7979,
+      "step": 3345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.463427281342825e-06,
+      "loss": 0.7827,
+      "step": 3346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.459154890209657e-06,
+      "loss": 0.3257,
+      "step": 3347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.454883679637262e-06,
+      "loss": 0.7305,
+      "step": 3348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.450613651283867e-06,
+      "loss": 0.8169,
+      "step": 3349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.446344806807245e-06,
+      "loss": 0.8604,
+      "step": 3350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.442077147864712e-06,
+      "loss": 0.7915,
+      "step": 3351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4378106761131165e-06,
+      "loss": 0.7686,
+      "step": 3352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.43354539320885e-06,
+      "loss": 0.8257,
+      "step": 3353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.429281300807846e-06,
+      "loss": 0.7944,
+      "step": 3354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.425018400565565e-06,
+      "loss": 0.8145,
+      "step": 3355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.420756694137016e-06,
+      "loss": 0.8335,
+      "step": 3356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.416496183176742e-06,
+      "loss": 0.7861,
+      "step": 3357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.412236869338818e-06,
+      "loss": 0.8193,
+      "step": 3358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.407978754276853e-06,
+      "loss": 0.3179,
+      "step": 3359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.403721839644001e-06,
+      "loss": 0.8125,
+      "step": 3360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.399466127092936e-06,
+      "loss": 0.7632,
+      "step": 3361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.395211618275875e-06,
+      "loss": 0.8286,
+      "step": 3362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.390958314844567e-06,
+      "loss": 0.8291,
+      "step": 3363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.386706218450286e-06,
+      "loss": 0.8184,
+      "step": 3364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.382455330743848e-06,
+      "loss": 0.8223,
+      "step": 3365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.378205653375589e-06,
+      "loss": 0.3374,
+      "step": 3366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.373957187995379e-06,
+      "loss": 0.8354,
+      "step": 3367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.369709936252627e-06,
+      "loss": 0.8218,
+      "step": 3368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3654638997962546e-06,
+      "loss": 0.8457,
+      "step": 3369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.361219080274719e-06,
+      "loss": 0.8257,
+      "step": 3370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3569754793360125e-06,
+      "loss": 0.35,
+      "step": 3371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.352733098627637e-06,
+      "loss": 0.8081,
+      "step": 3372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.348491939796639e-06,
+      "loss": 0.8315,
+      "step": 3373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.344252004489579e-06,
+      "loss": 0.8608,
+      "step": 3374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.340013294352547e-06,
+      "loss": 0.8604,
+      "step": 3375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.335775811031156e-06,
+      "loss": 0.8115,
+      "step": 3376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.331539556170543e-06,
+      "loss": 0.835,
+      "step": 3377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.327304531415367e-06,
+      "loss": 0.3481,
+      "step": 3378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.323070738409809e-06,
+      "loss": 0.8105,
+      "step": 3379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.318838178797579e-06,
+      "loss": 0.8184,
+      "step": 3380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.314606854221894e-06,
+      "loss": 0.7485,
+      "step": 3381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.310376766325506e-06,
+      "loss": 0.8418,
+      "step": 3382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.306147916750685e-06,
+      "loss": 0.7896,
+      "step": 3383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.301920307139202e-06,
+      "loss": 0.8184,
+      "step": 3384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.297693939132373e-06,
+      "loss": 0.877,
+      "step": 3385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.293468814371016e-06,
+      "loss": 0.8247,
+      "step": 3386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.289244934495468e-06,
+      "loss": 0.8325,
+      "step": 3387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.285022301145586e-06,
+      "loss": 0.8032,
+      "step": 3388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.280800915960743e-06,
+      "loss": 0.8613,
+      "step": 3389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.276580780579824e-06,
+      "loss": 0.8691,
+      "step": 3390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.27236189664123e-06,
+      "loss": 0.7578,
+      "step": 3391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.268144265782881e-06,
+      "loss": 0.7524,
+      "step": 3392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.2639278896422e-06,
+      "loss": 0.7915,
+      "step": 3393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.2597127698561395e-06,
+      "loss": 0.7607,
+      "step": 3394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.255498908061145e-06,
+      "loss": 0.8062,
+      "step": 3395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.251286305893184e-06,
+      "loss": 0.791,
+      "step": 3396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.247074964987739e-06,
+      "loss": 0.3263,
+      "step": 3397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.2428648869797905e-06,
+      "loss": 0.8232,
+      "step": 3398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.2386560735038406e-06,
+      "loss": 0.3323,
+      "step": 3399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.234448526193895e-06,
+      "loss": 0.8843,
+      "step": 3400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.230242246683468e-06,
+      "loss": 0.7729,
+      "step": 3401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.226037236605582e-06,
+      "loss": 0.8154,
+      "step": 3402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.221833497592766e-06,
+      "loss": 0.8232,
+      "step": 3403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.217631031277057e-06,
+      "loss": 0.7998,
+      "step": 3404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.213429839289994e-06,
+      "loss": 0.8379,
+      "step": 3405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.209229923262633e-06,
+      "loss": 0.3359,
+      "step": 3406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.205031284825515e-06,
+      "loss": 0.3014,
+      "step": 3407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.200833925608705e-06,
+      "loss": 0.8198,
+      "step": 3408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.1966378472417605e-06,
+      "loss": 0.2961,
+      "step": 3409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.192443051353736e-06,
+      "loss": 0.7651,
+      "step": 3410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.188249539573204e-06,
+      "loss": 0.7676,
+      "step": 3411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.184057313528229e-06,
+      "loss": 0.8354,
+      "step": 3412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.179866374846375e-06,
+      "loss": 0.7656,
+      "step": 3413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.17567672515471e-06,
+      "loss": 0.8154,
+      "step": 3414
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.1714883660798e-06,
+      "loss": 0.7646,
+      "step": 3415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.16730129924771e-06,
+      "loss": 0.8462,
+      "step": 3416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.163115526284006e-06,
+      "loss": 0.834,
+      "step": 3417
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.158931048813745e-06,
+      "loss": 0.6982,
+      "step": 3418
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.154747868461484e-06,
+      "loss": 0.7742,
+      "step": 3419
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.150565986851287e-06,
+      "loss": 0.7773,
+      "step": 3420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.146385405606696e-06,
+      "loss": 0.7354,
+      "step": 3421
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.142206126350756e-06,
+      "loss": 0.7876,
+      "step": 3422
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.138028150706015e-06,
+      "loss": 0.3162,
+      "step": 3423
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.133851480294497e-06,
+      "loss": 0.8564,
+      "step": 3424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.129676116737737e-06,
+      "loss": 0.772,
+      "step": 3425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.1255020616567515e-06,
+      "loss": 0.7178,
+      "step": 3426
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.121329316672054e-06,
+      "loss": 0.7871,
+      "step": 3427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.117157883403646e-06,
+      "loss": 0.7412,
+      "step": 3428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.112987763471022e-06,
+      "loss": 0.8042,
+      "step": 3429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.108818958493168e-06,
+      "loss": 0.3462,
+      "step": 3430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.1046514700885516e-06,
+      "loss": 0.814,
+      "step": 3431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.100485299875147e-06,
+      "loss": 0.7876,
+      "step": 3432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.096320449470391e-06,
+      "loss": 0.8101,
+      "step": 3433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.092156920491232e-06,
+      "loss": 0.8477,
+      "step": 3434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.087994714554094e-06,
+      "loss": 0.8579,
+      "step": 3435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.083833833274881e-06,
+      "loss": 0.8696,
+      "step": 3436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.079674278269e-06,
+      "loss": 0.8687,
+      "step": 3437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.0755160511513284e-06,
+      "loss": 0.8633,
+      "step": 3438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.071359153536235e-06,
+      "loss": 0.7812,
+      "step": 3439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.067203587037569e-06,
+      "loss": 0.8115,
+      "step": 3440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.063049353268664e-06,
+      "loss": 0.8555,
+      "step": 3441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.0588964538423395e-06,
+      "loss": 0.8105,
+      "step": 3442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.0547448903708915e-06,
+      "loss": 0.8047,
+      "step": 3443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.050594664466101e-06,
+      "loss": 0.8989,
+      "step": 3444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.046445777739226e-06,
+      "loss": 0.7915,
+      "step": 3445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.042298231801014e-06,
+      "loss": 0.7559,
+      "step": 3446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.038152028261676e-06,
+      "loss": 0.7324,
+      "step": 3447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.034007168730918e-06,
+      "loss": 0.8223,
+      "step": 3448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.029863654817918e-06,
+      "loss": 0.7598,
+      "step": 3449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.02572148813132e-06,
+      "loss": 0.8271,
+      "step": 3450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.021580670279269e-06,
+      "loss": 0.7725,
+      "step": 3451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.017441202869365e-06,
+      "loss": 0.8398,
+      "step": 3452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.013303087508694e-06,
+      "loss": 0.7266,
+      "step": 3453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.009166325803816e-06,
+      "loss": 0.7939,
+      "step": 3454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.005030919360762e-06,
+      "loss": 0.8008,
+      "step": 3455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.000896869785041e-06,
+      "loss": 0.8477,
+      "step": 3456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.996764178681631e-06,
+      "loss": 0.7725,
+      "step": 3457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.992632847654992e-06,
+      "loss": 0.8105,
+      "step": 3458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.988502878309036e-06,
+      "loss": 0.7949,
+      "step": 3459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.984374272247173e-06,
+      "loss": 0.7954,
+      "step": 3460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.980247031072265e-06,
+      "loss": 0.814,
+      "step": 3461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9761211563866435e-06,
+      "loss": 0.7461,
+      "step": 3462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.971996649792121e-06,
+      "loss": 0.8208,
+      "step": 3463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9678735128899735e-06,
+      "loss": 0.8257,
+      "step": 3464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.963751747280943e-06,
+      "loss": 0.8394,
+      "step": 3465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9596313545652405e-06,
+      "loss": 0.8252,
+      "step": 3466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.955512336342543e-06,
+      "loss": 0.813,
+      "step": 3467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.951394694211998e-06,
+      "loss": 0.2947,
+      "step": 3468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.947278429772214e-06,
+      "loss": 0.8311,
+      "step": 3469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9431635446212695e-06,
+      "loss": 0.8071,
+      "step": 3470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.939050040356698e-06,
+      "loss": 0.8462,
+      "step": 3471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.934937918575514e-06,
+      "loss": 0.8184,
+      "step": 3472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.930827180874173e-06,
+      "loss": 0.7705,
+      "step": 3473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9267178288486126e-06,
+      "loss": 0.8018,
+      "step": 3474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.922609864094227e-06,
+      "loss": 0.3229,
+      "step": 3475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.918503288205859e-06,
+      "loss": 0.7642,
+      "step": 3476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.914398102777832e-06,
+      "loss": 0.8091,
+      "step": 3477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.910294309403917e-06,
+      "loss": 0.7612,
+      "step": 3478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.906191909677349e-06,
+      "loss": 0.7949,
+      "step": 3479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.90209090519082e-06,
+      "loss": 0.7935,
+      "step": 3480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.897991297536481e-06,
+      "loss": 0.8145,
+      "step": 3481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8938930883059394e-06,
+      "loss": 0.749,
+      "step": 3482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889796279090263e-06,
+      "loss": 0.8398,
+      "step": 3483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.885700871479978e-06,
+      "loss": 0.7935,
+      "step": 3484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.881606867065053e-06,
+      "loss": 0.8157,
+      "step": 3485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.87751426743493e-06,
+      "loss": 0.8486,
+      "step": 3486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.873423074178495e-06,
+      "loss": 0.8467,
+      "step": 3487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.869333288884088e-06,
+      "loss": 0.7622,
+      "step": 3488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.865244913139509e-06,
+      "loss": 0.7876,
+      "step": 3489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8611579485319984e-06,
+      "loss": 0.7861,
+      "step": 3490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.857072396648263e-06,
+      "loss": 0.8013,
+      "step": 3491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8529882590744534e-06,
+      "loss": 0.7573,
+      "step": 3492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8489055373961715e-06,
+      "loss": 0.8652,
+      "step": 3493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.844824233198471e-06,
+      "loss": 0.833,
+      "step": 3494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.840744348065855e-06,
+      "loss": 0.8042,
+      "step": 3495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.836665883582274e-06,
+      "loss": 0.7588,
+      "step": 3496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.832588841331128e-06,
+      "loss": 0.812,
+      "step": 3497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.82851322289527e-06,
+      "loss": 0.8022,
+      "step": 3498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8244390298569875e-06,
+      "loss": 0.8159,
+      "step": 3499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8203662637980295e-06,
+      "loss": 0.8408,
+      "step": 3500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.816294926299585e-06,
+      "loss": 0.8364,
+      "step": 3501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8122250189422776e-06,
+      "loss": 0.8306,
+      "step": 3502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.808156543306194e-06,
+      "loss": 0.7456,
+      "step": 3503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8040895009708545e-06,
+      "loss": 0.8159,
+      "step": 3504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.800023893515224e-06,
+      "loss": 0.7847,
+      "step": 3505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7959597225177133e-06,
+      "loss": 0.8589,
+      "step": 3506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7918969895561716e-06,
+      "loss": 0.8076,
+      "step": 3507
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.787835696207892e-06,
+      "loss": 0.8013,
+      "step": 3508
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7837758440496058e-06,
+      "loss": 0.813,
+      "step": 3509
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7797174346574957e-06,
+      "loss": 0.7793,
+      "step": 3510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.775660469607166e-06,
+      "loss": 0.814,
+      "step": 3511
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.771604950473677e-06,
+      "loss": 0.8008,
+      "step": 3512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.767550878831518e-06,
+      "loss": 0.3386,
+      "step": 3513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7634982562546207e-06,
+      "loss": 0.8047,
+      "step": 3514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.759447084316354e-06,
+      "loss": 0.8042,
+      "step": 3515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7553973645895157e-06,
+      "loss": 0.8735,
+      "step": 3516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.751349098646352e-06,
+      "loss": 0.7573,
+      "step": 3517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7473022880585392e-06,
+      "loss": 0.8242,
+      "step": 3518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7432569343971874e-06,
+      "loss": 0.7993,
+      "step": 3519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.739213039232842e-06,
+      "loss": 0.7764,
+      "step": 3520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.735170604135482e-06,
+      "loss": 0.7803,
+      "step": 3521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.73112963067452e-06,
+      "loss": 0.7461,
+      "step": 3522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7270901204187994e-06,
+      "loss": 0.2886,
+      "step": 3523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.723052074936605e-06,
+      "loss": 0.894,
+      "step": 3524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7190154957956327e-06,
+      "loss": 0.8647,
+      "step": 3525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7149803845630317e-06,
+      "loss": 0.7803,
+      "step": 3526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.710946742805367e-06,
+      "loss": 0.7954,
+      "step": 3527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7069145720886373e-06,
+      "loss": 0.79,
+      "step": 3528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.702883873978272e-06,
+      "loss": 0.8159,
+      "step": 3529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.6988546500391235e-06,
+      "loss": 0.7393,
+      "step": 3530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.6948269018354784e-06,
+      "loss": 0.7524,
+      "step": 3531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.6908006309310453e-06,
+      "loss": 0.7915,
+      "step": 3532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.6867758388889613e-06,
+      "loss": 0.7822,
+      "step": 3533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.682752527271791e-06,
+      "loss": 0.7988,
+      "step": 3534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.678730697641518e-06,
+      "loss": 0.3176,
+      "step": 3535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.674710351559564e-06,
+      "loss": 0.7935,
+      "step": 3536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.6706914905867547e-06,
+      "loss": 0.7334,
+      "step": 3537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.666674116283358e-06,
+      "loss": 0.7979,
+      "step": 3538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.662658230209055e-06,
+      "loss": 0.8154,
+      "step": 3539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.6586438339229505e-06,
+      "loss": 0.7686,
+      "step": 3540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.654630928983574e-06,
+      "loss": 0.812,
+      "step": 3541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.6506195169488653e-06,
+      "loss": 0.8071,
+      "step": 3542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.646609599376201e-06,
+      "loss": 0.7275,
+      "step": 3543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.6426011778223665e-06,
+      "loss": 0.2821,
+      "step": 3544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.638594253843569e-06,
+      "loss": 0.811,
+      "step": 3545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.6345888289954336e-06,
+      "loss": 0.8081,
+      "step": 3546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.630584904833006e-06,
+      "loss": 0.7695,
+      "step": 3547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.6265824829107454e-06,
+      "loss": 0.3379,
+      "step": 3548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.622581564782527e-06,
+      "loss": 0.7793,
+      "step": 3549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.618582152001656e-06,
+      "loss": 0.7822,
+      "step": 3550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.614584246120828e-06,
+      "loss": 0.8135,
+      "step": 3551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.610587848692177e-06,
+      "loss": 0.73,
+      "step": 3552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.60659296126724e-06,
+      "loss": 0.8428,
+      "step": 3553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.6025995853969693e-06,
+      "loss": 0.7402,
+      "step": 3554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.5986077226317316e-06,
+      "loss": 0.8384,
+      "step": 3555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.594617374521304e-06,
+      "loss": 0.8359,
+      "step": 3556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.590628542614878e-06,
+      "loss": 0.8423,
+      "step": 3557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.5866412284610554e-06,
+      "loss": 0.7896,
+      "step": 3558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.582655433607849e-06,
+      "loss": 0.7876,
+      "step": 3559
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.5786711596026807e-06,
+      "loss": 0.7935,
+      "step": 3560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.574688407992382e-06,
+      "loss": 0.833,
+      "step": 3561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5707071803232004e-06,
+      "loss": 0.7842,
+      "step": 3562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.566727478140776e-06,
+      "loss": 0.8071,
+      "step": 3563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5627493029901757e-06,
+      "loss": 0.7666,
+      "step": 3564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5587726564158547e-06,
+      "loss": 0.8345,
+      "step": 3565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5547975399616915e-06,
+      "loss": 0.771,
+      "step": 3566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5508239551709636e-06,
+      "loss": 0.7852,
+      "step": 3567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.546851903586344e-06,
+      "loss": 0.8374,
+      "step": 3568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5428813867499304e-06,
+      "loss": 0.7871,
+      "step": 3569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.538912406203209e-06,
+      "loss": 0.8223,
+      "step": 3570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.534944963487077e-06,
+      "loss": 0.8491,
+      "step": 3571
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5309790601418293e-06,
+      "loss": 0.7969,
+      "step": 3572
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5270146977071693e-06,
+      "loss": 0.3059,
+      "step": 3573
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5230518777221964e-06,
+      "loss": 0.791,
+      "step": 3574
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5190906017254126e-06,
+      "loss": 0.8325,
+      "step": 3575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5151308712547302e-06,
+      "loss": 0.7515,
+      "step": 3576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.51117268784744e-06,
+      "loss": 0.8032,
+      "step": 3577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5072160530402554e-06,
+      "loss": 0.8071,
+      "step": 3578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.5032609683692736e-06,
+      "loss": 0.7417,
+      "step": 3579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.499307435369996e-06,
+      "loss": 0.7725,
+      "step": 3580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.49535545557732e-06,
+      "loss": 0.7812,
+      "step": 3581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4914050305255407e-06,
+      "loss": 0.8721,
+      "step": 3582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.487456161748348e-06,
+      "loss": 0.7603,
+      "step": 3583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.483508850778826e-06,
+      "loss": 0.793,
+      "step": 3584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4795630991494664e-06,
+      "loss": 0.8052,
+      "step": 3585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4756189083921362e-06,
+      "loss": 0.3137,
+      "step": 3586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4716762800381064e-06,
+      "loss": 0.751,
+      "step": 3587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4677352156180492e-06,
+      "loss": 0.8008,
+      "step": 3588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.463795716662011e-06,
+      "loss": 0.3137,
+      "step": 3589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.459857784699452e-06,
+      "loss": 0.8179,
+      "step": 3590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4559214212592017e-06,
+      "loss": 0.8311,
+      "step": 3591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4519866278694993e-06,
+      "loss": 0.8516,
+      "step": 3592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.448053406057966e-06,
+      "loss": 0.8164,
+      "step": 3593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4441217573516125e-06,
+      "loss": 0.7812,
+      "step": 3594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4401916832768414e-06,
+      "loss": 0.8306,
+      "step": 3595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4362631853594416e-06,
+      "loss": 0.8003,
+      "step": 3596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.432336265124592e-06,
+      "loss": 0.7227,
+      "step": 3597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4284109240968577e-06,
+      "loss": 0.8169,
+      "step": 3598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.424487163800192e-06,
+      "loss": 0.8091,
+      "step": 3599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.420564985757934e-06,
+      "loss": 0.7778,
+      "step": 3600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.416644391492803e-06,
+      "loss": 0.7793,
+      "step": 3601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4127253825269205e-06,
+      "loss": 0.3457,
+      "step": 3602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4088079603817666e-06,
+      "loss": 0.8608,
+      "step": 3603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4048921265782302e-06,
+      "loss": 0.8511,
+      "step": 3604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.4009778826365687e-06,
+      "loss": 0.812,
+      "step": 3605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.397065230076427e-06,
+      "loss": 0.7827,
+      "step": 3606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.393154170416831e-06,
+      "loss": 0.7954,
+      "step": 3607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.389244705176188e-06,
+      "loss": 0.7959,
+      "step": 3608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.385336835872288e-06,
+      "loss": 0.7778,
+      "step": 3609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.381430564022298e-06,
+      "loss": 0.7881,
+      "step": 3610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.3775258911427757e-06,
+      "loss": 0.8296,
+      "step": 3611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.3736228187496396e-06,
+      "loss": 0.8311,
+      "step": 3612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.369721348358199e-06,
+      "loss": 0.8184,
+      "step": 3613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.365821481483145e-06,
+      "loss": 0.7939,
+      "step": 3614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3619232196385307e-06,
+      "loss": 0.8013,
+      "step": 3615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.358026564337806e-06,
+      "loss": 0.8423,
+      "step": 3616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.354131517093777e-06,
+      "loss": 0.8003,
+      "step": 3617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3502380794186404e-06,
+      "loss": 0.8184,
+      "step": 3618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3463462528239633e-06,
+      "loss": 0.7705,
+      "step": 3619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.342456038820685e-06,
+      "loss": 0.3325,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3385674389191205e-06,
+      "loss": 0.8301,
+      "step": 3621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.334680454628959e-06,
+      "loss": 0.8491,
+      "step": 3622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.330795087459261e-06,
+      "loss": 0.7959,
+      "step": 3623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3269113389184595e-06,
+      "loss": 0.8154,
+      "step": 3624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3230292105143604e-06,
+      "loss": 0.8618,
+      "step": 3625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3191487037541387e-06,
+      "loss": 0.7554,
+      "step": 3626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3152698201443387e-06,
+      "loss": 0.7515,
+      "step": 3627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.311392561190884e-06,
+      "loss": 0.3009,
+      "step": 3628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.3075169283990504e-06,
+      "loss": 0.8159,
+      "step": 3629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.303642923273499e-06,
+      "loss": 0.7036,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.299770547318249e-06,
+      "loss": 0.7881,
+      "step": 3631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2958998020366913e-06,
+      "loss": 0.8413,
+      "step": 3632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2920306889315816e-06,
+      "loss": 0.7876,
+      "step": 3633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2881632095050434e-06,
+      "loss": 0.3442,
+      "step": 3634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.284297365258565e-06,
+      "loss": 0.8584,
+      "step": 3635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2804331576929973e-06,
+      "loss": 0.8218,
+      "step": 3636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2765705883085675e-06,
+      "loss": 0.7668,
+      "step": 3637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.272709658604849e-06,
+      "loss": 0.8496,
+      "step": 3638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2688503700807916e-06,
+      "loss": 0.8145,
+      "step": 3639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2649927242347035e-06,
+      "loss": 0.7827,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2611367225642514e-06,
+      "loss": 0.7375,
+      "step": 3641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.257282366566478e-06,
+      "loss": 0.7695,
+      "step": 3642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2534296577377657e-06,
+      "loss": 0.7905,
+      "step": 3643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.249578597573877e-06,
+      "loss": 0.8271,
+      "step": 3644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.245729187569923e-06,
+      "loss": 0.8066,
+      "step": 3645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2418814292203765e-06,
+      "loss": 0.7886,
+      "step": 3646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2380353240190705e-06,
+      "loss": 0.8252,
+      "step": 3647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.234190873459196e-06,
+      "loss": 0.8047,
+      "step": 3648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2303480790333005e-06,
+      "loss": 0.8071,
+      "step": 3649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.226506942233286e-06,
+      "loss": 0.7695,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2226674645504234e-06,
+      "loss": 0.8667,
+      "step": 3651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2188296474753202e-06,
+      "loss": 0.75,
+      "step": 3652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.214993492497951e-06,
+      "loss": 0.7856,
+      "step": 3653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2111590011076504e-06,
+      "loss": 0.7822,
+      "step": 3654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.207326174793089e-06,
+      "loss": 0.3259,
+      "step": 3655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2034950150423103e-06,
+      "loss": 0.8296,
+      "step": 3656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.1996655233427e-06,
+      "loss": 0.7842,
+      "step": 3657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.1958377011809985e-06,
+      "loss": 0.793,
+      "step": 3658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.1920115500432976e-06,
+      "loss": 0.8081,
+      "step": 3659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.1881870714150415e-06,
+      "loss": 0.7798,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.1843642667810247e-06,
+      "loss": 0.3328,
+      "step": 3661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.180543137625388e-06,
+      "loss": 0.7939,
+      "step": 3662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.1767236854316347e-06,
+      "loss": 0.7788,
+      "step": 3663
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.172905911682598e-06,
+      "loss": 0.7998,
+      "step": 3664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.1690898178604733e-06,
+      "loss": 0.7754,
+      "step": 3665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.1652754054467985e-06,
+      "loss": 0.772,
+      "step": 3666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.161462675922457e-06,
+      "loss": 0.7466,
+      "step": 3667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.1576516307676913e-06,
+      "loss": 0.8491,
+      "step": 3668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.153842271462068e-06,
+      "loss": 0.8091,
+      "step": 3669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.1500345994845204e-06,
+      "loss": 0.7769,
+      "step": 3670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.1462286163133146e-06,
+      "loss": 0.7832,
+      "step": 3671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.1424243234260656e-06,
+      "loss": 0.7593,
+      "step": 3672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.1386217222997287e-06,
+      "loss": 0.6851,
+      "step": 3673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.134820814410607e-06,
+      "loss": 0.7686,
+      "step": 3674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.131021601234342e-06,
+      "loss": 0.3251,
+      "step": 3675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.127224084245918e-06,
+      "loss": 0.7954,
+      "step": 3676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.1234282649196674e-06,
+      "loss": 0.7798,
+      "step": 3677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.1196341447292525e-06,
+      "loss": 0.7515,
+      "step": 3678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.1158417251476797e-06,
+      "loss": 0.8433,
+      "step": 3679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.1120510076473065e-06,
+      "loss": 0.8042,
+      "step": 3680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.1082619936998077e-06,
+      "loss": 0.7437,
+      "step": 3681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.1044746847762173e-06,
+      "loss": 0.7495,
+      "step": 3682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.100689082346897e-06,
+      "loss": 0.7544,
+      "step": 3683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0969051878815472e-06,
+      "loss": 0.8535,
+      "step": 3684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0931230028492076e-06,
+      "loss": 0.8027,
+      "step": 3685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0893425287182512e-06,
+      "loss": 0.7544,
+      "step": 3686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0855637669563877e-06,
+      "loss": 0.7837,
+      "step": 3687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.081786719030664e-06,
+      "loss": 0.8237,
+      "step": 3688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0780113864074593e-06,
+      "loss": 0.73,
+      "step": 3689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.074237770552485e-06,
+      "loss": 0.7627,
+      "step": 3690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0704658729307977e-06,
+      "loss": 0.8096,
+      "step": 3691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.066695695006768e-06,
+      "loss": 0.7979,
+      "step": 3692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0629272382441114e-06,
+      "loss": 0.8369,
+      "step": 3693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.059160504105878e-06,
+      "loss": 0.8379,
+      "step": 3694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.055395494054434e-06,
+      "loss": 0.7295,
+      "step": 3695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0516322095514955e-06,
+      "loss": 0.8086,
+      "step": 3696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.047870652058094e-06,
+      "loss": 0.8457,
+      "step": 3697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0441108230345973e-06,
+      "loss": 0.7437,
+      "step": 3698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0403527239406994e-06,
+      "loss": 0.832,
+      "step": 3699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.036596356235424e-06,
+      "loss": 0.7837,
+      "step": 3700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0328417213771224e-06,
+      "loss": 0.7749,
+      "step": 3701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0290888208234715e-06,
+      "loss": 0.7954,
+      "step": 3702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0253376560314832e-06,
+      "loss": 0.8188,
+      "step": 3703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.021588228457482e-06,
+      "loss": 0.8174,
+      "step": 3704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.017840539557123e-06,
+      "loss": 0.832,
+      "step": 3705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0140945907853983e-06,
+      "loss": 0.7852,
+      "step": 3706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.010350383596601e-06,
+      "loss": 0.7979,
+      "step": 3707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0066079194443715e-06,
+      "loss": 0.793,
+      "step": 3708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0028671997816596e-06,
+      "loss": 0.7749,
+      "step": 3709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9991282260607415e-06,
+      "loss": 0.7734,
+      "step": 3710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9953909997332157e-06,
+      "loss": 0.7881,
+      "step": 3711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.991655522250002e-06,
+      "loss": 0.8496,
+      "step": 3712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9879217950613415e-06,
+      "loss": 0.8091,
+      "step": 3713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.984189819616795e-06,
+      "loss": 0.7876,
+      "step": 3714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.980459597365245e-06,
+      "loss": 0.8472,
+      "step": 3715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.976731129754889e-06,
+      "loss": 0.832,
+      "step": 3716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.973004418233254e-06,
+      "loss": 0.8638,
+      "step": 3717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9692794642471707e-06,
+      "loss": 0.7954,
+      "step": 3718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9655562692427936e-06,
+      "loss": 0.7915,
+      "step": 3719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9618348346656037e-06,
+      "loss": 0.8613,
+      "step": 3720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9581151619603794e-06,
+      "loss": 0.793,
+      "step": 3721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9543972525712333e-06,
+      "loss": 0.8115,
+      "step": 3722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9506811079415836e-06,
+      "loss": 0.8428,
+      "step": 3723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9469667295141658e-06,
+      "loss": 0.8486,
+      "step": 3724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9432541187310285e-06,
+      "loss": 0.834,
+      "step": 3725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9395432770335363e-06,
+      "loss": 0.832,
+      "step": 3726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9358342058623645e-06,
+      "loss": 0.7568,
+      "step": 3727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9321269066575003e-06,
+      "loss": 0.7832,
+      "step": 3728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.928421380858253e-06,
+      "loss": 0.7876,
+      "step": 3729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9247176299032264e-06,
+      "loss": 0.7686,
+      "step": 3730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9210156552303453e-06,
+      "loss": 0.7632,
+      "step": 3731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9173154582768514e-06,
+      "loss": 0.3337,
+      "step": 3732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9136170404792774e-06,
+      "loss": 0.8267,
+      "step": 3733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9099204032734852e-06,
+      "loss": 0.8018,
+      "step": 3734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.906225548094633e-06,
+      "loss": 0.8472,
+      "step": 3735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9025324763771916e-06,
+      "loss": 0.8311,
+      "step": 3736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8988411895549378e-06,
+      "loss": 0.7866,
+      "step": 3737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8951516890609574e-06,
+      "loss": 0.7378,
+      "step": 3738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8914639763276403e-06,
+      "loss": 0.791,
+      "step": 3739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8877780527866845e-06,
+      "loss": 0.772,
+      "step": 3740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8840939198690923e-06,
+      "loss": 0.873,
+      "step": 3741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8804115790051683e-06,
+      "loss": 0.7764,
+      "step": 3742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.876731031624531e-06,
+      "loss": 0.8257,
+      "step": 3743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8730522791560886e-06,
+      "loss": 0.8374,
+      "step": 3744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8693753230280588e-06,
+      "loss": 0.812,
+      "step": 3745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8657001646679716e-06,
+      "loss": 0.7651,
+      "step": 3746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8620268055026387e-06,
+      "loss": 0.3387,
+      "step": 3747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.858355246958193e-06,
+      "loss": 0.8076,
+      "step": 3748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8546854904600563e-06,
+      "loss": 0.8081,
+      "step": 3749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.851017537432955e-06,
+      "loss": 0.7974,
+      "step": 3750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8473513893009154e-06,
+      "loss": 0.8135,
+      "step": 3751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8436870474872613e-06,
+      "loss": 0.7988,
+      "step": 3752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.840024513414617e-06,
+      "loss": 0.8145,
+      "step": 3753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.836363788504902e-06,
+      "loss": 0.8213,
+      "step": 3754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8327048741793437e-06,
+      "loss": 0.7524,
+      "step": 3755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8290477718584464e-06,
+      "loss": 0.8364,
+      "step": 3756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8253924829620335e-06,
+      "loss": 0.8604,
+      "step": 3757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8217390089092116e-06,
+      "loss": 0.8003,
+      "step": 3758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.818087351118379e-06,
+      "loss": 0.8325,
+      "step": 3759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.814437511007243e-06,
+      "loss": 0.8022,
+      "step": 3760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8107894899927926e-06,
+      "loss": 0.3132,
+      "step": 3761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8071432894913172e-06,
+      "loss": 0.8135,
+      "step": 3762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8034989109183963e-06,
+      "loss": 0.8013,
+      "step": 3763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.799856355688904e-06,
+      "loss": 0.7949,
+      "step": 3764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.796215625217004e-06,
+      "loss": 0.8174,
+      "step": 3765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7925767209161555e-06,
+      "loss": 0.812,
+      "step": 3766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7889396441991047e-06,
+      "loss": 0.814,
+      "step": 3767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7853043964778885e-06,
+      "loss": 0.856,
+      "step": 3768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7816709791638414e-06,
+      "loss": 0.8364,
+      "step": 3769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.778039393667575e-06,
+      "loss": 0.3154,
+      "step": 3770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.774409641398996e-06,
+      "loss": 0.8091,
+      "step": 3771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7707817237673052e-06,
+      "loss": 0.791,
+      "step": 3772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.767155642180977e-06,
+      "loss": 0.7788,
+      "step": 3773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7635313980477863e-06,
+      "loss": 0.7822,
+      "step": 3774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.759908992774789e-06,
+      "loss": 0.8081,
+      "step": 3775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.756288427768327e-06,
+      "loss": 0.7476,
+      "step": 3776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.752669704434028e-06,
+      "loss": 0.7847,
+      "step": 3777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7490528241768046e-06,
+      "loss": 0.8408,
+      "step": 3778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7454377884008552e-06,
+      "loss": 0.8467,
+      "step": 3779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7418245985096583e-06,
+      "loss": 0.8042,
+      "step": 3780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.738213255905986e-06,
+      "loss": 0.8228,
+      "step": 3781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7346037619918756e-06,
+      "loss": 0.7793,
+      "step": 3782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7309961181686645e-06,
+      "loss": 0.8423,
+      "step": 3783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.727390325836965e-06,
+      "loss": 0.769,
+      "step": 3784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7237863863966613e-06,
+      "loss": 0.8228,
+      "step": 3785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.720184301246935e-06,
+      "loss": 0.7993,
+      "step": 3786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7165840717862366e-06,
+      "loss": 0.7832,
+      "step": 3787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7129856994122997e-06,
+      "loss": 0.8613,
+      "step": 3788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.709389185522136e-06,
+      "loss": 0.8027,
+      "step": 3789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7057945315120362e-06,
+      "loss": 0.8379,
+      "step": 3790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.702201738777568e-06,
+      "loss": 0.8657,
+      "step": 3791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6986108087135784e-06,
+      "loss": 0.772,
+      "step": 3792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6950217427141897e-06,
+      "loss": 0.7915,
+      "step": 3793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.691434542172797e-06,
+      "loss": 0.7563,
+      "step": 3794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.687849208482085e-06,
+      "loss": 0.8408,
+      "step": 3795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6842657430339925e-06,
+      "loss": 0.8394,
+      "step": 3796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6806841472197506e-06,
+      "loss": 0.791,
+      "step": 3797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6771044224298607e-06,
+      "loss": 0.7983,
+      "step": 3798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6735265700540847e-06,
+      "loss": 0.7871,
+      "step": 3799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6699505914814775e-06,
+      "loss": 0.812,
+      "step": 3800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.666376488100355e-06,
+      "loss": 0.8193,
+      "step": 3801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6628042612983054e-06,
+      "loss": 0.8276,
+      "step": 3802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.659233912462191e-06,
+      "loss": 0.8804,
+      "step": 3803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.655665442978145e-06,
+      "loss": 0.8018,
+      "step": 3804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.652098854231569e-06,
+      "loss": 0.832,
+      "step": 3805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6485341476071333e-06,
+      "loss": 0.7783,
+      "step": 3806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.644971324488787e-06,
+      "loss": 0.7397,
+      "step": 3807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6414103862597312e-06,
+      "loss": 0.7808,
+      "step": 3808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6378513343024526e-06,
+      "loss": 0.8125,
+      "step": 3809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6342941699986974e-06,
+      "loss": 0.3391,
+      "step": 3810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.630738894729471e-06,
+      "loss": 0.7837,
+      "step": 3811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.627185509875062e-06,
+      "loss": 0.8145,
+      "step": 3812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6236340168150137e-06,
+      "loss": 0.7998,
+      "step": 3813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6200844169281384e-06,
+      "loss": 0.8091,
+      "step": 3814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6165367115925137e-06,
+      "loss": 0.3086,
+      "step": 3815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6129909021854787e-06,
+      "loss": 0.7451,
+      "step": 3816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.609446990083641e-06,
+      "loss": 0.7954,
+      "step": 3817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6059049766628682e-06,
+      "loss": 0.7812,
+      "step": 3818
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.602364863298292e-06,
+      "loss": 0.7979,
+      "step": 3819
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5988266513643036e-06,
+      "loss": 0.8306,
+      "step": 3820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.595290342234567e-06,
+      "loss": 0.3357,
+      "step": 3821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5917559372819887e-06,
+      "loss": 0.7593,
+      "step": 3822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5882234378787537e-06,
+      "loss": 0.3257,
+      "step": 3823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5846928453963002e-06,
+      "loss": 0.7441,
+      "step": 3824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5811641612053173e-06,
+      "loss": 0.7378,
+      "step": 3825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.577637386675771e-06,
+      "loss": 0.7476,
+      "step": 3826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5741125231768726e-06,
+      "loss": 0.8496,
+      "step": 3827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5705895720770964e-06,
+      "loss": 0.8032,
+      "step": 3828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5670685347441718e-06,
+      "loss": 0.8379,
+      "step": 3829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5635494125450888e-06,
+      "loss": 0.7773,
+      "step": 3830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5600322068460904e-06,
+      "loss": 0.7749,
+      "step": 3831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.556516919012676e-06,
+      "loss": 0.8271,
+      "step": 3832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.553003550409607e-06,
+      "loss": 0.8188,
+      "step": 3833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5494921024008836e-06,
+      "loss": 0.8457,
+      "step": 3834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.545982576349781e-06,
+      "loss": 0.8105,
+      "step": 3835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5424749736188134e-06,
+      "loss": 0.8042,
+      "step": 3836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5389692955697544e-06,
+      "loss": 0.8525,
+      "step": 3837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.53546554356363e-06,
+      "loss": 0.8115,
+      "step": 3838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5319637189607107e-06,
+      "loss": 0.7729,
+      "step": 3839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5284638231205325e-06,
+      "loss": 0.8003,
+      "step": 3840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.524965857401873e-06,
+      "loss": 0.7563,
+      "step": 3841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.521469823162763e-06,
+      "loss": 0.7654,
+      "step": 3842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5179757217604846e-06,
+      "loss": 0.7852,
+      "step": 3843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5144835545515663e-06,
+      "loss": 0.73,
+      "step": 3844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5109933228917888e-06,
+      "loss": 0.7905,
+      "step": 3845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.507505028136177e-06,
+      "loss": 0.8647,
+      "step": 3846
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.504018671639015e-06,
+      "loss": 0.7842,
+      "step": 3847
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.500534254753816e-06,
+      "loss": 0.7695,
+      "step": 3848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4970517788333593e-06,
+      "loss": 0.8135,
+      "step": 3849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.493571245229661e-06,
+      "loss": 0.8242,
+      "step": 3850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.490092655293976e-06,
+      "loss": 0.7964,
+      "step": 3851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.486616010376822e-06,
+      "loss": 0.7632,
+      "step": 3852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.483141311827947e-06,
+      "loss": 0.8379,
+      "step": 3853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4796685609963518e-06,
+      "loss": 0.8232,
+      "step": 3854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4761977592302754e-06,
+      "loss": 0.832,
+      "step": 3855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4727289078772034e-06,
+      "loss": 0.7661,
+      "step": 3856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4692620082838645e-06,
+      "loss": 0.8257,
+      "step": 3857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4657970617962232e-06,
+      "loss": 0.8281,
+      "step": 3858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4623340697595018e-06,
+      "loss": 0.7759,
+      "step": 3859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.45887303351814e-06,
+      "loss": 0.8179,
+      "step": 3860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.455413954415841e-06,
+      "loss": 0.7954,
+      "step": 3861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4519568337955346e-06,
+      "loss": 0.8052,
+      "step": 3862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.448501672999395e-06,
+      "loss": 0.7856,
+      "step": 3863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.445048473368837e-06,
+      "loss": 0.7671,
+      "step": 3864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.441597236244503e-06,
+      "loss": 0.7891,
+      "step": 3865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4381479629662907e-06,
+      "loss": 0.7983,
+      "step": 3866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.434700654873324e-06,
+      "loss": 0.7788,
+      "step": 3867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4312553133039657e-06,
+      "loss": 0.7896,
+      "step": 3868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4278119395958175e-06,
+      "loss": 0.7759,
+      "step": 3869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4243705350857146e-06,
+      "loss": 0.8491,
+      "step": 3870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4209311011097285e-06,
+      "loss": 0.7915,
+      "step": 3871
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.417493639003164e-06,
+      "loss": 0.7397,
+      "step": 3872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4140581501005703e-06,
+      "loss": 0.8174,
+      "step": 3873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4106246357357124e-06,
+      "loss": 0.7764,
+      "step": 3874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4071930972416046e-06,
+      "loss": 0.7979,
+      "step": 3875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4037635359504897e-06,
+      "loss": 0.3164,
+      "step": 3876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4003359531938337e-06,
+      "loss": 0.772,
+      "step": 3877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3969103503023505e-06,
+      "loss": 0.7871,
+      "step": 3878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3934867286059726e-06,
+      "loss": 0.3259,
+      "step": 3879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3900650894338703e-06,
+      "loss": 0.7983,
+      "step": 3880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3866454341144392e-06,
+      "loss": 0.3215,
+      "step": 3881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3832277639753092e-06,
+      "loss": 0.8101,
+      "step": 3882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.379812080343337e-06,
+      "loss": 0.8633,
+      "step": 3883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3763983845446056e-06,
+      "loss": 0.8037,
+      "step": 3884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3729866779044375e-06,
+      "loss": 0.7715,
+      "step": 3885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3695769617473638e-06,
+      "loss": 0.8076,
+      "step": 3886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3661692373971623e-06,
+      "loss": 0.7891,
+      "step": 3887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3627635061768258e-06,
+      "loss": 0.8008,
+      "step": 3888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.359359769408575e-06,
+      "loss": 0.3252,
+      "step": 3889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.355958028413863e-06,
+      "loss": 0.8394,
+      "step": 3890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.352558284513354e-06,
+      "loss": 0.8228,
+      "step": 3891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3491605390269526e-06,
+      "loss": 0.3376,
+      "step": 3892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.345764793273779e-06,
+      "loss": 0.7573,
+      "step": 3893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3423710485721778e-06,
+      "loss": 0.8364,
+      "step": 3894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3389793062397176e-06,
+      "loss": 0.8174,
+      "step": 3895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.335589567593191e-06,
+      "loss": 0.8188,
+      "step": 3896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3322018339486088e-06,
+      "loss": 0.7529,
+      "step": 3897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3288161066212053e-06,
+      "loss": 0.8564,
+      "step": 3898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.325432386925443e-06,
+      "loss": 0.8027,
+      "step": 3899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.322050676174989e-06,
+      "loss": 0.8027,
+      "step": 3900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.318670975682747e-06,
+      "loss": 0.7881,
+      "step": 3901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3152932867608305e-06,
+      "loss": 0.8047,
+      "step": 3902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3119176107205747e-06,
+      "loss": 0.7764,
+      "step": 3903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3085439488725335e-06,
+      "loss": 0.7568,
+      "step": 3904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3051723025264785e-06,
+      "loss": 0.7861,
+      "step": 3905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3018026729913984e-06,
+      "loss": 0.7671,
+      "step": 3906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2984350615755e-06,
+      "loss": 0.7822,
+      "step": 3907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2950694695862056e-06,
+      "loss": 0.7412,
+      "step": 3908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.291705898330155e-06,
+      "loss": 0.8018,
+      "step": 3909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2883443491131983e-06,
+      "loss": 0.7871,
+      "step": 3910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2849848232404123e-06,
+      "loss": 0.8364,
+      "step": 3911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.281627322016072e-06,
+      "loss": 0.7959,
+      "step": 3912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.278271846743684e-06,
+      "loss": 0.7764,
+      "step": 3913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2749183987259496e-06,
+      "loss": 0.3301,
+      "step": 3914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2715669792648e-06,
+      "loss": 0.7578,
+      "step": 3915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2682175896613713e-06,
+      "loss": 0.8628,
+      "step": 3916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.264870231216005e-06,
+      "loss": 0.7676,
+      "step": 3917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.261524905228269e-06,
+      "loss": 0.7637,
+      "step": 3918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.25818161299693e-06,
+      "loss": 0.7856,
+      "step": 3919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.25484035581997e-06,
+      "loss": 0.7881,
+      "step": 3920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.25150113499458e-06,
+      "loss": 0.8169,
+      "step": 3921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2481639518171616e-06,
+      "loss": 0.8145,
+      "step": 3922
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2448288075833216e-06,
+      "loss": 0.8579,
+      "step": 3923
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.241495703587878e-06,
+      "loss": 0.8193,
+      "step": 3924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.238164641124862e-06,
+      "loss": 0.8037,
+      "step": 3925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.234835621487498e-06,
+      "loss": 0.8018,
+      "step": 3926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.231508645968233e-06,
+      "loss": 0.8491,
+      "step": 3927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.2281837158587114e-06,
+      "loss": 0.7446,
+      "step": 3928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.224860832449785e-06,
+      "loss": 0.7583,
+      "step": 3929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.221539997031512e-06,
+      "loss": 0.7808,
+      "step": 3930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.2182212108931546e-06,
+      "loss": 0.8315,
+      "step": 3931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.21490447532318e-06,
+      "loss": 0.7847,
+      "step": 3932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.2115897916092594e-06,
+      "loss": 0.8022,
+      "step": 3933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.208277161038267e-06,
+      "loss": 0.7671,
+      "step": 3934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.2049665848962807e-06,
+      "loss": 0.8066,
+      "step": 3935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.2016580644685765e-06,
+      "loss": 0.7383,
+      "step": 3936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1983516010396433e-06,
+      "loss": 0.7607,
+      "step": 3937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.195047195893155e-06,
+      "loss": 0.3438,
+      "step": 3938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.191744850312006e-06,
+      "loss": 0.7998,
+      "step": 3939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1884445655782684e-06,
+      "loss": 0.8491,
+      "step": 3940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1851463429732364e-06,
+      "loss": 0.835,
+      "step": 3941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1818501837773887e-06,
+      "loss": 0.7666,
+      "step": 3942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.17855608927041e-06,
+      "loss": 0.3105,
+      "step": 3943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.175264060731179e-06,
+      "loss": 0.8174,
+      "step": 3944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1719740994377766e-06,
+      "loss": 0.8301,
+      "step": 3945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.168686206667477e-06,
+      "loss": 0.814,
+      "step": 3946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.165400383696756e-06,
+      "loss": 0.7671,
+      "step": 3947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1621166318012804e-06,
+      "loss": 0.8037,
+      "step": 3948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.158834952255918e-06,
+      "loss": 0.8315,
+      "step": 3949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.155555346334725e-06,
+      "loss": 0.7788,
+      "step": 3950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.152277815310967e-06,
+      "loss": 0.8198,
+      "step": 3951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1490023604570837e-06,
+      "loss": 0.8169,
+      "step": 3952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1457289830447274e-06,
+      "loss": 0.3318,
+      "step": 3953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1424576843447314e-06,
+      "loss": 0.7983,
+      "step": 3954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1391884656271294e-06,
+      "loss": 0.7603,
+      "step": 3955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.135921328161142e-06,
+      "loss": 0.8496,
+      "step": 3956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1326562732151866e-06,
+      "loss": 0.8247,
+      "step": 3957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.129393302056869e-06,
+      "loss": 0.8076,
+      "step": 3958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1261324159529834e-06,
+      "loss": 0.8354,
+      "step": 3959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1228736161695265e-06,
+      "loss": 0.7954,
+      "step": 3960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1196169039716705e-06,
+      "loss": 0.2518,
+      "step": 3961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.116362280623781e-06,
+      "loss": 0.8286,
+      "step": 3962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1131097473894234e-06,
+      "loss": 0.8423,
+      "step": 3963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1098593055313336e-06,
+      "loss": 0.8091,
+      "step": 3964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1066109563114534e-06,
+      "loss": 0.7637,
+      "step": 3965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.103364700990895e-06,
+      "loss": 0.833,
+      "step": 3966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1001205408299745e-06,
+      "loss": 0.7505,
+      "step": 3967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.096878477088183e-06,
+      "loss": 0.8076,
+      "step": 3968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.093638511024201e-06,
+      "loss": 0.7715,
+      "step": 3969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0904006438958964e-06,
+      "loss": 0.7568,
+      "step": 3970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0871648769603187e-06,
+      "loss": 0.8438,
+      "step": 3971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.083931211473706e-06,
+      "loss": 0.8193,
+      "step": 3972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0806996486914773e-06,
+      "loss": 0.7939,
+      "step": 3973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0774701898682373e-06,
+      "loss": 0.7925,
+      "step": 3974
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.074242836257773e-06,
+      "loss": 0.813,
+      "step": 3975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0710175891130513e-06,
+      "loss": 0.8521,
+      "step": 3976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.0677944496862314e-06,
+      "loss": 0.8242,
+      "step": 3977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.0645734192286374e-06,
+      "loss": 0.8018,
+      "step": 3978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.0613544989907916e-06,
+      "loss": 0.7749,
+      "step": 3979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.0581376902223885e-06,
+      "loss": 0.8154,
+      "step": 3980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.0549229941723015e-06,
+      "loss": 0.8374,
+      "step": 3981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.051710412088589e-06,
+      "loss": 0.8228,
+      "step": 3982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.0484999452184833e-06,
+      "loss": 0.7451,
+      "step": 3983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.0452915948084005e-06,
+      "loss": 0.7822,
+      "step": 3984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.0420853621039288e-06,
+      "loss": 0.333,
+      "step": 3985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.038881248349846e-06,
+      "loss": 0.8228,
+      "step": 3986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.035679254790092e-06,
+      "loss": 0.8052,
+      "step": 3987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.0324793826677936e-06,
+      "loss": 0.7651,
+      "step": 3988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.0292816332252506e-06,
+      "loss": 0.7407,
+      "step": 3989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.0260860077039383e-06,
+      "loss": 0.7612,
+      "step": 3990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.022892507344514e-06,
+      "loss": 0.8506,
+      "step": 3991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.0197011333867964e-06,
+      "loss": 0.751,
+      "step": 3992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.016511887069794e-06,
+      "loss": 0.8408,
+      "step": 3993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.0133247696316784e-06,
+      "loss": 0.7905,
+      "step": 3994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.010139782309799e-06,
+      "loss": 0.7729,
+      "step": 3995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.006956926340677e-06,
+      "loss": 0.7783,
+      "step": 3996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.0037762029600086e-06,
+      "loss": 0.7983,
+      "step": 3997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.000597613402657e-06,
+      "loss": 0.7988,
+      "step": 3998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.997421158902661e-06,
+      "loss": 0.8018,
+      "step": 3999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.994246840693235e-06,
+      "loss": 0.7407,
+      "step": 4000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9910746600067507e-06,
+      "loss": 0.7744,
+      "step": 4001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9879046180747596e-06,
+      "loss": 0.7549,
+      "step": 4002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.984736716127987e-06,
+      "loss": 0.7529,
+      "step": 4003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9815709553963136e-06,
+      "loss": 0.7632,
+      "step": 4004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.978407337108802e-06,
+      "loss": 0.8008,
+      "step": 4005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9752458624936765e-06,
+      "loss": 0.8062,
+      "step": 4006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9720865327783297e-06,
+      "loss": 0.759,
+      "step": 4007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.968929349189322e-06,
+      "loss": 0.814,
+      "step": 4008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.965774312952382e-06,
+      "loss": 0.835,
+      "step": 4009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.962621425292403e-06,
+      "loss": 0.7837,
+      "step": 4010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9594706874334417e-06,
+      "loss": 0.8252,
+      "step": 4011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.95632210059873e-06,
+      "loss": 0.7217,
+      "step": 4012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.953175666010651e-06,
+      "loss": 0.8433,
+      "step": 4013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9500313848907605e-06,
+      "loss": 0.8027,
+      "step": 4014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9468892584597785e-06,
+      "loss": 0.8184,
+      "step": 4015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9437492879375828e-06,
+      "loss": 0.8135,
+      "step": 4016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.940611474543225e-06,
+      "loss": 0.7671,
+      "step": 4017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.937475819494904e-06,
+      "loss": 0.7583,
+      "step": 4018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9343423240099955e-06,
+      "loss": 0.7837,
+      "step": 4019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9312109893050286e-06,
+      "loss": 0.8281,
+      "step": 4020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9280818165956936e-06,
+      "loss": 0.8237,
+      "step": 4021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.924954807096845e-06,
+      "loss": 0.8315,
+      "step": 4022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9218299620224942e-06,
+      "loss": 0.8091,
+      "step": 4023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9187072825858142e-06,
+      "loss": 0.8516,
+      "step": 4024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9155867699991348e-06,
+      "loss": 0.7739,
+      "step": 4025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9124684254739522e-06,
+      "loss": 0.7974,
+      "step": 4026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.909352250220909e-06,
+      "loss": 0.8452,
+      "step": 4027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9062382454498126e-06,
+      "loss": 0.8091,
+      "step": 4028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9031264123696315e-06,
+      "loss": 0.332,
+      "step": 4029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9000167521884786e-06,
+      "loss": 0.8066,
+      "step": 4030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8969092661136383e-06,
+      "loss": 0.7861,
+      "step": 4031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8938039553515414e-06,
+      "loss": 0.7769,
+      "step": 4032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8907008211077755e-06,
+      "loss": 0.8042,
+      "step": 4033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8875998645870852e-06,
+      "loss": 0.3359,
+      "step": 4034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8845010869933668e-06,
+      "loss": 0.7661,
+      "step": 4035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.881404489529675e-06,
+      "loss": 0.7964,
+      "step": 4036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8783100733982115e-06,
+      "loss": 0.8433,
+      "step": 4037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8752178398003425e-06,
+      "loss": 0.7695,
+      "step": 4038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.872127789936573e-06,
+      "loss": 0.8291,
+      "step": 4039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8690399250065682e-06,
+      "loss": 0.8062,
+      "step": 4040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.865954246209145e-06,
+      "loss": 0.7856,
+      "step": 4041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.862870754742266e-06,
+      "loss": 0.771,
+      "step": 4042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.859789451803057e-06,
+      "loss": 0.3379,
+      "step": 4043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8567103385877769e-06,
+      "loss": 0.7817,
+      "step": 4044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8536334162918492e-06,
+      "loss": 0.7729,
+      "step": 4045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8505586861098394e-06,
+      "loss": 0.8525,
+      "step": 4046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8474861492354645e-06,
+      "loss": 0.7729,
+      "step": 4047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8444158068615873e-06,
+      "loss": 0.8018,
+      "step": 4048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8413476601802214e-06,
+      "loss": 0.8047,
+      "step": 4049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8382817103825273e-06,
+      "loss": 0.7808,
+      "step": 4050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8352179586588093e-06,
+      "loss": 0.7964,
+      "step": 4051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8321564061985283e-06,
+      "loss": 0.7715,
+      "step": 4052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.829097054190278e-06,
+      "loss": 0.748,
+      "step": 4053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.826039903821804e-06,
+      "loss": 0.7432,
+      "step": 4054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8229849562800043e-06,
+      "loss": 0.812,
+      "step": 4055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8199322127509048e-06,
+      "loss": 0.769,
+      "step": 4056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8168816744196933e-06,
+      "loss": 0.8062,
+      "step": 4057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8138333424706915e-06,
+      "loss": 0.833,
+      "step": 4058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8107872180873667e-06,
+      "loss": 0.791,
+      "step": 4059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8077433024523285e-06,
+      "loss": 0.3142,
+      "step": 4060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.804701596747331e-06,
+      "loss": 0.7812,
+      "step": 4061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8016621021532679e-06,
+      "loss": 0.8359,
+      "step": 4062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7986248198501767e-06,
+      "loss": 0.8257,
+      "step": 4063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7955897510172346e-06,
+      "loss": 0.7568,
+      "step": 4064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7925568968327576e-06,
+      "loss": 0.8071,
+      "step": 4065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7895262584742103e-06,
+      "loss": 0.7954,
+      "step": 4066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7864978371181853e-06,
+      "loss": 0.7983,
+      "step": 4067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7834716339404188e-06,
+      "loss": 0.8174,
+      "step": 4068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7804476501157938e-06,
+      "loss": 0.7578,
+      "step": 4069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7774258868183175e-06,
+      "loss": 0.877,
+      "step": 4070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7744063452211482e-06,
+      "loss": 0.772,
+      "step": 4071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7713890264965732e-06,
+      "loss": 0.8237,
+      "step": 4072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7683739318160198e-06,
+      "loss": 0.8325,
+      "step": 4073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.765361062350052e-06,
+      "loss": 0.8237,
+      "step": 4074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.762350419268369e-06,
+      "loss": 0.8076,
+      "step": 4075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7593420037398062e-06,
+      "loss": 0.8438,
+      "step": 4076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7563358169323327e-06,
+      "loss": 0.7944,
+      "step": 4077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.753331860013059e-06,
+      "loss": 0.8184,
+      "step": 4078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7503301341482183e-06,
+      "loss": 0.8252,
+      "step": 4079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.747330640503185e-06,
+      "loss": 0.7979,
+      "step": 4080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7443333802424716e-06,
+      "loss": 0.7974,
+      "step": 4081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.741338354529709e-06,
+      "loss": 0.7827,
+      "step": 4082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7383455645276755e-06,
+      "loss": 0.8154,
+      "step": 4083
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7353550113982733e-06,
+      "loss": 0.8071,
+      "step": 4084
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7323666963025386e-06,
+      "loss": 0.7832,
+      "step": 4085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7293806204006376e-06,
+      "loss": 0.8213,
+      "step": 4086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726396784851869e-06,
+      "loss": 0.8037,
+      "step": 4087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.723415190814658e-06,
+      "loss": 0.8179,
+      "step": 4088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7204358394465644e-06,
+      "loss": 0.7607,
+      "step": 4089
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7174587319042741e-06,
+      "loss": 0.8169,
+      "step": 4090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7144838693436005e-06,
+      "loss": 0.7158,
+      "step": 4091
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7115112529194943e-06,
+      "loss": 0.8428,
+      "step": 4092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.708540883786021e-06,
+      "loss": 0.7847,
+      "step": 4093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7055727630963803e-06,
+      "loss": 0.8247,
+      "step": 4094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7026068920029056e-06,
+      "loss": 0.7695,
+      "step": 4095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6996432716570409e-06,
+      "loss": 0.7944,
+      "step": 4096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6966819032093725e-06,
+      "loss": 0.7993,
+      "step": 4097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6937227878096031e-06,
+      "loss": 0.8735,
+      "step": 4098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.690765926606564e-06,
+      "loss": 0.6875,
+      "step": 4099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6878113207482091e-06,
+      "loss": 0.7798,
+      "step": 4100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6848589713816202e-06,
+      "loss": 0.814,
+      "step": 4101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6819088796529985e-06,
+      "loss": 0.769,
+      "step": 4102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6789610467076707e-06,
+      "loss": 0.7673,
+      "step": 4103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6760154736900937e-06,
+      "loss": 0.7402,
+      "step": 4104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6730721617438311e-06,
+      "loss": 0.7568,
+      "step": 4105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6701311120115837e-06,
+      "loss": 0.7817,
+      "step": 4106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6671923256351698e-06,
+      "loss": 0.8428,
+      "step": 4107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6642558037555203e-06,
+      "loss": 0.7725,
+      "step": 4108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6613215475126997e-06,
+      "loss": 0.811,
+      "step": 4109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6583895580458868e-06,
+      "loss": 0.7959,
+      "step": 4110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6554598364933796e-06,
+      "loss": 0.8574,
+      "step": 4111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6525323839925972e-06,
+      "loss": 0.752,
+      "step": 4112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.649607201680078e-06,
+      "loss": 0.7583,
+      "step": 4113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6466842906914775e-06,
+      "loss": 0.8188,
+      "step": 4114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.643763652161571e-06,
+      "loss": 0.8335,
+      "step": 4115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.640845287224251e-06,
+      "loss": 0.7681,
+      "step": 4116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6379291970125248e-06,
+      "loss": 0.8281,
+      "step": 4117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.635015382658525e-06,
+      "loss": 0.8237,
+      "step": 4118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6321038452934902e-06,
+      "loss": 0.7852,
+      "step": 4119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6291945860477778e-06,
+      "loss": 0.8174,
+      "step": 4120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6262876060508683e-06,
+      "loss": 0.8018,
+      "step": 4121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6233829064313441e-06,
+      "loss": 0.7583,
+      "step": 4122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6204804883169156e-06,
+      "loss": 0.8291,
+      "step": 4123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6175803528343997e-06,
+      "loss": 0.3218,
+      "step": 4124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6146825011097282e-06,
+      "loss": 0.7593,
+      "step": 4125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6117869342679472e-06,
+      "loss": 0.7607,
+      "step": 4126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6088936534332158e-06,
+      "loss": 0.8037,
+      "step": 4127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6060026597288045e-06,
+      "loss": 0.8213,
+      "step": 4128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6031139542770945e-06,
+      "loss": 0.8115,
+      "step": 4129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6002275381995882e-06,
+      "loss": 0.7954,
+      "step": 4130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5973434126168815e-06,
+      "loss": 0.7417,
+      "step": 4131
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5944615786486997e-06,
+      "loss": 0.7446,
+      "step": 4132
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5915820374138676e-06,
+      "loss": 0.7573,
+      "step": 4133
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5887047900303178e-06,
+      "loss": 0.7798,
+      "step": 4134
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5858298376151027e-06,
+      "loss": 0.7803,
+      "step": 4135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.582957181284375e-06,
+      "loss": 0.7925,
+      "step": 4136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5800868221533993e-06,
+      "loss": 0.8032,
+      "step": 4137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5772187613365485e-06,
+      "loss": 0.8237,
+      "step": 4138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.574352999947301e-06,
+      "loss": 0.8179,
+      "step": 4139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5714895390982454e-06,
+      "loss": 0.8037,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5686283799010754e-06,
+      "loss": 0.7844,
+      "step": 4141
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5657695234665908e-06,
+      "loss": 0.8232,
+      "step": 4142
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5629129709046974e-06,
+      "loss": 0.7812,
+      "step": 4143
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.560058723324413e-06,
+      "loss": 0.7856,
+      "step": 4144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.557206781833845e-06,
+      "loss": 0.8276,
+      "step": 4145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.554357147540224e-06,
+      "loss": 0.7915,
+      "step": 4146
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5515098215498741e-06,
+      "loss": 0.7637,
+      "step": 4147
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.54866480496822e-06,
+      "loss": 0.8555,
+      "step": 4148
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.545822098899801e-06,
+      "loss": 0.3234,
+      "step": 4149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5429817044482522e-06,
+      "loss": 0.8013,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.540143622716311e-06,
+      "loss": 0.8345,
+      "step": 4151
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5373078548058201e-06,
+      "loss": 0.7256,
+      "step": 4152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5344744018177212e-06,
+      "loss": 0.7969,
+      "step": 4153
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5316432648520587e-06,
+      "loss": 0.8174,
+      "step": 4154
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5288144450079749e-06,
+      "loss": 0.8174,
+      "step": 4155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5259879433837215e-06,
+      "loss": 0.8262,
+      "step": 4156
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5231637610766364e-06,
+      "loss": 0.7769,
+      "step": 4157
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.520341899183169e-06,
+      "loss": 0.7954,
+      "step": 4158
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5175223587988644e-06,
+      "loss": 0.8135,
+      "step": 4159
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5147051410183588e-06,
+      "loss": 0.8091,
+      "step": 4160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5118902469353993e-06,
+      "loss": 0.8193,
+      "step": 4161
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5090776776428225e-06,
+      "loss": 0.7778,
+      "step": 4162
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5062674342325642e-06,
+      "loss": 0.7993,
+      "step": 4163
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.503459517795659e-06,
+      "loss": 0.8003,
+      "step": 4164
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5006539294222357e-06,
+      "loss": 0.8301,
+      "step": 4165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4978506702015207e-06,
+      "loss": 0.8013,
+      "step": 4166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.495049741221835e-06,
+      "loss": 0.8374,
+      "step": 4167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4922511435705963e-06,
+      "loss": 0.8613,
+      "step": 4168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4894548783343153e-06,
+      "loss": 0.8535,
+      "step": 4169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4866609465986031e-06,
+      "loss": 0.8188,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4838693494481526e-06,
+      "loss": 0.8521,
+      "step": 4171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4810800879667644e-06,
+      "loss": 0.8359,
+      "step": 4172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4782931632373257e-06,
+      "loss": 0.8394,
+      "step": 4173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.47550857634181e-06,
+      "loss": 0.7529,
+      "step": 4174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.472726328361297e-06,
+      "loss": 0.7456,
+      "step": 4175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.469946420375948e-06,
+      "loss": 0.8101,
+      "step": 4176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4671688534650193e-06,
+      "loss": 0.323,
+      "step": 4177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4643936287068582e-06,
+      "loss": 0.8398,
+      "step": 4178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.461620747178902e-06,
+      "loss": 0.8325,
+      "step": 4179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4588502099576803e-06,
+      "loss": 0.7944,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.456082018118807e-06,
+      "loss": 0.8105,
+      "step": 4181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4533161727369974e-06,
+      "loss": 0.7993,
+      "step": 4182
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4505526748860382e-06,
+      "loss": 0.751,
+      "step": 4183
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4477915256388212e-06,
+      "loss": 0.8643,
+      "step": 4184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4450327260673201e-06,
+      "loss": 0.8042,
+      "step": 4185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4422762772425887e-06,
+      "loss": 0.7212,
+      "step": 4186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4395221802347826e-06,
+      "loss": 0.7573,
+      "step": 4187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4367704361131342e-06,
+      "loss": 0.762,
+      "step": 4188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4340210459459655e-06,
+      "loss": 0.7812,
+      "step": 4189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.431274010800685e-06,
+      "loss": 0.7363,
+      "step": 4190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4285293317437855e-06,
+      "loss": 0.7354,
+      "step": 4191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4257870098408466e-06,
+      "loss": 0.8467,
+      "step": 4192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4230470461565312e-06,
+      "loss": 0.7598,
+      "step": 4193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4203094417545886e-06,
+      "loss": 0.3086,
+      "step": 4194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4175741976978487e-06,
+      "loss": 0.7134,
+      "step": 4195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4148413150482328e-06,
+      "loss": 0.7437,
+      "step": 4196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.412110794866733e-06,
+      "loss": 0.7715,
+      "step": 4197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4093826382134363e-06,
+      "loss": 0.8486,
+      "step": 4198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4066568461475082e-06,
+      "loss": 0.8013,
+      "step": 4199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4039334197271878e-06,
+      "loss": 0.8125,
+      "step": 4200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.40121236000981e-06,
+      "loss": 0.8257,
+      "step": 4201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.398493668051781e-06,
+      "loss": 0.7778,
+      "step": 4202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.395777344908592e-06,
+      "loss": 0.8228,
+      "step": 4203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3930633916348113e-06,
+      "loss": 0.7397,
+      "step": 4204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.39035180928409e-06,
+      "loss": 0.7832,
+      "step": 4205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.387642598909158e-06,
+      "loss": 0.7676,
+      "step": 4206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3849357615618218e-06,
+      "loss": 0.7188,
+      "step": 4207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3822312982929746e-06,
+      "loss": 0.8037,
+      "step": 4208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3795292101525747e-06,
+      "loss": 0.7642,
+      "step": 4209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3768294981896712e-06,
+      "loss": 0.7803,
+      "step": 4210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3741321634523831e-06,
+      "loss": 0.7896,
+      "step": 4211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.371437206987909e-06,
+      "loss": 0.7539,
+      "step": 4212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3687446298425262e-06,
+      "loss": 0.8032,
+      "step": 4213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3660544330615792e-06,
+      "loss": 0.7812,
+      "step": 4214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3633666176895008e-06,
+      "loss": 0.8135,
+      "step": 4215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3606811847697925e-06,
+      "loss": 0.8179,
+      "step": 4216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3579981353450307e-06,
+      "loss": 0.8179,
+      "step": 4217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3553174704568688e-06,
+      "loss": 0.8062,
+      "step": 4218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3526391911460318e-06,
+      "loss": 0.792,
+      "step": 4219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3499632984523212e-06,
+      "loss": 0.8003,
+      "step": 4220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3472897934146075e-06,
+      "loss": 0.7705,
+      "step": 4221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.344618677070844e-06,
+      "loss": 0.7695,
+      "step": 4222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.341949950458043e-06,
+      "loss": 0.7354,
+      "step": 4223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3392836146123003e-06,
+      "loss": 0.8047,
+      "step": 4224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3366196705687797e-06,
+      "loss": 0.8223,
+      "step": 4225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3339581193617115e-06,
+      "loss": 0.7983,
+      "step": 4226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3312989620244057e-06,
+      "loss": 0.7739,
+      "step": 4227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3286421995892374e-06,
+      "loss": 0.8071,
+      "step": 4228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3259878330876538e-06,
+      "loss": 0.7549,
+      "step": 4229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3233358635501708e-06,
+      "loss": 0.7983,
+      "step": 4230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3206862920063739e-06,
+      "loss": 0.8066,
+      "step": 4231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3180391194849186e-06,
+      "loss": 0.7715,
+      "step": 4232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3153943470135263e-06,
+      "loss": 0.7441,
+      "step": 4233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3127519756189946e-06,
+      "loss": 0.8188,
+      "step": 4234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3101120063271755e-06,
+      "loss": 0.792,
+      "step": 4235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3074744401630015e-06,
+      "loss": 0.769,
+      "step": 4236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3048392781504646e-06,
+      "loss": 0.8013,
+      "step": 4237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3022065213126253e-06,
+      "loss": 0.814,
+      "step": 4238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2995761706716128e-06,
+      "loss": 0.8472,
+      "step": 4239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2969482272486145e-06,
+      "loss": 0.8237,
+      "step": 4240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2943226920638923e-06,
+      "loss": 0.8149,
+      "step": 4241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2916995661367695e-06,
+      "loss": 0.7773,
+      "step": 4242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2890788504856339e-06,
+      "loss": 0.7959,
+      "step": 4243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2864605461279356e-06,
+      "loss": 0.7583,
+      "step": 4244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2838446540801923e-06,
+      "loss": 0.7983,
+      "step": 4245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2812311753579828e-06,
+      "loss": 0.8203,
+      "step": 4246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2786201109759476e-06,
+      "loss": 0.7476,
+      "step": 4247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2760114619477978e-06,
+      "loss": 0.8022,
+      "step": 4248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2734052292862922e-06,
+      "loss": 0.7886,
+      "step": 4249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2708014140032672e-06,
+      "loss": 0.699,
+      "step": 4250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2682000171096102e-06,
+      "loss": 0.7471,
+      "step": 4251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2656010396152742e-06,
+      "loss": 0.8291,
+      "step": 4252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.263004482529271e-06,
+      "loss": 0.79,
+      "step": 4253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2604103468596737e-06,
+      "loss": 0.791,
+      "step": 4254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2578186336136146e-06,
+      "loss": 0.8291,
+      "step": 4255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2552293437972874e-06,
+      "loss": 0.7642,
+      "step": 4256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2526424784159423e-06,
+      "loss": 0.834,
+      "step": 4257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2500580384738899e-06,
+      "loss": 0.8315,
+      "step": 4258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2474760249744976e-06,
+      "loss": 0.7964,
+      "step": 4259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2448964389201962e-06,
+      "loss": 0.7451,
+      "step": 4260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2423192813124625e-06,
+      "loss": 0.2991,
+      "step": 4261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2397445531518467e-06,
+      "loss": 0.7866,
+      "step": 4262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.237172255437938e-06,
+      "loss": 0.8081,
+      "step": 4263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2346023891693978e-06,
+      "loss": 0.7793,
+      "step": 4264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2320349553439353e-06,
+      "loss": 0.7939,
+      "step": 4265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2294699549583116e-06,
+      "loss": 0.8135,
+      "step": 4266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2269073890083544e-06,
+      "loss": 0.853,
+      "step": 4267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2243472584889377e-06,
+      "loss": 0.7822,
+      "step": 4268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2217895643939922e-06,
+      "loss": 0.7856,
+      "step": 4269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2192343077165022e-06,
+      "loss": 0.8389,
+      "step": 4270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2166814894485082e-06,
+      "loss": 0.8135,
+      "step": 4271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2141311105811e-06,
+      "loss": 0.8203,
+      "step": 4272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2115831721044226e-06,
+      "loss": 0.7939,
+      "step": 4273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.209037675007677e-06,
+      "loss": 0.2833,
+      "step": 4274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2064946202791075e-06,
+      "loss": 0.7783,
+      "step": 4275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2039540089060191e-06,
+      "loss": 0.7969,
+      "step": 4276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.201415841874764e-06,
+      "loss": 0.7915,
+      "step": 4277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1988801201707461e-06,
+      "loss": 0.7661,
+      "step": 4278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.196346844778419e-06,
+      "loss": 0.8032,
+      "step": 4279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1938160166812876e-06,
+      "loss": 0.7988,
+      "step": 4280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1912876368619075e-06,
+      "loss": 0.7676,
+      "step": 4281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1887617063018819e-06,
+      "loss": 0.813,
+      "step": 4282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.186238225981864e-06,
+      "loss": 0.8428,
+      "step": 4283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1837171968815556e-06,
+      "loss": 0.7505,
+      "step": 4284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1811986199797065e-06,
+      "loss": 0.8379,
+      "step": 4285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1786824962541186e-06,
+      "loss": 0.7803,
+      "step": 4286
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1761688266816318e-06,
+      "loss": 0.8037,
+      "step": 4287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1736576122381457e-06,
+      "loss": 0.7556,
+      "step": 4288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.171148853898593e-06,
+      "loss": 0.312,
+      "step": 4289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1686425526369668e-06,
+      "loss": 0.7729,
+      "step": 4290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1661387094262982e-06,
+      "loss": 0.811,
+      "step": 4291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1636373252386597e-06,
+      "loss": 0.8169,
+      "step": 4292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1611384010451806e-06,
+      "loss": 0.7881,
+      "step": 4293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.158641937816028e-06,
+      "loss": 0.8145,
+      "step": 4294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1561479365204148e-06,
+      "loss": 0.8159,
+      "step": 4295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1536563981265974e-06,
+      "loss": 0.7847,
+      "step": 4296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.151167323601877e-06,
+      "loss": 0.8311,
+      "step": 4297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1486807139125994e-06,
+      "loss": 0.769,
+      "step": 4298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1461965700241483e-06,
+      "loss": 0.833,
+      "step": 4299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1437148929009603e-06,
+      "loss": 0.8438,
+      "step": 4300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1412356835065002e-06,
+      "loss": 0.8169,
+      "step": 4301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1387589428032891e-06,
+      "loss": 0.7466,
+      "step": 4302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1362846717528793e-06,
+      "loss": 0.3372,
+      "step": 4303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1338128713158703e-06,
+      "loss": 0.8125,
+      "step": 4304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.131343542451898e-06,
+      "loss": 0.8735,
+      "step": 4305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.128876686119642e-06,
+      "loss": 0.8291,
+      "step": 4306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.126412303276821e-06,
+      "loss": 0.7227,
+      "step": 4307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1239503948801903e-06,
+      "loss": 0.8105,
+      "step": 4308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1214909618855546e-06,
+      "loss": 0.8008,
+      "step": 4309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1190340052477436e-06,
+      "loss": 0.7422,
+      "step": 4310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1165795259206333e-06,
+      "loss": 0.334,
+      "step": 4311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1141275248571422e-06,
+      "loss": 0.6748,
+      "step": 4312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1116780030092149e-06,
+      "loss": 0.8027,
+      "step": 4313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1092309613278457e-06,
+      "loss": 0.7812,
+      "step": 4314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.106786400763055e-06,
+      "loss": 0.3231,
+      "step": 4315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1043443222639103e-06,
+      "loss": 0.7837,
+      "step": 4316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1019047267785082e-06,
+      "loss": 0.7661,
+      "step": 4317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.099467615253985e-06,
+      "loss": 0.7896,
+      "step": 4318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0970329886365091e-06,
+      "loss": 0.7979,
+      "step": 4319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0946008478712882e-06,
+      "loss": 0.8184,
+      "step": 4320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.092171193902562e-06,
+      "loss": 0.79,
+      "step": 4321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0897440276736064e-06,
+      "loss": 0.7529,
+      "step": 4322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0873193501267314e-06,
+      "loss": 0.7476,
+      "step": 4323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0848971622032797e-06,
+      "loss": 0.7993,
+      "step": 4324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0824774648436264e-06,
+      "loss": 0.8486,
+      "step": 4325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.080060258987186e-06,
+      "loss": 0.7461,
+      "step": 4326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0776455455723946e-06,
+      "loss": 0.8232,
+      "step": 4327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.075233325536733e-06,
+      "loss": 0.3115,
+      "step": 4328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0728235998167046e-06,
+      "loss": 0.7866,
+      "step": 4329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0704163693478495e-06,
+      "loss": 0.8022,
+      "step": 4330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0680116350647374e-06,
+      "loss": 0.8125,
+      "step": 4331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0656093979009672e-06,
+      "loss": 0.7383,
+      "step": 4332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0632096587891727e-06,
+      "loss": 0.7852,
+      "step": 4333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0608124186610114e-06,
+      "loss": 0.7871,
+      "step": 4334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0584176784471797e-06,
+      "loss": 0.7666,
+      "step": 4335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0560254390773949e-06,
+      "loss": 0.7305,
+      "step": 4336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0536357014804066e-06,
+      "loss": 0.7788,
+      "step": 4337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0512484665839933e-06,
+      "loss": 0.7676,
+      "step": 4338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.0488637353149612e-06,
+      "loss": 0.772,
+      "step": 4339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.046481508599149e-06,
+      "loss": 0.8359,
+      "step": 4340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0441017873614136e-06,
+      "loss": 0.8184,
+      "step": 4341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.041724572525649e-06,
+      "loss": 0.8047,
+      "step": 4342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0393498650147707e-06,
+      "loss": 0.7944,
+      "step": 4343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0369776657507208e-06,
+      "loss": 0.79,
+      "step": 4344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0346079756544694e-06,
+      "loss": 0.7979,
+      "step": 4345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0322407956460122e-06,
+      "loss": 0.77,
+      "step": 4346
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0298761266443694e-06,
+      "loss": 0.7524,
+      "step": 4347
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.027513969567587e-06,
+      "loss": 0.7603,
+      "step": 4348
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0251543253327367e-06,
+      "loss": 0.7534,
+      "step": 4349
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0227971948559115e-06,
+      "loss": 0.792,
+      "step": 4350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.020442579052231e-06,
+      "loss": 0.7915,
+      "step": 4351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0180904788358422e-06,
+      "loss": 0.7661,
+      "step": 4352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0157408951199048e-06,
+      "loss": 0.8223,
+      "step": 4353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0133938288166134e-06,
+      "loss": 0.7773,
+      "step": 4354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.011049280837178e-06,
+      "loss": 0.7217,
+      "step": 4355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0087072520918322e-06,
+      "loss": 0.7446,
+      "step": 4356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0063677434898339e-06,
+      "loss": 0.8032,
+      "step": 4357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.0040307559394588e-06,
+      "loss": 0.835,
+      "step": 4358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.001696290348008e-06,
+      "loss": 0.8174,
+      "step": 4359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.993643476217987e-07,
+      "loss": 0.793,
+      "step": 4360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.970349286661762e-07,
+      "loss": 0.832,
+      "step": 4361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.94708034385496e-07,
+      "loss": 0.7334,
+      "step": 4362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.92383665683141e-07,
+      "loss": 0.8198,
+      "step": 4363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.900618234615103e-07,
+      "loss": 0.7378,
+      "step": 4364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.877425086220224e-07,
+      "loss": 0.7241,
+      "step": 4365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.854257220651195e-07,
+      "loss": 0.8066,
+      "step": 4366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.831114646902512e-07,
+      "loss": 0.8652,
+      "step": 4367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.80799737395897e-07,
+      "loss": 0.7954,
+      "step": 4368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.784905410795477e-07,
+      "loss": 0.7568,
+      "step": 4369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.761838766377119e-07,
+      "loss": 0.7842,
+      "step": 4370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.73879744965917e-07,
+      "loss": 0.7769,
+      "step": 4371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.715781469587063e-07,
+      "loss": 0.7803,
+      "step": 4372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.692790835096388e-07,
+      "loss": 0.8169,
+      "step": 4373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.669825555112891e-07,
+      "loss": 0.7163,
+      "step": 4374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.64688563855252e-07,
+      "loss": 0.7764,
+      "step": 4375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.623971094321315e-07,
+      "loss": 0.7905,
+      "step": 4376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.601081931315473e-07,
+      "loss": 0.8081,
+      "step": 4377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.5782181584214e-07,
+      "loss": 0.7822,
+      "step": 4378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.55537978451556e-07,
+      "loss": 0.7905,
+      "step": 4379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.532566818464611e-07,
+      "loss": 0.7642,
+      "step": 4380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.509779269125341e-07,
+      "loss": 0.7754,
+      "step": 4381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.487017145344654e-07,
+      "loss": 0.8408,
+      "step": 4382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.46428045595959e-07,
+      "loss": 0.8071,
+      "step": 4383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.441569209797312e-07,
+      "loss": 0.3342,
+      "step": 4384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.418883415675103e-07,
+      "loss": 0.7759,
+      "step": 4385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.396223082400365e-07,
+      "loss": 0.8081,
+      "step": 4386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.373588218770671e-07,
+      "loss": 0.8003,
+      "step": 4387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.35097883357359e-07,
+      "loss": 0.7788,
+      "step": 4388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.328394935586884e-07,
+      "loss": 0.7651,
+      "step": 4389
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.305836533578416e-07,
+      "loss": 0.7622,
+      "step": 4390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.283303636306096e-07,
+      "loss": 0.7598,
+      "step": 4391
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.260796252518025e-07,
+      "loss": 0.813,
+      "step": 4392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.23831439095229e-07,
+      "loss": 0.7695,
+      "step": 4393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.215858060337167e-07,
+      "loss": 0.7437,
+      "step": 4394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.193427269390958e-07,
+      "loss": 0.8218,
+      "step": 4395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.17102202682207e-07,
+      "loss": 0.3352,
+      "step": 4396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.148642341328989e-07,
+      "loss": 0.7397,
+      "step": 4397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.126288221600279e-07,
+      "loss": 0.7598,
+      "step": 4398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.103959676314583e-07,
+      "loss": 0.7637,
+      "step": 4399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.081656714140591e-07,
+      "loss": 0.7065,
+      "step": 4400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.059379343737126e-07,
+      "loss": 0.8237,
+      "step": 4401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.037127573752987e-07,
+      "loss": 0.8218,
+      "step": 4402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.014901412827073e-07,
+      "loss": 0.3252,
+      "step": 4403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.992700869588405e-07,
+      "loss": 0.7607,
+      "step": 4404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.970525952655915e-07,
+      "loss": 0.7788,
+      "step": 4405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.94837667063873e-07,
+      "loss": 0.751,
+      "step": 4406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.926253032135951e-07,
+      "loss": 0.7744,
+      "step": 4407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.904155045736729e-07,
+      "loss": 0.8071,
+      "step": 4408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.88208272002026e-07,
+      "loss": 0.2981,
+      "step": 4409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.860036063555791e-07,
+      "loss": 0.3275,
+      "step": 4410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.838015084902589e-07,
+      "loss": 0.8062,
+      "step": 4411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.816019792609952e-07,
+      "loss": 0.8477,
+      "step": 4412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.794050195217216e-07,
+      "loss": 0.8062,
+      "step": 4413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.772106301253705e-07,
+      "loss": 0.7705,
+      "step": 4414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.750188119238864e-07,
+      "loss": 0.7744,
+      "step": 4415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.72829565768202e-07,
+      "loss": 0.7285,
+      "step": 4416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.706428925082591e-07,
+      "loss": 0.8276,
+      "step": 4417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.684587929930043e-07,
+      "loss": 0.793,
+      "step": 4418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.662772680703729e-07,
+      "loss": 0.7939,
+      "step": 4419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.640983185873139e-07,
+      "loss": 0.7573,
+      "step": 4420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.619219453897681e-07,
+      "loss": 0.7935,
+      "step": 4421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.597481493226789e-07,
+      "loss": 0.8418,
+      "step": 4422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.575769312299886e-07,
+      "loss": 0.7974,
+      "step": 4423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.554082919546397e-07,
+      "loss": 0.812,
+      "step": 4424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.532422323385711e-07,
+      "loss": 0.7729,
+      "step": 4425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.51078753222721e-07,
+      "loss": 0.7744,
+      "step": 4426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.489178554470309e-07,
+      "loss": 0.7822,
+      "step": 4427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.467595398504313e-07,
+      "loss": 0.8174,
+      "step": 4428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.446038072708545e-07,
+      "loss": 0.7583,
+      "step": 4429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.424506585452351e-07,
+      "loss": 0.7598,
+      "step": 4430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.40300094509493e-07,
+      "loss": 0.833,
+      "step": 4431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.381521159985558e-07,
+      "loss": 0.7979,
+      "step": 4432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.360067238463417e-07,
+      "loss": 0.8125,
+      "step": 4433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.338639188857653e-07,
+      "loss": 0.8423,
+      "step": 4434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.317237019487376e-07,
+      "loss": 0.7734,
+      "step": 4435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.29586073866164e-07,
+      "loss": 0.833,
+      "step": 4436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.274510354679449e-07,
+      "loss": 0.8345,
+      "step": 4437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.253185875829771e-07,
+      "loss": 0.7729,
+      "step": 4438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.231887310391494e-07,
+      "loss": 0.2908,
+      "step": 4439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.210614666633446e-07,
+      "loss": 0.7646,
+      "step": 4440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.189367952814441e-07,
+      "loss": 0.3171,
+      "step": 4441
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.168147177183136e-07,
+      "loss": 0.7354,
+      "step": 4442
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.14695234797817e-07,
+      "loss": 0.7637,
+      "step": 4443
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.125783473428147e-07,
+      "loss": 0.8145,
+      "step": 4444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.1046405617515e-07,
+      "loss": 0.7954,
+      "step": 4445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.08352362115668e-07,
+      "loss": 0.8315,
+      "step": 4446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.062432659841986e-07,
+      "loss": 0.771,
+      "step": 4447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.041367685995663e-07,
+      "loss": 0.8184,
+      "step": 4448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.020328707795857e-07,
+      "loss": 0.77,
+      "step": 4449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.999315733410628e-07,
+      "loss": 0.7964,
+      "step": 4450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.978328770997925e-07,
+      "loss": 0.7607,
+      "step": 4451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.957367828705614e-07,
+      "loss": 0.8037,
+      "step": 4452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.936432914671476e-07,
+      "loss": 0.8584,
+      "step": 4453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.91552403702312e-07,
+      "loss": 0.7617,
+      "step": 4454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.894641203878134e-07,
+      "loss": 0.8198,
+      "step": 4455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.873784423343958e-07,
+      "loss": 0.811,
+      "step": 4456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.852953703517857e-07,
+      "loss": 0.7407,
+      "step": 4457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.832149052487086e-07,
+      "loss": 0.7925,
+      "step": 4458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.811370478328703e-07,
+      "loss": 0.7666,
+      "step": 4459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.790617989109677e-07,
+      "loss": 0.7695,
+      "step": 4460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.769891592886841e-07,
+      "loss": 0.7925,
+      "step": 4461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.749191297706881e-07,
+      "loss": 0.8047,
+      "step": 4462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.728517111606379e-07,
+      "loss": 0.8057,
+      "step": 4463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.707869042611753e-07,
+      "loss": 0.77,
+      "step": 4464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.687247098739303e-07,
+      "loss": 0.8232,
+      "step": 4465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.666651287995163e-07,
+      "loss": 0.3038,
+      "step": 4466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.646081618375372e-07,
+      "loss": 0.8003,
+      "step": 4467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.62553809786575e-07,
+      "loss": 0.8408,
+      "step": 4468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.605020734441992e-07,
+      "loss": 0.8906,
+      "step": 4469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.584529536069692e-07,
+      "loss": 0.7964,
+      "step": 4470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.564064510704181e-07,
+      "loss": 0.8027,
+      "step": 4471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.543625666290724e-07,
+      "loss": 0.7427,
+      "step": 4472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.523213010764377e-07,
+      "loss": 0.7681,
+      "step": 4473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.50282655205004e-07,
+      "loss": 0.7827,
+      "step": 4474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.482466298062418e-07,
+      "loss": 0.793,
+      "step": 4475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.462132256706086e-07,
+      "loss": 0.813,
+      "step": 4476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.441824435875405e-07,
+      "loss": 0.7495,
+      "step": 4477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.42154284345456e-07,
+      "loss": 0.7661,
+      "step": 4478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.40128748731761e-07,
+      "loss": 0.8677,
+      "step": 4479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.381058375328314e-07,
+      "loss": 0.7935,
+      "step": 4480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.360855515340367e-07,
+      "loss": 0.814,
+      "step": 4481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.340678915197202e-07,
+      "loss": 0.8433,
+      "step": 4482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.320528582732036e-07,
+      "loss": 0.7822,
+      "step": 4483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.300404525767953e-07,
+      "loss": 0.7812,
+      "step": 4484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.280306752117804e-07,
+      "loss": 0.8325,
+      "step": 4485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.260235269584231e-07,
+      "loss": 0.8228,
+      "step": 4486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.240190085959675e-07,
+      "loss": 0.7651,
+      "step": 4487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.220171209026368e-07,
+      "loss": 0.7646,
+      "step": 4488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.200178646556321e-07,
+      "loss": 0.8354,
+      "step": 4489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.180212406311349e-07,
+      "loss": 0.7529,
+      "step": 4490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.160272496043016e-07,
+      "loss": 0.7905,
+      "step": 4491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.140358923492674e-07,
+      "loss": 0.8037,
+      "step": 4492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.120471696391512e-07,
+      "loss": 0.8218,
+      "step": 4493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.100610822460371e-07,
+      "loss": 0.3323,
+      "step": 4494
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.080776309409951e-07,
+      "loss": 0.7686,
+      "step": 4495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.060968164940715e-07,
+      "loss": 0.8281,
+      "step": 4496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.041186396742822e-07,
+      "loss": 0.7974,
+      "step": 4497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.021431012496277e-07,
+      "loss": 0.7485,
+      "step": 4498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.001702019870779e-07,
+      "loss": 0.812,
+      "step": 4499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.981999426525809e-07,
+      "loss": 0.772,
+      "step": 4500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.962323240110591e-07,
+      "loss": 0.7383,
+      "step": 4501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.942673468264096e-07,
+      "loss": 0.7969,
+      "step": 4502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.923050118615041e-07,
+      "loss": 0.7949,
+      "step": 4503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.903453198781878e-07,
+      "loss": 0.8071,
+      "step": 4504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.883882716372842e-07,
+      "loss": 0.7812,
+      "step": 4505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.86433867898581e-07,
+      "loss": 0.8018,
+      "step": 4506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.844821094208506e-07,
+      "loss": 0.7993,
+      "step": 4507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.82532996961831e-07,
+      "loss": 0.3116,
+      "step": 4508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.80586531278232e-07,
+      "loss": 0.7905,
+      "step": 4509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.786427131257422e-07,
+      "loss": 0.7778,
+      "step": 4510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.767015432590165e-07,
+      "loss": 0.7549,
+      "step": 4511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.747630224316861e-07,
+      "loss": 0.7827,
+      "step": 4512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.728271513963491e-07,
+      "loss": 0.8101,
+      "step": 4513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.708939309045792e-07,
+      "loss": 0.7495,
+      "step": 4514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.689633617069176e-07,
+      "loss": 0.8267,
+      "step": 4515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.670354445528792e-07,
+      "loss": 0.7739,
+      "step": 4516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.651101801909468e-07,
+      "loss": 0.7598,
+      "step": 4517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.631875693685724e-07,
+      "loss": 0.752,
+      "step": 4518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.612676128321857e-07,
+      "loss": 0.7876,
+      "step": 4519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.593503113271729e-07,
+      "loss": 0.7607,
+      "step": 4520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.574356655979011e-07,
+      "loss": 0.7197,
+      "step": 4521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.555236763877024e-07,
+      "loss": 0.7334,
+      "step": 4522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.536143444388712e-07,
+      "loss": 0.77,
+      "step": 4523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.517076704926814e-07,
+      "loss": 0.7964,
+      "step": 4524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.498036552893675e-07,
+      "loss": 0.792,
+      "step": 4525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.47902299568133e-07,
+      "loss": 0.7793,
+      "step": 4526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.460036040671511e-07,
+      "loss": 0.7549,
+      "step": 4527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.441075695235601e-07,
+      "loss": 0.7988,
+      "step": 4528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.422141966734657e-07,
+      "loss": 0.7559,
+      "step": 4529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.403234862519388e-07,
+      "loss": 0.7808,
+      "step": 4530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.384354389930233e-07,
+      "loss": 0.792,
+      "step": 4531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.365500556297186e-07,
+      "loss": 0.769,
+      "step": 4532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.346673368939989e-07,
+      "loss": 0.833,
+      "step": 4533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.327872835168008e-07,
+      "loss": 0.7559,
+      "step": 4534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.309098962280219e-07,
+      "loss": 0.8682,
+      "step": 4535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.290351757565315e-07,
+      "loss": 0.79,
+      "step": 4536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.271631228301602e-07,
+      "loss": 0.7412,
+      "step": 4537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.25293738175704e-07,
+      "loss": 0.8003,
+      "step": 4538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.234270225189215e-07,
+      "loss": 0.8018,
+      "step": 4539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.215629765845362e-07,
+      "loss": 0.7935,
+      "step": 4540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.197016010962344e-07,
+      "loss": 0.791,
+      "step": 4541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.178428967766672e-07,
+      "loss": 0.8062,
+      "step": 4542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.159868643474473e-07,
+      "loss": 0.7983,
+      "step": 4543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.141335045291482e-07,
+      "loss": 0.8145,
+      "step": 4544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.122828180413128e-07,
+      "loss": 0.7437,
+      "step": 4545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.104348056024365e-07,
+      "loss": 0.7681,
+      "step": 4546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.085894679299842e-07,
+      "loss": 0.7241,
+      "step": 4547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.067468057403796e-07,
+      "loss": 0.8086,
+      "step": 4548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.04906819749004e-07,
+      "loss": 0.7856,
+      "step": 4549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.030695106702083e-07,
+      "loss": 0.8086,
+      "step": 4550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.012348792172961e-07,
+      "loss": 0.8013,
+      "step": 4551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.994029261025352e-07,
+      "loss": 0.8418,
+      "step": 4552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.975736520371533e-07,
+      "loss": 0.752,
+      "step": 4553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.957470577313365e-07,
+      "loss": 0.7495,
+      "step": 4554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.939231438942325e-07,
+      "loss": 0.8115,
+      "step": 4555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.921019112339465e-07,
+      "loss": 0.7773,
+      "step": 4556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.902833604575464e-07,
+      "loss": 0.7837,
+      "step": 4557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.884674922710523e-07,
+      "loss": 0.8286,
+      "step": 4558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.866543073794495e-07,
+      "loss": 0.7827,
+      "step": 4559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.848438064866782e-07,
+      "loss": 0.7783,
+      "step": 4560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.83035990295637e-07,
+      "loss": 0.7925,
+      "step": 4561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.812308595081836e-07,
+      "loss": 0.8042,
+      "step": 4562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.794284148251283e-07,
+      "loss": 0.7266,
+      "step": 4563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.776286569462463e-07,
+      "loss": 0.749,
+      "step": 4564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.758315865702637e-07,
+      "loss": 0.8105,
+      "step": 4565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.740372043948649e-07,
+      "loss": 0.8232,
+      "step": 4566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.722455111166918e-07,
+      "loss": 0.811,
+      "step": 4567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.704565074313406e-07,
+      "loss": 0.792,
+      "step": 4568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.686701940333638e-07,
+      "loss": 0.3257,
+      "step": 4569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.668865716162703e-07,
+      "loss": 0.7896,
+      "step": 4570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.65105640872525e-07,
+      "loss": 0.793,
+      "step": 4571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.633274024935442e-07,
+      "loss": 0.8179,
+      "step": 4572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.615518571697023e-07,
+      "loss": 0.8428,
+      "step": 4573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.597790055903295e-07,
+      "loss": 0.7871,
+      "step": 4574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.580088484437021e-07,
+      "loss": 0.7842,
+      "step": 4575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.562413864170618e-07,
+      "loss": 0.7578,
+      "step": 4576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.544766201965948e-07,
+      "loss": 0.7939,
+      "step": 4577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.527145504674463e-07,
+      "loss": 0.8013,
+      "step": 4578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.5095517791371e-07,
+      "loss": 0.8062,
+      "step": 4579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.49198503218438e-07,
+      "loss": 0.8057,
+      "step": 4580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.474445270636286e-07,
+      "loss": 0.7485,
+      "step": 4581
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.456932501302372e-07,
+      "loss": 0.7554,
+      "step": 4582
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.439446730981718e-07,
+      "loss": 0.8335,
+      "step": 4583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.421987966462858e-07,
+      "loss": 0.7944,
+      "step": 4584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.40455621452392e-07,
+      "loss": 0.7305,
+      "step": 4585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.387151481932509e-07,
+      "loss": 0.7974,
+      "step": 4586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.369773775445727e-07,
+      "loss": 0.8281,
+      "step": 4587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.352423101810219e-07,
+      "loss": 0.8545,
+      "step": 4588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.335099467762073e-07,
+      "loss": 0.8252,
+      "step": 4589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.317802880026956e-07,
+      "loss": 0.8115,
+      "step": 4590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.300533345319999e-07,
+      "loss": 0.8193,
+      "step": 4591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.283290870345814e-07,
+      "loss": 0.7871,
+      "step": 4592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.266075461798527e-07,
+      "loss": 0.7578,
+      "step": 4593
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.248887126361762e-07,
+      "loss": 0.7437,
+      "step": 4594
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.231725870708612e-07,
+      "loss": 0.7881,
+      "step": 4595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.214591701501656e-07,
+      "loss": 0.7998,
+      "step": 4596
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.197484625393014e-07,
+      "loss": 0.8403,
+      "step": 4597
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.180404649024184e-07,
+      "loss": 0.7803,
+      "step": 4598
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.163351779026235e-07,
+      "loss": 0.7368,
+      "step": 4599
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.146326022019695e-07,
+      "loss": 0.7886,
+      "step": 4600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.129327384614496e-07,
+      "loss": 0.7607,
+      "step": 4601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.112355873410132e-07,
+      "loss": 0.8013,
+      "step": 4602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.095411494995525e-07,
+      "loss": 0.3552,
+      "step": 4603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.078494255949062e-07,
+      "loss": 0.7783,
+      "step": 4604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.061604162838591e-07,
+      "loss": 0.7798,
+      "step": 4605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.044741222221439e-07,
+      "loss": 0.8037,
+      "step": 4606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.027905440644371e-07,
+      "loss": 0.7847,
+      "step": 4607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.011096824643616e-07,
+      "loss": 0.7871,
+      "step": 4608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.99431538074488e-07,
+      "loss": 0.7861,
+      "step": 4609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.97756111546325e-07,
+      "loss": 0.7009,
+      "step": 4610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.960834035303357e-07,
+      "loss": 0.7944,
+      "step": 4611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.944134146759216e-07,
+      "loss": 0.7915,
+      "step": 4612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.927461456314281e-07,
+      "loss": 0.811,
+      "step": 4613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.910815970441499e-07,
+      "loss": 0.7065,
+      "step": 4614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.894197695603167e-07,
+      "loss": 0.731,
+      "step": 4615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.877606638251125e-07,
+      "loss": 0.7974,
+      "step": 4616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.861042804826563e-07,
+      "loss": 0.7935,
+      "step": 4617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.844506201760143e-07,
+      "loss": 0.7446,
+      "step": 4618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.827996835471932e-07,
+      "loss": 0.7729,
+      "step": 4619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.811514712371451e-07,
+      "loss": 0.7529,
+      "step": 4620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.795059838857615e-07,
+      "loss": 0.8071,
+      "step": 4621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.778632221318766e-07,
+      "loss": 0.7559,
+      "step": 4622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7622318661327073e-07,
+      "loss": 0.8242,
+      "step": 4623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7458587796665734e-07,
+      "loss": 0.7588,
+      "step": 4624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.729512968276986e-07,
+      "loss": 0.7852,
+      "step": 4625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7131944383099534e-07,
+      "loss": 0.7769,
+      "step": 4626
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.6969031961008825e-07,
+      "loss": 0.8096,
+      "step": 4627
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.6806392479745967e-07,
+      "loss": 0.8228,
+      "step": 4628
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.664402600245321e-07,
+      "loss": 0.7764,
+      "step": 4629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.6481932592166877e-07,
+      "loss": 0.7666,
+      "step": 4630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.632011231181713e-07,
+      "loss": 0.3143,
+      "step": 4631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.615856522422812e-07,
+      "loss": 0.8174,
+      "step": 4632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5997291392118114e-07,
+      "loss": 0.7734,
+      "step": 4633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5836290878099016e-07,
+      "loss": 0.7949,
+      "step": 4634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5675563744677073e-07,
+      "loss": 0.8477,
+      "step": 4635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5515110054251715e-07,
+      "loss": 0.7578,
+      "step": 4636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5354929869116933e-07,
+      "loss": 0.3474,
+      "step": 4637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5195023251459746e-07,
+      "loss": 0.7808,
+      "step": 4638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5035390263361815e-07,
+      "loss": 0.7832,
+      "step": 4639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.487603096679815e-07,
+      "loss": 0.7866,
+      "step": 4640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.471694542363719e-07,
+      "loss": 0.8115,
+      "step": 4641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.455813369564177e-07,
+      "loss": 0.771,
+      "step": 4642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.43995958444679e-07,
+      "loss": 0.7866,
+      "step": 4643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4241331931665435e-07,
+      "loss": 0.7471,
+      "step": 4644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.408334201867789e-07,
+      "loss": 0.7466,
+      "step": 4645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.392562616684237e-07,
+      "loss": 0.748,
+      "step": 4646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.376818443738964e-07,
+      "loss": 0.7573,
+      "step": 4647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3611016891443896e-07,
+      "loss": 0.7812,
+      "step": 4648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.345412359002332e-07,
+      "loss": 0.79,
+      "step": 4649
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3297504594038776e-07,
+      "loss": 0.7402,
+      "step": 4650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.314115996429563e-07,
+      "loss": 0.748,
+      "step": 4651
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.2985089761491985e-07,
+      "loss": 0.8159,
+      "step": 4652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.2829294046219714e-07,
+      "loss": 0.8096,
+      "step": 4653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.2673772878964094e-07,
+      "loss": 0.8511,
+      "step": 4654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.2518526320103833e-07,
+      "loss": 0.813,
+      "step": 4655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.2363554429910966e-07,
+      "loss": 0.8325,
+      "step": 4656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.2208857268550784e-07,
+      "loss": 0.8154,
+      "step": 4657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.2054434896082407e-07,
+      "loss": 0.7754,
+      "step": 4658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1900287372457546e-07,
+      "loss": 0.8096,
+      "step": 4659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1746414757521583e-07,
+      "loss": 0.7827,
+      "step": 4660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1592817111013565e-07,
+      "loss": 0.769,
+      "step": 4661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.143949449256487e-07,
+      "loss": 0.3154,
+      "step": 4662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.128644696170106e-07,
+      "loss": 0.7998,
+      "step": 4663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.113367457784009e-07,
+      "loss": 0.7993,
+      "step": 4664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.0981177400293695e-07,
+      "loss": 0.8218,
+      "step": 4665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.082895548826651e-07,
+      "loss": 0.7979,
+      "step": 4666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.0677008900856264e-07,
+      "loss": 0.7803,
+      "step": 4667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.0525337697053853e-07,
+      "loss": 0.812,
+      "step": 4668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.037394193574334e-07,
+      "loss": 0.8184,
+      "step": 4669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.022282167570163e-07,
+      "loss": 0.7256,
+      "step": 4670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.0071976975598947e-07,
+      "loss": 0.7729,
+      "step": 4671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9921407893998286e-07,
+      "loss": 0.7524,
+      "step": 4672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.977111448935589e-07,
+      "loss": 0.8428,
+      "step": 4673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9621096820020576e-07,
+      "loss": 0.707,
+      "step": 4674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9471354944234763e-07,
+      "loss": 0.7651,
+      "step": 4675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.932188892013294e-07,
+      "loss": 0.7935,
+      "step": 4676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9172698805743366e-07,
+      "loss": 0.8003,
+      "step": 4677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.902378465898654e-07,
+      "loss": 0.7349,
+      "step": 4678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.887514653767621e-07,
+      "loss": 0.7485,
+      "step": 4679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.872678449951864e-07,
+      "loss": 0.7402,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.8578698602113325e-07,
+      "loss": 0.855,
+      "step": 4681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.8430888902952045e-07,
+      "loss": 0.8052,
+      "step": 4682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.828335545941972e-07,
+      "loss": 0.749,
+      "step": 4683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.8136098328794026e-07,
+      "loss": 0.7266,
+      "step": 4684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.798911756824516e-07,
+      "loss": 0.3365,
+      "step": 4685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.7842413234835975e-07,
+      "loss": 0.8052,
+      "step": 4686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.769598538552241e-07,
+      "loss": 0.3092,
+      "step": 4687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.754983407715257e-07,
+      "loss": 0.7554,
+      "step": 4688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.740395936646765e-07,
+      "loss": 0.8179,
+      "step": 4689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.7258361310100936e-07,
+      "loss": 0.7661,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.7113039964578796e-07,
+      "loss": 0.2942,
+      "step": 4691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.696799538632009e-07,
+      "loss": 0.7891,
+      "step": 4692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.682322763163587e-07,
+      "loss": 0.7344,
+      "step": 4693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.667873675673017e-07,
+      "loss": 0.7896,
+      "step": 4694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.65345228176992e-07,
+      "loss": 0.7842,
+      "step": 4695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.639058587053176e-07,
+      "loss": 0.7778,
+      "step": 4696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.6246925971109237e-07,
+      "loss": 0.7505,
+      "step": 4697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.6103543175205255e-07,
+      "loss": 0.7407,
+      "step": 4698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5960437538485964e-07,
+      "loss": 0.77,
+      "step": 4699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5817609116509917e-07,
+      "loss": 0.7979,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.567505796472817e-07,
+      "loss": 0.7734,
+      "step": 4701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.553278413848363e-07,
+      "loss": 0.8062,
+      "step": 4702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5390787693012264e-07,
+      "loss": 0.3232,
+      "step": 4703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5249068683441745e-07,
+      "loss": 0.8047,
+      "step": 4704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.510762716479246e-07,
+      "loss": 0.8408,
+      "step": 4705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4966463191976744e-07,
+      "loss": 0.7417,
+      "step": 4706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.482557681979934e-07,
+      "loss": 0.3068,
+      "step": 4707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4684968102957153e-07,
+      "loss": 0.7612,
+      "step": 4708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4544637096039277e-07,
+      "loss": 0.7734,
+      "step": 4709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.440458385352738e-07,
+      "loss": 0.7627,
+      "step": 4710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4264808429794543e-07,
+      "loss": 0.3014,
+      "step": 4711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4125310879106616e-07,
+      "loss": 0.8247,
+      "step": 4712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.398609125562127e-07,
+      "loss": 0.8086,
+      "step": 4713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.384714961338828e-07,
+      "loss": 0.8584,
+      "step": 4714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.3708486006349904e-07,
+      "loss": 0.7632,
+      "step": 4715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.3570100488339666e-07,
+      "loss": 0.8042,
+      "step": 4716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.3431993113083923e-07,
+      "loss": 0.7896,
+      "step": 4717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.329416393420062e-07,
+      "loss": 0.288,
+      "step": 4718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.3156613005199786e-07,
+      "loss": 0.8271,
+      "step": 4719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.301934037948337e-07,
+      "loss": 0.7505,
+      "step": 4720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.2882346110345407e-07,
+      "loss": 0.8442,
+      "step": 4721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.274563025097185e-07,
+      "loss": 0.728,
+      "step": 4722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.260919285444025e-07,
+      "loss": 0.717,
+      "step": 4723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.247303397372064e-07,
+      "loss": 0.8696,
+      "step": 4724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.233715366167442e-07,
+      "loss": 0.7729,
+      "step": 4725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.220155197105487e-07,
+      "loss": 0.771,
+      "step": 4726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.2066228954507476e-07,
+      "loss": 0.7607,
+      "step": 4727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.1931184664569125e-07,
+      "loss": 0.7764,
+      "step": 4728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.179641915366871e-07,
+      "loss": 0.8032,
+      "step": 4729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.1661932474126884e-07,
+      "loss": 0.7729,
+      "step": 4730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.152772467815604e-07,
+      "loss": 0.7297,
+      "step": 4731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.1393795817860174e-07,
+      "loss": 0.8374,
+      "step": 4732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.126014594523502e-07,
+      "loss": 0.8037,
+      "step": 4733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.112677511216824e-07,
+      "loss": 0.7246,
+      "step": 4734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.099368337043876e-07,
+      "loss": 0.7656,
+      "step": 4735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.086087077171773e-07,
+      "loss": 0.8115,
+      "step": 4736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.072833736756725e-07,
+      "loss": 0.7939,
+      "step": 4737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.0596083209441407e-07,
+      "loss": 0.8345,
+      "step": 4738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.046410834868588e-07,
+      "loss": 0.3037,
+      "step": 4739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.033241283653776e-07,
+      "loss": 0.7554,
+      "step": 4740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.0200996724126074e-07,
+      "loss": 0.8242,
+      "step": 4741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.0069860062470675e-07,
+      "loss": 0.7646,
+      "step": 4742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.9939002902483584e-07,
+      "loss": 0.791,
+      "step": 4743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.9808425294967995e-07,
+      "loss": 0.7236,
+      "step": 4744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.967812729061861e-07,
+      "loss": 0.7837,
+      "step": 4745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.954810894002163e-07,
+      "loss": 0.7603,
+      "step": 4746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.9418370293654685e-07,
+      "loss": 0.8022,
+      "step": 4747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.928891140188672e-07,
+      "loss": 0.7998,
+      "step": 4748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.915973231497804e-07,
+      "loss": 0.7271,
+      "step": 4749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.903083308308077e-07,
+      "loss": 0.8174,
+      "step": 4750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.890221375623764e-07,
+      "loss": 0.813,
+      "step": 4751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8773874384383196e-07,
+      "loss": 0.8145,
+      "step": 4752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8645815017343423e-07,
+      "loss": 0.7812,
+      "step": 4753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.851803570483505e-07,
+      "loss": 0.3406,
+      "step": 4754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.839053649646664e-07,
+      "loss": 0.7681,
+      "step": 4755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8263317441737766e-07,
+      "loss": 0.7539,
+      "step": 4756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8136378590039093e-07,
+      "loss": 0.8052,
+      "step": 4757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.800971999065277e-07,
+      "loss": 0.8481,
+      "step": 4758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.788334169275206e-07,
+      "loss": 0.7534,
+      "step": 4759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7757243745401285e-07,
+      "loss": 0.7734,
+      "step": 4760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.763142619755588e-07,
+      "loss": 0.8042,
+      "step": 4761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7505889098062923e-07,
+      "loss": 0.772,
+      "step": 4762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.738063249565986e-07,
+      "loss": 0.8403,
+      "step": 4763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7255656438975954e-07,
+      "loss": 0.7432,
+      "step": 4764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7130960976530975e-07,
+      "loss": 0.7568,
+      "step": 4765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.700654615673587e-07,
+      "loss": 0.7632,
+      "step": 4766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6882412027893156e-07,
+      "loss": 0.793,
+      "step": 4767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.675855863819562e-07,
+      "loss": 0.8257,
+      "step": 4768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6634986035727645e-07,
+      "loss": 0.7852,
+      "step": 4769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6511694268464255e-07,
+      "loss": 0.7354,
+      "step": 4770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.6388683384271585e-07,
+      "loss": 0.7324,
+      "step": 4771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.626595343090668e-07,
+      "loss": 0.8247,
+      "step": 4772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.61435044560176e-07,
+      "loss": 0.814,
+      "step": 4773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.602133650714314e-07,
+      "loss": 0.7363,
+      "step": 4774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.5899449631713117e-07,
+      "loss": 0.8145,
+      "step": 4775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.577784387704835e-07,
+      "loss": 0.7715,
+      "step": 4776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.565651929036031e-07,
+      "loss": 0.7827,
+      "step": 4777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.553547591875116e-07,
+      "loss": 0.7832,
+      "step": 4778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.541471380921456e-07,
+      "loss": 0.7368,
+      "step": 4779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.5294233008634095e-07,
+      "loss": 0.7837,
+      "step": 4780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.517403356378475e-07,
+      "loss": 0.7319,
+      "step": 4781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.505411552133213e-07,
+      "loss": 0.8198,
+      "step": 4782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.493447892783249e-07,
+      "loss": 0.7729,
+      "step": 4783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.481512382973283e-07,
+      "loss": 0.7725,
+      "step": 4784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.4696050273370937e-07,
+      "loss": 0.7231,
+      "step": 4785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.457725830497523e-07,
+      "loss": 0.7168,
+      "step": 4786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.4458747970664995e-07,
+      "loss": 0.7856,
+      "step": 4787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.434051931644984e-07,
+      "loss": 0.772,
+      "step": 4788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.4222572388230225e-07,
+      "loss": 0.7769,
+      "step": 4789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.4104907231797424e-07,
+      "loss": 0.751,
+      "step": 4790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.398752389283282e-07,
+      "loss": 0.7207,
+      "step": 4791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3870422416908706e-07,
+      "loss": 0.8184,
+      "step": 4792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3753602849488131e-07,
+      "loss": 0.7651,
+      "step": 4793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3637065235924237e-07,
+      "loss": 0.7705,
+      "step": 4794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.352080962146108e-07,
+      "loss": 0.7822,
+      "step": 4795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3404836051232975e-07,
+      "loss": 0.7959,
+      "step": 4796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3289144570264987e-07,
+      "loss": 0.7866,
+      "step": 4797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3173735223472436e-07,
+      "loss": 0.7676,
+      "step": 4798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3058608055661235e-07,
+      "loss": 0.7988,
+      "step": 4799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.294376311152771e-07,
+      "loss": 0.8311,
+      "step": 4800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2829200435658616e-07,
+      "loss": 0.8057,
+      "step": 4801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2714920072531204e-07,
+      "loss": 0.8301,
+      "step": 4802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2600922066512902e-07,
+      "loss": 0.7095,
+      "step": 4803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2487206461861725e-07,
+      "loss": 0.811,
+      "step": 4804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.237377330272611e-07,
+      "loss": 0.792,
+      "step": 4805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.226062263314449e-07,
+      "loss": 0.7603,
+      "step": 4806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.214775449704598e-07,
+      "loss": 0.8145,
+      "step": 4807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.203516893824986e-07,
+      "loss": 0.8154,
+      "step": 4808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.192286600046575e-07,
+      "loss": 0.3057,
+      "step": 4809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1810845727293527e-07,
+      "loss": 0.8364,
+      "step": 4810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.169910816222323e-07,
+      "loss": 0.7808,
+      "step": 4811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.158765334863533e-07,
+      "loss": 0.7505,
+      "step": 4812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.147648132980029e-07,
+      "loss": 0.7593,
+      "step": 4813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1365592148879092e-07,
+      "loss": 0.7588,
+      "step": 4814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1254985848922465e-07,
+      "loss": 0.8037,
+      "step": 4815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1144662472871972e-07,
+      "loss": 0.7842,
+      "step": 4816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1034622063558607e-07,
+      "loss": 0.7808,
+      "step": 4817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.092486466370394e-07,
+      "loss": 0.7651,
+      "step": 4818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0815390315919724e-07,
+      "loss": 0.7944,
+      "step": 4819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.070619906270746e-07,
+      "loss": 0.7852,
+      "step": 4820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.059729094645907e-07,
+      "loss": 0.7974,
+      "step": 4821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0488666009456396e-07,
+      "loss": 0.7759,
+      "step": 4822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0380324293871377e-07,
+      "loss": 0.8271,
+      "step": 4823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.027226584176603e-07,
+      "loss": 0.3149,
+      "step": 4824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.016449069509238e-07,
+      "loss": 0.7832,
+      "step": 4825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0056998895692376e-07,
+      "loss": 0.7808,
+      "step": 4826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9949790485298047e-07,
+      "loss": 0.7944,
+      "step": 4827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9842865505531522e-07,
+      "loss": 0.792,
+      "step": 4828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9736223997904505e-07,
+      "loss": 0.8159,
+      "step": 4829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9629866003819135e-07,
+      "loss": 0.7505,
+      "step": 4830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9523791564567212e-07,
+      "loss": 0.7686,
+      "step": 4831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9418000721330292e-07,
+      "loss": 0.7808,
+      "step": 4832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9312493515180274e-07,
+      "loss": 0.8184,
+      "step": 4833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9207269987078557e-07,
+      "loss": 0.7197,
+      "step": 4834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.910233017787655e-07,
+      "loss": 0.7944,
+      "step": 4835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8997674128315572e-07,
+      "loss": 0.7632,
+      "step": 4836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.889330187902655e-07,
+      "loss": 0.7959,
+      "step": 4837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8789213470530558e-07,
+      "loss": 0.8076,
+      "step": 4838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8685408943238196e-07,
+      "loss": 0.7793,
+      "step": 4839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8581888337449977e-07,
+      "loss": 0.7832,
+      "step": 4840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.847865169335608e-07,
+      "loss": 0.7769,
+      "step": 4841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8375699051036783e-07,
+      "loss": 0.8105,
+      "step": 4842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8273030450461608e-07,
+      "loss": 0.3164,
+      "step": 4843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8170645931490093e-07,
+      "loss": 0.7549,
+      "step": 4844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8068545533871523e-07,
+      "loss": 0.8213,
+      "step": 4845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7966729297244522e-07,
+      "loss": 0.8164,
+      "step": 4846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7865197261138055e-07,
+      "loss": 0.7456,
+      "step": 4847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.776394946497001e-07,
+      "loss": 0.7856,
+      "step": 4848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7662985948048522e-07,
+      "loss": 0.8198,
+      "step": 4849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7562306749570905e-07,
+      "loss": 0.769,
+      "step": 4850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7461911908624478e-07,
+      "loss": 0.7847,
+      "step": 4851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7361801464185813e-07,
+      "loss": 0.8027,
+      "step": 4852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7261975455121148e-07,
+      "loss": 0.3435,
+      "step": 4853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7162433920186737e-07,
+      "loss": 0.8564,
+      "step": 4854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.706317689802775e-07,
+      "loss": 0.7004,
+      "step": 4855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6964204427179191e-07,
+      "loss": 0.7925,
+      "step": 4856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6865516546065745e-07,
+      "loss": 0.8193,
+      "step": 4857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6767113293001178e-07,
+      "loss": 0.8052,
+      "step": 4858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.666899470618935e-07,
+      "loss": 0.812,
+      "step": 4859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6571160823722953e-07,
+      "loss": 0.8091,
+      "step": 4860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6473611683584772e-07,
+      "loss": 0.7485,
+      "step": 4861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6376347323646429e-07,
+      "loss": 0.749,
+      "step": 4862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6279367781669552e-07,
+      "loss": 0.7998,
+      "step": 4863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6182673095304855e-07,
+      "loss": 0.7915,
+      "step": 4864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608626330209248e-07,
+      "loss": 0.7671,
+      "step": 4865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.599013843946215e-07,
+      "loss": 0.7397,
+      "step": 4866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5894298544732678e-07,
+      "loss": 0.814,
+      "step": 4867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5798743655112635e-07,
+      "loss": 0.7598,
+      "step": 4868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5703473807699432e-07,
+      "loss": 0.8291,
+      "step": 4869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5608489039480395e-07,
+      "loss": 0.8184,
+      "step": 4870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.551378938733178e-07,
+      "loss": 0.8062,
+      "step": 4871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.541937488801909e-07,
+      "loss": 0.7837,
+      "step": 4872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5325245578197506e-07,
+      "loss": 0.7734,
+      "step": 4873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5231401494411213e-07,
+      "loss": 0.7896,
+      "step": 4874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.513784267309365e-07,
+      "loss": 0.7734,
+      "step": 4875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5044569150567593e-07,
+      "loss": 0.8501,
+      "step": 4876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4951580963044998e-07,
+      "loss": 0.8301,
+      "step": 4877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.485887814662723e-07,
+      "loss": 0.8369,
+      "step": 4878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4766460737304506e-07,
+      "loss": 0.7983,
+      "step": 4879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4674328770956703e-07,
+      "loss": 0.7251,
+      "step": 4880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4582482283352294e-07,
+      "loss": 0.3525,
+      "step": 4881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.449092131014959e-07,
+      "loss": 0.8188,
+      "step": 4882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4399645886895573e-07,
+      "loss": 0.7529,
+      "step": 4883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4308656049026313e-07,
+      "loss": 0.8081,
+      "step": 4884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.421795183186747e-07,
+      "loss": 0.8115,
+      "step": 4885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.412753327063346e-07,
+      "loss": 0.7441,
+      "step": 4886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4037400400427863e-07,
+      "loss": 0.3013,
+      "step": 4887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.394755325624336e-07,
+      "loss": 0.7563,
+      "step": 4888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3857991872961635e-07,
+      "loss": 0.7188,
+      "step": 4889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.37687162853537e-07,
+      "loss": 0.8345,
+      "step": 4890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3679726528079167e-07,
+      "loss": 0.3381,
+      "step": 4891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.359102263568715e-07,
+      "loss": 0.3378,
+      "step": 4892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3502604642615351e-07,
+      "loss": 0.7861,
+      "step": 4893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3414472583190896e-07,
+      "loss": 0.7842,
+      "step": 4894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3326626491629412e-07,
+      "loss": 0.7842,
+      "step": 4895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3239066402036038e-07,
+      "loss": 0.8223,
+      "step": 4896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3151792348404578e-07,
+      "loss": 0.2891,
+      "step": 4897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.30648043646176e-07,
+      "loss": 0.752,
+      "step": 4898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2978102484447085e-07,
+      "loss": 0.7539,
+      "step": 4899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.289168674155361e-07,
+      "loss": 0.7969,
+      "step": 4900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.280555716948667e-07,
+      "loss": 0.8174,
+      "step": 4901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.271971380168477e-07,
+      "loss": 0.7524,
+      "step": 4902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.263415667147533e-07,
+      "loss": 0.8203,
+      "step": 4903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.254888581207453e-07,
+      "loss": 0.7769,
+      "step": 4904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2463901256587307e-07,
+      "loss": 0.7954,
+      "step": 4905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2379203038007936e-07,
+      "loss": 0.7969,
+      "step": 4906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.229479118921878e-07,
+      "loss": 0.8564,
+      "step": 4907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2210665742991705e-07,
+      "loss": 0.769,
+      "step": 4908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.212682673198709e-07,
+      "loss": 0.7998,
+      "step": 4909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2043274188753895e-07,
+      "loss": 0.7959,
+      "step": 4910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.196000814573042e-07,
+      "loss": 0.8198,
+      "step": 4911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1877028635243147e-07,
+      "loss": 0.7993,
+      "step": 4912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1794335689507718e-07,
+      "loss": 0.7603,
+      "step": 4913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1711929340628286e-07,
+      "loss": 0.7939,
+      "step": 4914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1629809620597925e-07,
+      "loss": 0.7441,
+      "step": 4915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1547976561298301e-07,
+      "loss": 0.8491,
+      "step": 4916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1466430194499832e-07,
+      "loss": 0.7139,
+      "step": 4917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1385170551861613e-07,
+      "loss": 0.7607,
+      "step": 4918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1304197664931404e-07,
+      "loss": 0.771,
+      "step": 4919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1223511565145811e-07,
+      "loss": 0.7998,
+      "step": 4920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1143112283829693e-07,
+      "loss": 0.772,
+      "step": 4921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.106299985219708e-07,
+      "loss": 0.7988,
+      "step": 4922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0983174301350263e-07,
+      "loss": 0.8398,
+      "step": 4923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0903635662280198e-07,
+      "loss": 0.7715,
+      "step": 4924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0824383965866686e-07,
+      "loss": 0.8018,
+      "step": 4925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0745419242877862e-07,
+      "loss": 0.7715,
+      "step": 4926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0666741523970541e-07,
+      "loss": 0.7983,
+      "step": 4927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0588350839690202e-07,
+      "loss": 0.8042,
+      "step": 4928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.051024722047067e-07,
+      "loss": 0.3037,
+      "step": 4929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0432430696634603e-07,
+      "loss": 0.8188,
+      "step": 4930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0354901298393004e-07,
+      "loss": 0.793,
+      "step": 4931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0277659055845462e-07,
+      "loss": 0.7949,
+      "step": 4932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0200703998979988e-07,
+      "loss": 0.7861,
+      "step": 4933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.012403615767335e-07,
+      "loss": 0.7964,
+      "step": 4934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0047655561690489e-07,
+      "loss": 0.7949,
+      "step": 4935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.971562240685017e-08,
+      "loss": 0.7979,
+      "step": 4936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.895756224199137e-08,
+      "loss": 0.7808,
+      "step": 4937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.820237541663141e-08,
+      "loss": 0.7642,
+      "step": 4938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.745006222396074e-08,
+      "loss": 0.7461,
+      "step": 4939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.670062295605242e-08,
+      "loss": 0.813,
+      "step": 4940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.595405790386619e-08,
+      "loss": 0.7949,
+      "step": 4941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.521036735724353e-08,
+      "loss": 0.8096,
+      "step": 4942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.446955160491016e-08,
+      "loss": 0.7622,
+      "step": 4943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.373161093447769e-08,
+      "loss": 0.7847,
+      "step": 4944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.299654563243776e-08,
+      "loss": 0.8525,
+      "step": 4945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.226435598417043e-08,
+      "loss": 0.7715,
+      "step": 4946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.1535042273935e-08,
+      "loss": 0.7876,
+      "step": 4947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.080860478487829e-08,
+      "loss": 0.8042,
+      "step": 4948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.008504379902721e-08,
+      "loss": 0.811,
+      "step": 4949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.936435959729122e-08,
+      "loss": 0.7886,
+      "step": 4950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.864655245946818e-08,
+      "loss": 0.7437,
+      "step": 4951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.793162266423266e-08,
+      "loss": 0.8076,
+      "step": 4952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.721957048914681e-08,
+      "loss": 0.7559,
+      "step": 4953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.651039621065366e-08,
+      "loss": 0.7988,
+      "step": 4954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.580410010407797e-08,
+      "loss": 0.7881,
+      "step": 4955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.510068244362873e-08,
+      "loss": 0.8413,
+      "step": 4956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.440014350239749e-08,
+      "loss": 0.7754,
+      "step": 4957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.370248355235754e-08,
+      "loss": 0.8149,
+      "step": 4958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.300770286436305e-08,
+      "loss": 0.7515,
+      "step": 4959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.231580170815323e-08,
+      "loss": 0.8091,
+      "step": 4960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.162678035234739e-08,
+      "loss": 0.793,
+      "step": 4961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.09406390644482e-08,
+      "loss": 0.7153,
+      "step": 4962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.025737811083844e-08,
+      "loss": 0.7886,
+      "step": 4963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.957699775678257e-08,
+      "loss": 0.7603,
+      "step": 4964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.889949826643011e-08,
+      "loss": 0.7612,
+      "step": 4965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.822487990280819e-08,
+      "loss": 0.7236,
+      "step": 4966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.755314292782645e-08,
+      "loss": 0.7861,
+      "step": 4967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.688428760227794e-08,
+      "loss": 0.7754,
+      "step": 4968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.62183141858333e-08,
+      "loss": 0.7891,
+      "step": 4969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.55552229370482e-08,
+      "loss": 0.7651,
+      "step": 4970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.489501411335586e-08,
+      "loss": 0.8213,
+      "step": 4971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.423768797107377e-08,
+      "loss": 0.7661,
+      "step": 4972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.358324476539696e-08,
+      "loss": 0.7764,
+      "step": 4973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.293168475040385e-08,
+      "loss": 0.8052,
+      "step": 4974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.228300817905209e-08,
+      "loss": 0.3179,
+      "step": 4975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.163721530318106e-08,
+      "loss": 0.8032,
+      "step": 4976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.099430637350773e-08,
+      "loss": 0.8257,
+      "step": 4977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.03542816396341e-08,
+      "loss": 0.7773,
+      "step": 4978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.971714135003804e-08,
+      "loss": 0.7881,
+      "step": 4979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.908288575208005e-08,
+      "loss": 0.8003,
+      "step": 4980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.845151509199899e-08,
+      "loss": 0.8237,
+      "step": 4981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.782302961491627e-08,
+      "loss": 0.8247,
+      "step": 4982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.719742956483005e-08,
+      "loss": 0.7563,
+      "step": 4983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.657471518462188e-08,
+      "loss": 0.7861,
+      "step": 4984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.595488671604838e-08,
+      "loss": 0.7808,
+      "step": 4985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.533794439975038e-08,
+      "loss": 0.8379,
+      "step": 4986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.472388847524463e-08,
+      "loss": 0.2893,
+      "step": 4987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.411271918092959e-08,
+      "loss": 0.8379,
+      "step": 4988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.350443675408296e-08,
+      "loss": 0.7158,
+      "step": 4989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.289904143085833e-08,
+      "loss": 0.7798,
+      "step": 4990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.229653344629355e-08,
+      "loss": 0.8057,
+      "step": 4991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.169691303430231e-08,
+      "loss": 0.7817,
+      "step": 4992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.110018042767756e-08,
+      "loss": 0.7871,
+      "step": 4993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.050633585809152e-08,
+      "loss": 0.811,
+      "step": 4994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.991537955609556e-08,
+      "loss": 0.7603,
+      "step": 4995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.9327311751118664e-08,
+      "loss": 0.3401,
+      "step": 4996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.8742132671468216e-08,
+      "loss": 0.791,
+      "step": 4997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.815984254433332e-08,
+      "loss": 0.7866,
+      "step": 4998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.758044159577647e-08,
+      "loss": 0.8105,
+      "step": 4999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.7003930050743556e-08,
+      "loss": 0.7886,
+      "step": 5000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.643030813305389e-08,
+      "loss": 0.7632,
+      "step": 5001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.585957606540848e-08,
+      "loss": 0.8115,
+      "step": 5002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.529173406938509e-08,
+      "loss": 0.7539,
+      "step": 5003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.4726782365439045e-08,
+      "loss": 0.7397,
+      "step": 5004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.416472117290405e-08,
+      "loss": 0.79,
+      "step": 5005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3605550709992234e-08,
+      "loss": 0.7881,
+      "step": 5006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3049271193792434e-08,
+      "loss": 0.7817,
+      "step": 5007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.24958828402719e-08,
+      "loss": 0.7778,
+      "step": 5008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.194538586427461e-08,
+      "loss": 0.8301,
+      "step": 5009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1397780479522925e-08,
+      "loss": 0.7549,
+      "step": 5010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0853066898614296e-08,
+      "loss": 0.9019,
+      "step": 5011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.031124533302872e-08,
+      "loss": 0.7983,
+      "step": 5012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.977231599311627e-08,
+      "loss": 0.7808,
+      "step": 5013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.92362790881104e-08,
+      "loss": 0.7988,
+      "step": 5014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.870313482611882e-08,
+      "loss": 0.3087,
+      "step": 5015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8172883414125125e-08,
+      "loss": 0.7979,
+      "step": 5016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7645525057992137e-08,
+      "loss": 0.7725,
+      "step": 5017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.712105996245858e-08,
+      "loss": 0.8677,
+      "step": 5018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6599488331139917e-08,
+      "loss": 0.7686,
+      "step": 5019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.60808103665275e-08,
+      "loss": 0.7705,
+      "step": 5020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5565026269991097e-08,
+      "loss": 0.7632,
+      "step": 5021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.505213624177468e-08,
+      "loss": 0.7915,
+      "step": 5022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.454214048100064e-08,
+      "loss": 0.7832,
+      "step": 5023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4035039185666425e-08,
+      "loss": 0.7632,
+      "step": 5024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.353083255264623e-08,
+      "loss": 0.7656,
+      "step": 5025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3029520777690954e-08,
+      "loss": 0.8374,
+      "step": 5026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.253110405542659e-08,
+      "loss": 0.8066,
+      "step": 5027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.203558257935586e-08,
+      "loss": 0.7979,
+      "step": 5028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.154295654185653e-08,
+      "loss": 0.7524,
+      "step": 5029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.105322613418477e-08,
+      "loss": 0.7998,
+      "step": 5030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.056639154646935e-08,
+      "loss": 0.7393,
+      "step": 5031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.008245296771573e-08,
+      "loss": 0.7881,
+      "step": 5032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9601410585806964e-08,
+      "loss": 0.2821,
+      "step": 5033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9123264587500316e-08,
+      "loss": 0.7935,
+      "step": 5034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.864801515842647e-08,
+      "loss": 0.7612,
+      "step": 5035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.817566248309618e-08,
+      "loss": 0.7905,
+      "step": 5036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.770620674489106e-08,
+      "loss": 0.7866,
+      "step": 5037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7239648126072e-08,
+      "loss": 0.79,
+      "step": 5038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.677598680777078e-08,
+      "loss": 0.7539,
+      "step": 5039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.63152229699984e-08,
+      "loss": 0.7563,
+      "step": 5040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.585735679163843e-08,
+      "loss": 0.79,
+      "step": 5041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5402388450450354e-08,
+      "loss": 0.7915,
+      "step": 5042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.49503181230687e-08,
+      "loss": 0.7896,
+      "step": 5043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.450114598500309e-08,
+      "loss": 0.8145,
+      "step": 5044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.405487221063652e-08,
+      "loss": 0.7842,
+      "step": 5045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.361149697322957e-08,
+      "loss": 0.8325,
+      "step": 5046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3171020444914544e-08,
+      "loss": 0.8018,
+      "step": 5047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.273344279669882e-08,
+      "loss": 0.813,
+      "step": 5048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.229876419846733e-08,
+      "loss": 0.7693,
+      "step": 5049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.186698481897593e-08,
+      "loss": 0.7686,
+      "step": 5050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1438104825855516e-08,
+      "loss": 0.8076,
+      "step": 5051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.101212438561374e-08,
+      "loss": 0.7646,
+      "step": 5052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0589043663629966e-08,
+      "loss": 0.8276,
+      "step": 5053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.016886282415865e-08,
+      "loss": 0.793,
+      "step": 5054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.975158203032846e-08,
+      "loss": 0.8076,
+      "step": 5055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9337201444142313e-08,
+      "loss": 0.7578,
+      "step": 5056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8925721226477354e-08,
+      "loss": 0.791,
+      "step": 5057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.851714153708329e-08,
+      "loss": 0.8145,
+      "step": 5058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8111462534586575e-08,
+      "loss": 0.7959,
+      "step": 5059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.770868437648455e-08,
+      "loss": 0.7661,
+      "step": 5060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7308807219149633e-08,
+      "loss": 0.7266,
+      "step": 5061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6911831217827644e-08,
+      "loss": 0.7769,
+      "step": 5062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6517756526639457e-08,
+      "loss": 0.8496,
+      "step": 5063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6126583298576866e-08,
+      "loss": 0.8164,
+      "step": 5064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5738311685508388e-08,
+      "loss": 0.7954,
+      "step": 5065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.535294183817344e-08,
+      "loss": 0.772,
+      "step": 5066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4970473906185675e-08,
+      "loss": 0.8262,
+      "step": 5067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4590908038033823e-08,
+      "loss": 0.7603,
+      "step": 5068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4214244381076668e-08,
+      "loss": 0.8018,
+      "step": 5069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.384048308154807e-08,
+      "loss": 0.8335,
+      "step": 5070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3469624284556957e-08,
+      "loss": 0.8066,
+      "step": 5071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3101668134080654e-08,
+      "loss": 0.7773,
+      "step": 5072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2736614772975728e-08,
+      "loss": 0.8228,
+      "step": 5073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.237446434296547e-08,
+      "loss": 0.8037,
+      "step": 5074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2015216984649922e-08,
+      "loss": 0.771,
+      "step": 5075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.165887283750334e-08,
+      "loss": 0.8057,
+      "step": 5076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1305432039867568e-08,
+      "loss": 0.7871,
+      "step": 5077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0954894728962838e-08,
+      "loss": 0.3352,
+      "step": 5078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.060726104087862e-08,
+      "loss": 0.8076,
+      "step": 5079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0262531110579463e-08,
+      "loss": 0.7993,
+      "step": 5080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.992070507189997e-08,
+      "loss": 0.7695,
+      "step": 5081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9581783057549817e-08,
+      "loss": 0.7939,
+      "step": 5082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.924576519910959e-08,
+      "loss": 0.7744,
+      "step": 5083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8912651627032428e-08,
+      "loss": 0.8354,
+      "step": 5084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8582442470646554e-08,
+      "loss": 0.8325,
+      "step": 5085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.825513785814942e-08,
+      "loss": 0.8403,
+      "step": 5086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.793073791661104e-08,
+      "loss": 0.7466,
+      "step": 5087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7609242771975677e-08,
+      "loss": 0.7539,
+      "step": 5088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.729065254905765e-08,
+      "loss": 0.7393,
+      "step": 5089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6974967371547188e-08,
+      "loss": 0.8066,
+      "step": 5090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.666218736200126e-08,
+      "loss": 0.7852,
+      "step": 5091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.635231264185355e-08,
+      "loss": 0.7734,
+      "step": 5092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6045343331407003e-08,
+      "loss": 0.7373,
+      "step": 5093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5741279549838783e-08,
+      "loss": 0.7671,
+      "step": 5094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5440121415196962e-08,
+      "loss": 0.8003,
+      "step": 5095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5141869044400513e-08,
+      "loss": 0.77,
+      "step": 5096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4846522553240983e-08,
+      "loss": 0.7563,
+      "step": 5097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4554082056383311e-08,
+      "loss": 0.8462,
+      "step": 5098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4264547667362516e-08,
+      "loss": 0.8208,
+      "step": 5099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3977919498586178e-08,
+      "loss": 0.793,
+      "step": 5100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3694197661331954e-08,
+      "loss": 0.811,
+      "step": 5101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3413382265751728e-08,
+      "loss": 0.7935,
+      "step": 5102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3135473420867461e-08,
+      "loss": 0.811,
+      "step": 5103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2860471234572014e-08,
+      "loss": 0.8262,
+      "step": 5104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.258837581363248e-08,
+      "loss": 0.8789,
+      "step": 5105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.231918726368353e-08,
+      "loss": 0.7852,
+      "step": 5106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2052905689234895e-08,
+      "loss": 0.7607,
+      "step": 5107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1789531193665548e-08,
+      "loss": 0.8354,
+      "step": 5108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.152906387922703e-08,
+      "loss": 0.7378,
+      "step": 5109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1271503847041787e-08,
+      "loss": 0.8071,
+      "step": 5110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1016851197103161e-08,
+      "loss": 0.7495,
+      "step": 5111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.076510602827624e-08,
+      "loss": 0.8691,
+      "step": 5112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0516268438297006e-08,
+      "loss": 0.8018,
+      "step": 5113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.027033852377235e-08,
+      "loss": 0.73,
+      "step": 5114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0027316380181728e-08,
+      "loss": 0.7856,
+      "step": 5115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.787202101873837e-09,
+      "loss": 0.3442,
+      "step": 5116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.549995782069942e-09,
+      "loss": 0.793,
+      "step": 5117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.315697512861376e-09,
+      "loss": 0.7725,
+      "step": 5118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.084307385210377e-09,
+      "loss": 0.7646,
+      "step": 5119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.855825488951751e-09,
+      "loss": 0.7822,
+      "step": 5120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.630251912789543e-09,
+      "loss": 0.8413,
+      "step": 5121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.407586744297868e-09,
+      "loss": 0.7441,
+      "step": 5122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.187830069925074e-09,
+      "loss": 0.7676,
+      "step": 5123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.970981974987079e-09,
+      "loss": 0.7578,
+      "step": 5124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.757042543671545e-09,
+      "loss": 0.728,
+      "step": 5125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.54601185903786e-09,
+      "loss": 0.8032,
+      "step": 5126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.337890003015491e-09,
+      "loss": 0.833,
+      "step": 5127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.132677056403969e-09,
+      "loss": 0.8047,
+      "step": 5128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.930373098873732e-09,
+      "loss": 0.8257,
+      "step": 5129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.7309782089669535e-09,
+      "loss": 0.791,
+      "step": 5130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.534492464095043e-09,
+      "loss": 0.7593,
+      "step": 5131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.3409159405411475e-09,
+      "loss": 0.8071,
+      "step": 5132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.1502487134576515e-09,
+      "loss": 0.792,
+      "step": 5133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9624908568686746e-09,
+      "loss": 0.8174,
+      "step": 5134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.777642443667575e-09,
+      "loss": 0.7793,
+      "step": 5135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.595703545619446e-09,
+      "loss": 0.8462,
+      "step": 5136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.416674233359453e-09,
+      "loss": 0.8442,
+      "step": 5137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.240554576391998e-09,
+      "loss": 0.77,
+      "step": 5138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.067344643094052e-09,
+      "loss": 0.7832,
+      "step": 5139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.897044500710157e-09,
+      "loss": 0.8853,
+      "step": 5140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.729654215358259e-09,
+      "loss": 0.7617,
+      "step": 5141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.565173852023874e-09,
+      "loss": 0.7759,
+      "step": 5142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.403603474564255e-09,
+      "loss": 0.8179,
+      "step": 5143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.244943145707558e-09,
+      "loss": 0.8027,
+      "step": 5144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.089192927049512e-09,
+      "loss": 0.8047,
+      "step": 5145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.936352879059247e-09,
+      "loss": 0.8145,
+      "step": 5146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.786423061073463e-09,
+      "loss": 0.7402,
+      "step": 5147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6394035312997674e-09,
+      "loss": 0.7935,
+      "step": 5148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.495294346818334e-09,
+      "loss": 0.7661,
+      "step": 5149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3540955635744107e-09,
+      "loss": 0.79,
+      "step": 5150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2158072363883127e-09,
+      "loss": 0.8052,
+      "step": 5151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0804294189479278e-09,
+      "loss": 0.7095,
+      "step": 5152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.947962163811213e-09,
+      "loss": 0.7207,
+      "step": 5153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8184055224061976e-09,
+      "loss": 0.8096,
+      "step": 5154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.691759545031813e-09,
+      "loss": 0.7939,
+      "step": 5155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.568024280857062e-09,
+      "loss": 0.7905,
+      "step": 5156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4471997779185185e-09,
+      "loss": 0.8125,
+      "step": 5157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3292860831261586e-09,
+      "loss": 0.7344,
+      "step": 5158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2142832422566983e-09,
+      "loss": 0.751,
+      "step": 5159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.102191299958589e-09,
+      "loss": 0.7983,
+      "step": 5160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.993010299750353e-09,
+      "loss": 0.7744,
+      "step": 5161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.88674028401975e-09,
+      "loss": 0.698,
+      "step": 5162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7833812940237782e-09,
+      "loss": 0.7979,
+      "step": 5163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6829333698903382e-09,
+      "loss": 0.8169,
+      "step": 5164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5853965506165691e-09,
+      "loss": 0.3159,
+      "step": 5165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4907708740705127e-09,
+      "loss": 0.7734,
+      "step": 5166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.399056376987784e-09,
+      "loss": 0.8452,
+      "step": 5167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3102530949765657e-09,
+      "loss": 0.8169,
+      "step": 5168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2243610625117807e-09,
+      "loss": 0.7524,
+      "step": 5169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1413803129409207e-09,
+      "loss": 0.7603,
+      "step": 5170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0613108784790493e-09,
+      "loss": 0.7642,
+      "step": 5171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.84152790212134e-10,
+      "loss": 0.7646,
+      "step": 5172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.099060780962121e-10,
+      "loss": 0.7637,
+      "step": 5173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.385707709565594e-10,
+      "loss": 0.814,
+      "step": 5174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.701468964868563e-10,
+      "loss": 0.7627,
+      "step": 5175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.04634481251687e-10,
+      "loss": 0.769,
+      "step": 5176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.420335506865382e-10,
+      "loss": 0.7515,
+      "step": 5177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.823441290936371e-10,
+      "loss": 0.7812,
+      "step": 5178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.255662396477789e-10,
+      "loss": 0.7896,
+      "step": 5179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.716999043913317e-10,
+      "loss": 0.8042,
+      "step": 5180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.207451442375665e-10,
+      "loss": 0.7451,
+      "step": 5181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7270197896899226e-10,
+      "loss": 0.8032,
+      "step": 5182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.275704272365232e-10,
+      "loss": 0.8242,
+      "step": 5183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8535050656280927e-10,
+      "loss": 0.8462,
+      "step": 5184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4604223333807296e-10,
+      "loss": 0.7939,
+      "step": 5185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0964562282510536e-10,
+      "loss": 0.7793,
+      "step": 5186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7616068915177197e-10,
+      "loss": 0.8477,
+      "step": 5187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.455874453201722e-10,
+      "loss": 0.7861,
+      "step": 5188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.179259031991453e-10,
+      "loss": 0.7856,
+      "step": 5189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.317607352760105e-11,
+      "loss": 0.77,
+      "step": 5190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.133796591451968e-11,
+      "loss": 0.8315,
+      "step": 5191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.241158883811936e-11,
+      "loss": 0.2783,
+      "step": 5192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6396949645856045e-11,
+      "loss": 0.811,
+      "step": 5193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3294054556088906e-11,
+      "loss": 0.8721,
+      "step": 5194
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.310290865558228e-11,
+      "loss": 0.8096,
+      "step": 5195
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.823515899505694e-12,
+      "loss": 0.7715,
+      "step": 5196
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4558791164298591e-12,
+      "loss": 0.8159,
+      "step": 5197
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.4614,
+      "step": 5198
+    },
+    {
+      "epoch": 1.0,
+      "step": 5198,
+      "total_flos": 6201972658798592.0,
+      "train_loss": 0.8074982405717944,
+      "train_runtime": 42095.523,
+      "train_samples_per_second": 15.804,
+      "train_steps_per_second": 0.123
+    }
+  ],
+  "max_steps": 5198,
+  "num_train_epochs": 1,
+  "total_flos": 6201972658798592.0,
+  "trial_name": null,
+  "trial_params": null
+}