diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,20815 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998557206752272,
+  "global_step": 3465,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769234e-07,
+      "loss": 2.1133,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 2.2914,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.76923076923077e-07,
+      "loss": 2.3982,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 2.2418,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.615384615384617e-07,
+      "loss": 2.0468,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 2.2895,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3461538461538462e-06,
+      "loss": 2.3783,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 1.8635,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7307692307692308e-06,
+      "loss": 1.6769,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 1.4746,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1153846153846155e-06,
+      "loss": 1.2418,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 1.2507,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 1.2161,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 1.2891,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8846153846153845e-06,
+      "loss": 1.1935,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.0324,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2692307692307696e-06,
+      "loss": 1.1402,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 1.1039,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.653846153846154e-06,
+      "loss": 1.147,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.1163,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0384615384615385e-06,
+      "loss": 1.1599,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 1.0765,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.423076923076924e-06,
+      "loss": 1.1147,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.1031,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.807692307692308e-06,
+      "loss": 0.9705,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 0.9361,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.192307692307693e-06,
+      "loss": 1.0581,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 1.1584,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.576923076923077e-06,
+      "loss": 1.0384,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.769230769230769e-06,
+      "loss": 1.1184,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.961538461538462e-06,
+      "loss": 0.9641,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.023,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.3461538461538466e-06,
+      "loss": 1.0255,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 0.9245,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.730769230769232e-06,
+      "loss": 0.931,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 0.9401,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.115384615384616e-06,
+      "loss": 0.9134,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 0.9559,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.0784,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.9733,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.884615384615384e-06,
+      "loss": 0.9864,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 0.986,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.26923076923077e-06,
+      "loss": 0.9168,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 0.9884,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.653846153846155e-06,
+      "loss": 1.0024,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 1.0106,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.03846153846154e-06,
+      "loss": 0.9228,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.9525,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.423076923076923e-06,
+      "loss": 0.9217,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 0.9097,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.807692307692308e-06,
+      "loss": 0.9185,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.9252,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0192307692307692e-05,
+      "loss": 1.0239,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0384615384615386e-05,
+      "loss": 0.9696,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0576923076923078e-05,
+      "loss": 0.9682,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 0.9051,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0961538461538464e-05,
+      "loss": 0.9426,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1153846153846154e-05,
+      "loss": 0.8904,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1346153846153847e-05,
+      "loss": 1.0092,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 0.8674,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1730769230769232e-05,
+      "loss": 0.937,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1923076923076925e-05,
+      "loss": 0.949,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2115384615384615e-05,
+      "loss": 1.0449,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.982,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.25e-05,
+      "loss": 0.8711,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2692307692307693e-05,
+      "loss": 0.8843,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2884615384615386e-05,
+      "loss": 0.9749,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 0.9287,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3269230769230769e-05,
+      "loss": 0.8926,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3461538461538463e-05,
+      "loss": 0.8982,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3653846153846156e-05,
+      "loss": 0.8666,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.91,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.403846153846154e-05,
+      "loss": 0.9252,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4230769230769232e-05,
+      "loss": 0.9092,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4423076923076924e-05,
+      "loss": 0.8963,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 0.9255,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.480769230769231e-05,
+      "loss": 0.8679,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.9028,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5192307692307693e-05,
+      "loss": 0.9155,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.9032,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.557692307692308e-05,
+      "loss": 0.8442,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.576923076923077e-05,
+      "loss": 0.925,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5961538461538465e-05,
+      "loss": 0.849,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 0.9574,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6346153846153847e-05,
+      "loss": 0.8664,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.653846153846154e-05,
+      "loss": 0.8914,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.673076923076923e-05,
+      "loss": 0.8364,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.9792,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7115384615384617e-05,
+      "loss": 0.9171,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 0.9561,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.9074,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 0.8456,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7884615384615387e-05,
+      "loss": 0.8826,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.807692307692308e-05,
+      "loss": 0.9649,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.826923076923077e-05,
+      "loss": 0.8894,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.8584,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8653846153846157e-05,
+      "loss": 0.923,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8846153846153846e-05,
+      "loss": 0.9164,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.903846153846154e-05,
+      "loss": 0.884,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 0.9153,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9423076923076924e-05,
+      "loss": 0.8958,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9615384615384617e-05,
+      "loss": 0.8594,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.980769230769231e-05,
+      "loss": 0.9121,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.8843,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999563149645e-05,
+      "loss": 0.849,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998252598961e-05,
+      "loss": 0.8286,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999960683490935e-05,
+      "loss": 0.8114,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999930104019505e-05,
+      "loss": 0.8944,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999989078760204e-05,
+      "loss": 0.844,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999842734272892e-05,
+      "loss": 0.872,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999978594407404e-05,
+      "loss": 0.9448,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999972041705511e-05,
+      "loss": 0.9729,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999646153273343e-05,
+      "loss": 0.9652,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999956315279363e-05,
+      "loss": 0.8678,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999471415688484e-05,
+      "loss": 0.9621,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999370942038065e-05,
+      "loss": 0.913,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999926173193014e-05,
+      "loss": 0.9392,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999143785460148e-05,
+      "loss": 0.9066,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999901710273112e-05,
+      "loss": 0.9399,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999888168385375e-05,
+      "loss": 0.9298,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999873752894635e-05,
+      "loss": 0.8743,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998584638134864e-05,
+      "loss": 0.8695,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998423011552877e-05,
+      "loss": 0.8628,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998252649341607e-05,
+      "loss": 0.8696,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999807355164989e-05,
+      "loss": 0.9365,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999788571863421e-05,
+      "loss": 0.8831,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997689150458676e-05,
+      "loss": 0.859,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997483847295026e-05,
+      "loss": 0.9316,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997269809322638e-05,
+      "loss": 0.7522,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997047036728518e-05,
+      "loss": 0.8298,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996815529707296e-05,
+      "loss": 0.8648,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996575288461247e-05,
+      "loss": 0.8048,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996326313200266e-05,
+      "loss": 0.931,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996068604141885e-05,
+      "loss": 0.877,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995802161511265e-05,
+      "loss": 0.8761,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999552698554119e-05,
+      "loss": 0.9342,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999524307647209e-05,
+      "loss": 0.898,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994950434552016e-05,
+      "loss": 0.902,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994649060036642e-05,
+      "loss": 0.881,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994338953189293e-05,
+      "loss": 0.8378,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994020114280892e-05,
+      "loss": 0.8513,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999369254359002e-05,
+      "loss": 0.8735,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993356241402876e-05,
+      "loss": 0.8454,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999301120801328e-05,
+      "loss": 0.8435,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992657443722697e-05,
+      "loss": 0.9212,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992294948840204e-05,
+      "loss": 0.9061,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991923723682514e-05,
+      "loss": 0.8657,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991543768573974e-05,
+      "loss": 0.7814,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991155083846537e-05,
+      "loss": 0.859,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990757669839812e-05,
+      "loss": 0.8426,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990351526901012e-05,
+      "loss": 0.8551,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998993665538499e-05,
+      "loss": 0.8477,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998951305565421e-05,
+      "loss": 0.9107,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989080728078778e-05,
+      "loss": 0.8713,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998863967303642e-05,
+      "loss": 0.9989,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988189890912482e-05,
+      "loss": 0.9748,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998773138209994e-05,
+      "loss": 0.7796,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987264146999396e-05,
+      "loss": 0.9066,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998678818601907e-05,
+      "loss": 0.8681,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998630349957481e-05,
+      "loss": 0.8976,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985810088090093e-05,
+      "loss": 0.84,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985307951996005e-05,
+      "loss": 0.8933,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984797091731267e-05,
+      "loss": 0.888,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998427750774222e-05,
+      "loss": 0.8556,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983749200482816e-05,
+      "loss": 0.8603,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983212170414648e-05,
+      "loss": 0.8629,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982666418006912e-05,
+      "loss": 0.8418,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982111943736435e-05,
+      "loss": 0.9596,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998154874808766e-05,
+      "loss": 0.8687,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980976831552653e-05,
+      "loss": 0.8282,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980396194631104e-05,
+      "loss": 0.866,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979806837830305e-05,
+      "loss": 0.9005,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997920876166518e-05,
+      "loss": 0.8392,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978601966658277e-05,
+      "loss": 0.8587,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997798645333974e-05,
+      "loss": 0.9529,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977362222247353e-05,
+      "loss": 0.8541,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997672927392651e-05,
+      "loss": 0.8844,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976087608930208e-05,
+      "loss": 0.7995,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975437227819076e-05,
+      "loss": 0.9145,
+      "step": 179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997477813116135e-05,
+      "loss": 0.8396,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974110319532886e-05,
+      "loss": 0.9262,
+      "step": 181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997343379351715e-05,
+      "loss": 0.9768,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972748553705225e-05,
+      "loss": 0.8461,
+      "step": 183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972054600695802e-05,
+      "loss": 0.8451,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971351935095194e-05,
+      "loss": 0.8871,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970640557517314e-05,
+      "loss": 0.8674,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969920468583693e-05,
+      "loss": 1.012,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996919166892348e-05,
+      "loss": 0.872,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968454159173426e-05,
+      "loss": 0.9623,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967707939977887e-05,
+      "loss": 0.9008,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966953011988842e-05,
+      "loss": 0.8059,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966189375865868e-05,
+      "loss": 0.8701,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965417032276157e-05,
+      "loss": 0.8675,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964635981894505e-05,
+      "loss": 0.8151,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996384622540332e-05,
+      "loss": 0.8966,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996304776349261e-05,
+      "loss": 0.8324,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962240596859987e-05,
+      "loss": 0.8274,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996142472621068e-05,
+      "loss": 0.8869,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960600152257513e-05,
+      "loss": 0.8727,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959766875720915e-05,
+      "loss": 0.8002,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958924897328925e-05,
+      "loss": 0.8856,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995807421781718e-05,
+      "loss": 0.8281,
+      "step": 202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957214837928913e-05,
+      "loss": 0.888,
+      "step": 203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956346758414968e-05,
+      "loss": 0.828,
+      "step": 204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955469980033787e-05,
+      "loss": 0.8967,
+      "step": 205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954584503551414e-05,
+      "loss": 0.9155,
+      "step": 206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995369032974149e-05,
+      "loss": 0.7656,
+      "step": 207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952787459385254e-05,
+      "loss": 0.8712,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995187589327154e-05,
+      "loss": 0.948,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950955632196792e-05,
+      "loss": 0.7853,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950026676965037e-05,
+      "loss": 0.93,
+      "step": 211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949089028387908e-05,
+      "loss": 0.9025,
+      "step": 212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948142687284625e-05,
+      "loss": 0.8411,
+      "step": 213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994718765448201e-05,
+      "loss": 0.9089,
+      "step": 214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946223930814475e-05,
+      "loss": 0.8467,
+      "step": 215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945251517124026e-05,
+      "loss": 0.7875,
+      "step": 216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944270414260263e-05,
+      "loss": 0.8651,
+      "step": 217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994328062308037e-05,
+      "loss": 0.8644,
+      "step": 218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942282144449138e-05,
+      "loss": 0.7928,
+      "step": 219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994127497923893e-05,
+      "loss": 0.8667,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940259128329714e-05,
+      "loss": 0.8416,
+      "step": 221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939234592609034e-05,
+      "loss": 0.8774,
+      "step": 222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993820137297203e-05,
+      "loss": 0.8216,
+      "step": 223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937159470321424e-05,
+      "loss": 0.8572,
+      "step": 224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936108885567534e-05,
+      "loss": 0.8082,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935049619628252e-05,
+      "loss": 0.9025,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933981673429056e-05,
+      "loss": 0.8697,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932905047903017e-05,
+      "loss": 0.8621,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993181974399078e-05,
+      "loss": 0.7811,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993072576264058e-05,
+      "loss": 0.836,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929623104808224e-05,
+      "loss": 0.866,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928511771457108e-05,
+      "loss": 0.8151,
+      "step": 232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927391763558203e-05,
+      "loss": 0.8871,
+      "step": 233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926263082090064e-05,
+      "loss": 0.9282,
+      "step": 234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925125728038816e-05,
+      "loss": 0.8828,
+      "step": 235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923979702398172e-05,
+      "loss": 0.8311,
+      "step": 236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992282500616941e-05,
+      "loss": 0.876,
+      "step": 237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992166164036139e-05,
+      "loss": 0.8145,
+      "step": 238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992048960599055e-05,
+      "loss": 0.8565,
+      "step": 239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919308904080893e-05,
+      "loss": 0.8416,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918119535664e-05,
+      "loss": 0.7961,
+      "step": 241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916921501779018e-05,
+      "loss": 0.8689,
+      "step": 242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991571480347268e-05,
+      "loss": 0.8753,
+      "step": 243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914499441799274e-05,
+      "loss": 0.8053,
+      "step": 244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991327541782066e-05,
+      "loss": 0.8576,
+      "step": 245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912042732606274e-05,
+      "loss": 0.8343,
+      "step": 246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991080138723311e-05,
+      "loss": 0.8184,
+      "step": 247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990955138278573e-05,
+      "loss": 0.8237,
+      "step": 248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990829272035627e-05,
+      "loss": 0.7348,
+      "step": 249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907025401044425e-05,
+      "loss": 0.8496,
+      "step": 250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905749425957445e-05,
+      "loss": 0.8677,
+      "step": 251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904464796210158e-05,
+      "loss": 0.8085,
+      "step": 252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903171512924938e-05,
+      "loss": 0.8274,
+      "step": 253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901869577231732e-05,
+      "loss": 0.8316,
+      "step": 254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900558990268047e-05,
+      "loss": 0.8379,
+      "step": 255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899239753178936e-05,
+      "loss": 0.9052,
+      "step": 256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989791186711702e-05,
+      "loss": 0.8482,
+      "step": 257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896575333242477e-05,
+      "loss": 0.8228,
+      "step": 258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895230152723033e-05,
+      "loss": 0.8094,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893876326733974e-05,
+      "loss": 0.9142,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989251385645814e-05,
+      "loss": 0.8605,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891142743085923e-05,
+      "loss": 0.8658,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889762987815264e-05,
+      "loss": 0.815,
+      "step": 263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888374591851658e-05,
+      "loss": 0.834,
+      "step": 264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886977556408145e-05,
+      "loss": 0.8742,
+      "step": 265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885571882705322e-05,
+      "loss": 0.7851,
+      "step": 266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884157571971318e-05,
+      "loss": 0.8627,
+      "step": 267
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988273462544182e-05,
+      "loss": 0.77,
+      "step": 268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881303044360065e-05,
+      "loss": 0.8008,
+      "step": 269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879862829976816e-05,
+      "loss": 0.7971,
+      "step": 270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878413983550397e-05,
+      "loss": 0.7756,
+      "step": 271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987695650634666e-05,
+      "loss": 0.8432,
+      "step": 272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875490399639007e-05,
+      "loss": 0.8693,
+      "step": 273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874015664708378e-05,
+      "loss": 0.8367,
+      "step": 274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987253230284325e-05,
+      "loss": 0.8385,
+      "step": 275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871040315339632e-05,
+      "loss": 0.8745,
+      "step": 276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869539703501083e-05,
+      "loss": 0.8424,
+      "step": 277
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868030468638677e-05,
+      "loss": 0.8116,
+      "step": 278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866512612071046e-05,
+      "loss": 0.8753,
+      "step": 279
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864986135124336e-05,
+      "loss": 0.8103,
+      "step": 280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863451039132227e-05,
+      "loss": 0.9089,
+      "step": 281
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861907325435946e-05,
+      "loss": 0.8848,
+      "step": 282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986035499538422e-05,
+      "loss": 0.8425,
+      "step": 283
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858794050333338e-05,
+      "loss": 0.8709,
+      "step": 284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857224491647087e-05,
+      "loss": 0.8552,
+      "step": 285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855646320696797e-05,
+      "loss": 0.8963,
+      "step": 286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854059538861312e-05,
+      "loss": 0.9245,
+      "step": 287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985246414752701e-05,
+      "loss": 0.778,
+      "step": 288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985086014808778e-05,
+      "loss": 0.9172,
+      "step": 289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849247541945044e-05,
+      "loss": 0.8675,
+      "step": 290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847626330507734e-05,
+      "loss": 0.8164,
+      "step": 291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98459965151923e-05,
+      "loss": 0.8662,
+      "step": 292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984435809742272e-05,
+      "loss": 0.8296,
+      "step": 293
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842711078630477e-05,
+      "loss": 0.8334,
+      "step": 294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841055460254572e-05,
+      "loss": 0.8695,
+      "step": 295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839391243741514e-05,
+      "loss": 0.8656,
+      "step": 296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837718430545343e-05,
+      "loss": 0.869,
+      "step": 297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836037022127593e-05,
+      "loss": 0.8691,
+      "step": 298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834347019957302e-05,
+      "loss": 0.8762,
+      "step": 299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832648425511038e-05,
+      "loss": 0.8989,
+      "step": 300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830941240272858e-05,
+      "loss": 0.8624,
+      "step": 301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829225465734335e-05,
+      "loss": 0.861,
+      "step": 302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827501103394536e-05,
+      "loss": 0.8278,
+      "step": 303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825768154760046e-05,
+      "loss": 0.8456,
+      "step": 304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824026621344935e-05,
+      "loss": 0.7952,
+      "step": 305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982227650467079e-05,
+      "loss": 0.8947,
+      "step": 306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820517806266682e-05,
+      "loss": 0.867,
+      "step": 307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818750527669194e-05,
+      "loss": 0.7872,
+      "step": 308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816974670422392e-05,
+      "loss": 0.8537,
+      "step": 309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815190236077846e-05,
+      "loss": 0.9384,
+      "step": 310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981339722619462e-05,
+      "loss": 0.7777,
+      "step": 311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981159564233926e-05,
+      "loss": 0.798,
+      "step": 312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809785486085825e-05,
+      "loss": 0.8961,
+      "step": 313
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980796675901584e-05,
+      "loss": 0.8864,
+      "step": 314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980613946271833e-05,
+      "loss": 0.9028,
+      "step": 315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804303598789804e-05,
+      "loss": 0.7944,
+      "step": 316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980245916883426e-05,
+      "loss": 0.7573,
+      "step": 317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800606174463177e-05,
+      "loss": 0.8445,
+      "step": 318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979874461729552e-05,
+      "loss": 0.7594,
+      "step": 319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979687449895772e-05,
+      "loss": 0.8429,
+      "step": 320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979499582108372e-05,
+      "loss": 0.8307,
+      "step": 321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979310858531491e-05,
+      "loss": 0.8715,
+      "step": 322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791212793300173e-05,
+      "loss": 0.8253,
+      "step": 323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978930844669586e-05,
+      "loss": 0.8499,
+      "step": 324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787395547165804e-05,
+      "loss": 0.8378,
+      "step": 325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978547409638131e-05,
+      "loss": 0.8403,
+      "step": 326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978354409602114e-05,
+      "loss": 0.8591,
+      "step": 327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781605547771546e-05,
+      "loss": 0.8097,
+      "step": 328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779658453326234e-05,
+      "loss": 0.932,
+      "step": 329
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777702814386387e-05,
+      "loss": 0.8979,
+      "step": 330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775738632660647e-05,
+      "loss": 0.791,
+      "step": 331
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773765909865114e-05,
+      "loss": 0.8203,
+      "step": 332
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771784647723363e-05,
+      "loss": 0.862,
+      "step": 333
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769794847966427e-05,
+      "loss": 0.9075,
+      "step": 334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976779651233279e-05,
+      "loss": 0.8218,
+      "step": 335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97657896425684e-05,
+      "loss": 0.8856,
+      "step": 336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763774240426664e-05,
+      "loss": 0.8357,
+      "step": 337
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761750307668436e-05,
+      "loss": 0.8319,
+      "step": 338
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975971784606203e-05,
+      "loss": 0.7696,
+      "step": 339
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975767685738321e-05,
+      "loss": 0.8502,
+      "step": 340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975562734341519e-05,
+      "loss": 0.8541,
+      "step": 341
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753569305948624e-05,
+      "loss": 0.9388,
+      "step": 342
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975150274678163e-05,
+      "loss": 0.9062,
+      "step": 343
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749427667719757e-05,
+      "loss": 0.8186,
+      "step": 344
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747344070576005e-05,
+      "loss": 0.8534,
+      "step": 345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745251957170817e-05,
+      "loss": 0.7781,
+      "step": 346
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743151329332072e-05,
+      "loss": 0.8352,
+      "step": 347
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741042188895084e-05,
+      "loss": 0.8215,
+      "step": 348
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738924537702618e-05,
+      "loss": 0.7844,
+      "step": 349
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736798377604864e-05,
+      "loss": 0.7324,
+      "step": 350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734663710459454e-05,
+      "loss": 0.8866,
+      "step": 351
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732520538131442e-05,
+      "loss": 0.7701,
+      "step": 352
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973036886249332e-05,
+      "loss": 0.8431,
+      "step": 353
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972820868542501e-05,
+      "loss": 0.7694,
+      "step": 354
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726040008813866e-05,
+      "loss": 0.8285,
+      "step": 355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723862834554656e-05,
+      "loss": 0.8507,
+      "step": 356
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972167716454958e-05,
+      "loss": 0.7988,
+      "step": 357
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719483000708255e-05,
+      "loss": 0.8517,
+      "step": 358
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971728034494773e-05,
+      "loss": 0.8149,
+      "step": 359
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715069199192465e-05,
+      "loss": 0.796,
+      "step": 360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712849565374336e-05,
+      "loss": 0.8521,
+      "step": 361
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710621445432646e-05,
+      "loss": 0.8591,
+      "step": 362
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708384841314098e-05,
+      "loss": 0.8539,
+      "step": 363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970613975497282e-05,
+      "loss": 0.9508,
+      "step": 364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970388618837034e-05,
+      "loss": 0.8141,
+      "step": 365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701624143475608e-05,
+      "loss": 0.8802,
+      "step": 366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699353622264967e-05,
+      "loss": 0.8651,
+      "step": 367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969707462672218e-05,
+      "loss": 0.8139,
+      "step": 368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694787158838394e-05,
+      "loss": 0.8693,
+      "step": 369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692491220612188e-05,
+      "loss": 0.8683,
+      "step": 370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690186814049513e-05,
+      "loss": 0.7973,
+      "step": 371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687873941163737e-05,
+      "loss": 0.772,
+      "step": 372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685552603975612e-05,
+      "loss": 0.8692,
+      "step": 373
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96832228045133e-05,
+      "loss": 0.8847,
+      "step": 374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680884544812347e-05,
+      "loss": 0.8078,
+      "step": 375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678537826915687e-05,
+      "loss": 0.8824,
+      "step": 376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676182652873653e-05,
+      "loss": 0.8276,
+      "step": 377
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967381902474396e-05,
+      "loss": 0.8667,
+      "step": 378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671446944591714e-05,
+      "loss": 0.8048,
+      "step": 379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96690664144894e-05,
+      "loss": 0.838,
+      "step": 380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666677436516894e-05,
+      "loss": 0.7842,
+      "step": 381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664280012761445e-05,
+      "loss": 0.9122,
+      "step": 382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966187414531768e-05,
+      "loss": 0.803,
+      "step": 383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659459836287614e-05,
+      "loss": 0.8032,
+      "step": 384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657037087780626e-05,
+      "loss": 0.751,
+      "step": 385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654605901913472e-05,
+      "loss": 0.9136,
+      "step": 386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652166280810285e-05,
+      "loss": 0.7958,
+      "step": 387
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649718226602562e-05,
+      "loss": 0.8514,
+      "step": 388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964726174142917e-05,
+      "loss": 0.7875,
+      "step": 389
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964479682743634e-05,
+      "loss": 0.8115,
+      "step": 390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642323486777668e-05,
+      "loss": 0.8381,
+      "step": 391
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963984172161412e-05,
+      "loss": 0.8925,
+      "step": 392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637351534114008e-05,
+      "loss": 0.8745,
+      "step": 393
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634852926453016e-05,
+      "loss": 0.8117,
+      "step": 394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632345900814178e-05,
+      "loss": 0.8066,
+      "step": 395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629830459387884e-05,
+      "loss": 0.8207,
+      "step": 396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627306604371873e-05,
+      "loss": 0.869,
+      "step": 397
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962477433797125e-05,
+      "loss": 0.7733,
+      "step": 398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622233662398444e-05,
+      "loss": 0.8675,
+      "step": 399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619684579873254e-05,
+      "loss": 0.8275,
+      "step": 400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961712709262281e-05,
+      "loss": 0.8363,
+      "step": 401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614561202881596e-05,
+      "loss": 0.8647,
+      "step": 402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961198691289143e-05,
+      "loss": 0.8862,
+      "step": 403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960940422490147e-05,
+      "loss": 0.7727,
+      "step": 404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606813141168214e-05,
+      "loss": 0.8656,
+      "step": 405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604213663955483e-05,
+      "loss": 0.7863,
+      "step": 406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960160579553446e-05,
+      "loss": 0.8558,
+      "step": 407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959898953818363e-05,
+      "loss": 0.8058,
+      "step": 408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596364894188822e-05,
+      "loss": 0.8372,
+      "step": 409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593731865843187e-05,
+      "loss": 0.7997,
+      "step": 410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959109045544721e-05,
+      "loss": 0.862,
+      "step": 411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588440665308682e-05,
+      "loss": 0.8429,
+      "step": 412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585782497742735e-05,
+      "loss": 0.8367,
+      "step": 413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583115955071813e-05,
+      "loss": 0.7883,
+      "step": 414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958044103962567e-05,
+      "loss": 0.839,
+      "step": 415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577757753741385e-05,
+      "loss": 0.8046,
+      "step": 416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957506609976335e-05,
+      "loss": 0.8156,
+      "step": 417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957236608004326e-05,
+      "loss": 0.8072,
+      "step": 418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569657696940124e-05,
+      "loss": 0.9518,
+      "step": 419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956694095282026e-05,
+      "loss": 0.9026,
+      "step": 420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564215850057294e-05,
+      "loss": 0.79,
+      "step": 421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956148239103214e-05,
+      "loss": 0.8025,
+      "step": 422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558740578133033e-05,
+      "loss": 0.7769,
+      "step": 423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955599041375549e-05,
+      "loss": 0.7616,
+      "step": 424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553231900302334e-05,
+      "loss": 0.7518,
+      "step": 425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955046504018368e-05,
+      "loss": 0.8027,
+      "step": 426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547689835816936e-05,
+      "loss": 0.8803,
+      "step": 427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544906289626802e-05,
+      "loss": 0.83,
+      "step": 428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954211440404526e-05,
+      "loss": 0.784,
+      "step": 429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539314181511584e-05,
+      "loss": 0.8093,
+      "step": 430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536505624472332e-05,
+      "loss": 0.7447,
+      "step": 431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953368873538134e-05,
+      "loss": 0.7753,
+      "step": 432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953086351669973e-05,
+      "loss": 0.7922,
+      "step": 433
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528029970895892e-05,
+      "loss": 0.8251,
+      "step": 434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95251881004455e-05,
+      "loss": 0.814,
+      "step": 435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522337907831497e-05,
+      "loss": 0.7714,
+      "step": 436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519479395544103e-05,
+      "loss": 0.8749,
+      "step": 437
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516612566080797e-05,
+      "loss": 0.7759,
+      "step": 438
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513737421946332e-05,
+      "loss": 0.8566,
+      "step": 439
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510853965652723e-05,
+      "loss": 0.8519,
+      "step": 440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507962199719245e-05,
+      "loss": 0.8546,
+      "step": 441
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505062126672442e-05,
+      "loss": 0.7844,
+      "step": 442
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502153749046105e-05,
+      "loss": 0.781,
+      "step": 443
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499237069381288e-05,
+      "loss": 0.7881,
+      "step": 444
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496312090226296e-05,
+      "loss": 0.8275,
+      "step": 445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493378814136683e-05,
+      "loss": 0.9251,
+      "step": 446
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949043724367526e-05,
+      "loss": 0.8534,
+      "step": 447
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487487381412065e-05,
+      "loss": 0.8251,
+      "step": 448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484529229924416e-05,
+      "loss": 0.7924,
+      "step": 449
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481562791796834e-05,
+      "loss": 0.7764,
+      "step": 450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947858806962111e-05,
+      "loss": 0.8311,
+      "step": 451
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475605065996258e-05,
+      "loss": 0.7689,
+      "step": 452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472613783528522e-05,
+      "loss": 0.7813,
+      "step": 453
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469614224831397e-05,
+      "loss": 0.789,
+      "step": 454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466606392525595e-05,
+      "loss": 0.8117,
+      "step": 455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946359028923907e-05,
+      "loss": 0.8088,
+      "step": 456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460565917606976e-05,
+      "loss": 0.8328,
+      "step": 457
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457533280271725e-05,
+      "loss": 0.7925,
+      "step": 458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454492379882926e-05,
+      "loss": 0.8267,
+      "step": 459
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945144321909742e-05,
+      "loss": 0.8175,
+      "step": 460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448385800579262e-05,
+      "loss": 0.829,
+      "step": 461
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445320126999715e-05,
+      "loss": 0.7696,
+      "step": 462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442246201037264e-05,
+      "loss": 0.846,
+      "step": 463
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94391640253776e-05,
+      "loss": 0.8316,
+      "step": 464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436073602713617e-05,
+      "loss": 0.8146,
+      "step": 465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432974935745426e-05,
+      "loss": 0.934,
+      "step": 466
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429868027180333e-05,
+      "loss": 0.8222,
+      "step": 467
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426752879732848e-05,
+      "loss": 0.8134,
+      "step": 468
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423629496124674e-05,
+      "loss": 0.7906,
+      "step": 469
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420497879084715e-05,
+      "loss": 0.8501,
+      "step": 470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941735803134907e-05,
+      "loss": 0.7614,
+      "step": 471
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414209955661015e-05,
+      "loss": 0.8381,
+      "step": 472
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941105365477104e-05,
+      "loss": 0.8088,
+      "step": 473
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94078891314368e-05,
+      "loss": 0.822,
+      "step": 474
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940471638842314e-05,
+      "loss": 0.7999,
+      "step": 475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401535428502096e-05,
+      "loss": 0.8467,
+      "step": 476
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398346254452864e-05,
+      "loss": 0.8318,
+      "step": 477
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395148869061836e-05,
+      "loss": 0.7509,
+      "step": 478
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391943275122567e-05,
+      "loss": 0.8273,
+      "step": 479
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388729475435784e-05,
+      "loss": 0.832,
+      "step": 480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938550747280939e-05,
+      "loss": 0.7456,
+      "step": 481
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938227727005845e-05,
+      "loss": 0.762,
+      "step": 482
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379038870005195e-05,
+      "loss": 0.8576,
+      "step": 483
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375792275479014e-05,
+      "loss": 0.7226,
+      "step": 484
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372537489316462e-05,
+      "loss": 0.8421,
+      "step": 485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936927451436125e-05,
+      "loss": 0.7868,
+      "step": 486
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366003353464235e-05,
+      "loss": 0.7895,
+      "step": 487
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936272400948344e-05,
+      "loss": 0.7403,
+      "step": 488
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359436485284025e-05,
+      "loss": 0.8109,
+      "step": 489
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356140783738304e-05,
+      "loss": 0.7478,
+      "step": 490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352836907725735e-05,
+      "loss": 0.807,
+      "step": 491
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349524860132916e-05,
+      "loss": 0.7652,
+      "step": 492
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934620464385358e-05,
+      "loss": 0.8738,
+      "step": 493
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342876261788612e-05,
+      "loss": 0.7818,
+      "step": 494
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339539716846016e-05,
+      "loss": 0.7477,
+      "step": 495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336195011940938e-05,
+      "loss": 0.8935,
+      "step": 496
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933284214999564e-05,
+      "loss": 0.7914,
+      "step": 497
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329481133939527e-05,
+      "loss": 0.8615,
+      "step": 498
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326111966709122e-05,
+      "loss": 0.8486,
+      "step": 499
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932273465124806e-05,
+      "loss": 0.7555,
+      "step": 500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931934919050712e-05,
+      "loss": 0.7938,
+      "step": 501
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931595558744417e-05,
+      "loss": 0.7892,
+      "step": 502
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.93125538450242e-05,
+      "loss": 0.8701,
+      "step": 503
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930914396621932e-05,
+      "loss": 0.8219,
+      "step": 504
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930572595400875e-05,
+      "loss": 0.7807,
+      "step": 505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302299811378797e-05,
+      "loss": 0.8077,
+      "step": 506
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298865541322893e-05,
+      "loss": 0.7478,
+      "step": 507
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295423146841562e-05,
+      "loss": 0.8832,
+      "step": 508
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929197263094242e-05,
+      "loss": 0.9003,
+      "step": 509
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928851399664019e-05,
+      "loss": 0.8736,
+      "step": 510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928504724695669e-05,
+      "loss": 0.7743,
+      "step": 511
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281572384920806e-05,
+      "loss": 0.771,
+      "step": 512
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278089413568537e-05,
+      "loss": 0.8341,
+      "step": 513
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274598335942958e-05,
+      "loss": 0.8833,
+      "step": 514
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271099155094223e-05,
+      "loss": 0.8262,
+      "step": 515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926759187407957e-05,
+      "loss": 0.8755,
+      "step": 516
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264076495963312e-05,
+      "loss": 0.9714,
+      "step": 517
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926055302381684e-05,
+      "loss": 0.827,
+      "step": 518
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257021460718618e-05,
+      "loss": 0.8689,
+      "step": 519
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253481809754163e-05,
+      "loss": 0.9614,
+      "step": 520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249934074016077e-05,
+      "loss": 0.7755,
+      "step": 521
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246378256604023e-05,
+      "loss": 0.8096,
+      "step": 522
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242814360624714e-05,
+      "loss": 0.8625,
+      "step": 523
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239242389191936e-05,
+      "loss": 0.8783,
+      "step": 524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235662345426513e-05,
+      "loss": 0.8011,
+      "step": 525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232074232456345e-05,
+      "loss": 0.8282,
+      "step": 526
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228478053416358e-05,
+      "loss": 0.8365,
+      "step": 527
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922487381144854e-05,
+      "loss": 0.8043,
+      "step": 528
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221261509701922e-05,
+      "loss": 0.928,
+      "step": 529
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921764115133257e-05,
+      "loss": 0.8437,
+      "step": 530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214012739503594e-05,
+      "loss": 0.9264,
+      "step": 531
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210376277385143e-05,
+      "loss": 0.8093,
+      "step": 532
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92067317681544e-05,
+      "loss": 0.8684,
+      "step": 533
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203079214995565e-05,
+      "loss": 0.8283,
+      "step": 534
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199418621099886e-05,
+      "loss": 0.8232,
+      "step": 535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919574998966562e-05,
+      "loss": 0.8197,
+      "step": 536
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919207332389806e-05,
+      "loss": 0.7902,
+      "step": 537
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91883886270095e-05,
+      "loss": 0.7723,
+      "step": 538
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184695902219274e-05,
+      "loss": 0.8545,
+      "step": 539
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180995152753713e-05,
+      "loss": 0.8374,
+      "step": 540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177286381846165e-05,
+      "loss": 0.8246,
+      "step": 541
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917356959273698e-05,
+      "loss": 0.7973,
+      "step": 542
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169844788673534e-05,
+      "loss": 0.8033,
+      "step": 543
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9166111972910178e-05,
+      "loss": 0.8308,
+      "step": 544
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9162371148708278e-05,
+      "loss": 0.8504,
+      "step": 545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91586223193362e-05,
+      "loss": 0.8051,
+      "step": 546
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154865488069294e-05,
+      "loss": 0.74,
+      "step": 547
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151100658189906e-05,
+      "loss": 0.8413,
+      "step": 548
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147327832987375e-05,
+      "loss": 0.7907,
+      "step": 549
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143547015758014e-05,
+      "loss": 0.8075,
+      "step": 550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9139758209805132e-05,
+      "loss": 0.8701,
+      "step": 551
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913596141843901e-05,
+      "loss": 0.7277,
+      "step": 552
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132156644976906e-05,
+      "loss": 0.813,
+      "step": 553
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128343892743055e-05,
+      "loss": 0.8403,
+      "step": 554
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912452316506866e-05,
+      "loss": 0.8057,
+      "step": 555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120694465291893e-05,
+      "loss": 0.8533,
+      "step": 556
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9116857796757892e-05,
+      "loss": 0.816,
+      "step": 557
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113013162818757e-05,
+      "loss": 0.8511,
+      "step": 558
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109160566833548e-05,
+      "loss": 0.8202,
+      "step": 559
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910530001216828e-05,
+      "loss": 0.8195,
+      "step": 560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101431502195925e-05,
+      "loss": 0.7783,
+      "step": 561
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90975550402964e-05,
+      "loss": 0.838,
+      "step": 562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093670629856577e-05,
+      "loss": 0.8107,
+      "step": 563
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089778274270262e-05,
+      "loss": 0.7803,
+      "step": 564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908587797693821e-05,
+      "loss": 0.7855,
+      "step": 565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081969741268117e-05,
+      "loss": 0.8774,
+      "step": 566
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907805357067461e-05,
+      "loss": 0.831,
+      "step": 567
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074129468579247e-05,
+      "loss": 0.8364,
+      "step": 568
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907019743841052e-05,
+      "loss": 0.8852,
+      "step": 569
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906625748360385e-05,
+      "loss": 0.7808,
+      "step": 570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9062309607601575e-05,
+      "loss": 0.8108,
+      "step": 571
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9058353813852958e-05,
+      "loss": 0.8771,
+      "step": 572
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905439010581418e-05,
+      "loss": 0.8619,
+      "step": 573
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050418486948334e-05,
+      "loss": 0.8289,
+      "step": 574
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046438960725428e-05,
+      "loss": 0.7921,
+      "step": 575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042451530622372e-05,
+      "loss": 0.878,
+      "step": 576
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903845620012299e-05,
+      "loss": 0.7907,
+      "step": 577
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034452972718008e-05,
+      "loss": 0.8523,
+      "step": 578
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9030441851905046e-05,
+      "loss": 0.8111,
+      "step": 579
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902642284118862e-05,
+      "loss": 0.7945,
+      "step": 580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022395944080147e-05,
+      "loss": 0.8208,
+      "step": 581
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901836116409793e-05,
+      "loss": 0.8127,
+      "step": 582
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014318504767155e-05,
+      "loss": 0.8113,
+      "step": 583
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90102679696199e-05,
+      "loss": 0.8624,
+      "step": 584
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006209562195116e-05,
+      "loss": 0.8053,
+      "step": 585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9002143286038642e-05,
+      "loss": 0.7986,
+      "step": 586
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998069144703187e-05,
+      "loss": 0.7855,
+      "step": 587
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993987141748324e-05,
+      "loss": 0.7731,
+      "step": 588
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898989728074051e-05,
+      "loss": 0.7639,
+      "step": 589
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985799565253052e-05,
+      "loss": 0.8112,
+      "step": 590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898169399886613e-05,
+      "loss": 0.8523,
+      "step": 591
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897758058516678e-05,
+      "loss": 0.7769,
+      "step": 592
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973459327748897e-05,
+      "loss": 0.8022,
+      "step": 593
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969330230213225e-05,
+      "loss": 0.7833,
+      "step": 594
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896519329616736e-05,
+      "loss": 0.8152,
+      "step": 595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896104852922574e-05,
+      "loss": 0.8492,
+      "step": 596
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895689593300966e-05,
+      "loss": 0.9262,
+      "step": 597
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8952735511147234e-05,
+      "loss": 0.796,
+      "step": 598
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948567267273433e-05,
+      "loss": 0.8211,
+      "step": 599
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944391205030057e-05,
+      "loss": 0.7989,
+      "step": 600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940207328065728e-05,
+      "loss": 0.7636,
+      "step": 601
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8936015640035907e-05,
+      "loss": 0.7781,
+      "step": 602
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931816144602874e-05,
+      "loss": 0.8351,
+      "step": 603
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927608845435728e-05,
+      "loss": 0.7707,
+      "step": 604
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923393746210395e-05,
+      "loss": 0.8935,
+      "step": 605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8919170850609606e-05,
+      "loss": 0.779,
+      "step": 606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8914940162322907e-05,
+      "loss": 0.8601,
+      "step": 607
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8910701685046656e-05,
+      "loss": 0.7677,
+      "step": 608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890645542248401e-05,
+      "loss": 0.836,
+      "step": 609
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902201378344938e-05,
+      "loss": 0.831,
+      "step": 610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8897939556346196e-05,
+      "loss": 0.7759,
+      "step": 611
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8893669960211342e-05,
+      "loss": 0.8365,
+      "step": 612
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888939259367073e-05,
+      "loss": 0.8158,
+      "step": 613
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888510746046149e-05,
+      "loss": 0.8181,
+      "step": 614
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880814564327548e-05,
+      "loss": 0.8094,
+      "step": 615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8876513909019615e-05,
+      "loss": 0.8247,
+      "step": 616
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872205498295174e-05,
+      "loss": 0.8129,
+      "step": 617
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8867889335918487e-05,
+      "loss": 0.7943,
+      "step": 618
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886356542566059e-05,
+      "loss": 0.8555,
+      "step": 619
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885923377129928e-05,
+      "loss": 0.8807,
+      "step": 620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885489437661913e-05,
+      "loss": 0.7921,
+      "step": 621
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8850547245411476e-05,
+      "loss": 0.8223,
+      "step": 622
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846192381474404e-05,
+      "loss": 0.7728,
+      "step": 623
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8841829788612762e-05,
+      "loss": 0.8123,
+      "step": 624
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8837459470638158e-05,
+      "loss": 0.7862,
+      "step": 625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8833081431368932e-05,
+      "loss": 0.838,
+      "step": 626
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828695674630183e-05,
+      "loss": 0.8463,
+      "step": 627
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824302204253756e-05,
+      "loss": 0.8076,
+      "step": 628
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819901024078225e-05,
+      "loss": 0.7509,
+      "step": 629
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88154921379489e-05,
+      "loss": 0.7903,
+      "step": 630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811075549717835e-05,
+      "loss": 0.7636,
+      "step": 631
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88066512632438e-05,
+      "loss": 0.8209,
+      "step": 632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88022192823923e-05,
+      "loss": 0.8218,
+      "step": 633
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879777961103556e-05,
+      "loss": 0.7835,
+      "step": 634
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793332253052525e-05,
+      "loss": 0.8124,
+      "step": 635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788877212328852e-05,
+      "loss": 0.8335,
+      "step": 636
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878441449275692e-05,
+      "loss": 0.8765,
+      "step": 637
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87799440982358e-05,
+      "loss": 0.8376,
+      "step": 638
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775466032671288e-05,
+      "loss": 0.7673,
+      "step": 639
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770980299975867e-05,
+      "loss": 0.8251,
+      "step": 640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876648690406873e-05,
+      "loss": 0.8515,
+      "step": 641
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8761985848875752e-05,
+      "loss": 0.7838,
+      "step": 642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8757477138329515e-05,
+      "loss": 0.8654,
+      "step": 643
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752960776369283e-05,
+      "loss": 0.8561,
+      "step": 644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748436766941e-05,
+      "loss": 0.7795,
+      "step": 645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87439051139973e-05,
+      "loss": 0.8359,
+      "step": 646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739365821497496e-05,
+      "loss": 0.8186,
+      "step": 647
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873481889340756e-05,
+      "loss": 0.8566,
+      "step": 648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730264333700153e-05,
+      "loss": 0.8926,
+      "step": 649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8725702146354593e-05,
+      "loss": 0.8328,
+      "step": 650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872113233535687e-05,
+      "loss": 0.7654,
+      "step": 651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716554904699635e-05,
+      "loss": 0.8019,
+      "step": 652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871196985838218e-05,
+      "loss": 0.7773,
+      "step": 653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870737720041048e-05,
+      "loss": 0.8025,
+      "step": 654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702776934797125e-05,
+      "loss": 0.8328,
+      "step": 655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698169065561385e-05,
+      "loss": 0.8254,
+      "step": 656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8693553596729153e-05,
+      "loss": 0.7993,
+      "step": 657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868893053233297e-05,
+      "loss": 0.8491,
+      "step": 658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684299876412e-05,
+      "loss": 0.8223,
+      "step": 659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8679661633012065e-05,
+      "loss": 0.7742,
+      "step": 660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675015806185594e-05,
+      "loss": 0.8909,
+      "step": 661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8670362399991654e-05,
+      "loss": 0.8915,
+      "step": 662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8665701418495922e-05,
+      "loss": 0.8027,
+      "step": 663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8661032865770706e-05,
+      "loss": 0.7598,
+      "step": 664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656356745894922e-05,
+      "loss": 0.8808,
+      "step": 665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8651673062954097e-05,
+      "loss": 0.7739,
+      "step": 666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8646981821040377e-05,
+      "loss": 0.8288,
+      "step": 667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864228302425249e-05,
+      "loss": 0.8047,
+      "step": 668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863757667669579e-05,
+      "loss": 0.7955,
+      "step": 669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8632862782482208e-05,
+      "loss": 0.7761,
+      "step": 670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862814134573028e-05,
+      "loss": 0.8961,
+      "step": 671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8623412370565135e-05,
+      "loss": 0.9151,
+      "step": 672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861867586111847e-05,
+      "loss": 0.7695,
+      "step": 673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8613931821528588e-05,
+      "loss": 0.8467,
+      "step": 674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860918025594035e-05,
+      "loss": 0.8353,
+      "step": 675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860442116850521e-05,
+      "loss": 0.8582,
+      "step": 676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599654563381178e-05,
+      "loss": 0.7394,
+      "step": 677
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594880444732846e-05,
+      "loss": 0.8958,
+      "step": 678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590098816731365e-05,
+      "loss": 0.6966,
+      "step": 679
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8585309683554444e-05,
+      "loss": 0.8277,
+      "step": 680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580513049386356e-05,
+      "loss": 0.8489,
+      "step": 681
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857570891841792e-05,
+      "loss": 0.8564,
+      "step": 682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570897294846508e-05,
+      "loss": 0.8404,
+      "step": 683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8566078182876045e-05,
+      "loss": 0.7869,
+      "step": 684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561251586716985e-05,
+      "loss": 0.886,
+      "step": 685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855641751058633e-05,
+      "loss": 0.8459,
+      "step": 686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551575958707623e-05,
+      "loss": 0.6976,
+      "step": 687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8546726935310925e-05,
+      "loss": 0.7391,
+      "step": 688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854187044463283e-05,
+      "loss": 0.8875,
+      "step": 689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853700649091646e-05,
+      "loss": 0.7763,
+      "step": 690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8532135078411453e-05,
+      "loss": 0.8636,
+      "step": 691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852725621137397e-05,
+      "loss": 0.8354,
+      "step": 692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8522369894066674e-05,
+      "loss": 0.7973,
+      "step": 693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851747613075875e-05,
+      "loss": 0.6865,
+      "step": 694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8512574925725878e-05,
+      "loss": 0.7451,
+      "step": 695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507666283250246e-05,
+      "loss": 0.8075,
+      "step": 696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850275020762054e-05,
+      "loss": 0.8343,
+      "step": 697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8497826703131938e-05,
+      "loss": 0.8782,
+      "step": 698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8492895774086107e-05,
+      "loss": 0.8418,
+      "step": 699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8487957424791206e-05,
+      "loss": 0.874,
+      "step": 700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848301165956187e-05,
+      "loss": 0.7918,
+      "step": 701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478058482719222e-05,
+      "loss": 0.8142,
+      "step": 702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847309789859086e-05,
+      "loss": 0.8412,
+      "step": 703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846812991151084e-05,
+      "loss": 0.8195,
+      "step": 704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84631545258197e-05,
+      "loss": 0.9244,
+      "step": 705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458171745864443e-05,
+      "loss": 0.855,
+      "step": 706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8453181575998518e-05,
+      "loss": 0.7973,
+      "step": 707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448184020581847e-05,
+      "loss": 0.7469,
+      "step": 708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84431790839808e-05,
+      "loss": 0.7345,
+      "step": 709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8438166770568182e-05,
+      "loss": 0.7317,
+      "step": 710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8433147084723268e-05,
+      "loss": 0.8505,
+      "step": 711
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842812003083175e-05,
+      "loss": 0.8373,
+      "step": 712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423085613285775e-05,
+      "loss": 0.8248,
+      "step": 713
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841804383648392e-05,
+      "loss": 0.777,
+      "step": 714
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841299470483118e-05,
+      "loss": 0.8157,
+      "step": 715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407938222738993e-05,
+      "loss": 0.8538,
+      "step": 716
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8402874394625208e-05,
+      "loss": 0.8426,
+      "step": 717
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8397803224914095e-05,
+      "loss": 0.8266,
+      "step": 718
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8392724718036338e-05,
+      "loss": 0.9536,
+      "step": 719
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387638878429034e-05,
+      "loss": 0.835,
+      "step": 720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838254571053568e-05,
+      "loss": 0.7792,
+      "step": 721
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8377445218806186e-05,
+      "loss": 0.8441,
+      "step": 722
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8372337407696858e-05,
+      "loss": 0.8629,
+      "step": 723
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367222281670384e-05,
+      "loss": 0.7404,
+      "step": 724
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362099845195857e-05,
+      "loss": 0.7713,
+      "step": 725
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356970102748758e-05,
+      "loss": 0.8506,
+      "step": 726
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835183305881094e-05,
+      "loss": 0.8176,
+      "step": 727
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834668871787065e-05,
+      "loss": 0.7623,
+      "step": 728
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8341537084422495e-05,
+      "loss": 0.8825,
+      "step": 729
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8336378162967465e-05,
+      "loss": 0.8637,
+      "step": 730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833121195801291e-05,
+      "loss": 0.8669,
+      "step": 731
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832603847407255e-05,
+      "loss": 0.8293,
+      "step": 732
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832085771566646e-05,
+      "loss": 0.8616,
+      "step": 733
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831566968732107e-05,
+      "loss": 0.8078,
+      "step": 734
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310474393569172e-05,
+      "loss": 0.7798,
+      "step": 735
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830527183894989e-05,
+      "loss": 0.8431,
+      "step": 736
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300062028008704e-05,
+      "loss": 0.8313,
+      "step": 737
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8294844965297426e-05,
+      "loss": 0.8077,
+      "step": 738
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8289620655374213e-05,
+      "loss": 0.7811,
+      "step": 739
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284389102803542e-05,
+      "loss": 0.8636,
+      "step": 740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279150312156225e-05,
+      "loss": 0.8191,
+      "step": 741
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8273904288009402e-05,
+      "loss": 0.8101,
+      "step": 742
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8268651034946522e-05,
+      "loss": 0.8162,
+      "step": 743
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8263390557557355e-05,
+      "loss": 0.8265,
+      "step": 744
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8258122860437993e-05,
+      "loss": 0.7793,
+      "step": 745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825284794819082e-05,
+      "loss": 0.7977,
+      "step": 746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824756582542454e-05,
+      "loss": 0.7974,
+      "step": 747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242276496754128e-05,
+      "loss": 0.775,
+      "step": 748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823697996680089e-05,
+      "loss": 0.8356,
+      "step": 749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231676240192402e-05,
+      "loss": 0.8061,
+      "step": 750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8226365321562536e-05,
+      "loss": 0.7821,
+      "step": 751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8221047215551446e-05,
+      "loss": 0.7683,
+      "step": 752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821572192680556e-05,
+      "loss": 0.7749,
+      "step": 753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8210389459977594e-05,
+      "loss": 0.7923,
+      "step": 754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8205049819726523e-05,
+      "loss": 0.7904,
+      "step": 755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81997030107176e-05,
+      "loss": 0.7297,
+      "step": 756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194349037622324e-05,
+      "loss": 0.8261,
+      "step": 757
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8188987905118475e-05,
+      "loss": 0.7839,
+      "step": 758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818361961789008e-05,
+      "loss": 0.8278,
+      "step": 759
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178244180627406e-05,
+      "loss": 0.7455,
+      "step": 760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172861598026986e-05,
+      "loss": 0.787,
+      "step": 761
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8167471874791578e-05,
+      "loss": 0.7823,
+      "step": 762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8162075015630192e-05,
+      "loss": 0.8161,
+      "step": 763
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8156671025258068e-05,
+      "loss": 0.809,
+      "step": 764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151259908396674e-05,
+      "loss": 0.7941,
+      "step": 765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8145841669773706e-05,
+      "loss": 0.7923,
+      "step": 766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8140416314123087e-05,
+      "loss": 0.8475,
+      "step": 767
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813498384618495e-05,
+      "loss": 0.7922,
+      "step": 768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812954427070565e-05,
+      "loss": 0.7731,
+      "step": 769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812409759243774e-05,
+      "loss": 0.7889,
+      "step": 770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8118643816139996e-05,
+      "loss": 0.7699,
+      "step": 771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8113182946577383e-05,
+      "loss": 0.7536,
+      "step": 772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8107714988521064e-05,
+      "loss": 0.7943,
+      "step": 773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81022399467484e-05,
+      "loss": 0.8706,
+      "step": 774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096757826042944e-05,
+      "loss": 0.7929,
+      "step": 775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8091268631194417e-05,
+      "loss": 0.7598,
+      "step": 776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808577236699874e-05,
+      "loss": 0.719,
+      "step": 777
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8080269038258004e-05,
+      "loss": 0.8328,
+      "step": 778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807475864978047e-05,
+      "loss": 0.8055,
+      "step": 779
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8069241206380565e-05,
+      "loss": 0.7841,
+      "step": 780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8063716712878885e-05,
+      "loss": 0.7389,
+      "step": 781
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8058185174102186e-05,
+      "loss": 0.8245,
+      "step": 782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8052646594883375e-05,
+      "loss": 0.7739,
+      "step": 783
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8047100980061517e-05,
+      "loss": 0.8351,
+      "step": 784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041548334481813e-05,
+      "loss": 0.8568,
+      "step": 785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8035988662995617e-05,
+      "loss": 0.8352,
+      "step": 786
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8030421970460417e-05,
+      "loss": 0.7208,
+      "step": 787
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8024848261739837e-05,
+      "loss": 0.83,
+      "step": 788
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801926754170363e-05,
+      "loss": 0.8714,
+      "step": 789
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8013679815227676e-05,
+      "loss": 0.8319,
+      "step": 790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008085087193975e-05,
+      "loss": 0.825,
+      "step": 791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8002483362490643e-05,
+      "loss": 0.8252,
+      "step": 792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799687464601191e-05,
+      "loss": 0.8048,
+      "step": 793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7991258942658122e-05,
+      "loss": 0.8415,
+      "step": 794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7985636257335716e-05,
+      "loss": 0.7298,
+      "step": 795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798000659495724e-05,
+      "loss": 0.7302,
+      "step": 796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7974369960441335e-05,
+      "loss": 0.8342,
+      "step": 797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7968726358712728e-05,
+      "loss": 0.8623,
+      "step": 798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796307579470224e-05,
+      "loss": 0.8621,
+      "step": 799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7957418273346773e-05,
+      "loss": 0.7612,
+      "step": 800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7951753799589312e-05,
+      "loss": 0.7587,
+      "step": 801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7946082378378906e-05,
+      "loss": 0.7763,
+      "step": 802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7940404014670678e-05,
+      "loss": 0.8063,
+      "step": 803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7934718713425825e-05,
+      "loss": 0.8148,
+      "step": 804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792902647961159e-05,
+      "loss": 0.8133,
+      "step": 805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7923327318201295e-05,
+      "loss": 0.7549,
+      "step": 806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791762123417429e-05,
+      "loss": 0.8946,
+      "step": 807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7911908232515988e-05,
+      "loss": 0.8262,
+      "step": 808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7906188318217842e-05,
+      "loss": 0.8212,
+      "step": 809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7900461496277345e-05,
+      "loss": 0.8377,
+      "step": 810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7894727771698027e-05,
+      "loss": 0.7912,
+      "step": 811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788898714948945e-05,
+      "loss": 0.7774,
+      "step": 812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788323963466719e-05,
+      "loss": 0.8432,
+      "step": 813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7877485232252865e-05,
+      "loss": 0.7624,
+      "step": 814
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871723947274096e-05,
+      "loss": 0.7982,
+      "step": 815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7865955784764525e-05,
+      "loss": 0.8377,
+      "step": 816
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7860180749763787e-05,
+      "loss": 0.7993,
+      "step": 817
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7854398847317555e-05,
+      "loss": 0.778,
+      "step": 818
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7848610082477463e-05,
+      "loss": 0.7692,
+      "step": 819
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7842814460301173e-05,
+      "loss": 0.8254,
+      "step": 820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837011985852315e-05,
+      "loss": 0.7767,
+      "step": 821
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7831202664200518e-05,
+      "loss": 0.7502,
+      "step": 822
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7825386500421386e-05,
+      "loss": 0.8202,
+      "step": 823
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7819563499596516e-05,
+      "loss": 0.7318,
+      "step": 824
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7813733666813456e-05,
+      "loss": 0.7962,
+      "step": 825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780789700716574e-05,
+      "loss": 0.7591,
+      "step": 826
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780205352575287e-05,
+      "loss": 0.8069,
+      "step": 827
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796203227680283e-05,
+      "loss": 0.7067,
+      "step": 828
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7790346118059402e-05,
+      "loss": 0.7801,
+      "step": 829
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7784482202007583e-05,
+      "loss": 0.7915,
+      "step": 830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7778611484648135e-05,
+      "loss": 0.7768,
+      "step": 831
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7772733971110305e-05,
+      "loss": 0.7598,
+      "step": 832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7766849666529286e-05,
+      "loss": 0.7777,
+      "step": 833
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7760958576046192e-05,
+      "loss": 0.7664,
+      "step": 834
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775506070480808e-05,
+      "loss": 0.8032,
+      "step": 835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774915605796792e-05,
+      "loss": 0.8085,
+      "step": 836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743244640684605e-05,
+      "loss": 0.825,
+      "step": 837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7737326458122948e-05,
+      "loss": 0.7679,
+      "step": 838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731401515453667e-05,
+      "loss": 0.8027,
+      "step": 839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7725469817853394e-05,
+      "loss": 0.7754,
+      "step": 840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771953137050465e-05,
+      "loss": 0.8391,
+      "step": 841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7713586178595865e-05,
+      "loss": 0.792,
+      "step": 842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7707634247321355e-05,
+      "loss": 0.7653,
+      "step": 843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7701675581881326e-05,
+      "loss": 0.8053,
+      "step": 844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769571018748187e-05,
+      "loss": 0.7722,
+      "step": 845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768973806933496e-05,
+      "loss": 0.8182,
+      "step": 846
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7683759232658432e-05,
+      "loss": 0.8099,
+      "step": 847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7677773682676007e-05,
+      "loss": 0.8502,
+      "step": 848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767178142461726e-05,
+      "loss": 0.8331,
+      "step": 849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766578246371763e-05,
+      "loss": 0.8405,
+      "step": 850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765977680521842e-05,
+      "loss": 0.8089,
+      "step": 851
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7653764454366774e-05,
+      "loss": 0.7632,
+      "step": 852
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764774541641568e-05,
+      "loss": 0.7881,
+      "step": 853
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764171969662399e-05,
+      "loss": 0.7956,
+      "step": 854
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635687300256374e-05,
+      "loss": 0.7301,
+      "step": 855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7629648232583334e-05,
+      "loss": 0.6577,
+      "step": 856
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7623602498881217e-05,
+      "loss": 0.6924,
+      "step": 857
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7617550104432175e-05,
+      "loss": 0.7468,
+      "step": 858
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76114910545242e-05,
+      "loss": 0.8257,
+      "step": 859
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7605425354451085e-05,
+      "loss": 0.8391,
+      "step": 860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7599353009512432e-05,
+      "loss": 0.7344,
+      "step": 861
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7593274025013657e-05,
+      "loss": 0.756,
+      "step": 862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758718840626597e-05,
+      "loss": 0.6993,
+      "step": 863
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7581096158586386e-05,
+      "loss": 0.7509,
+      "step": 864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75749972872977e-05,
+      "loss": 0.8015,
+      "step": 865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7568891797728504e-05,
+      "loss": 0.8503,
+      "step": 866
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7562779695213167e-05,
+      "loss": 0.8258,
+      "step": 867
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755666098509184e-05,
+      "loss": 0.8454,
+      "step": 868
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755053567271044e-05,
+      "loss": 0.7598,
+      "step": 869
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7544403763420657e-05,
+      "loss": 0.8129,
+      "step": 870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753826526257995e-05,
+      "loss": 0.7526,
+      "step": 871
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7532120175551528e-05,
+      "loss": 0.8033,
+      "step": 872
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7525968507704356e-05,
+      "loss": 0.817,
+      "step": 873
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7519810264413154e-05,
+      "loss": 0.825,
+      "step": 874
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751364545105838e-05,
+      "loss": 0.818,
+      "step": 875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750747407302624e-05,
+      "loss": 0.7752,
+      "step": 876
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7501296135708668e-05,
+      "loss": 0.7747,
+      "step": 877
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7495111644503336e-05,
+      "loss": 0.8174,
+      "step": 878
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7488920604813633e-05,
+      "loss": 0.8298,
+      "step": 879
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7482723022048675e-05,
+      "loss": 0.8332,
+      "step": 880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74765189016233e-05,
+      "loss": 0.773,
+      "step": 881
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747030824895805e-05,
+      "loss": 0.8024,
+      "step": 882
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7464091069479175e-05,
+      "loss": 0.7868,
+      "step": 883
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745786736861863e-05,
+      "loss": 0.7944,
+      "step": 884
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745163715181407e-05,
+      "loss": 0.8286,
+      "step": 885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744540042450883e-05,
+      "loss": 0.8102,
+      "step": 886
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7439157192151958e-05,
+      "loss": 0.7882,
+      "step": 887
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7432907460198157e-05,
+      "loss": 0.7595,
+      "step": 888
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742665123410783e-05,
+      "loss": 0.8535,
+      "step": 889
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742038851934704e-05,
+      "loss": 0.8357,
+      "step": 890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7414119321387538e-05,
+      "loss": 0.785,
+      "step": 891
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740784364570671e-05,
+      "loss": 0.7914,
+      "step": 892
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401561497787632e-05,
+      "loss": 0.7757,
+      "step": 893
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739527288311901e-05,
+      "loss": 0.7895,
+      "step": 894
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738897780719522e-05,
+      "loss": 0.8067,
+      "step": 895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7382676275516267e-05,
+      "loss": 0.8142,
+      "step": 896
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737636829358781e-05,
+      "loss": 0.7956,
+      "step": 897
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7370053866921135e-05,
+      "loss": 0.7586,
+      "step": 898
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736373300103316e-05,
+      "loss": 0.748,
+      "step": 899
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7357405701446433e-05,
+      "loss": 0.9262,
+      "step": 900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7351071973689118e-05,
+      "loss": 0.7438,
+      "step": 901
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7344731823294996e-05,
+      "loss": 0.8238,
+      "step": 902
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733838525580346e-05,
+      "loss": 0.7677,
+      "step": 903
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7332032276759515e-05,
+      "loss": 0.7855,
+      "step": 904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732567289171376e-05,
+      "loss": 0.799,
+      "step": 905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7319307106222402e-05,
+      "loss": 0.8357,
+      "step": 906
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731293492584722e-05,
+      "loss": 0.7936,
+      "step": 907
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73065563561556e-05,
+      "loss": 0.8721,
+      "step": 908
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7300171402720498e-05,
+      "loss": 0.7377,
+      "step": 909
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729378007112046e-05,
+      "loss": 0.7635,
+      "step": 910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7287382366939586e-05,
+      "loss": 0.7732,
+      "step": 911
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7280978295767567e-05,
+      "loss": 0.8032,
+      "step": 912
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7274567863199637e-05,
+      "loss": 0.7969,
+      "step": 913
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7268151074836595e-05,
+      "loss": 0.7683,
+      "step": 914
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72617279362848e-05,
+      "loss": 0.7743,
+      "step": 915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7255298453156147e-05,
+      "loss": 0.7352,
+      "step": 916
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724886263106808e-05,
+      "loss": 0.7429,
+      "step": 917
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7242420475643583e-05,
+      "loss": 0.7661,
+      "step": 918
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723597199251117e-05,
+      "loss": 0.8381,
+      "step": 919
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722951718730489e-05,
+      "loss": 0.7739,
+      "step": 920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7223056065664306e-05,
+      "loss": 0.8411,
+      "step": 921
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7216588633234507e-05,
+      "loss": 0.7819,
+      "step": 922
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721011489566609e-05,
+      "loss": 0.8067,
+      "step": 923
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720363485861517e-05,
+      "loss": 0.768,
+      "step": 924
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7197148527743356e-05,
+      "loss": 0.7977,
+      "step": 925
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7190655908717757e-05,
+      "loss": 0.861,
+      "step": 926
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7184157007210984e-05,
+      "loss": 0.748,
+      "step": 927
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7177651828901136e-05,
+      "loss": 0.7709,
+      "step": 928
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717114037947178e-05,
+      "loss": 0.7296,
+      "step": 929
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716462266461198e-05,
+      "loss": 0.8653,
+      "step": 930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715809869001627e-05,
+      "loss": 0.783,
+      "step": 931
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715156846138465e-05,
+      "loss": 0.8323,
+      "step": 932
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145031984422583e-05,
+      "loss": 0.7734,
+      "step": 933
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7138489264840997e-05,
+      "loss": 0.6981,
+      "step": 934
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7131940308356267e-05,
+      "loss": 0.8228,
+      "step": 935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7125385120690227e-05,
+      "loss": 0.7702,
+      "step": 936
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7118823707570144e-05,
+      "loss": 0.7757,
+      "step": 937
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7112256074728728e-05,
+      "loss": 0.8479,
+      "step": 938
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7105682227904127e-05,
+      "loss": 0.9079,
+      "step": 939
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7099102172839918e-05,
+      "loss": 0.8685,
+      "step": 940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70925159152851e-05,
+      "loss": 0.8605,
+      "step": 941
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708592346099408e-05,
+      "loss": 0.7583,
+      "step": 942
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7079324815726708e-05,
+      "loss": 0.8292,
+      "step": 943
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707271998524821e-05,
+      "loss": 0.8065,
+      "step": 944
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7066108975329237e-05,
+      "loss": 0.8104,
+      "step": 945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7059491791745833e-05,
+      "loss": 0.8519,
+      "step": 946
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7052868440279435e-05,
+      "loss": 0.8221,
+      "step": 947
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704623892671687e-05,
+      "loss": 0.7562,
+      "step": 948
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703960325685035e-05,
+      "loss": 0.7959,
+      "step": 949
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7032961436477468e-05,
+      "loss": 0.764,
+      "step": 950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702631347140118e-05,
+      "loss": 0.7204,
+      "step": 951
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7019659367429815e-05,
+      "loss": 0.9451,
+      "step": 952
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701299913037708e-05,
+      "loss": 0.8056,
+      "step": 953
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700633276606202e-05,
+      "loss": 0.804,
+      "step": 954
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6999660280309047e-05,
+      "loss": 0.8622,
+      "step": 955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699298167894791e-05,
+      "loss": 0.8046,
+      "step": 956
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698629696781372e-05,
+      "loss": 0.754,
+      "step": 957
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69796061527469e-05,
+      "loss": 0.8228,
+      "step": 958
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697290923959323e-05,
+      "loss": 0.7298,
+      "step": 959
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69662062342038e-05,
+      "loss": 0.7494,
+      "step": 960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695949714243504e-05,
+      "loss": 0.8578,
+      "step": 961
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6952781970148677e-05,
+      "loss": 0.8333,
+      "step": 962
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694606072321177e-05,
+      "loss": 0.8991,
+      "step": 963
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6939333407496678e-05,
+      "loss": 0.8224,
+      "step": 964
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6932600028881053e-05,
+      "loss": 0.8321,
+      "step": 965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692586059324786e-05,
+      "loss": 0.8174,
+      "step": 966
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6919115106485346e-05,
+      "loss": 0.6674,
+      "step": 967
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691236357448705e-05,
+      "loss": 0.7733,
+      "step": 968
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6905606003151785e-05,
+      "loss": 0.8742,
+      "step": 969
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689884239838365e-05,
+      "loss": 0.7887,
+      "step": 970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6892072766092014e-05,
+      "loss": 0.8267,
+      "step": 971
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6885297112191504e-05,
+      "loss": 0.8104,
+      "step": 972
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687851544260202e-05,
+      "loss": 0.7244,
+      "step": 973
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6871727763248697e-05,
+      "loss": 0.7844,
+      "step": 974
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6864934080061955e-05,
+      "loss": 0.9305,
+      "step": 975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6858134398977428e-05,
+      "loss": 0.7851,
+      "step": 976
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6851328725936003e-05,
+      "loss": 0.9336,
+      "step": 977
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6844517066883806e-05,
+      "loss": 0.717,
+      "step": 978
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683769942777218e-05,
+      "loss": 0.837,
+      "step": 979
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683087581455771e-05,
+      "loss": 0.7731,
+      "step": 980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682404623320219e-05,
+      "loss": 0.743,
+      "step": 981
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817210689672626e-05,
+      "loss": 0.8237,
+      "step": 982
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810369189941242e-05,
+      "loss": 0.806,
+      "step": 983
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6803521739985455e-05,
+      "loss": 0.8469,
+      "step": 984
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6796668345787896e-05,
+      "loss": 0.7595,
+      "step": 985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678980901333637e-05,
+      "loss": 0.7491,
+      "step": 986
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6782943748623886e-05,
+      "loss": 0.8043,
+      "step": 987
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776072557648634e-05,
+      "loss": 0.8483,
+      "step": 988
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676919544641397e-05,
+      "loss": 0.7908,
+      "step": 989
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6762312420928436e-05,
+      "loss": 0.7287,
+      "step": 990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675542348720574e-05,
+      "loss": 0.7858,
+      "step": 991
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674852865126474e-05,
+      "loss": 0.7655,
+      "step": 992
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6741627919129467e-05,
+      "loss": 0.7828,
+      "step": 993
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673472129682909e-05,
+      "loss": 0.7158,
+      "step": 994
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6727808790397934e-05,
+      "loss": 0.8019,
+      "step": 995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6720890405875458e-05,
+      "loss": 0.7816,
+      "step": 996
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713966149306257e-05,
+      "loss": 0.8023,
+      "step": 997
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670703602674006e-05,
+      "loss": 0.7531,
+      "step": 998
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6700100044231728e-05,
+      "loss": 0.7819,
+      "step": 999
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6693158207841224e-05,
+      "loss": 0.7658,
+      "step": 1000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668621052363364e-05,
+      "loss": 0.8202,
+      "step": 1001
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667925699767917e-05,
+      "loss": 0.8617,
+      "step": 1002
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6672297636053118e-05,
+      "loss": 0.8469,
+      "step": 1003
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666533244483588e-05,
+      "loss": 0.7637,
+      "step": 1004
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6658361430112956e-05,
+      "loss": 0.7797,
+      "step": 1005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665138459797491e-05,
+      "loss": 0.7619,
+      "step": 1006
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6644401954517426e-05,
+      "loss": 0.7517,
+      "step": 1007
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637413505841225e-05,
+      "loss": 0.8885,
+      "step": 1008
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6630419258052136e-05,
+      "loss": 0.8229,
+      "step": 1009
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6623419217261026e-05,
+      "loss": 0.7969,
+      "step": 1010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6616413389583844e-05,
+      "loss": 0.7192,
+      "step": 1011
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6609401781141583e-05,
+      "loss": 0.8438,
+      "step": 1012
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660238439806029e-05,
+      "loss": 0.804,
+      "step": 1013
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6595361246471057e-05,
+      "loss": 0.746,
+      "step": 1014
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658833233251002e-05,
+      "loss": 0.9336,
+      "step": 1015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6581297662318343e-05,
+      "loss": 0.752,
+      "step": 1016
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657425724204222e-05,
+      "loss": 0.8783,
+      "step": 1017
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6567211077832878e-05,
+      "loss": 0.792,
+      "step": 1018
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656015917584655e-05,
+      "loss": 0.7441,
+      "step": 1019
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655310154224449e-05,
+      "loss": 0.7541,
+      "step": 1020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654603818319296e-05,
+      "loss": 0.8094,
+      "step": 1021
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6538969104863215e-05,
+      "loss": 0.809,
+      "step": 1022
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6531894313431522e-05,
+      "loss": 0.8077,
+      "step": 1023
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6524813815079124e-05,
+      "loss": 0.7793,
+      "step": 1024
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517727615992263e-05,
+      "loss": 0.8526,
+      "step": 1025
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6510635722362152e-05,
+      "loss": 0.7775,
+      "step": 1026
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6503538140384984e-05,
+      "loss": 0.8373,
+      "step": 1027
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6496434876261926e-05,
+      "loss": 0.7561,
+      "step": 1028
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64893259361991e-05,
+      "loss": 0.7425,
+      "step": 1029
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6482211326407595e-05,
+      "loss": 0.7976,
+      "step": 1030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6475091053103446e-05,
+      "loss": 0.8252,
+      "step": 1031
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646796512250765e-05,
+      "loss": 0.8179,
+      "step": 1032
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6460833540846123e-05,
+      "loss": 0.7787,
+      "step": 1033
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453696314349746e-05,
+      "loss": 0.8582,
+      "step": 1034
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6446553449254316e-05,
+      "loss": 0.8349,
+      "step": 1035
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439404951800557e-05,
+      "loss": 0.8786,
+      "step": 1036
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6432250828234117e-05,
+      "loss": 0.8323,
+      "step": 1037
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6425091084805557e-05,
+      "loss": 0.8883,
+      "step": 1038
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6417925727770355e-05,
+      "loss": 0.7703,
+      "step": 1039
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6410754763388884e-05,
+      "loss": 0.7133,
+      "step": 1040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640357819792642e-05,
+      "loss": 0.787,
+      "step": 1041
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639639603765314e-05,
+      "loss": 0.794,
+      "step": 1042
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6389208288844094e-05,
+      "loss": 0.7754,
+      "step": 1043
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6382014957779227e-05,
+      "loss": 0.7651,
+      "step": 1044
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6374816050743357e-05,
+      "loss": 0.7244,
+      "step": 1045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636761157402618e-05,
+      "loss": 0.7572,
+      "step": 1046
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6360401533922244e-05,
+      "loss": 0.8021,
+      "step": 1047
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353185936730972e-05,
+      "loss": 0.7756,
+      "step": 1048
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634596478875663e-05,
+      "loss": 0.8106,
+      "step": 1049
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6338738096308345e-05,
+      "loss": 0.8832,
+      "step": 1050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6331505865700084e-05,
+      "loss": 0.8046,
+      "step": 1051
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6324268103250648e-05,
+      "loss": 0.7462,
+      "step": 1052
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631702481528368e-05,
+      "loss": 0.7464,
+      "step": 1053
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6309776008127642e-05,
+      "loss": 0.7718,
+      "step": 1054
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6302521688115823e-05,
+      "loss": 0.813,
+      "step": 1055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6295261861586328e-05,
+      "loss": 0.7459,
+      "step": 1056
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628799653488207e-05,
+      "loss": 0.7226,
+      "step": 1057
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628072571435078e-05,
+      "loss": 0.8472,
+      "step": 1058
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6273449406344964e-05,
+      "loss": 0.7515,
+      "step": 1059
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6266167617221948e-05,
+      "loss": 0.7948,
+      "step": 1060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6258880353343832e-05,
+      "loss": 0.7151,
+      "step": 1061
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6251587621077506e-05,
+      "loss": 0.717,
+      "step": 1062
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6244289426794636e-05,
+      "loss": 0.7252,
+      "step": 1063
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6236985776871656e-05,
+      "loss": 0.7702,
+      "step": 1064
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6229676677689773e-05,
+      "loss": 0.8816,
+      "step": 1065
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622236213563495e-05,
+      "loss": 0.8192,
+      "step": 1066
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621504215709791e-05,
+      "loss": 0.8301,
+      "step": 1067
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6207716748474122e-05,
+      "loss": 0.7527,
+      "step": 1068
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6200385916163803e-05,
+      "loss": 0.8572,
+      "step": 1069
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61930496665719e-05,
+      "loss": 0.8022,
+      "step": 1070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618570800610811e-05,
+      "loss": 0.7938,
+      "step": 1071
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6178360941186834e-05,
+      "loss": 0.8529,
+      "step": 1072
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6171008478227216e-05,
+      "loss": 0.8072,
+      "step": 1073
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616365062365311e-05,
+      "loss": 0.8222,
+      "step": 1074
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6156287383893073e-05,
+      "loss": 0.8391,
+      "step": 1075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6148918765380377e-05,
+      "loss": 0.7406,
+      "step": 1076
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6141544774552987e-05,
+      "loss": 0.8177,
+      "step": 1077
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6134165417853563e-05,
+      "loss": 0.8398,
+      "step": 1078
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6126780701729458e-05,
+      "loss": 0.8228,
+      "step": 1079
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61193906326327e-05,
+      "loss": 0.7923,
+      "step": 1080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611199521702e-05,
+      "loss": 0.7928,
+      "step": 1081
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6104594461352733e-05,
+      "loss": 0.8137,
+      "step": 1082
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6097188372096955e-05,
+      "loss": 0.7491,
+      "step": 1083
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6089776955723362e-05,
+      "loss": 0.8255,
+      "step": 1084
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6082360218707316e-05,
+      "loss": 0.8053,
+      "step": 1085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6074938167528827e-05,
+      "loss": 0.8247,
+      "step": 1086
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6067510808672545e-05,
+      "loss": 0.7846,
+      "step": 1087
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606007814862776e-05,
+      "loss": 0.751,
+      "step": 1088
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605264019388839e-05,
+      "loss": 0.7349,
+      "step": 1089
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6045196950952983e-05,
+      "loss": 0.7222,
+      "step": 1090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6037748426324705e-05,
+      "loss": 0.7601,
+      "step": 1091
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603029462651134e-05,
+      "loss": 0.7817,
+      "step": 1092
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6022835558025273e-05,
+      "loss": 0.7557,
+      "step": 1093
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6015371227383504e-05,
+      "loss": 0.7499,
+      "step": 1094
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600790164110762e-05,
+      "loss": 0.7761,
+      "step": 1095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6000426805723802e-05,
+      "loss": 0.7768,
+      "step": 1096
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599294672776282e-05,
+      "loss": 0.8232,
+      "step": 1097
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5985461413760028e-05,
+      "loss": 0.7805,
+      "step": 1098
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5977970870255345e-05,
+      "loss": 0.783,
+      "step": 1099
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5970475103793266e-05,
+      "loss": 0.746,
+      "step": 1100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596297412092285e-05,
+      "loss": 0.7905,
+      "step": 1101
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5955467928197704e-05,
+      "loss": 0.7627,
+      "step": 1102
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5947956532176e-05,
+      "loss": 0.7931,
+      "step": 1103
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5940439939420448e-05,
+      "loss": 0.7173,
+      "step": 1104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5932918156498306e-05,
+      "loss": 0.8073,
+      "step": 1105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592539118998135e-05,
+      "loss": 0.7998,
+      "step": 1106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5917859046445906e-05,
+      "loss": 0.8477,
+      "step": 1107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5910321732472805e-05,
+      "loss": 0.8442,
+      "step": 1108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590277925464741e-05,
+      "loss": 0.762,
+      "step": 1109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5895231619559588e-05,
+      "loss": 0.8351,
+      "step": 1110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588767883380371e-05,
+      "loss": 0.7519,
+      "step": 1111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5880120903978652e-05,
+      "loss": 0.8622,
+      "step": 1112
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587255783668779e-05,
+      "loss": 0.7802,
+      "step": 1113
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5864989638538967e-05,
+      "loss": 0.8054,
+      "step": 1114
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585741631614453e-05,
+      "loss": 0.909,
+      "step": 1115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5849837876121295e-05,
+      "loss": 0.6857,
+      "step": 1116
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5842254325090557e-05,
+      "loss": 0.7285,
+      "step": 1117
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5834665669678065e-05,
+      "loss": 0.7571,
+      "step": 1118
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582707191651403e-05,
+      "loss": 0.818,
+      "step": 1119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5819473072233123e-05,
+      "loss": 0.7909,
+      "step": 1120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5811869143474458e-05,
+      "loss": 0.6899,
+      "step": 1121
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5804260136881596e-05,
+      "loss": 0.8204,
+      "step": 1122
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5796646059102527e-05,
+      "loss": 0.7609,
+      "step": 1123
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5789026916789677e-05,
+      "loss": 0.7714,
+      "step": 1124
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5781402716599902e-05,
+      "loss": 0.7849,
+      "step": 1125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577377346519446e-05,
+      "loss": 0.8239,
+      "step": 1126
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5766139169239044e-05,
+      "loss": 0.7631,
+      "step": 1127
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575849983540374e-05,
+      "loss": 0.8314,
+      "step": 1128
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575085547036304e-05,
+      "loss": 0.7567,
+      "step": 1129
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5743206080795825e-05,
+      "loss": 0.7496,
+      "step": 1130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573555167338538e-05,
+      "loss": 0.6943,
+      "step": 1131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5727892254819363e-05,
+      "loss": 0.806,
+      "step": 1132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5720227831789816e-05,
+      "loss": 0.8079,
+      "step": 1133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5712558410993147e-05,
+      "loss": 0.7533,
+      "step": 1134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5704883999130137e-05,
+      "loss": 0.7921,
+      "step": 1135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5697204602905926e-05,
+      "loss": 0.7497,
+      "step": 1136
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5689520229030002e-05,
+      "loss": 0.8297,
+      "step": 1137
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5681830884216215e-05,
+      "loss": 0.7995,
+      "step": 1138
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567413657518275e-05,
+      "loss": 0.7455,
+      "step": 1139
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5666437308652122e-05,
+      "loss": 0.8105,
+      "step": 1140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5658733091351195e-05,
+      "loss": 0.8085,
+      "step": 1141
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565102393001115e-05,
+      "loss": 0.7464,
+      "step": 1142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5643309831367477e-05,
+      "loss": 0.7321,
+      "step": 1143
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563559080216e-05,
+      "loss": 0.793,
+      "step": 1144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5627866849132837e-05,
+      "loss": 0.8627,
+      "step": 1145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5620137979034403e-05,
+      "loss": 0.8261,
+      "step": 1146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561240419861743e-05,
+      "loss": 0.8254,
+      "step": 1147
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560466551463892e-05,
+      "loss": 0.77,
+      "step": 1148
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5596921933860167e-05,
+      "loss": 0.8718,
+      "step": 1149
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5589173463046745e-05,
+      "loss": 0.8452,
+      "step": 1150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55814201089685e-05,
+      "loss": 0.7696,
+      "step": 1151
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557366187839954e-05,
+      "loss": 0.7958,
+      "step": 1152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5565898778118236e-05,
+      "loss": 0.7837,
+      "step": 1153
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5558130814907215e-05,
+      "loss": 0.8007,
+      "step": 1154
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5550357995553353e-05,
+      "loss": 0.7241,
+      "step": 1155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554258032684777e-05,
+      "loss": 0.7359,
+      "step": 1156
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5534797815585813e-05,
+      "loss": 0.7525,
+      "step": 1157
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552701046856708e-05,
+      "loss": 0.7933,
+      "step": 1158
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551921829259537e-05,
+      "loss": 0.7215,
+      "step": 1159
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5511421294478715e-05,
+      "loss": 0.7096,
+      "step": 1160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550361948102936e-05,
+      "loss": 0.7825,
+      "step": 1161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549581285906376e-05,
+      "loss": 0.7753,
+      "step": 1162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548800143540256e-05,
+      "loss": 0.8385,
+      "step": 1163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54801852168706e-05,
+      "loss": 0.728,
+      "step": 1164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5472364210296928e-05,
+      "loss": 0.6918,
+      "step": 1165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5464538422514757e-05,
+      "loss": 0.7072,
+      "step": 1166
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5456707860361485e-05,
+      "loss": 0.6752,
+      "step": 1167
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5448872530678678e-05,
+      "loss": 0.8231,
+      "step": 1168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544103244031207e-05,
+      "loss": 0.832,
+      "step": 1169
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5433187596111557e-05,
+      "loss": 0.856,
+      "step": 1170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542533800493118e-05,
+      "loss": 0.7748,
+      "step": 1171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541748367362913e-05,
+      "loss": 0.7258,
+      "step": 1172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540962460906775e-05,
+      "loss": 0.7263,
+      "step": 1173
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5401760818113504e-05,
+      "loss": 0.7517,
+      "step": 1174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5393892307636996e-05,
+      "loss": 0.7572,
+      "step": 1175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5386019084512946e-05,
+      "loss": 0.8224,
+      "step": 1176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5378141155620196e-05,
+      "loss": 0.7707,
+      "step": 1177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5370258527841696e-05,
+      "loss": 0.7461,
+      "step": 1178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5362371208064507e-05,
+      "loss": 0.8226,
+      "step": 1179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5354479203179785e-05,
+      "loss": 0.7191,
+      "step": 1180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5346582520082777e-05,
+      "loss": 0.8071,
+      "step": 1181
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5338681165672825e-05,
+      "loss": 0.7079,
+      "step": 1182
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5330775146853345e-05,
+      "loss": 0.7826,
+      "step": 1183
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5322864470531834e-05,
+      "loss": 0.6778,
+      "step": 1184
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5314949143619853e-05,
+      "loss": 0.8466,
+      "step": 1185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5307029173033026e-05,
+      "loss": 0.8357,
+      "step": 1186
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5299104565691047e-05,
+      "loss": 0.8063,
+      "step": 1187
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5291175328517638e-05,
+      "loss": 0.8193,
+      "step": 1188
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528324146844059e-05,
+      "loss": 0.8017,
+      "step": 1189
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5275302992391717e-05,
+      "loss": 0.7202,
+      "step": 1190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526735990730687e-05,
+      "loss": 0.7469,
+      "step": 1191
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525941222012593e-05,
+      "loss": 0.7968,
+      "step": 1192
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52514599377928e-05,
+      "loss": 0.7518,
+      "step": 1193
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5243503067255389e-05,
+      "loss": 0.8074,
+      "step": 1194
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5235541615465628e-05,
+      "loss": 0.7342,
+      "step": 1195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5227575589379433e-05,
+      "loss": 0.7424,
+      "step": 1196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5219604995956732e-05,
+      "loss": 0.7857,
+      "step": 1197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5211629842161442e-05,
+      "loss": 0.8523,
+      "step": 1198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5203650134961454e-05,
+      "loss": 0.7243,
+      "step": 1199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5195665881328646e-05,
+      "loss": 0.8693,
+      "step": 1200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5187677088238869e-05,
+      "loss": 0.8163,
+      "step": 1201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5179683762671933e-05,
+      "loss": 0.8152,
+      "step": 1202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5171685911611613e-05,
+      "loss": 0.8196,
+      "step": 1203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516368354204564e-05,
+      "loss": 0.8847,
+      "step": 1204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5155676660965686e-05,
+      "loss": 0.8068,
+      "step": 1205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5147665275367374e-05,
+      "loss": 0.8135,
+      "step": 1206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513964939225025e-05,
+      "loss": 0.7574,
+      "step": 1207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5131629018617803e-05,
+      "loss": 0.8414,
+      "step": 1208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5123604161477436e-05,
+      "loss": 0.9241,
+      "step": 1209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5115574827840475e-05,
+      "loss": 0.727,
+      "step": 1210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5107541024722152e-05,
+      "loss": 0.8064,
+      "step": 1211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5099502759141609e-05,
+      "loss": 0.8009,
+      "step": 1212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5091460038121878e-05,
+      "loss": 0.821,
+      "step": 1213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5083412868689899e-05,
+      "loss": 0.786,
+      "step": 1214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5075361257876484e-05,
+      "loss": 0.7325,
+      "step": 1215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5067305212716332e-05,
+      "loss": 0.7479,
+      "step": 1216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505924474024802e-05,
+      "loss": 0.7734,
+      "step": 1217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5051179847513979e-05,
+      "loss": 0.8238,
+      "step": 1218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043110541560519e-05,
+      "loss": 0.7355,
+      "step": 1219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5035036829437797e-05,
+      "loss": 0.8415,
+      "step": 1220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5026958718199823e-05,
+      "loss": 0.7636,
+      "step": 1221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5018876214904441e-05,
+      "loss": 0.7171,
+      "step": 1222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5010789326613347e-05,
+      "loss": 0.6961,
+      "step": 1223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5002698060392056e-05,
+      "loss": 0.7769,
+      "step": 1224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4994602423309919e-05,
+      "loss": 0.8787,
+      "step": 1225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4986502422440095e-05,
+      "loss": 0.7822,
+      "step": 1226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497839806485956e-05,
+      "loss": 0.6975,
+      "step": 1227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4970289357649104e-05,
+      "loss": 0.7841,
+      "step": 1228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.49621763078933e-05,
+      "loss": 0.852,
+      "step": 1229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4954058922680535e-05,
+      "loss": 0.8055,
+      "step": 1230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494593720910297e-05,
+      "loss": 0.7774,
+      "step": 1231
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4937811174256549e-05,
+      "loss": 0.7506,
+      "step": 1232
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4929680825241e-05,
+      "loss": 0.739,
+      "step": 1233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4921546169159808e-05,
+      "loss": 0.7721,
+      "step": 1234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913407213120235e-05,
+      "loss": 0.7517,
+      "step": 1235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490526396423329e-05,
+      "loss": 0.7933,
+      "step": 1236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4897116429613734e-05,
+      "loss": 0.8189,
+      "step": 1237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4888964616380073e-05,
+      "loss": 0.8115,
+      "step": 1238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4880808531654554e-05,
+      "loss": 0.7911,
+      "step": 1239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4872648182563154e-05,
+      "loss": 0.7939,
+      "step": 1240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4864483576235577e-05,
+      "loss": 0.7591,
+      "step": 1241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4856314719805243e-05,
+      "loss": 0.8912,
+      "step": 1242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4848141620409289e-05,
+      "loss": 0.8192,
+      "step": 1243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4839964285188556e-05,
+      "loss": 0.7562,
+      "step": 1244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483178272128759e-05,
+      "loss": 0.7729,
+      "step": 1245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4823596935854625e-05,
+      "loss": 0.8112,
+      "step": 1246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4815406936041592e-05,
+      "loss": 0.8464,
+      "step": 1247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48072127290041e-05,
+      "loss": 0.7869,
+      "step": 1248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4799014321901431e-05,
+      "loss": 0.7779,
+      "step": 1249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4790811721896539e-05,
+      "loss": 0.6942,
+      "step": 1250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478260493615604e-05,
+      "loss": 0.7879,
+      "step": 1251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4774393971850214e-05,
+      "loss": 0.7652,
+      "step": 1252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4766178836152983e-05,
+      "loss": 0.7394,
+      "step": 1253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4757959536241917e-05,
+      "loss": 0.8186,
+      "step": 1254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4749736079298222e-05,
+      "loss": 0.8569,
+      "step": 1255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4741508472506743e-05,
+      "loss": 0.7815,
+      "step": 1256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733276723055941e-05,
+      "loss": 0.7544,
+      "step": 1257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4725040838137905e-05,
+      "loss": 0.8598,
+      "step": 1258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4716800824948331e-05,
+      "loss": 0.7399,
+      "step": 1259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4708556690686526e-05,
+      "loss": 0.8044,
+      "step": 1260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4700308442555395e-05,
+      "loss": 0.775,
+      "step": 1261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692056087761439e-05,
+      "loss": 0.8601,
+      "step": 1262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4683799633514748e-05,
+      "loss": 0.728,
+      "step": 1263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4675539087028987e-05,
+      "loss": 0.8295,
+      "step": 1264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4667274455521405e-05,
+      "loss": 0.7515,
+      "step": 1265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4659005746212816e-05,
+      "loss": 0.8037,
+      "step": 1266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4650732966327598e-05,
+      "loss": 0.8117,
+      "step": 1267
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464245612309368e-05,
+      "loss": 0.7877,
+      "step": 1268
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4634175223742553e-05,
+      "loss": 0.7743,
+      "step": 1269
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4625890275509237e-05,
+      "loss": 0.7755,
+      "step": 1270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4617601285632308e-05,
+      "loss": 0.7915,
+      "step": 1271
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460930826135385e-05,
+      "loss": 0.728,
+      "step": 1272
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4601011209919489e-05,
+      "loss": 0.8351,
+      "step": 1273
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4592710138578365e-05,
+      "loss": 0.7416,
+      "step": 1274
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4584405054583133e-05,
+      "loss": 0.7784,
+      "step": 1275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457609596518995e-05,
+      "loss": 0.7706,
+      "step": 1276
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4567782877658467e-05,
+      "loss": 0.7992,
+      "step": 1277
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455946579925184e-05,
+      "loss": 0.797,
+      "step": 1278
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4551144737236707e-05,
+      "loss": 0.7703,
+      "step": 1279
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542819698883184e-05,
+      "loss": 0.7772,
+      "step": 1280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4534490691464862e-05,
+      "loss": 0.8464,
+      "step": 1281
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45261577222588e-05,
+      "loss": 0.7848,
+      "step": 1282
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4517820798545524e-05,
+      "loss": 0.7593,
+      "step": 1283
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4509479927609001e-05,
+      "loss": 0.7551,
+      "step": 1284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450113511673667e-05,
+      "loss": 0.8283,
+      "step": 1285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4492786373219382e-05,
+      "loss": 0.7697,
+      "step": 1286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4484433704351453e-05,
+      "loss": 0.7113,
+      "step": 1287
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4476077117430609e-05,
+      "loss": 0.751,
+      "step": 1288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446771661975801e-05,
+      "loss": 0.6971,
+      "step": 1289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4459352218638226e-05,
+      "loss": 0.7905,
+      "step": 1290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4450983921379238e-05,
+      "loss": 0.802,
+      "step": 1291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4442611735292435e-05,
+      "loss": 0.8705,
+      "step": 1292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4434235667692607e-05,
+      "loss": 0.8147,
+      "step": 1293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4425855725897924e-05,
+      "loss": 0.7286,
+      "step": 1294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441747191722995e-05,
+      "loss": 0.758,
+      "step": 1295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409084249013623e-05,
+      "loss": 0.7975,
+      "step": 1296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4400692728577257e-05,
+      "loss": 0.8442,
+      "step": 1297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4392297363252525e-05,
+      "loss": 0.7742,
+      "step": 1298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438389816037447e-05,
+      "loss": 0.7571,
+      "step": 1299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4375495127281474e-05,
+      "loss": 0.7553,
+      "step": 1300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4367088271315279e-05,
+      "loss": 0.8215,
+      "step": 1301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4358677599820961e-05,
+      "loss": 0.8783,
+      "step": 1302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4350263120146925e-05,
+      "loss": 0.7894,
+      "step": 1303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4341844839644913e-05,
+      "loss": 0.776,
+      "step": 1304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433342276566998e-05,
+      "loss": 0.7127,
+      "step": 1305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324996905580493e-05,
+      "loss": 0.7415,
+      "step": 1306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4316567266738142e-05,
+      "loss": 0.7894,
+      "step": 1307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4308133856507905e-05,
+      "loss": 0.7498,
+      "step": 1308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4299696682258053e-05,
+      "loss": 0.7526,
+      "step": 1309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4291255751360158e-05,
+      "loss": 0.7167,
+      "step": 1310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428281107118906e-05,
+      "loss": 0.7981,
+      "step": 1311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4274362649122892e-05,
+      "loss": 0.731,
+      "step": 1312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4265910492543041e-05,
+      "loss": 0.7755,
+      "step": 1313
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425745460883416e-05,
+      "loss": 0.7084,
+      "step": 1314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4248995005384163e-05,
+      "loss": 0.7717,
+      "step": 1315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424053168958421e-05,
+      "loss": 0.7102,
+      "step": 1316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423206466882871e-05,
+      "loss": 0.7825,
+      "step": 1317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42235939505153e-05,
+      "loss": 0.7786,
+      "step": 1318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4215119542044858e-05,
+      "loss": 0.8189,
+      "step": 1319
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4206641450821474e-05,
+      "loss": 0.7651,
+      "step": 1320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4198159684252468e-05,
+      "loss": 0.8048,
+      "step": 1321
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4189674249748364e-05,
+      "loss": 0.7624,
+      "step": 1322
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4181185154722892e-05,
+      "loss": 0.7494,
+      "step": 1323
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4172692406592981e-05,
+      "loss": 0.8054,
+      "step": 1324
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4164196012778745e-05,
+      "loss": 0.7997,
+      "step": 1325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4155695980703502e-05,
+      "loss": 0.7763,
+      "step": 1326
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4147192317793721e-05,
+      "loss": 0.7531,
+      "step": 1327
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.413868503147907e-05,
+      "loss": 0.8119,
+      "step": 1328
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4130174129192367e-05,
+      "loss": 0.9078,
+      "step": 1329
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4121659618369591e-05,
+      "loss": 0.8169,
+      "step": 1330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411314150644988e-05,
+      "loss": 0.7401,
+      "step": 1331
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4104619800875514e-05,
+      "loss": 0.7767,
+      "step": 1332
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096094509091911e-05,
+      "loss": 0.7947,
+      "step": 1333
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4087565638547625e-05,
+      "loss": 0.7587,
+      "step": 1334
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4079033196694339e-05,
+      "loss": 0.7917,
+      "step": 1335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407049719098685e-05,
+      "loss": 0.7787,
+      "step": 1336
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4061957628883072e-05,
+      "loss": 0.8023,
+      "step": 1337
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4053414517844028e-05,
+      "loss": 0.783,
+      "step": 1338
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4044867865333843e-05,
+      "loss": 0.7319,
+      "step": 1339
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4036317678819729e-05,
+      "loss": 0.8148,
+      "step": 1340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4027763965771993e-05,
+      "loss": 0.7176,
+      "step": 1341
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4019206733664016e-05,
+      "loss": 0.6843,
+      "step": 1342
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4010645989972262e-05,
+      "loss": 0.8315,
+      "step": 1343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400208174217626e-05,
+      "loss": 0.7582,
+      "step": 1344
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3993513997758592e-05,
+      "loss": 0.756,
+      "step": 1345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398494276420491e-05,
+      "loss": 0.7068,
+      "step": 1346
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3976368049003903e-05,
+      "loss": 0.7649,
+      "step": 1347
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967789859647308e-05,
+      "loss": 0.7569,
+      "step": 1348
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3959208203629892e-05,
+      "loss": 0.7728,
+      "step": 1349
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.395062308844946e-05,
+      "loss": 0.84,
+      "step": 1350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3942034521606824e-05,
+      "loss": 0.8525,
+      "step": 1351
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393344251060583e-05,
+      "loss": 0.7777,
+      "step": 1352
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3924847062953318e-05,
+      "loss": 0.6963,
+      "step": 1353
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3916248186159143e-05,
+      "loss": 0.7493,
+      "step": 1354
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3907645887736138e-05,
+      "loss": 0.8636,
+      "step": 1355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3899040175200152e-05,
+      "loss": 0.823,
+      "step": 1356
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3890431056069991e-05,
+      "loss": 0.7835,
+      "step": 1357
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3881818537867455e-05,
+      "loss": 0.8629,
+      "step": 1358
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3873202628117306e-05,
+      "loss": 0.7672,
+      "step": 1359
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3864583334347263e-05,
+      "loss": 0.7402,
+      "step": 1360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3855960664088022e-05,
+      "loss": 0.7373,
+      "step": 1361
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3847334624873205e-05,
+      "loss": 0.7875,
+      "step": 1362
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3838705224239397e-05,
+      "loss": 0.7306,
+      "step": 1363
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3830072469726105e-05,
+      "loss": 0.8185,
+      "step": 1364
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3821436368875779e-05,
+      "loss": 0.752,
+      "step": 1365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3812796929233776e-05,
+      "loss": 0.8252,
+      "step": 1366
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3804154158348395e-05,
+      "loss": 0.8139,
+      "step": 1367
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3795508063770819e-05,
+      "loss": 0.7574,
+      "step": 1368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3786858653055154e-05,
+      "loss": 0.8715,
+      "step": 1369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377820593375839e-05,
+      "loss": 0.7727,
+      "step": 1370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3769549913440422e-05,
+      "loss": 0.7474,
+      "step": 1371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3760890599664012e-05,
+      "loss": 0.7791,
+      "step": 1372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3752227999994816e-05,
+      "loss": 0.8,
+      "step": 1373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374356212200135e-05,
+      "loss": 0.7002,
+      "step": 1374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3734892973254998e-05,
+      "loss": 0.7813,
+      "step": 1375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726220561330004e-05,
+      "loss": 0.7505,
+      "step": 1376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3717544893803458e-05,
+      "loss": 0.7861,
+      "step": 1377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3708865978255298e-05,
+      "loss": 0.8114,
+      "step": 1378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3700183822268296e-05,
+      "loss": 0.7801,
+      "step": 1379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369149843342806e-05,
+      "loss": 0.7435,
+      "step": 1380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3682809819323023e-05,
+      "loss": 0.6973,
+      "step": 1381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3674117987544429e-05,
+      "loss": 0.751,
+      "step": 1382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3665422945686339e-05,
+      "loss": 0.7301,
+      "step": 1383
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3656724701345616e-05,
+      "loss": 0.7557,
+      "step": 1384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3648023262121925e-05,
+      "loss": 0.7716,
+      "step": 1385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3639318635617722e-05,
+      "loss": 0.7415,
+      "step": 1386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3630610829438235e-05,
+      "loss": 0.6926,
+      "step": 1387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3621899851191494e-05,
+      "loss": 0.8679,
+      "step": 1388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3613185708488275e-05,
+      "loss": 0.8117,
+      "step": 1389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360446840894214e-05,
+      "loss": 0.819,
+      "step": 1390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3595747960169393e-05,
+      "loss": 0.7341,
+      "step": 1391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3587024369789097e-05,
+      "loss": 0.7815,
+      "step": 1392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3578297645423062e-05,
+      "loss": 0.8208,
+      "step": 1393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3569567794695835e-05,
+      "loss": 0.8194,
+      "step": 1394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3560834825234689e-05,
+      "loss": 0.9297,
+      "step": 1395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3552098744669624e-05,
+      "loss": 0.7633,
+      "step": 1396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3543359560633365e-05,
+      "loss": 0.7699,
+      "step": 1397
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3534617280761334e-05,
+      "loss": 0.6959,
+      "step": 1398
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3525871912691683e-05,
+      "loss": 0.7081,
+      "step": 1399
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3517123464065234e-05,
+      "loss": 0.7871,
+      "step": 1400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3508371942525516e-05,
+      "loss": 0.8405,
+      "step": 1401
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3499617355718739e-05,
+      "loss": 0.6722,
+      "step": 1402
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3490859711293794e-05,
+      "loss": 0.7857,
+      "step": 1403
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3482099016902234e-05,
+      "loss": 0.9143,
+      "step": 1404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3473335280198295e-05,
+      "loss": 0.7412,
+      "step": 1405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3464568508838852e-05,
+      "loss": 0.7462,
+      "step": 1406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345579871048344e-05,
+      "loss": 0.7946,
+      "step": 1407
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3447025892794243e-05,
+      "loss": 0.8215,
+      "step": 1408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3438250063436074e-05,
+      "loss": 0.8425,
+      "step": 1409
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3429471230076381e-05,
+      "loss": 0.7685,
+      "step": 1410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3420689400385239e-05,
+      "loss": 0.8245,
+      "step": 1411
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3411904582035336e-05,
+      "loss": 0.8602,
+      "step": 1412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3403116782701977e-05,
+      "loss": 0.7377,
+      "step": 1413
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3394326010063067e-05,
+      "loss": 0.8121,
+      "step": 1414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338553227179911e-05,
+      "loss": 0.6717,
+      "step": 1415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3376735575593202e-05,
+      "loss": 0.7157,
+      "step": 1416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3367935929131028e-05,
+      "loss": 0.7167,
+      "step": 1417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3359133340100836e-05,
+      "loss": 0.7618,
+      "step": 1418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3350327816193457e-05,
+      "loss": 0.7991,
+      "step": 1419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3341519365102286e-05,
+      "loss": 0.7647,
+      "step": 1420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3332707994523273e-05,
+      "loss": 0.6942,
+      "step": 1421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3323893712154918e-05,
+      "loss": 0.7785,
+      "step": 1422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3315076525698264e-05,
+      "loss": 0.7584,
+      "step": 1423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3306256442856897e-05,
+      "loss": 0.8412,
+      "step": 1424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3297433471336923e-05,
+      "loss": 0.8342,
+      "step": 1425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3288607618846987e-05,
+      "loss": 0.7112,
+      "step": 1426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3279778893098237e-05,
+      "loss": 0.7379,
+      "step": 1427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3270947301804336e-05,
+      "loss": 0.8264,
+      "step": 1428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3262112852681455e-05,
+      "loss": 0.8171,
+      "step": 1429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3253275553448257e-05,
+      "loss": 0.7805,
+      "step": 1430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3244435411825898e-05,
+      "loss": 0.7399,
+      "step": 1431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3235592435538013e-05,
+      "loss": 0.7417,
+      "step": 1432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3226746632310721e-05,
+      "loss": 0.7992,
+      "step": 1433
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3217898009872601e-05,
+      "loss": 0.8564,
+      "step": 1434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3209046575954707e-05,
+      "loss": 0.7545,
+      "step": 1435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3200192338290538e-05,
+      "loss": 0.7946,
+      "step": 1436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3191335304616049e-05,
+      "loss": 0.731,
+      "step": 1437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3182475482669639e-05,
+      "loss": 0.7602,
+      "step": 1438
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173612880192137e-05,
+      "loss": 0.8021,
+      "step": 1439
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316474750492681e-05,
+      "loss": 0.805,
+      "step": 1440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.315587936461934e-05,
+      "loss": 0.7291,
+      "step": 1441
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3147008467017824e-05,
+      "loss": 0.6976,
+      "step": 1442
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3138134819872775e-05,
+      "loss": 0.8156,
+      "step": 1443
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3129258430937104e-05,
+      "loss": 0.8367,
+      "step": 1444
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120379307966124e-05,
+      "loss": 0.7554,
+      "step": 1445
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3111497458717521e-05,
+      "loss": 0.8202,
+      "step": 1446
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3102612890951379e-05,
+      "loss": 0.713,
+      "step": 1447
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309372561243015e-05,
+      "loss": 0.7651,
+      "step": 1448
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3084835630918657e-05,
+      "loss": 0.7936,
+      "step": 1449
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3075942954184082e-05,
+      "loss": 0.8248,
+      "step": 1450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306704758999596e-05,
+      "loss": 0.7404,
+      "step": 1451
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3058149546126182e-05,
+      "loss": 0.7757,
+      "step": 1452
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304924883034897e-05,
+      "loss": 0.8285,
+      "step": 1453
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040345450440897e-05,
+      "loss": 0.8012,
+      "step": 1454
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3031439414180838e-05,
+      "loss": 0.7027,
+      "step": 1455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302253072935001e-05,
+      "loss": 0.6946,
+      "step": 1456
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3013619403731936e-05,
+      "loss": 0.7715,
+      "step": 1457
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3004705445112449e-05,
+      "loss": 0.7386,
+      "step": 1458
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299578886127968e-05,
+      "loss": 0.7747,
+      "step": 1459
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2986869660024052e-05,
+      "loss": 0.7618,
+      "step": 1460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977947849138283e-05,
+      "loss": 0.858,
+      "step": 1461
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969023436417363e-05,
+      "loss": 0.7662,
+      "step": 1462
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2960096429658557e-05,
+      "loss": 0.791,
+      "step": 1463
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2951166836661397e-05,
+      "loss": 0.8141,
+      "step": 1464
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2942234665227675e-05,
+      "loss": 0.8179,
+      "step": 1465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2933299923161437e-05,
+      "loss": 0.7459,
+      "step": 1466
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2924362618268971e-05,
+      "loss": 0.7383,
+      "step": 1467
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2915422758358811e-05,
+      "loss": 0.8274,
+      "step": 1468
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2906480351241713e-05,
+      "loss": 0.7816,
+      "step": 1469
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2897535404730667e-05,
+      "loss": 0.7164,
+      "step": 1470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2888587926640881e-05,
+      "loss": 0.8131,
+      "step": 1471
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2879637924789775e-05,
+      "loss": 0.8321,
+      "step": 1472
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2870685406996964e-05,
+      "loss": 0.8672,
+      "step": 1473
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2861730381084275e-05,
+      "loss": 0.8483,
+      "step": 1474
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2852772854875715e-05,
+      "loss": 0.7875,
+      "step": 1475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2843812836197488e-05,
+      "loss": 0.7876,
+      "step": 1476
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2834850332877968e-05,
+      "loss": 0.7812,
+      "step": 1477
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2825885352747695e-05,
+      "loss": 0.8027,
+      "step": 1478
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2816917903639386e-05,
+      "loss": 0.7018,
+      "step": 1479
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2807947993387898e-05,
+      "loss": 0.7844,
+      "step": 1480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2798975629830258e-05,
+      "loss": 0.6859,
+      "step": 1481
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2790000820805618e-05,
+      "loss": 0.8164,
+      "step": 1482
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2781023574155283e-05,
+      "loss": 0.7282,
+      "step": 1483
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2772043897722672e-05,
+      "loss": 0.772,
+      "step": 1484
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2763061799353342e-05,
+      "loss": 0.7392,
+      "step": 1485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2754077286894953e-05,
+      "loss": 0.8323,
+      "step": 1486
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274509036819728e-05,
+      "loss": 0.7349,
+      "step": 1487
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2736101051112207e-05,
+      "loss": 0.7957,
+      "step": 1488
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2727109343493698e-05,
+      "loss": 0.7676,
+      "step": 1489
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271811525319782e-05,
+      "loss": 0.7608,
+      "step": 1490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2709118788082713e-05,
+      "loss": 0.8144,
+      "step": 1491
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2700119956008597e-05,
+      "loss": 0.7724,
+      "step": 1492
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2691118764837758e-05,
+      "loss": 0.7854,
+      "step": 1493
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2682115222434541e-05,
+      "loss": 0.8115,
+      "step": 1494
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673109336665353e-05,
+      "loss": 0.7842,
+      "step": 1495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2664101115398633e-05,
+      "loss": 0.7823,
+      "step": 1496
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2655090566504877e-05,
+      "loss": 0.7721,
+      "step": 1497
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264607769785661e-05,
+      "loss": 0.8272,
+      "step": 1498
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2637062517328373e-05,
+      "loss": 0.7743,
+      "step": 1499
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262804503279675e-05,
+      "loss": 0.7728,
+      "step": 1500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261902525214031e-05,
+      "loss": 0.7533,
+      "step": 1501
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2610003183239643e-05,
+      "loss": 0.817,
+      "step": 1502
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2600978833977344e-05,
+      "loss": 0.8404,
+      "step": 1503
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2591952212237992e-05,
+      "loss": 0.6833,
+      "step": 1504
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258292332590815e-05,
+      "loss": 0.7348,
+      "step": 1505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2573892182876361e-05,
+      "loss": 0.7772,
+      "step": 1506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2564858791033143e-05,
+      "loss": 0.7873,
+      "step": 1507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2555823158270982e-05,
+      "loss": 0.8299,
+      "step": 1508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2546785292484307e-05,
+      "loss": 0.756,
+      "step": 1509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2537745201569516e-05,
+      "loss": 0.8592,
+      "step": 1510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2528702893424938e-05,
+      "loss": 0.8058,
+      "step": 1511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2519658375950845e-05,
+      "loss": 0.8619,
+      "step": 1512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251061165704944e-05,
+      "loss": 0.8485,
+      "step": 1513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2501562744624846e-05,
+      "loss": 0.73,
+      "step": 1514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2492511646583108e-05,
+      "loss": 0.7928,
+      "step": 1515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2483458370832171e-05,
+      "loss": 0.7595,
+      "step": 1516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247440292528189e-05,
+      "loss": 0.7286,
+      "step": 1517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2465345317844018e-05,
+      "loss": 0.7804,
+      "step": 1518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2456285556432188e-05,
+      "loss": 0.7871,
+      "step": 1519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2447223648961924e-05,
+      "loss": 0.6983,
+      "step": 1520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.243815960335062e-05,
+      "loss": 0.7067,
+      "step": 1521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2429093427517537e-05,
+      "loss": 0.7534,
+      "step": 1522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2420025129383803e-05,
+      "loss": 0.7853,
+      "step": 1523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2410954716872392e-05,
+      "loss": 0.7826,
+      "step": 1524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2401882197908133e-05,
+      "loss": 0.7776,
+      "step": 1525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392807580417689e-05,
+      "loss": 0.7762,
+      "step": 1526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2383730872329566e-05,
+      "loss": 0.7558,
+      "step": 1527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2374652081574086e-05,
+      "loss": 0.8306,
+      "step": 1528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2365571216083393e-05,
+      "loss": 0.766,
+      "step": 1529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2356488283791448e-05,
+      "loss": 0.8573,
+      "step": 1530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2347403292634016e-05,
+      "loss": 0.7869,
+      "step": 1531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2338316250548665e-05,
+      "loss": 0.7951,
+      "step": 1532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232922716547474e-05,
+      "loss": 0.714,
+      "step": 1533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2320136045353384e-05,
+      "loss": 0.7933,
+      "step": 1534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2311042898127519e-05,
+      "loss": 0.7976,
+      "step": 1535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2301947731741834e-05,
+      "loss": 0.7579,
+      "step": 1536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229285055414278e-05,
+      "loss": 0.6958,
+      "step": 1537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283751373278565e-05,
+      "loss": 0.8297,
+      "step": 1538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2274650197099154e-05,
+      "loss": 0.8065,
+      "step": 1539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.226554703355625e-05,
+      "loss": 0.749,
+      "step": 1540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2256441890603294e-05,
+      "loss": 0.7168,
+      "step": 1541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2247334776195457e-05,
+      "loss": 0.8486,
+      "step": 1542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2238225698289626e-05,
+      "loss": 0.6962,
+      "step": 1543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2229114664844415e-05,
+      "loss": 0.6891,
+      "step": 1544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2220001683820135e-05,
+      "loss": 0.7761,
+      "step": 1545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2210886763178808e-05,
+      "loss": 0.7655,
+      "step": 1546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2201769910884147e-05,
+      "loss": 0.7855,
+      "step": 1547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2192651134901549e-05,
+      "loss": 0.8439,
+      "step": 1548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2183530443198097e-05,
+      "loss": 0.7023,
+      "step": 1549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217440784374255e-05,
+      "loss": 0.7959,
+      "step": 1550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2165283344505319e-05,
+      "loss": 0.7362,
+      "step": 1551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2156156953458494e-05,
+      "loss": 0.7993,
+      "step": 1552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2147028678575808e-05,
+      "loss": 0.7223,
+      "step": 1553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2137898527832639e-05,
+      "loss": 0.7932,
+      "step": 1554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212876650920601e-05,
+      "loss": 0.7328,
+      "step": 1555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2119632630674565e-05,
+      "loss": 0.7918,
+      "step": 1556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2110496900218588e-05,
+      "loss": 0.6984,
+      "step": 1557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2101359325819968e-05,
+      "loss": 0.7557,
+      "step": 1558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2092219915462216e-05,
+      "loss": 0.8,
+      "step": 1559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2083078677130435e-05,
+      "loss": 0.7187,
+      "step": 1560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2073935618811335e-05,
+      "loss": 0.8099,
+      "step": 1561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2064790748493208e-05,
+      "loss": 0.7778,
+      "step": 1562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2055644074165941e-05,
+      "loss": 0.8012,
+      "step": 1563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2046495603820984e-05,
+      "loss": 0.6982,
+      "step": 1564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2037345345451368e-05,
+      "loss": 0.7952,
+      "step": 1565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2028193307051671e-05,
+      "loss": 0.8025,
+      "step": 1566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2019039496618044e-05,
+      "loss": 0.7774,
+      "step": 1567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2009883922148173e-05,
+      "loss": 0.8049,
+      "step": 1568
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2000726591641293e-05,
+      "loss": 0.8225,
+      "step": 1569
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1991567513098168e-05,
+      "loss": 0.7865,
+      "step": 1570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1982406694521091e-05,
+      "loss": 0.7685,
+      "step": 1571
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1973244143913879e-05,
+      "loss": 0.6765,
+      "step": 1572
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1964079869281855e-05,
+      "loss": 0.8349,
+      "step": 1573
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1954913878631855e-05,
+      "loss": 0.7286,
+      "step": 1574
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1945746179972208e-05,
+      "loss": 0.7269,
+      "step": 1575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1936576781312743e-05,
+      "loss": 0.7969,
+      "step": 1576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1927405690664769e-05,
+      "loss": 0.7943,
+      "step": 1577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1918232916041075e-05,
+      "loss": 0.6451,
+      "step": 1578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1909058465455917e-05,
+      "loss": 0.7801,
+      "step": 1579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1899882346925022e-05,
+      "loss": 0.7832,
+      "step": 1580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1890704568465572e-05,
+      "loss": 0.7389,
+      "step": 1581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1881525138096198e-05,
+      "loss": 0.8235,
+      "step": 1582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1872344063836972e-05,
+      "loss": 0.7782,
+      "step": 1583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1863161353709407e-05,
+      "loss": 0.7283,
+      "step": 1584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1853977015736445e-05,
+      "loss": 0.7256,
+      "step": 1585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1844791057942448e-05,
+      "loss": 0.7455,
+      "step": 1586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1835603488353194e-05,
+      "loss": 0.7451,
+      "step": 1587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1826414314995867e-05,
+      "loss": 0.7679,
+      "step": 1588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1817223545899052e-05,
+      "loss": 0.7794,
+      "step": 1589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1808031189092737e-05,
+      "loss": 0.8566,
+      "step": 1590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179883725260829e-05,
+      "loss": 0.7408,
+      "step": 1591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1789641744478455e-05,
+      "loss": 0.6656,
+      "step": 1592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1780444672737357e-05,
+      "loss": 0.8453,
+      "step": 1593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177124604542048e-05,
+      "loss": 0.7787,
+      "step": 1594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1762045870564683e-05,
+      "loss": 0.7964,
+      "step": 1595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1752844156208152e-05,
+      "loss": 0.786,
+      "step": 1596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1743640910390437e-05,
+      "loss": 0.7799,
+      "step": 1597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1734436141152419e-05,
+      "loss": 0.8012,
+      "step": 1598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1725229856536313e-05,
+      "loss": 0.6708,
+      "step": 1599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1716022064585655e-05,
+      "loss": 0.745,
+      "step": 1600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1706812773345301e-05,
+      "loss": 0.7832,
+      "step": 1601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1697601990861416e-05,
+      "loss": 0.7582,
+      "step": 1602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1688389725181464e-05,
+      "loss": 0.8301,
+      "step": 1603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167917598435421e-05,
+      "loss": 0.8135,
+      "step": 1604
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1669960776429706e-05,
+      "loss": 0.7408,
+      "step": 1605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1660744109459283e-05,
+      "loss": 0.7017,
+      "step": 1606
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1651525991495553e-05,
+      "loss": 0.8012,
+      "step": 1607
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1642306430592392e-05,
+      "loss": 0.7991,
+      "step": 1608
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1633085434804939e-05,
+      "loss": 0.8091,
+      "step": 1609
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1623863012189576e-05,
+      "loss": 0.8192,
+      "step": 1610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1614639170803947e-05,
+      "loss": 0.8165,
+      "step": 1611
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1605413918706927e-05,
+      "loss": 0.7421,
+      "step": 1612
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1596187263958626e-05,
+      "loss": 0.7741,
+      "step": 1613
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1586959214620384e-05,
+      "loss": 0.8059,
+      "step": 1614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1577729778754743e-05,
+      "loss": 0.7703,
+      "step": 1615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1568498964425472e-05,
+      "loss": 0.8185,
+      "step": 1616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1559266779697545e-05,
+      "loss": 0.7713,
+      "step": 1617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1550033232637126e-05,
+      "loss": 0.6998,
+      "step": 1618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154079833131157e-05,
+      "loss": 0.8164,
+      "step": 1619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1531562083789413e-05,
+      "loss": 0.784,
+      "step": 1620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152232449814038e-05,
+      "loss": 0.8096,
+      "step": 1621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1513085582435349e-05,
+      "loss": 0.8199,
+      "step": 1622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1503845344746372e-05,
+      "loss": 0.7349,
+      "step": 1623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1494603793146646e-05,
+      "loss": 0.7353,
+      "step": 1624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1485360935710528e-05,
+      "loss": 0.8026,
+      "step": 1625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1476116780513502e-05,
+      "loss": 0.7413,
+      "step": 1626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1466871335632197e-05,
+      "loss": 0.8841,
+      "step": 1627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1457624609144365e-05,
+      "loss": 0.8089,
+      "step": 1628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1448376609128875e-05,
+      "loss": 0.8695,
+      "step": 1629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1439127343665715e-05,
+      "loss": 0.7012,
+      "step": 1630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.142987682083597e-05,
+      "loss": 0.766,
+      "step": 1631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1420625048721832e-05,
+      "loss": 0.7319,
+      "step": 1632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1411372035406578e-05,
+      "loss": 0.7674,
+      "step": 1633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1402117788974575e-05,
+      "loss": 0.8065,
+      "step": 1634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1392862317511266e-05,
+      "loss": 0.8104,
+      "step": 1635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1383605629103158e-05,
+      "loss": 0.7343,
+      "step": 1636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1374347731837834e-05,
+      "loss": 0.7884,
+      "step": 1637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1365088633803913e-05,
+      "loss": 0.8585,
+      "step": 1638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1355828343091087e-05,
+      "loss": 0.7397,
+      "step": 1639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1346566867790073e-05,
+      "loss": 0.7482,
+      "step": 1640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1337304215992636e-05,
+      "loss": 0.6808,
+      "step": 1641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1328040395791549e-05,
+      "loss": 0.7678,
+      "step": 1642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1318775415280622e-05,
+      "loss": 0.7212,
+      "step": 1643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1309509282554682e-05,
+      "loss": 0.7017,
+      "step": 1644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1300242005709551e-05,
+      "loss": 0.7642,
+      "step": 1645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.129097359284206e-05,
+      "loss": 0.8317,
+      "step": 1646
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128170405205002e-05,
+      "loss": 0.7941,
+      "step": 1647
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127243339143224e-05,
+      "loss": 0.7732,
+      "step": 1648
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1263161619088498e-05,
+      "loss": 0.7547,
+      "step": 1649
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1253888743119558e-05,
+      "loss": 0.7774,
+      "step": 1650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1244614771627129e-05,
+      "loss": 0.8017,
+      "step": 1651
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1235339712713894e-05,
+      "loss": 0.7243,
+      "step": 1652
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.122606357448347e-05,
+      "loss": 0.7778,
+      "step": 1653
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121678636504043e-05,
+      "loss": 0.8233,
+      "step": 1654
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120750809249028e-05,
+      "loss": 0.773,
+      "step": 1655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1198228764939451e-05,
+      "loss": 0.7747,
+      "step": 1656
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11889483904953e-05,
+      "loss": 0.7969,
+      "step": 1657
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1179666977266095e-05,
+      "loss": 0.782,
+      "step": 1658
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1170384533361014e-05,
+      "loss": 0.7953,
+      "step": 1659
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1161101066890134e-05,
+      "loss": 0.7556,
+      "step": 1660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1151816585964428e-05,
+      "loss": 0.7987,
+      "step": 1661
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1142531098695752e-05,
+      "loss": 0.7796,
+      "step": 1662
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1133244613196843e-05,
+      "loss": 0.7534,
+      "step": 1663
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1123957137581313e-05,
+      "loss": 0.7763,
+      "step": 1664
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111466867996363e-05,
+      "loss": 0.7569,
+      "step": 1665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1105379248459132e-05,
+      "loss": 0.7801,
+      "step": 1666
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1096088851183997e-05,
+      "loss": 0.76,
+      "step": 1667
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086797496255256e-05,
+      "loss": 0.7068,
+      "step": 1668
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1077505191790772e-05,
+      "loss": 0.7698,
+      "step": 1669
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1068211945909233e-05,
+      "loss": 0.7122,
+      "step": 1670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1058917766730158e-05,
+      "loss": 0.7204,
+      "step": 1671
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.104962266237388e-05,
+      "loss": 0.7555,
+      "step": 1672
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1040326640961537e-05,
+      "loss": 0.773,
+      "step": 1673
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1031029710615067e-05,
+      "loss": 0.8256,
+      "step": 1674
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1021731879457207e-05,
+      "loss": 0.7441,
+      "step": 1675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1012433155611476e-05,
+      "loss": 0.7672,
+      "step": 1676
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1003133547202181e-05,
+      "loss": 0.7345,
+      "step": 1677
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0993833062354393e-05,
+      "loss": 0.8046,
+      "step": 1678
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.098453170919395e-05,
+      "loss": 0.6734,
+      "step": 1679
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0975229495847455e-05,
+      "loss": 0.7214,
+      "step": 1680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0965926430442256e-05,
+      "loss": 0.772,
+      "step": 1681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0956622521106448e-05,
+      "loss": 0.7798,
+      "step": 1682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0947317775968864e-05,
+      "loss": 0.8169,
+      "step": 1683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0938012203159067e-05,
+      "loss": 0.7422,
+      "step": 1684
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092870581080734e-05,
+      "loss": 0.8353,
+      "step": 1685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0919398607044688e-05,
+      "loss": 0.7216,
+      "step": 1686
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091009060000282e-05,
+      "loss": 0.7562,
+      "step": 1687
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0900781797814148e-05,
+      "loss": 0.7917,
+      "step": 1688
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089147220861178e-05,
+      "loss": 0.8565,
+      "step": 1689
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0882161840529508e-05,
+      "loss": 0.7963,
+      "step": 1690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0872850701701811e-05,
+      "loss": 0.8017,
+      "step": 1691
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0863538800263836e-05,
+      "loss": 0.7307,
+      "step": 1692
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0854226144351397e-05,
+      "loss": 0.6708,
+      "step": 1693
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0844912742100969e-05,
+      "loss": 0.743,
+      "step": 1694
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0835598601649677e-05,
+      "loss": 0.8235,
+      "step": 1695
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0826283731135296e-05,
+      "loss": 0.8176,
+      "step": 1696
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0816968138696232e-05,
+      "loss": 0.7594,
+      "step": 1697
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080765183247152e-05,
+      "loss": 0.7195,
+      "step": 1698
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0798334820600831e-05,
+      "loss": 0.7875,
+      "step": 1699
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0789017111224443e-05,
+      "loss": 0.7737,
+      "step": 1700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0779698712483245e-05,
+      "loss": 0.7212,
+      "step": 1701
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0770379632518726e-05,
+      "loss": 0.6972,
+      "step": 1702
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0761059879472975e-05,
+      "loss": 0.7517,
+      "step": 1703
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0751739461488667e-05,
+      "loss": 0.7903,
+      "step": 1704
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074241838670906e-05,
+      "loss": 0.7468,
+      "step": 1705
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0733096663277977e-05,
+      "loss": 0.6954,
+      "step": 1706
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723774299339822e-05,
+      "loss": 0.779,
+      "step": 1707
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0714451303039545e-05,
+      "loss": 0.8031,
+      "step": 1708
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0705127682522658e-05,
+      "loss": 0.8216,
+      "step": 1709
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0695803445935213e-05,
+      "loss": 0.8213,
+      "step": 1710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0686478601423804e-05,
+      "loss": 0.7548,
+      "step": 1711
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0677153157135553e-05,
+      "loss": 0.8582,
+      "step": 1712
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0667827121218108e-05,
+      "loss": 0.8555,
+      "step": 1713
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0658500501819633e-05,
+      "loss": 0.7728,
+      "step": 1714
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.06491733070888e-05,
+      "loss": 0.782,
+      "step": 1715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063984554517479e-05,
+      "loss": 0.7826,
+      "step": 1716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063051722422727e-05,
+      "loss": 0.8226,
+      "step": 1717
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0621188352396406e-05,
+      "loss": 0.7451,
+      "step": 1718
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0611858937832835e-05,
+      "loss": 0.8136,
+      "step": 1719
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0602528988687675e-05,
+      "loss": 0.8422,
+      "step": 1720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059319851311251e-05,
+      "loss": 0.7484,
+      "step": 1721
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0583867519259383e-05,
+      "loss": 0.6775,
+      "step": 1722
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.057453601528079e-05,
+      "loss": 0.7781,
+      "step": 1723
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0565204009329674e-05,
+      "loss": 0.7279,
+      "step": 1724
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055587150955941e-05,
+      "loss": 0.8335,
+      "step": 1725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0546538524123815e-05,
+      "loss": 0.7176,
+      "step": 1726
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0537205061177122e-05,
+      "loss": 0.838,
+      "step": 1727
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052787112887399e-05,
+      "loss": 0.734,
+      "step": 1728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0518536735369476e-05,
+      "loss": 0.7139,
+      "step": 1729
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0509201888819045e-05,
+      "loss": 0.8491,
+      "step": 1730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0499866597378561e-05,
+      "loss": 0.7843,
+      "step": 1731
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0490530869204281e-05,
+      "loss": 0.8204,
+      "step": 1732
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048119471245283e-05,
+      "loss": 0.7783,
+      "step": 1733
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0471858135281215e-05,
+      "loss": 0.851,
+      "step": 1734
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0462521145846811e-05,
+      "loss": 0.6796,
+      "step": 1735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0453183752307354e-05,
+      "loss": 0.81,
+      "step": 1736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0443845962820936e-05,
+      "loss": 0.7302,
+      "step": 1737
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043450778554598e-05,
+      "loss": 0.7019,
+      "step": 1738
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0425169228641264e-05,
+      "loss": 0.7219,
+      "step": 1739
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.041583030026589e-05,
+      "loss": 0.6651,
+      "step": 1740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040649100857929e-05,
+      "loss": 0.7981,
+      "step": 1741
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0397151361741204e-05,
+      "loss": 0.7162,
+      "step": 1742
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0387811367911694e-05,
+      "loss": 0.8142,
+      "step": 1743
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0378471035251112e-05,
+      "loss": 0.6959,
+      "step": 1744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0369130371920121e-05,
+      "loss": 0.7954,
+      "step": 1745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0359789386079662e-05,
+      "loss": 0.7818,
+      "step": 1746
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0350448085890958e-05,
+      "loss": 0.8068,
+      "step": 1747
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0341106479515517e-05,
+      "loss": 0.7785,
+      "step": 1748
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.03317645751151e-05,
+      "loss": 0.777,
+      "step": 1749
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0322422380851737e-05,
+      "loss": 0.7435,
+      "step": 1750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0313079904887712e-05,
+      "loss": 0.7964,
+      "step": 1751
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0303737155385552e-05,
+      "loss": 0.707,
+      "step": 1752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0294394140508022e-05,
+      "loss": 0.7046,
+      "step": 1753
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0285050868418123e-05,
+      "loss": 0.716,
+      "step": 1754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027570734727908e-05,
+      "loss": 0.7983,
+      "step": 1755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0266363585254331e-05,
+      "loss": 0.9082,
+      "step": 1756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0257019590507527e-05,
+      "loss": 0.7609,
+      "step": 1757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0247675371202521e-05,
+      "loss": 0.7793,
+      "step": 1758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023833093550337e-05,
+      "loss": 0.7251,
+      "step": 1759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0228986291574317e-05,
+      "loss": 0.7541,
+      "step": 1760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0219641447579772e-05,
+      "loss": 0.707,
+      "step": 1761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021029641168434e-05,
+      "loss": 0.7775,
+      "step": 1762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0200951192052781e-05,
+      "loss": 0.7958,
+      "step": 1763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0191605796850027e-05,
+      "loss": 0.8246,
+      "step": 1764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.018226023424115e-05,
+      "loss": 0.7646,
+      "step": 1765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0172914512391379e-05,
+      "loss": 0.7641,
+      "step": 1766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0163568639466073e-05,
+      "loss": 0.8103,
+      "step": 1767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0154222623630732e-05,
+      "loss": 0.8404,
+      "step": 1768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0144876473050973e-05,
+      "loss": 0.7618,
+      "step": 1769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0135530195892537e-05,
+      "loss": 0.7533,
+      "step": 1770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0126183800321274e-05,
+      "loss": 0.6961,
+      "step": 1771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0116837294503134e-05,
+      "loss": 0.6742,
+      "step": 1772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0107490686604169e-05,
+      "loss": 0.7622,
+      "step": 1773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0098143984790513e-05,
+      "loss": 0.7499,
+      "step": 1774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0088797197228389e-05,
+      "loss": 0.756,
+      "step": 1775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0079450332084089e-05,
+      "loss": 0.7567,
+      "step": 1776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0070103397523979e-05,
+      "loss": 0.7324,
+      "step": 1777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0060756401714479e-05,
+      "loss": 0.7495,
+      "step": 1778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005140935282207e-05,
+      "loss": 0.7412,
+      "step": 1779
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0042062259013269e-05,
+      "loss": 0.8401,
+      "step": 1780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0032715128454646e-05,
+      "loss": 0.7695,
+      "step": 1781
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002336796931279e-05,
+      "loss": 0.6942,
+      "step": 1782
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0014020789754327e-05,
+      "loss": 0.7146,
+      "step": 1783
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0004673597945886e-05,
+      "loss": 0.7334,
+      "step": 1784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.995326402054117e-06,
+      "loss": 0.708,
+      "step": 1785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.985979210245678e-06,
+      "loss": 0.8109,
+      "step": 1786
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.976632030687212e-06,
+      "loss": 0.7818,
+      "step": 1787
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.967284871545356e-06,
+      "loss": 0.7422,
+      "step": 1788
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.957937740986732e-06,
+      "loss": 0.6691,
+      "step": 1789
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.948590647177933e-06,
+      "loss": 0.7615,
+      "step": 1790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.939243598285524e-06,
+      "loss": 0.8014,
+      "step": 1791
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.929896602476025e-06,
+      "loss": 0.7414,
+      "step": 1792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.920549667915911e-06,
+      "loss": 0.7267,
+      "step": 1793
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.911202802771615e-06,
+      "loss": 0.7839,
+      "step": 1794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.901856015209488e-06,
+      "loss": 0.8008,
+      "step": 1795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.892509313395835e-06,
+      "loss": 0.8429,
+      "step": 1796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883162705496867e-06,
+      "loss": 0.74,
+      "step": 1797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.873816199678731e-06,
+      "loss": 0.7111,
+      "step": 1798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.864469804107464e-06,
+      "loss": 0.7908,
+      "step": 1799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85512352694903e-06,
+      "loss": 0.7138,
+      "step": 1800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.845777376369272e-06,
+      "loss": 0.8226,
+      "step": 1801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83643136053393e-06,
+      "loss": 0.7095,
+      "step": 1802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.827085487608628e-06,
+      "loss": 0.775,
+      "step": 1803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.817739765758853e-06,
+      "loss": 0.7362,
+      "step": 1804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.808394203149976e-06,
+      "loss": 0.7758,
+      "step": 1805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.799048807947222e-06,
+      "loss": 0.6979,
+      "step": 1806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789703588315663e-06,
+      "loss": 0.7756,
+      "step": 1807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.780358552420233e-06,
+      "loss": 0.8432,
+      "step": 1808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.771013708425688e-06,
+      "loss": 0.7351,
+      "step": 1809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76166906449663e-06,
+      "loss": 0.8066,
+      "step": 1810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75232462879748e-06,
+      "loss": 0.7587,
+      "step": 1811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.742980409492476e-06,
+      "loss": 0.7256,
+      "step": 1812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.733636414745674e-06,
+      "loss": 0.8456,
+      "step": 1813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.724292652720922e-06,
+      "loss": 0.782,
+      "step": 1814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.714949131581877e-06,
+      "loss": 0.7429,
+      "step": 1815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.705605859491981e-06,
+      "loss": 0.8351,
+      "step": 1816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.69626284461445e-06,
+      "loss": 0.6959,
+      "step": 1817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.686920095112291e-06,
+      "loss": 0.7311,
+      "step": 1818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.677577619148266e-06,
+      "loss": 0.787,
+      "step": 1819
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.668235424884902e-06,
+      "loss": 0.762,
+      "step": 1820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658893520484486e-06,
+      "loss": 0.7784,
+      "step": 1821
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.649551914109042e-06,
+      "loss": 0.7646,
+      "step": 1822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.640210613920341e-06,
+      "loss": 0.7365,
+      "step": 1823
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.630869628079882e-06,
+      "loss": 0.7529,
+      "step": 1824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.621528964748891e-06,
+      "loss": 0.6946,
+      "step": 1825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.612188632088311e-06,
+      "loss": 0.81,
+      "step": 1826
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.602848638258798e-06,
+      "loss": 0.7378,
+      "step": 1827
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.593508991420715e-06,
+      "loss": 0.6714,
+      "step": 1828
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.584169699734113e-06,
+      "loss": 0.7616,
+      "step": 1829
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.574830771358741e-06,
+      "loss": 0.8139,
+      "step": 1830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.565492214454022e-06,
+      "loss": 0.7369,
+      "step": 1831
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.556154037179069e-06,
+      "loss": 0.7878,
+      "step": 1832
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54681624769265e-06,
+      "loss": 0.6652,
+      "step": 1833
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53747885415319e-06,
+      "loss": 0.778,
+      "step": 1834
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.52814186471879e-06,
+      "loss": 0.8166,
+      "step": 1835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.518805287547175e-06,
+      "loss": 0.6591,
+      "step": 1836
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.509469130795724e-06,
+      "loss": 0.8181,
+      "step": 1837
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.500133402621442e-06,
+      "loss": 0.7789,
+      "step": 1838
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.490798111180957e-06,
+      "loss": 0.7631,
+      "step": 1839
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48146326463053e-06,
+      "loss": 0.7842,
+      "step": 1840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.472128871126014e-06,
+      "loss": 0.8203,
+      "step": 1841
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.462794938822878e-06,
+      "loss": 0.8108,
+      "step": 1842
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.453461475876188e-06,
+      "loss": 0.7643,
+      "step": 1843
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44412849044059e-06,
+      "loss": 0.8104,
+      "step": 1844
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43479599067033e-06,
+      "loss": 0.7747,
+      "step": 1845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.425463984719212e-06,
+      "loss": 0.6947,
+      "step": 1846
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.416132480740619e-06,
+      "loss": 0.7257,
+      "step": 1847
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.406801486887493e-06,
+      "loss": 0.7902,
+      "step": 1848
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397471011312325e-06,
+      "loss": 0.7704,
+      "step": 1849
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388141062167168e-06,
+      "loss": 0.6797,
+      "step": 1850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.378811647603598e-06,
+      "loss": 0.7254,
+      "step": 1851
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.369482775772734e-06,
+      "loss": 0.7162,
+      "step": 1852
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.360154454825215e-06,
+      "loss": 0.7273,
+      "step": 1853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.350826692911201e-06,
+      "loss": 0.7175,
+      "step": 1854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34149949818037e-06,
+      "loss": 0.7777,
+      "step": 1855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.332172878781896e-06,
+      "loss": 0.7814,
+      "step": 1856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.322846842864452e-06,
+      "loss": 0.7356,
+      "step": 1857
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.313521398576199e-06,
+      "loss": 0.7581,
+      "step": 1858
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.304196554064787e-06,
+      "loss": 0.6901,
+      "step": 1859
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.294872317477345e-06,
+      "loss": 0.7854,
+      "step": 1860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.285548696960458e-06,
+      "loss": 0.7651,
+      "step": 1861
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.276225700660185e-06,
+      "loss": 0.7256,
+      "step": 1862
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.266903336722025e-06,
+      "loss": 0.7512,
+      "step": 1863
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.257581613290943e-06,
+      "loss": 0.7464,
+      "step": 1864
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.248260538511337e-06,
+      "loss": 0.7861,
+      "step": 1865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.238940120527027e-06,
+      "loss": 0.7845,
+      "step": 1866
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.229620367481279e-06,
+      "loss": 0.7687,
+      "step": 1867
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.220301287516757e-06,
+      "loss": 0.7351,
+      "step": 1868
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.210982888775559e-06,
+      "loss": 0.6627,
+      "step": 1869
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.201665179399172e-06,
+      "loss": 0.6819,
+      "step": 1870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.192348167528482e-06,
+      "loss": 0.7716,
+      "step": 1871
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.183031861303773e-06,
+      "loss": 0.6913,
+      "step": 1872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.173716268864706e-06,
+      "loss": 0.7564,
+      "step": 1873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.164401398350326e-06,
+      "loss": 0.8416,
+      "step": 1874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.155087257899034e-06,
+      "loss": 0.7331,
+      "step": 1875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.145773855648605e-06,
+      "loss": 0.8182,
+      "step": 1876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.136461199736167e-06,
+      "loss": 0.7148,
+      "step": 1877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12714929829819e-06,
+      "loss": 0.7764,
+      "step": 1878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.117838159470495e-06,
+      "loss": 0.7115,
+      "step": 1879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.108527791388224e-06,
+      "loss": 0.8359,
+      "step": 1880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.099218202185852e-06,
+      "loss": 0.7158,
+      "step": 1881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.089909399997183e-06,
+      "loss": 0.6849,
+      "step": 1882
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.080601392955313e-06,
+      "loss": 0.7076,
+      "step": 1883
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.071294189192663e-06,
+      "loss": 0.7922,
+      "step": 1884
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061987796840936e-06,
+      "loss": 0.8058,
+      "step": 1885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.052682224031136e-06,
+      "loss": 0.7541,
+      "step": 1886
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.043377478893554e-06,
+      "loss": 0.8065,
+      "step": 1887
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.034073569557746e-06,
+      "loss": 0.7827,
+      "step": 1888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02477050415255e-06,
+      "loss": 0.7937,
+      "step": 1889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.015468290806051e-06,
+      "loss": 0.6625,
+      "step": 1890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00616693764561e-06,
+      "loss": 0.7681,
+      "step": 1891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.99686645279782e-06,
+      "loss": 0.6699,
+      "step": 1892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.987566844388524e-06,
+      "loss": 0.7486,
+      "step": 1893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.978268120542798e-06,
+      "loss": 0.7092,
+      "step": 1894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.968970289384935e-06,
+      "loss": 0.7476,
+      "step": 1895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.959673359038468e-06,
+      "loss": 0.7208,
+      "step": 1896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.950377337626124e-06,
+      "loss": 0.8028,
+      "step": 1897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.941082233269842e-06,
+      "loss": 0.7698,
+      "step": 1898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93178805409077e-06,
+      "loss": 0.7544,
+      "step": 1899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.922494808209231e-06,
+      "loss": 0.7287,
+      "step": 1900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913202503744748e-06,
+      "loss": 0.7777,
+      "step": 1901
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.903911148816006e-06,
+      "loss": 0.7729,
+      "step": 1902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89462075154087e-06,
+      "loss": 0.717,
+      "step": 1903
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.885331320036374e-06,
+      "loss": 0.8442,
+      "step": 1904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.876042862418688e-06,
+      "loss": 0.7172,
+      "step": 1905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.86675538680316e-06,
+      "loss": 0.7814,
+      "step": 1906
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.857468901304251e-06,
+      "loss": 0.8169,
+      "step": 1907
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.848183414035574e-06,
+      "loss": 0.6696,
+      "step": 1908
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.838898933109869e-06,
+      "loss": 0.7605,
+      "step": 1909
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82961546663899e-06,
+      "loss": 0.7544,
+      "step": 1910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.820333022733911e-06,
+      "loss": 0.8522,
+      "step": 1911
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811051609504704e-06,
+      "loss": 0.7379,
+      "step": 1912
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80177123506055e-06,
+      "loss": 0.737,
+      "step": 1913
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.792491907509722e-06,
+      "loss": 0.7419,
+      "step": 1914
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.783213634959573e-06,
+      "loss": 0.7713,
+      "step": 1915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.773936425516536e-06,
+      "loss": 0.7701,
+      "step": 1916
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.764660287286111e-06,
+      "loss": 0.7586,
+      "step": 1917
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.755385228372871e-06,
+      "loss": 0.7844,
+      "step": 1918
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.746111256880445e-06,
+      "loss": 0.7872,
+      "step": 1919
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.736838380911502e-06,
+      "loss": 0.7761,
+      "step": 1920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.727566608567766e-06,
+      "loss": 0.7161,
+      "step": 1921
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718295947949983e-06,
+      "loss": 0.7368,
+      "step": 1922
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.709026407157945e-06,
+      "loss": 0.7589,
+      "step": 1923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69975799429045e-06,
+      "loss": 0.7928,
+      "step": 1924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.690490717445318e-06,
+      "loss": 0.7713,
+      "step": 1925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68122458471938e-06,
+      "loss": 0.7564,
+      "step": 1926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.671959604208455e-06,
+      "loss": 0.7904,
+      "step": 1927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.662695784007371e-06,
+      "loss": 0.769,
+      "step": 1928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65343313220993e-06,
+      "loss": 0.7329,
+      "step": 1929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.644171656908914e-06,
+      "loss": 0.784,
+      "step": 1930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63491136619609e-06,
+      "loss": 0.7336,
+      "step": 1931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.625652268162169e-06,
+      "loss": 0.6907,
+      "step": 1932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.616394370896843e-06,
+      "loss": 0.7867,
+      "step": 1933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.607137682488736e-06,
+      "loss": 0.8516,
+      "step": 1934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.597882211025423e-06,
+      "loss": 0.718,
+      "step": 1935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.588627964593424e-06,
+      "loss": 0.7836,
+      "step": 1936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.579374951278171e-06,
+      "loss": 0.7511,
+      "step": 1937
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.570123179164034e-06,
+      "loss": 0.703,
+      "step": 1938
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.560872656334289e-06,
+      "loss": 0.7548,
+      "step": 1939
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.551623390871125e-06,
+      "loss": 0.8205,
+      "step": 1940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.542375390855638e-06,
+      "loss": 0.8571,
+      "step": 1941
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.533128664367807e-06,
+      "loss": 0.7665,
+      "step": 1942
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.523883219486504e-06,
+      "loss": 0.7628,
+      "step": 1943
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.514639064289477e-06,
+      "loss": 0.737,
+      "step": 1944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.505396206853358e-06,
+      "loss": 0.8251,
+      "step": 1945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.496154655253632e-06,
+      "loss": 0.7264,
+      "step": 1946
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.486914417564653e-06,
+      "loss": 0.8052,
+      "step": 1947
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.477675501859625e-06,
+      "loss": 0.8206,
+      "step": 1948
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.468437916210588e-06,
+      "loss": 0.8107,
+      "step": 1949
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.459201668688437e-06,
+      "loss": 0.7411,
+      "step": 1950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.449966767362877e-06,
+      "loss": 0.6971,
+      "step": 1951
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.440733220302455e-06,
+      "loss": 0.6729,
+      "step": 1952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.431501035574532e-06,
+      "loss": 0.8064,
+      "step": 1953
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42227022124526e-06,
+      "loss": 0.7884,
+      "step": 1954
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.413040785379623e-06,
+      "loss": 0.7175,
+      "step": 1955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.403812736041375e-06,
+      "loss": 0.7232,
+      "step": 1956
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.394586081293073e-06,
+      "loss": 0.6831,
+      "step": 1957
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.385360829196056e-06,
+      "loss": 0.7621,
+      "step": 1958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.376136987810426e-06,
+      "loss": 0.7553,
+      "step": 1959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.366914565195066e-06,
+      "loss": 0.7278,
+      "step": 1960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35769356940761e-06,
+      "loss": 0.7373,
+      "step": 1961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.348474008504447e-06,
+      "loss": 0.7079,
+      "step": 1962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33925589054072e-06,
+      "loss": 0.7769,
+      "step": 1963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.330039223570296e-06,
+      "loss": 0.7525,
+      "step": 1964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.320824015645794e-06,
+      "loss": 0.7126,
+      "step": 1965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31161027481854e-06,
+      "loss": 0.709,
+      "step": 1966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.302398009138586e-06,
+      "loss": 0.7708,
+      "step": 1967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2931872266547e-06,
+      "loss": 0.7435,
+      "step": 1968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.283977935414346e-06,
+      "loss": 0.7455,
+      "step": 1969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.274770143463692e-06,
+      "loss": 0.7532,
+      "step": 1970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.265563858847584e-06,
+      "loss": 0.7618,
+      "step": 1971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25635908960957e-06,
+      "loss": 0.7154,
+      "step": 1972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247155843791852e-06,
+      "loss": 0.7019,
+      "step": 1973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.237954129435322e-06,
+      "loss": 0.7615,
+      "step": 1974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22875395457952e-06,
+      "loss": 0.7057,
+      "step": 1975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.219555327262647e-06,
+      "loss": 0.7894,
+      "step": 1976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21035825552155e-06,
+      "loss": 0.7471,
+      "step": 1977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.201162747391713e-06,
+      "loss": 0.7637,
+      "step": 1978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.191968810907264e-06,
+      "loss": 0.7057,
+      "step": 1979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.182776454100951e-06,
+      "loss": 0.7505,
+      "step": 1980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.173585685004137e-06,
+      "loss": 0.7128,
+      "step": 1981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.164396511646811e-06,
+      "loss": 0.7513,
+      "step": 1982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.155208942057555e-06,
+      "loss": 0.7622,
+      "step": 1983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.146022984263557e-06,
+      "loss": 0.7893,
+      "step": 1984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.136838646290596e-06,
+      "loss": 0.7436,
+      "step": 1985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12765593616303e-06,
+      "loss": 0.7164,
+      "step": 1986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.118474861903807e-06,
+      "loss": 0.7321,
+      "step": 1987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.109295431534432e-06,
+      "loss": 0.7142,
+      "step": 1988
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10011765307498e-06,
+      "loss": 0.8018,
+      "step": 1989
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.090941534544087e-06,
+      "loss": 0.8212,
+      "step": 1990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.081767083958926e-06,
+      "loss": 0.8006,
+      "step": 1991
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.072594309335233e-06,
+      "loss": 0.7441,
+      "step": 1992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.063423218687259e-06,
+      "loss": 0.7526,
+      "step": 1993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.054253820027797e-06,
+      "loss": 0.7616,
+      "step": 1994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045086121368148e-06,
+      "loss": 0.743,
+      "step": 1995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.035920130718146e-06,
+      "loss": 0.7479,
+      "step": 1996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.026755856086123e-06,
+      "loss": 0.7194,
+      "step": 1997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01759330547891e-06,
+      "loss": 0.7308,
+      "step": 1998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.008432486901837e-06,
+      "loss": 0.8014,
+      "step": 1999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999273408358708e-06,
+      "loss": 0.7294,
+      "step": 2000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.990116077851829e-06,
+      "loss": 0.7277,
+      "step": 2001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.980960503381961e-06,
+      "loss": 0.7283,
+      "step": 2002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97180669294833e-06,
+      "loss": 0.7251,
+      "step": 2003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.962654654548639e-06,
+      "loss": 0.8354,
+      "step": 2004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.953504396179017e-06,
+      "loss": 0.8029,
+      "step": 2005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.94435592583406e-06,
+      "loss": 0.7277,
+      "step": 2006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.935209251506794e-06,
+      "loss": 0.7635,
+      "step": 2007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.926064381188669e-06,
+      "loss": 0.763,
+      "step": 2008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91692132286957e-06,
+      "loss": 0.727,
+      "step": 2009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.907780084537787e-06,
+      "loss": 0.7248,
+      "step": 2010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.898640674180032e-06,
+      "loss": 0.7582,
+      "step": 2011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.889503099781416e-06,
+      "loss": 0.7418,
+      "step": 2012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.880367369325436e-06,
+      "loss": 0.7356,
+      "step": 2013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.871233490793995e-06,
+      "loss": 0.7746,
+      "step": 2014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.862101472167364e-06,
+      "loss": 0.7647,
+      "step": 2015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.852971321424197e-06,
+      "loss": 0.7943,
+      "step": 2016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84384304654151e-06,
+      "loss": 0.7431,
+      "step": 2017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.834716655494681e-06,
+      "loss": 0.7434,
+      "step": 2018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.825592156257455e-06,
+      "loss": 0.7163,
+      "step": 2019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.816469556801904e-06,
+      "loss": 0.7501,
+      "step": 2020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.807348865098456e-06,
+      "loss": 0.7879,
+      "step": 2021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.798230089115855e-06,
+      "loss": 0.7828,
+      "step": 2022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789113236821191e-06,
+      "loss": 0.742,
+      "step": 2023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.779998316179867e-06,
+      "loss": 0.7747,
+      "step": 2024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.770885335155588e-06,
+      "loss": 0.8158,
+      "step": 2025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76177430171038e-06,
+      "loss": 0.7373,
+      "step": 2026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.752665223804547e-06,
+      "loss": 0.7071,
+      "step": 2027
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.743558109396706e-06,
+      "loss": 0.7314,
+      "step": 2028
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.734452966443751e-06,
+      "loss": 0.721,
+      "step": 2029
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.725349802900848e-06,
+      "loss": 0.7408,
+      "step": 2030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71624862672144e-06,
+      "loss": 0.8019,
+      "step": 2031
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707149445857225e-06,
+      "loss": 0.7919,
+      "step": 2032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.698052268258169e-06,
+      "loss": 0.6732,
+      "step": 2033
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688957101872484e-06,
+      "loss": 0.6278,
+      "step": 2034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.679863954646618e-06,
+      "loss": 0.8055,
+      "step": 2035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670772834525265e-06,
+      "loss": 0.7422,
+      "step": 2036
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.661683749451338e-06,
+      "loss": 0.6889,
+      "step": 2037
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.652596707365982e-06,
+      "loss": 0.7094,
+      "step": 2038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643511716208553e-06,
+      "loss": 0.7448,
+      "step": 2039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.634428783916608e-06,
+      "loss": 0.681,
+      "step": 2040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.625347918425918e-06,
+      "loss": 0.7574,
+      "step": 2041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.616269127670435e-06,
+      "loss": 0.7223,
+      "step": 2042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607192419582312e-06,
+      "loss": 0.7588,
+      "step": 2043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.598117802091871e-06,
+      "loss": 0.7403,
+      "step": 2044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5890452831276095e-06,
+      "loss": 0.7757,
+      "step": 2045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.579974870616201e-06,
+      "loss": 0.7684,
+      "step": 2046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.570906572482466e-06,
+      "loss": 0.7168,
+      "step": 2047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.561840396649386e-06,
+      "loss": 0.7678,
+      "step": 2048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.552776351038078e-06,
+      "loss": 0.7803,
+      "step": 2049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.543714443567812e-06,
+      "loss": 0.6961,
+      "step": 2050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.534654682155985e-06,
+      "loss": 0.8046,
+      "step": 2051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.525597074718112e-06,
+      "loss": 0.7135,
+      "step": 2052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.516541629167835e-06,
+      "loss": 0.756,
+      "step": 2053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.507488353416896e-06,
+      "loss": 0.747,
+      "step": 2054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.498437255375153e-06,
+      "loss": 0.8098,
+      "step": 2055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.489388342950562e-06,
+      "loss": 0.7115,
+      "step": 2056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.480341624049156e-06,
+      "loss": 0.7792,
+      "step": 2057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.471297106575067e-06,
+      "loss": 0.749,
+      "step": 2058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.462254798430487e-06,
+      "loss": 0.745,
+      "step": 2059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.453214707515695e-06,
+      "loss": 0.7848,
+      "step": 2060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.444176841729022e-06,
+      "loss": 0.7673,
+      "step": 2061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.435141208966858e-06,
+      "loss": 0.6737,
+      "step": 2062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.426107817123644e-06,
+      "loss": 0.7446,
+      "step": 2063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.417076674091854e-06,
+      "loss": 0.754,
+      "step": 2064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408047787762013e-06,
+      "loss": 0.7724,
+      "step": 2065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3990211660226595e-06,
+      "loss": 0.7277,
+      "step": 2066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3899968167603585e-06,
+      "loss": 0.8112,
+      "step": 2067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.380974747859697e-06,
+      "loss": 0.7333,
+      "step": 2068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3719549672032546e-06,
+      "loss": 0.7791,
+      "step": 2069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.362937482671628e-06,
+      "loss": 0.8069,
+      "step": 2070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.353922302143395e-06,
+      "loss": 0.7944,
+      "step": 2071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.344909433495124e-06,
+      "loss": 0.7191,
+      "step": 2072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3358988846013715e-06,
+      "loss": 0.7069,
+      "step": 2073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3268906633346515e-06,
+      "loss": 0.7013,
+      "step": 2074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.317884777565462e-06,
+      "loss": 0.805,
+      "step": 2075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.308881235162246e-06,
+      "loss": 0.7591,
+      "step": 2076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2998800439914045e-06,
+      "loss": 0.7271,
+      "step": 2077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29088121191729e-06,
+      "loss": 0.7038,
+      "step": 2078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2818847468021834e-06,
+      "loss": 0.6669,
+      "step": 2079
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2728906565063074e-06,
+      "loss": 0.7719,
+      "step": 2080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.263898948887797e-06,
+      "loss": 0.7212,
+      "step": 2081
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25490963180272e-06,
+      "loss": 0.8204,
+      "step": 2082
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24592271310505e-06,
+      "loss": 0.7275,
+      "step": 2083
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236938200646661e-06,
+      "loss": 0.7116,
+      "step": 2084
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.227956102277332e-06,
+      "loss": 0.7542,
+      "step": 2085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.218976425844719e-06,
+      "loss": 0.6928,
+      "step": 2086
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.209999179194381e-06,
+      "loss": 0.8024,
+      "step": 2087
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.201024370169745e-06,
+      "loss": 0.6953,
+      "step": 2088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.192052006612102e-06,
+      "loss": 0.8396,
+      "step": 2089
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.183082096360621e-06,
+      "loss": 0.7726,
+      "step": 2090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.174114647252306e-06,
+      "loss": 0.7981,
+      "step": 2091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.165149667122036e-06,
+      "loss": 0.7881,
+      "step": 2092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.156187163802515e-06,
+      "loss": 0.7729,
+      "step": 2093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.147227145124287e-06,
+      "loss": 0.7642,
+      "step": 2094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.138269618915732e-06,
+      "loss": 0.7925,
+      "step": 2095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1293145930030395e-06,
+      "loss": 0.7317,
+      "step": 2096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.120362075210232e-06,
+      "loss": 0.769,
+      "step": 2097
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.111412073359121e-06,
+      "loss": 0.7268,
+      "step": 2098
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1024645952693325e-06,
+      "loss": 0.6839,
+      "step": 2099
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.09351964875829e-06,
+      "loss": 0.8888,
+      "step": 2100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.084577241641191e-06,
+      "loss": 0.7774,
+      "step": 2101
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0756373817310306e-06,
+      "loss": 0.7167,
+      "step": 2102
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.066700076838566e-06,
+      "loss": 0.8067,
+      "step": 2103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.057765334772325e-06,
+      "loss": 0.6828,
+      "step": 2104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.048833163338607e-06,
+      "loss": 0.7589,
+      "step": 2105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.039903570341447e-06,
+      "loss": 0.7174,
+      "step": 2106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.03097656358264e-06,
+      "loss": 0.8082,
+      "step": 2107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022052150861719e-06,
+      "loss": 0.6995,
+      "step": 2108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.013130339975948e-06,
+      "loss": 0.7412,
+      "step": 2109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.004211138720325e-06,
+      "loss": 0.7908,
+      "step": 2110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.995294554887555e-06,
+      "loss": 0.7027,
+      "step": 2111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.98638059626807e-06,
+      "loss": 0.7532,
+      "step": 2112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.977469270649995e-06,
+      "loss": 0.7251,
+      "step": 2113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.968560585819168e-06,
+      "loss": 0.8098,
+      "step": 2114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.959654549559108e-06,
+      "loss": 0.7217,
+      "step": 2115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9507511696510285e-06,
+      "loss": 0.742,
+      "step": 2116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.941850453873822e-06,
+      "loss": 0.7358,
+      "step": 2117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9329524100040415e-06,
+      "loss": 0.7294,
+      "step": 2118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9240570458159225e-06,
+      "loss": 0.76,
+      "step": 2119
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.915164369081344e-06,
+      "loss": 0.7251,
+      "step": 2120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906274387569851e-06,
+      "loss": 0.7521,
+      "step": 2121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.897387109048625e-06,
+      "loss": 0.6683,
+      "step": 2122
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.88850254128248e-06,
+      "loss": 0.7701,
+      "step": 2123
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.879620692033882e-06,
+      "loss": 0.8026,
+      "step": 2124
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.870741569062898e-06,
+      "loss": 0.7034,
+      "step": 2125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.861865180127226e-06,
+      "loss": 0.692,
+      "step": 2126
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.852991532982181e-06,
+      "loss": 0.7298,
+      "step": 2127
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.844120635380664e-06,
+      "loss": 0.7823,
+      "step": 2128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.835252495073194e-06,
+      "loss": 0.7536,
+      "step": 2129
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.826387119807864e-06,
+      "loss": 0.7903,
+      "step": 2130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.817524517330363e-06,
+      "loss": 0.7221,
+      "step": 2131
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.808664695383953e-06,
+      "loss": 0.7901,
+      "step": 2132
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.799807661709463e-06,
+      "loss": 0.7259,
+      "step": 2133
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.790953424045297e-06,
+      "loss": 0.744,
+      "step": 2134
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.782101990127401e-06,
+      "loss": 0.6742,
+      "step": 2135
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.773253367689281e-06,
+      "loss": 0.6772,
+      "step": 2136
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.76440756446199e-06,
+      "loss": 0.6522,
+      "step": 2137
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.755564588174106e-06,
+      "loss": 0.6854,
+      "step": 2138
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.746724446551748e-06,
+      "loss": 0.7307,
+      "step": 2139
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.737887147318548e-06,
+      "loss": 0.8371,
+      "step": 2140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.729052698195669e-06,
+      "loss": 0.8031,
+      "step": 2141
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.720221106901768e-06,
+      "loss": 0.7329,
+      "step": 2142
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7113923811530165e-06,
+      "loss": 0.647,
+      "step": 2143
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.70256652866308e-06,
+      "loss": 0.7371,
+      "step": 2144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6937435571431065e-06,
+      "loss": 0.7395,
+      "step": 2145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.684923474301739e-06,
+      "loss": 0.7336,
+      "step": 2146
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.676106287845083e-06,
+      "loss": 0.7819,
+      "step": 2147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.667292005476729e-06,
+      "loss": 0.7376,
+      "step": 2148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.658480634897717e-06,
+      "loss": 0.6525,
+      "step": 2149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6496721838065435e-06,
+      "loss": 0.7676,
+      "step": 2150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.640866659899169e-06,
+      "loss": 0.7621,
+      "step": 2151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.632064070868975e-06,
+      "loss": 0.7836,
+      "step": 2152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.623264424406796e-06,
+      "loss": 0.8002,
+      "step": 2153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.614467728200893e-06,
+      "loss": 0.7166,
+      "step": 2154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.605673989936934e-06,
+      "loss": 0.7376,
+      "step": 2155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596883217298027e-06,
+      "loss": 0.7174,
+      "step": 2156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.588095417964668e-06,
+      "loss": 0.7711,
+      "step": 2157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.579310599614764e-06,
+      "loss": 0.7928,
+      "step": 2158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.570528769923623e-06,
+      "loss": 0.7726,
+      "step": 2159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.561749936563928e-06,
+      "loss": 0.8536,
+      "step": 2160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.55297410720576e-06,
+      "loss": 0.6487,
+      "step": 2161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544201289516562e-06,
+      "loss": 0.7708,
+      "step": 2162
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.535431491161152e-06,
+      "loss": 0.7917,
+      "step": 2163
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.526664719801708e-06,
+      "loss": 0.7092,
+      "step": 2164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.517900983097764e-06,
+      "loss": 0.6749,
+      "step": 2165
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.509140288706211e-06,
+      "loss": 0.7298,
+      "step": 2166
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.500382644281265e-06,
+      "loss": 0.7369,
+      "step": 2167
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.491628057474489e-06,
+      "loss": 0.7785,
+      "step": 2168
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.482876535934768e-06,
+      "loss": 0.7412,
+      "step": 2169
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.474128087308319e-06,
+      "loss": 0.7896,
+      "step": 2170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.465382719238666e-06,
+      "loss": 0.7863,
+      "step": 2171
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.456640439366639e-06,
+      "loss": 0.6909,
+      "step": 2172
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.447901255330381e-06,
+      "loss": 0.7652,
+      "step": 2173
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.439165174765315e-06,
+      "loss": 0.7532,
+      "step": 2174
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.430432205304168e-06,
+      "loss": 0.7453,
+      "step": 2175
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.421702354576941e-06,
+      "loss": 0.7338,
+      "step": 2176
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.412975630210905e-06,
+      "loss": 0.8353,
+      "step": 2177
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4042520398306115e-06,
+      "loss": 0.7616,
+      "step": 2178
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.395531591057863e-06,
+      "loss": 0.6424,
+      "step": 2179
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.386814291511726e-06,
+      "loss": 0.7197,
+      "step": 2180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.37810014880851e-06,
+      "loss": 0.7957,
+      "step": 2181
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.369389170561764e-06,
+      "loss": 0.71,
+      "step": 2182
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.360681364382282e-06,
+      "loss": 0.7735,
+      "step": 2183
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.351976737878076e-06,
+      "loss": 0.8117,
+      "step": 2184
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.343275298654384e-06,
+      "loss": 0.6896,
+      "step": 2185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.334577054313664e-06,
+      "loss": 0.6595,
+      "step": 2186
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.325882012455573e-06,
+      "loss": 0.7026,
+      "step": 2187
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.31719018067698e-06,
+      "loss": 0.7495,
+      "step": 2188
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.308501566571942e-06,
+      "loss": 0.7135,
+      "step": 2189
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.299816177731708e-06,
+      "loss": 0.8079,
+      "step": 2190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.291134021744706e-06,
+      "loss": 0.7503,
+      "step": 2191
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.282455106196543e-06,
+      "loss": 0.7442,
+      "step": 2192
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273779438669998e-06,
+      "loss": 0.8,
+      "step": 2193
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.265107026745003e-06,
+      "loss": 0.7989,
+      "step": 2194
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.256437877998654e-06,
+      "loss": 0.7475,
+      "step": 2195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.247772000005188e-06,
+      "loss": 0.6992,
+      "step": 2196
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.239109400335989e-06,
+      "loss": 0.7363,
+      "step": 2197
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230450086559583e-06,
+      "loss": 0.7692,
+      "step": 2198
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.221794066241613e-06,
+      "loss": 0.7945,
+      "step": 2199
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.213141346944853e-06,
+      "loss": 0.7677,
+      "step": 2200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204491936229184e-06,
+      "loss": 0.8098,
+      "step": 2201
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.195845841651609e-06,
+      "loss": 0.7154,
+      "step": 2202
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.187203070766225e-06,
+      "loss": 0.7313,
+      "step": 2203
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1785636311242245e-06,
+      "loss": 0.7539,
+      "step": 2204
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.169927530273898e-06,
+      "loss": 0.705,
+      "step": 2205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.161294775760604e-06,
+      "loss": 0.7432,
+      "step": 2206
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.152665375126795e-06,
+      "loss": 0.6779,
+      "step": 2207
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144039335911982e-06,
+      "loss": 0.7599,
+      "step": 2208
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.135416665652737e-06,
+      "loss": 0.7441,
+      "step": 2209
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1267973718827e-06,
+      "loss": 0.7523,
+      "step": 2210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.118181462132546e-06,
+      "loss": 0.6919,
+      "step": 2211
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.109568943930011e-06,
+      "loss": 0.7906,
+      "step": 2212
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.100959824799853e-06,
+      "loss": 0.729,
+      "step": 2213
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.092354112263861e-06,
+      "loss": 0.775,
+      "step": 2214
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.083751813840864e-06,
+      "loss": 0.7488,
+      "step": 2215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.075152937046685e-06,
+      "loss": 0.7667,
+      "step": 2216
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.066557489394176e-06,
+      "loss": 0.6816,
+      "step": 2217
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.057965478393177e-06,
+      "loss": 0.7426,
+      "step": 2218
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.049376911550542e-06,
+      "loss": 0.6362,
+      "step": 2219
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.040791796370109e-06,
+      "loss": 0.7768,
+      "step": 2220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032210140352695e-06,
+      "loss": 0.758,
+      "step": 2221
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.023631950996101e-06,
+      "loss": 0.8162,
+      "step": 2222
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.015057235795093e-06,
+      "loss": 0.7399,
+      "step": 2223
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0064860022414085e-06,
+      "loss": 0.7668,
+      "step": 2224
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.997918257823745e-06,
+      "loss": 0.7146,
+      "step": 2225
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.98935401002774e-06,
+      "loss": 0.8534,
+      "step": 2226
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9807932663359895e-06,
+      "loss": 0.7415,
+      "step": 2227
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.972236034228011e-06,
+      "loss": 0.6858,
+      "step": 2228
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.963682321180274e-06,
+      "loss": 0.6736,
+      "step": 2229
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.955132134666159e-06,
+      "loss": 0.7864,
+      "step": 2230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.946585482155973e-06,
+      "loss": 0.7663,
+      "step": 2231
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938042371116932e-06,
+      "loss": 0.7213,
+      "step": 2232
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9295028090131545e-06,
+      "loss": 0.759,
+      "step": 2233
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.920966803305664e-06,
+      "loss": 0.6998,
+      "step": 2234
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.912434361452379e-06,
+      "loss": 0.7928,
+      "step": 2235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.903905490908091e-06,
+      "loss": 0.7548,
+      "step": 2236
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8953801991244895e-06,
+      "loss": 0.7927,
+      "step": 2237
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.88685849355012e-06,
+      "loss": 0.6782,
+      "step": 2238
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.878340381630412e-06,
+      "loss": 0.7974,
+      "step": 2239
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.869825870807638e-06,
+      "loss": 0.7108,
+      "step": 2240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8613149685209305e-06,
+      "loss": 0.7962,
+      "step": 2241
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.85280768220628e-06,
+      "loss": 0.8078,
+      "step": 2242
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.844304019296501e-06,
+      "loss": 0.7922,
+      "step": 2243
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8358039872212554e-06,
+      "loss": 0.8068,
+      "step": 2244
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.827307593407021e-06,
+      "loss": 0.7026,
+      "step": 2245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.818814845277111e-06,
+      "loss": 0.7262,
+      "step": 2246
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.81032575025164e-06,
+      "loss": 0.6821,
+      "step": 2247
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.801840315747534e-06,
+      "loss": 0.7636,
+      "step": 2248
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.79335854917853e-06,
+      "loss": 0.7312,
+      "step": 2249
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7848804579551466e-06,
+      "loss": 0.7849,
+      "step": 2250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7764060494847e-06,
+      "loss": 0.7259,
+      "step": 2251
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.767935331171293e-06,
+      "loss": 0.7933,
+      "step": 2252
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.75946831041579e-06,
+      "loss": 0.7172,
+      "step": 2253
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.751004994615841e-06,
+      "loss": 0.7719,
+      "step": 2254
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.742545391165845e-06,
+      "loss": 0.7298,
+      "step": 2255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.734089507456963e-06,
+      "loss": 0.7373,
+      "step": 2256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.725637350877111e-06,
+      "loss": 0.7358,
+      "step": 2257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.71718892881094e-06,
+      "loss": 0.7604,
+      "step": 2258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.708744248639848e-06,
+      "loss": 0.7191,
+      "step": 2259
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.700303317741949e-06,
+      "loss": 0.8008,
+      "step": 2260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.691866143492101e-06,
+      "loss": 0.758,
+      "step": 2261
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68343273326186e-06,
+      "loss": 0.7564,
+      "step": 2262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.675003094419504e-06,
+      "loss": 0.7252,
+      "step": 2263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.666577234330024e-06,
+      "loss": 0.7654,
+      "step": 2264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.658155160355092e-06,
+      "loss": 0.6504,
+      "step": 2265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.649736879853081e-06,
+      "loss": 0.7758,
+      "step": 2266
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.641322400179044e-06,
+      "loss": 0.804,
+      "step": 2267
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.63291172868472e-06,
+      "loss": 0.7701,
+      "step": 2268
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.624504872718528e-06,
+      "loss": 0.7011,
+      "step": 2269
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.616101839625532e-06,
+      "loss": 0.6976,
+      "step": 2270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607702636747477e-06,
+      "loss": 0.6781,
+      "step": 2271
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5993072714227445e-06,
+      "loss": 0.7427,
+      "step": 2272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.59091575098638e-06,
+      "loss": 0.7229,
+      "step": 2273
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.582528082770056e-06,
+      "loss": 0.8141,
+      "step": 2274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.574144274102078e-06,
+      "loss": 0.8001,
+      "step": 2275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5657643323073975e-06,
+      "loss": 0.7639,
+      "step": 2276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.557388264707566e-06,
+      "loss": 0.8136,
+      "step": 2277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.549016078620764e-06,
+      "loss": 0.7217,
+      "step": 2278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.540647781361778e-06,
+      "loss": 0.7832,
+      "step": 2279
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.532283380241992e-06,
+      "loss": 0.7305,
+      "step": 2280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.523922882569393e-06,
+      "loss": 0.7278,
+      "step": 2281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5155662956485464e-06,
+      "loss": 0.7205,
+      "step": 2282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5072136267806184e-06,
+      "loss": 0.645,
+      "step": 2283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.498864883263337e-06,
+      "loss": 0.7427,
+      "step": 2284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.490520072390999e-06,
+      "loss": 0.7262,
+      "step": 2285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.482179201454483e-06,
+      "loss": 0.7239,
+      "step": 2286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.473842277741201e-06,
+      "loss": 0.7049,
+      "step": 2287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465509308535144e-06,
+      "loss": 0.6465,
+      "step": 2288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457180301116819e-06,
+      "loss": 0.7303,
+      "step": 2289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.448855262763293e-06,
+      "loss": 0.7342,
+      "step": 2290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.440534200748161e-06,
+      "loss": 0.7187,
+      "step": 2291
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432217122341537e-06,
+      "loss": 0.7279,
+      "step": 2292
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.423904034810057e-06,
+      "loss": 0.6537,
+      "step": 2293
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.415594945416869e-06,
+      "loss": 0.7155,
+      "step": 2294
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.407289861421634e-06,
+      "loss": 0.7611,
+      "step": 2295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.398988790080515e-06,
+      "loss": 0.7759,
+      "step": 2296
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.390691738646153e-06,
+      "loss": 0.6972,
+      "step": 2297
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.382398714367698e-06,
+      "loss": 0.8183,
+      "step": 2298
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3741097244907625e-06,
+      "loss": 0.7808,
+      "step": 2299
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.365824776257451e-06,
+      "loss": 0.7325,
+      "step": 2300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.357543876906324e-06,
+      "loss": 0.7302,
+      "step": 2301
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.349267033672406e-06,
+      "loss": 0.7741,
+      "step": 2302
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340994253787189e-06,
+      "loss": 0.7682,
+      "step": 2303
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3327255444785974e-06,
+      "loss": 0.7688,
+      "step": 2304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.324460912971014e-06,
+      "loss": 0.6406,
+      "step": 2305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.316200366485256e-06,
+      "loss": 0.7084,
+      "step": 2306
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.30794391223856e-06,
+      "loss": 0.6754,
+      "step": 2307
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.299691557444608e-06,
+      "loss": 0.6276,
+      "step": 2308
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.291443309313475e-06,
+      "loss": 0.7652,
+      "step": 2309
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.283199175051672e-06,
+      "loss": 0.7407,
+      "step": 2310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2749591618621e-06,
+      "loss": 0.7187,
+      "step": 2311
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2667232769440615e-06,
+      "loss": 0.763,
+      "step": 2312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.258491527493262e-06,
+      "loss": 0.7217,
+      "step": 2313
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.25026392070178e-06,
+      "loss": 0.7595,
+      "step": 2314
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.242040463758089e-06,
+      "loss": 0.749,
+      "step": 2315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.233821163847019e-06,
+      "loss": 0.8275,
+      "step": 2316
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.225606028149786e-06,
+      "loss": 0.7462,
+      "step": 2317
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2173950638439605e-06,
+      "loss": 0.8212,
+      "step": 2318
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.209188278103465e-06,
+      "loss": 0.7284,
+      "step": 2319
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.200985678098576e-06,
+      "loss": 0.7673,
+      "step": 2320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.192787270995905e-06,
+      "loss": 0.6937,
+      "step": 2321
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.184593063958409e-06,
+      "loss": 0.7445,
+      "step": 2322
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.176403064145379e-06,
+      "loss": 0.7485,
+      "step": 2323
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.168217278712413e-06,
+      "loss": 0.7159,
+      "step": 2324
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.160035714811448e-06,
+      "loss": 0.7821,
+      "step": 2325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.151858379590713e-06,
+      "loss": 0.8346,
+      "step": 2326
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.143685280194757e-06,
+      "loss": 0.7576,
+      "step": 2327
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135516423764424e-06,
+      "loss": 0.8022,
+      "step": 2328
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.127351817436847e-06,
+      "loss": 0.7562,
+      "step": 2329
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1191914683454505e-06,
+      "loss": 0.7678,
+      "step": 2330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.111035383619929e-06,
+      "loss": 0.8246,
+      "step": 2331
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1028835703862676e-06,
+      "loss": 0.8579,
+      "step": 2332
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.094736035766714e-06,
+      "loss": 0.7668,
+      "step": 2333
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.086592786879765e-06,
+      "loss": 0.8159,
+      "step": 2334
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.078453830840193e-06,
+      "loss": 0.7517,
+      "step": 2335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0703191747590015e-06,
+      "loss": 0.7723,
+      "step": 2336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0621888257434535e-06,
+      "loss": 0.6661,
+      "step": 2337
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.054062790897036e-06,
+      "loss": 0.7988,
+      "step": 2338
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.045941077319467e-06,
+      "loss": 0.7443,
+      "step": 2339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.037823692106703e-06,
+      "loss": 0.7225,
+      "step": 2340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0297106423509e-06,
+      "loss": 0.7908,
+      "step": 2341
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.021601935140443e-06,
+      "loss": 0.7883,
+      "step": 2342
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0134975775599095e-06,
+      "loss": 0.7899,
+      "step": 2343
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.005397576690083e-06,
+      "loss": 0.7374,
+      "step": 2344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.997301939607946e-06,
+      "loss": 0.7337,
+      "step": 2345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.989210673386658e-06,
+      "loss": 0.8393,
+      "step": 2346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.981123785095566e-06,
+      "loss": 0.7518,
+      "step": 2347
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9730412818001825e-06,
+      "loss": 0.6383,
+      "step": 2348
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.964963170562204e-06,
+      "loss": 0.7097,
+      "step": 2349
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9568894584394835e-06,
+      "loss": 0.7103,
+      "step": 2350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.948820152486021e-06,
+      "loss": 0.7028,
+      "step": 2351
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.940755259751984e-06,
+      "loss": 0.7137,
+      "step": 2352
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.932694787283667e-06,
+      "loss": 0.7827,
+      "step": 2353
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.924638742123514e-06,
+      "loss": 0.7281,
+      "step": 2354
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.916587131310101e-06,
+      "loss": 0.6416,
+      "step": 2355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9085399618781226e-06,
+      "loss": 0.7585,
+      "step": 2356
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.900497240858398e-06,
+      "loss": 0.7511,
+      "step": 2357
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.892458975277851e-06,
+      "loss": 0.7197,
+      "step": 2358
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.884425172159529e-06,
+      "loss": 0.7239,
+      "step": 2359
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.876395838522565e-06,
+      "loss": 0.7205,
+      "step": 2360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.868370981382198e-06,
+      "loss": 0.7349,
+      "step": 2361
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.860350607749752e-06,
+      "loss": 0.6973,
+      "step": 2362
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852334724632628e-06,
+      "loss": 0.7518,
+      "step": 2363
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844323339034315e-06,
+      "loss": 0.6813,
+      "step": 2364
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.836316457954365e-06,
+      "loss": 0.7624,
+      "step": 2365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8283140883883895e-06,
+      "loss": 0.7193,
+      "step": 2366
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.820316237328072e-06,
+      "loss": 0.6936,
+      "step": 2367
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.812322911761135e-06,
+      "loss": 0.813,
+      "step": 2368
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.804334118671358e-06,
+      "loss": 0.7938,
+      "step": 2369
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7963498650385485e-06,
+      "loss": 0.7196,
+      "step": 2370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.788370157838559e-06,
+      "loss": 0.8063,
+      "step": 2371
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.780395004043269e-06,
+      "loss": 0.7106,
+      "step": 2372
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772424410620567e-06,
+      "loss": 0.743,
+      "step": 2373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764458384534376e-06,
+      "loss": 0.7086,
+      "step": 2374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.756496932744613e-06,
+      "loss": 0.7197,
+      "step": 2375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.748540062207202e-06,
+      "loss": 0.793,
+      "step": 2376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.740587779874072e-06,
+      "loss": 0.7862,
+      "step": 2377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732640092693132e-06,
+      "loss": 0.7727,
+      "step": 2378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.724697007608288e-06,
+      "loss": 0.7235,
+      "step": 2379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.716758531559412e-06,
+      "loss": 0.7408,
+      "step": 2380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.708824671482361e-06,
+      "loss": 0.7528,
+      "step": 2381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7008954343089565e-06,
+      "loss": 0.7297,
+      "step": 2382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.692970826966976e-06,
+      "loss": 0.6712,
+      "step": 2383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.685050856380153e-06,
+      "loss": 0.8119,
+      "step": 2384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.67713552946817e-06,
+      "loss": 0.7704,
+      "step": 2385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.669224853146659e-06,
+      "loss": 0.7637,
+      "step": 2386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6613188343271776e-06,
+      "loss": 0.7356,
+      "step": 2387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653417479917223e-06,
+      "loss": 0.7324,
+      "step": 2388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.645520796820218e-06,
+      "loss": 0.7573,
+      "step": 2389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637628791935492e-06,
+      "loss": 0.7266,
+      "step": 2390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629741472158304e-06,
+      "loss": 0.7346,
+      "step": 2391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.621858844379809e-06,
+      "loss": 0.6969,
+      "step": 2392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.613980915487056e-06,
+      "loss": 0.8106,
+      "step": 2393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.606107692363008e-06,
+      "loss": 0.7718,
+      "step": 2394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.598239181886497e-06,
+      "loss": 0.7111,
+      "step": 2395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.590375390932254e-06,
+      "loss": 0.6834,
+      "step": 2396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.582516326370871e-06,
+      "loss": 0.7346,
+      "step": 2397
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.574661995068822e-06,
+      "loss": 0.7638,
+      "step": 2398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.566812403888446e-06,
+      "loss": 0.7446,
+      "step": 2399
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.558967559687929e-06,
+      "loss": 0.762,
+      "step": 2400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.551127469321324e-06,
+      "loss": 0.6867,
+      "step": 2401
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543292139638519e-06,
+      "loss": 0.6634,
+      "step": 2402
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.535461577485245e-06,
+      "loss": 0.6935,
+      "step": 2403
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.527635789703076e-06,
+      "loss": 0.8816,
+      "step": 2404
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5198147831294015e-06,
+      "loss": 0.7349,
+      "step": 2405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.511998564597448e-06,
+      "loss": 0.7083,
+      "step": 2406
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504187140936244e-06,
+      "loss": 0.7469,
+      "step": 2407
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.496380518970642e-06,
+      "loss": 0.7812,
+      "step": 2408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.488578705521287e-06,
+      "loss": 0.7438,
+      "step": 2409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4807817074046354e-06,
+      "loss": 0.707,
+      "step": 2410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4729895314329275e-06,
+      "loss": 0.6949,
+      "step": 2411
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.465202184414188e-06,
+      "loss": 0.7671,
+      "step": 2412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.457419673152237e-06,
+      "loss": 0.6973,
+      "step": 2413
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.449642004446649e-06,
+      "loss": 0.7541,
+      "step": 2414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.441869185092787e-06,
+      "loss": 0.8195,
+      "step": 2415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.434101221881768e-06,
+      "loss": 0.743,
+      "step": 2416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.426338121600463e-06,
+      "loss": 0.7185,
+      "step": 2417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.418579891031504e-06,
+      "loss": 0.7137,
+      "step": 2418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.410826536953256e-06,
+      "loss": 0.7381,
+      "step": 2419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403078066139835e-06,
+      "loss": 0.6945,
+      "step": 2420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.395334485361086e-06,
+      "loss": 0.8369,
+      "step": 2421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.387595801382574e-06,
+      "loss": 0.7704,
+      "step": 2422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.379862020965601e-06,
+      "loss": 0.6521,
+      "step": 2423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372133150867168e-06,
+      "loss": 0.6704,
+      "step": 2424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36440919784e-06,
+      "loss": 0.7057,
+      "step": 2425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.356690168632524e-06,
+      "loss": 0.841,
+      "step": 2426
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.348976069988851e-06,
+      "loss": 0.7538,
+      "step": 2427
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.341266908648806e-06,
+      "loss": 0.7436,
+      "step": 2428
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.333562691347881e-06,
+      "loss": 0.6571,
+      "step": 2429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3258634248172596e-06,
+      "loss": 0.7547,
+      "step": 2430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.318169115783789e-06,
+      "loss": 0.6972,
+      "step": 2431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31047977097e-06,
+      "loss": 0.7528,
+      "step": 2432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.30279539709408e-06,
+      "loss": 0.6916,
+      "step": 2433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.295116000869864e-06,
+      "loss": 0.7204,
+      "step": 2434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2874415890068565e-06,
+      "loss": 0.6966,
+      "step": 2435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279772168210185e-06,
+      "loss": 0.7412,
+      "step": 2436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272107745180639e-06,
+      "loss": 0.7243,
+      "step": 2437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.264448326614625e-06,
+      "loss": 0.7781,
+      "step": 2438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.256793919204178e-06,
+      "loss": 0.7843,
+      "step": 2439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2491445296369685e-06,
+      "loss": 0.7468,
+      "step": 2440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.241500164596264e-06,
+      "loss": 0.7881,
+      "step": 2441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2338608307609564e-06,
+      "loss": 0.7572,
+      "step": 2442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.226226534805542e-06,
+      "loss": 0.7831,
+      "step": 2443
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.218597283400101e-06,
+      "loss": 0.7659,
+      "step": 2444
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.210973083210325e-06,
+      "loss": 0.7318,
+      "step": 2445
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203353940897474e-06,
+      "loss": 0.7486,
+      "step": 2446
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.195739863118407e-06,
+      "loss": 0.7542,
+      "step": 2447
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.188130856525545e-06,
+      "loss": 0.6898,
+      "step": 2448
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.180526927766879e-06,
+      "loss": 0.7201,
+      "step": 2449
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172928083485974e-06,
+      "loss": 0.7395,
+      "step": 2450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.165334330321937e-06,
+      "loss": 0.7721,
+      "step": 2451
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.157745674909443e-06,
+      "loss": 0.799,
+      "step": 2452
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.150162123878705e-06,
+      "loss": 0.7168,
+      "step": 2453
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.142583683855471e-06,
+      "loss": 0.7401,
+      "step": 2454
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.135010361461037e-06,
+      "loss": 0.7533,
+      "step": 2455
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1274421633122174e-06,
+      "loss": 0.6936,
+      "step": 2456
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.119879096021352e-06,
+      "loss": 0.7143,
+      "step": 2457
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1123211661962934e-06,
+      "loss": 0.7547,
+      "step": 2458
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.104768380440415e-06,
+      "loss": 0.6943,
+      "step": 2459
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.097220745352592e-06,
+      "loss": 0.6865,
+      "step": 2460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.089678267527196e-06,
+      "loss": 0.6752,
+      "step": 2461
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.082140953554099e-06,
+      "loss": 0.7825,
+      "step": 2462
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.074608810018651e-06,
+      "loss": 0.7132,
+      "step": 2463
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.067081843501696e-06,
+      "loss": 0.7234,
+      "step": 2464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0595600605795506e-06,
+      "loss": 0.7736,
+      "step": 2465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.052043467824002e-06,
+      "loss": 0.7331,
+      "step": 2466
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0445320718023e-06,
+      "loss": 0.7389,
+      "step": 2467
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0370258790771546e-06,
+      "loss": 0.8986,
+      "step": 2468
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.029524896206734e-06,
+      "loss": 0.7291,
+      "step": 2469
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.022029129744658e-06,
+      "loss": 0.6997,
+      "step": 2470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.014538586239973e-06,
+      "loss": 0.6697,
+      "step": 2471
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.007053272237181e-06,
+      "loss": 0.7468,
+      "step": 2472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999573194276199e-06,
+      "loss": 0.7151,
+      "step": 2473
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.992098358892384e-06,
+      "loss": 0.6902,
+      "step": 2474
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984628772616501e-06,
+      "loss": 0.7497,
+      "step": 2475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.977164441974728e-06,
+      "loss": 0.6944,
+      "step": 2476
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9697053734886645e-06,
+      "loss": 0.7177,
+      "step": 2477
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.962251573675297e-06,
+      "loss": 0.7781,
+      "step": 2478
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9548030490470215e-06,
+      "loss": 0.7142,
+      "step": 2479
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.947359806111613e-06,
+      "loss": 0.7284,
+      "step": 2480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.939921851372243e-06,
+      "loss": 0.6711,
+      "step": 2481
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932489191327457e-06,
+      "loss": 0.8224,
+      "step": 2482
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925061832471178e-06,
+      "loss": 0.74,
+      "step": 2483
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.917639781292687e-06,
+      "loss": 0.6648,
+      "step": 2484
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.910223044276643e-06,
+      "loss": 0.7528,
+      "step": 2485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.902811627903048e-06,
+      "loss": 0.7122,
+      "step": 2486
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.895405538647269e-06,
+      "loss": 0.7076,
+      "step": 2487
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.888004782980004e-06,
+      "loss": 0.7486,
+      "step": 2488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.880609367367304e-06,
+      "loss": 0.774,
+      "step": 2489
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.873219298270545e-06,
+      "loss": 0.7343,
+      "step": 2490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.865834582146437e-06,
+      "loss": 0.6147,
+      "step": 2491
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.858455225447017e-06,
+      "loss": 0.7132,
+      "step": 2492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.851081234619627e-06,
+      "loss": 0.7328,
+      "step": 2493
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.843712616106931e-06,
+      "loss": 0.7967,
+      "step": 2494
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836349376346894e-06,
+      "loss": 0.7399,
+      "step": 2495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.828991521772785e-06,
+      "loss": 0.6936,
+      "step": 2496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.821639058813169e-06,
+      "loss": 0.7231,
+      "step": 2497
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8142919938918945e-06,
+      "loss": 0.7302,
+      "step": 2498
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806950333428102e-06,
+      "loss": 0.7172,
+      "step": 2499
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7996140838361994e-06,
+      "loss": 0.6939,
+      "step": 2500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79228325152588e-06,
+      "loss": 0.735,
+      "step": 2501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.784957842902094e-06,
+      "loss": 0.7633,
+      "step": 2502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7776378643650515e-06,
+      "loss": 0.7419,
+      "step": 2503
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7703233223102307e-06,
+      "loss": 0.7616,
+      "step": 2504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7630142231283452e-06,
+      "loss": 0.7293,
+      "step": 2505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.755710573205368e-06,
+      "loss": 0.7091,
+      "step": 2506
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7484123789224945e-06,
+      "loss": 0.6751,
+      "step": 2507
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.741119646656167e-06,
+      "loss": 0.6367,
+      "step": 2508
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.733832382778054e-06,
+      "loss": 0.7527,
+      "step": 2509
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7265505936550363e-06,
+      "loss": 0.7635,
+      "step": 2510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.719274285649225e-06,
+      "loss": 0.6934,
+      "step": 2511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7120034651179315e-06,
+      "loss": 0.6888,
+      "step": 2512
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.704738138413674e-06,
+      "loss": 0.6878,
+      "step": 2513
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.697478311884181e-06,
+      "loss": 0.7694,
+      "step": 2514
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6902239918723602e-06,
+      "loss": 0.7719,
+      "step": 2515
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6829751847163242e-06,
+      "loss": 0.6879,
+      "step": 2516
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6757318967493526e-06,
+      "loss": 0.7012,
+      "step": 2517
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.668494134299916e-06,
+      "loss": 0.7118,
+      "step": 2518
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.661261903691656e-06,
+      "loss": 0.7236,
+      "step": 2519
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.654035211243373e-06,
+      "loss": 0.764,
+      "step": 2520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.646814063269034e-06,
+      "loss": 0.7511,
+      "step": 2521
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6395984660777585e-06,
+      "loss": 0.7513,
+      "step": 2522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6323884259738206e-06,
+      "loss": 0.8015,
+      "step": 2523
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.625183949256643e-06,
+      "loss": 0.6944,
+      "step": 2524
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.617985042220773e-06,
+      "loss": 0.7236,
+      "step": 2525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.61079171115591e-06,
+      "loss": 0.711,
+      "step": 2526
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6036039623468632e-06,
+      "loss": 0.7145,
+      "step": 2527
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.596421802073582e-06,
+      "loss": 0.7283,
+      "step": 2528
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5892452366111187e-06,
+      "loss": 0.6984,
+      "step": 2529
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5820742722296487e-06,
+      "loss": 0.6713,
+      "step": 2530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.574908915194447e-06,
+      "loss": 0.7063,
+      "step": 2531
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5677491717658875e-06,
+      "loss": 0.7256,
+      "step": 2532
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.560595048199449e-06,
+      "loss": 0.6757,
+      "step": 2533
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5534465507456874e-06,
+      "loss": 0.7299,
+      "step": 2534
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546303685650254e-06,
+      "loss": 0.7074,
+      "step": 2535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5391664591538787e-06,
+      "loss": 0.7205,
+      "step": 2536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5320348774923553e-06,
+      "loss": 0.7058,
+      "step": 2537
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.524908946896556e-06,
+      "loss": 0.7935,
+      "step": 2538
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5177886735924105e-06,
+      "loss": 0.7631,
+      "step": 2539
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.510674063800903e-06,
+      "loss": 0.756,
+      "step": 2540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5035651237380787e-06,
+      "loss": 0.7838,
+      "step": 2541
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4964618596150168e-06,
+      "loss": 0.6873,
+      "step": 2542
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.489364277637852e-06,
+      "loss": 0.7477,
+      "step": 2543
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.48227238400774e-06,
+      "loss": 0.7221,
+      "step": 2544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4751861849208756e-06,
+      "loss": 0.7239,
+      "step": 2545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.46810568656848e-06,
+      "loss": 0.7125,
+      "step": 2546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4610308951367864e-06,
+      "loss": 0.7528,
+      "step": 2547
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4539618168070456e-06,
+      "loss": 0.7355,
+      "step": 2548
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.446898457755512e-06,
+      "loss": 0.7163,
+      "step": 2549
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.439840824153451e-06,
+      "loss": 0.7427,
+      "step": 2550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.432788922167125e-06,
+      "loss": 0.7826,
+      "step": 2551
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4257427579577805e-06,
+      "loss": 0.7464,
+      "step": 2552
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.418702337681662e-06,
+      "loss": 0.7044,
+      "step": 2553
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.411667667489982e-06,
+      "loss": 0.6747,
+      "step": 2554
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.404638753528945e-06,
+      "loss": 0.7717,
+      "step": 2555
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.397615601939711e-06,
+      "loss": 0.6533,
+      "step": 2556
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3905982188584195e-06,
+      "loss": 0.7109,
+      "step": 2557
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3835866104161596e-06,
+      "loss": 0.6902,
+      "step": 2558
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.376580782738975e-06,
+      "loss": 0.6041,
+      "step": 2559
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3695807419478686e-06,
+      "loss": 0.6605,
+      "step": 2560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3625864941587748e-06,
+      "loss": 0.6796,
+      "step": 2561
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3555980454825764e-06,
+      "loss": 0.6597,
+      "step": 2562
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3486154020250893e-06,
+      "loss": 0.7875,
+      "step": 2563
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.341638569887048e-06,
+      "loss": 0.7026,
+      "step": 2564
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3346675551641204e-06,
+      "loss": 0.6796,
+      "step": 2565
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.327702363946885e-06,
+      "loss": 0.7239,
+      "step": 2566
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.320743002320831e-06,
+      "loss": 0.7134,
+      "step": 2567
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3137894763663636e-06,
+      "loss": 0.7719,
+      "step": 2568
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3068417921587783e-06,
+      "loss": 0.7103,
+      "step": 2569
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2998999557682766e-06,
+      "loss": 0.7005,
+      "step": 2570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2929639732599393e-06,
+      "loss": 0.6997,
+      "step": 2571
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.286033850693744e-06,
+      "loss": 0.6925,
+      "step": 2572
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2791095941245455e-06,
+      "loss": 0.8469,
+      "step": 2573
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.27219120960207e-06,
+      "loss": 0.7837,
+      "step": 2574
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.26527870317091e-06,
+      "loss": 0.7819,
+      "step": 2575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2583720808705355e-06,
+      "loss": 0.7571,
+      "step": 2576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2514713487352622e-06,
+      "loss": 0.7676,
+      "step": 2577
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.244576512794263e-06,
+      "loss": 0.7659,
+      "step": 2578
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2376875790715623e-06,
+      "loss": 0.767,
+      "step": 2579
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.230804553586032e-06,
+      "loss": 0.6727,
+      "step": 2580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2239274423513687e-06,
+      "loss": 0.6617,
+      "step": 2581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.217056251376116e-06,
+      "loss": 0.6749,
+      "step": 2582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2101909866636316e-06,
+      "loss": 0.7678,
+      "step": 2583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2033316542121086e-06,
+      "loss": 0.7387,
+      "step": 2584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1964782600145484e-06,
+      "loss": 0.7406,
+      "step": 2585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1896308100587613e-06,
+      "loss": 0.7555,
+      "step": 2586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1827893103273768e-06,
+      "loss": 0.7793,
+      "step": 2587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1759537667978122e-06,
+      "loss": 0.7181,
+      "step": 2588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1691241854422882e-06,
+      "loss": 0.693,
+      "step": 2589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.16230057222782e-06,
+      "loss": 0.74,
+      "step": 2590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1554829331161963e-06,
+      "loss": 0.7023,
+      "step": 2591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.148671274063999e-06,
+      "loss": 0.6646,
+      "step": 2592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.141865601022577e-06,
+      "loss": 0.7276,
+      "step": 2593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1350659199380475e-06,
+      "loss": 0.7194,
+      "step": 2594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.128272236751304e-06,
+      "loss": 0.6956,
+      "step": 2595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.121484557397986e-06,
+      "loss": 0.8467,
+      "step": 2596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1147028878084993e-06,
+      "loss": 0.7071,
+      "step": 2597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.107927233907988e-06,
+      "loss": 0.7344,
+      "step": 2598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1011576016163493e-06,
+      "loss": 0.7881,
+      "step": 2599
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0943939968482173e-06,
+      "loss": 0.6676,
+      "step": 2600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.087636425512952e-06,
+      "loss": 0.8551,
+      "step": 2601
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.080884893514656e-06,
+      "loss": 0.7772,
+      "step": 2602
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0741394067521444e-06,
+      "loss": 0.7963,
+      "step": 2603
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0673999711189527e-06,
+      "loss": 0.758,
+      "step": 2604
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0606665925033274e-06,
+      "loss": 0.6937,
+      "step": 2605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0539392767882305e-06,
+      "loss": 0.7834,
+      "step": 2606
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0472180298513256e-06,
+      "loss": 0.7748,
+      "step": 2607
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.040502857564963e-06,
+      "loss": 0.6895,
+      "step": 2608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0337937657962015e-06,
+      "loss": 0.7542,
+      "step": 2609
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.027090760406771e-06,
+      "loss": 0.6968,
+      "step": 2610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0203938472531012e-06,
+      "loss": 0.746,
+      "step": 2611
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.013703032186286e-06,
+      "loss": 0.7762,
+      "step": 2612
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0070183210520896e-06,
+      "loss": 0.7366,
+      "step": 2613
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.000339719690958e-06,
+      "loss": 0.7053,
+      "step": 2614
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9936672339379825e-06,
+      "loss": 0.7017,
+      "step": 2615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.987000869622921e-06,
+      "loss": 0.7019,
+      "step": 2616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9803406325701867e-06,
+      "loss": 0.7739,
+      "step": 2617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9736865285988238e-06,
+      "loss": 0.6482,
+      "step": 2618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9670385635225363e-06,
+      "loss": 0.75,
+      "step": 2619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9603967431496485e-06,
+      "loss": 0.7076,
+      "step": 2620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9537610732831302e-06,
+      "loss": 0.6705,
+      "step": 2621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9471315597205675e-06,
+      "loss": 0.7545,
+      "step": 2622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9405082082541694e-06,
+      "loss": 0.7012,
+      "step": 2623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9338910246707663e-06,
+      "loss": 0.7277,
+      "step": 2624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.927280014751792e-06,
+      "loss": 0.7671,
+      "step": 2625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9206751842732973e-06,
+      "loss": 0.8022,
+      "step": 2626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.914076539005919e-06,
+      "loss": 0.6632,
+      "step": 2627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.907484084714902e-06,
+      "loss": 0.7333,
+      "step": 2628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9008978271600828e-06,
+      "loss": 0.7541,
+      "step": 2629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.894317772095875e-06,
+      "loss": 0.782,
+      "step": 2630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.887743925271277e-06,
+      "loss": 0.6605,
+      "step": 2631
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8811762924298616e-06,
+      "loss": 0.7887,
+      "step": 2632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.874614879309775e-06,
+      "loss": 0.7349,
+      "step": 2633
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8680596916437354e-06,
+      "loss": 0.75,
+      "step": 2634
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8615107351590056e-06,
+      "loss": 0.7429,
+      "step": 2635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8549680155774206e-06,
+      "loss": 0.6996,
+      "step": 2636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.848431538615353e-06,
+      "loss": 0.7035,
+      "step": 2637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8419013099837333e-06,
+      "loss": 0.7329,
+      "step": 2638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.835377335388024e-06,
+      "loss": 0.6559,
+      "step": 2639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8288596205282248e-06,
+      "loss": 0.7881,
+      "step": 2640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.822348171098872e-06,
+      "loss": 0.6895,
+      "step": 2641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8158429927890174e-06,
+      "loss": 0.6813,
+      "step": 2642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8093440912822436e-06,
+      "loss": 0.6895,
+      "step": 2643
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8028514722566493e-06,
+      "loss": 0.663,
+      "step": 2644
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7963651413848326e-06,
+      "loss": 0.6955,
+      "step": 2645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7898851043339126e-06,
+      "loss": 0.6878,
+      "step": 2646
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.783411366765495e-06,
+      "loss": 0.6747,
+      "step": 2647
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7769439343356973e-06,
+      "loss": 0.7971,
+      "step": 2648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7704828126951145e-06,
+      "loss": 0.6737,
+      "step": 2649
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7640280074888314e-06,
+      "loss": 0.8054,
+      "step": 2650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7575795243564216e-06,
+      "loss": 0.7693,
+      "step": 2651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7511373689319233e-06,
+      "loss": 0.7817,
+      "step": 2652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7447015468438575e-06,
+      "loss": 0.6685,
+      "step": 2653
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7382720637152025e-06,
+      "loss": 0.7519,
+      "step": 2654
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.731848925163403e-06,
+      "loss": 0.6732,
+      "step": 2655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7254321368003643e-06,
+      "loss": 0.6805,
+      "step": 2656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.719021704232436e-06,
+      "loss": 0.672,
+      "step": 2657
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.712617633060416e-06,
+      "loss": 0.7121,
+      "step": 2658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7062199288795445e-06,
+      "loss": 0.7937,
+      "step": 2659
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6998285972795025e-06,
+      "loss": 0.7461,
+      "step": 2660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6934436438444045e-06,
+      "loss": 0.7576,
+      "step": 2661
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6870650741527817e-06,
+      "loss": 0.6869,
+      "step": 2662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.680692893777602e-06,
+      "loss": 0.7089,
+      "step": 2663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6743271082862375e-06,
+      "loss": 0.7282,
+      "step": 2664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.667967723240482e-06,
+      "loss": 0.6969,
+      "step": 2665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.661614744196539e-06,
+      "loss": 0.7363,
+      "step": 2666
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.655268176705007e-06,
+      "loss": 0.6327,
+      "step": 2667
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6489280263108865e-06,
+      "loss": 0.7428,
+      "step": 2668
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6425942985535693e-06,
+      "loss": 0.815,
+      "step": 2669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.636266998966839e-06,
+      "loss": 0.7409,
+      "step": 2670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6299461330788655e-06,
+      "loss": 0.7322,
+      "step": 2671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6236317064121897e-06,
+      "loss": 0.756,
+      "step": 2672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6173237244837346e-06,
+      "loss": 0.687,
+      "step": 2673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6110221928047818e-06,
+      "loss": 0.7547,
+      "step": 2674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.604727116880993e-06,
+      "loss": 0.7539,
+      "step": 2675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.598438502212375e-06,
+      "loss": 0.733,
+      "step": 2676
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5921563542932913e-06,
+      "loss": 0.7944,
+      "step": 2677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.585880678612468e-06,
+      "loss": 0.6988,
+      "step": 2678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5796114806529595e-06,
+      "loss": 0.6704,
+      "step": 2679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.573348765892174e-06,
+      "loss": 0.7165,
+      "step": 2680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5670925398018455e-06,
+      "loss": 0.7066,
+      "step": 2681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.560842807848044e-06,
+      "loss": 0.7732,
+      "step": 2682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55459957549117e-06,
+      "loss": 0.7485,
+      "step": 2683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5483628481859355e-06,
+      "loss": 0.6684,
+      "step": 2684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5421326313813745e-06,
+      "loss": 0.7031,
+      "step": 2685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5359089305208284e-06,
+      "loss": 0.7165,
+      "step": 2686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5296917510419518e-06,
+      "loss": 0.7086,
+      "step": 2687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.523481098376702e-06,
+      "loss": 0.732,
+      "step": 2688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.517276977951325e-06,
+      "loss": 0.6712,
+      "step": 2689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.511079395186372e-06,
+      "loss": 0.6667,
+      "step": 2690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5048883554966672e-06,
+      "loss": 0.7384,
+      "step": 2691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.498703864291331e-06,
+      "loss": 0.7391,
+      "step": 2692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.492525926973761e-06,
+      "loss": 0.6412,
+      "step": 2693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4863545489416206e-06,
+      "loss": 0.7297,
+      "step": 2694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.48018973558685e-06,
+      "loss": 0.7084,
+      "step": 2695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4740314922956452e-06,
+      "loss": 0.7546,
+      "step": 2696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4678798244484724e-06,
+      "loss": 0.7001,
+      "step": 2697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4617347374200506e-06,
+      "loss": 0.7344,
+      "step": 2698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.455596236579342e-06,
+      "loss": 0.767,
+      "step": 2699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4494643272895623e-06,
+      "loss": 0.687,
+      "step": 2700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4433390149081614e-06,
+      "loss": 0.7509,
+      "step": 2701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4372203047868336e-06,
+      "loss": 0.7286,
+      "step": 2702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4311082022714993e-06,
+      "loss": 0.7908,
+      "step": 2703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4250027127023014e-06,
+      "loss": 0.7287,
+      "step": 2704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4189038414136177e-06,
+      "loss": 0.7235,
+      "step": 2705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.412811593734031e-06,
+      "loss": 0.7357,
+      "step": 2706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.406725974986348e-06,
+      "loss": 0.7085,
+      "step": 2707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4006469904875707e-06,
+      "loss": 0.6879,
+      "step": 2708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3945746455489173e-06,
+      "loss": 0.7095,
+      "step": 2709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3885089454758015e-06,
+      "loss": 0.7173,
+      "step": 2710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3824498955678243e-06,
+      "loss": 0.767,
+      "step": 2711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.376397501118788e-06,
+      "loss": 0.7829,
+      "step": 2712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3703517674166697e-06,
+      "loss": 0.7304,
+      "step": 2713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3643126997436305e-06,
+      "loss": 0.732,
+      "step": 2714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3582803033760116e-06,
+      "loss": 0.7137,
+      "step": 2715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.352254583584318e-06,
+      "loss": 0.7128,
+      "step": 2716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.346235545633231e-06,
+      "loss": 0.8093,
+      "step": 2717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.340223194781581e-06,
+      "loss": 0.8605,
+      "step": 2718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3342175362823683e-06,
+      "loss": 0.7634,
+      "step": 2719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3282185753827413e-06,
+      "loss": 0.7357,
+      "step": 2720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.322226317323997e-06,
+      "loss": 0.8018,
+      "step": 2721
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.316240767341571e-06,
+      "loss": 0.6954,
+      "step": 2722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3102619306650444e-06,
+      "loss": 0.7768,
+      "step": 2723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3042898125181334e-06,
+      "loss": 0.7148,
+      "step": 2724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.298324418118677e-06,
+      "loss": 0.7686,
+      "step": 2725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.292365752678648e-06,
+      "loss": 0.6843,
+      "step": 2726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2864138214041396e-06,
+      "loss": 0.7322,
+      "step": 2727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2804686294953517e-06,
+      "loss": 0.6702,
+      "step": 2728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.274530182146609e-06,
+      "loss": 0.6677,
+      "step": 2729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2685984845463347e-06,
+      "loss": 0.7482,
+      "step": 2730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.262673541877054e-06,
+      "loss": 0.7163,
+      "step": 2731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.256755359315399e-06,
+      "loss": 0.7398,
+      "step": 2732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.250843942032084e-06,
+      "loss": 0.7217,
+      "step": 2733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2449392951919248e-06,
+      "loss": 0.7798,
+      "step": 2734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2390414239538096e-06,
+      "loss": 0.7621,
+      "step": 2735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.233150333470716e-06,
+      "loss": 0.7459,
+      "step": 2736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.227266028889696e-06,
+      "loss": 0.6346,
+      "step": 2737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2213885153518655e-06,
+      "loss": 0.6563,
+      "step": 2738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.215517797992418e-06,
+      "loss": 0.6772,
+      "step": 2739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2096538819406e-06,
+      "loss": 0.7313,
+      "step": 2740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2037967723197174e-06,
+      "loss": 0.792,
+      "step": 2741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.197946474247136e-06,
+      "loss": 0.8242,
+      "step": 2742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1921029928342584e-06,
+      "loss": 0.753,
+      "step": 2743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.186266333186546e-06,
+      "loss": 0.6832,
+      "step": 2744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1804365004034877e-06,
+      "loss": 0.7071,
+      "step": 2745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1746134995786127e-06,
+      "loss": 0.7087,
+      "step": 2746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1687973357994852e-06,
+      "loss": 0.7766,
+      "step": 2747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1629880141476876e-06,
+      "loss": 0.8079,
+      "step": 2748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1571855396988306e-06,
+      "loss": 0.6994,
+      "step": 2749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.151389917522536e-06,
+      "loss": 0.6592,
+      "step": 2750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1456011526824496e-06,
+      "loss": 0.751,
+      "step": 2751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1398192502362112e-06,
+      "loss": 0.7511,
+      "step": 2752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.134044215235479e-06,
+      "loss": 0.701,
+      "step": 2753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.128276052725907e-06,
+      "loss": 0.7725,
+      "step": 2754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.122514767747136e-06,
+      "loss": 0.746,
+      "step": 2755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.116760365332812e-06,
+      "loss": 0.7513,
+      "step": 2756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1110128505105534e-06,
+      "loss": 0.6858,
+      "step": 2757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1052722283019745e-06,
+      "loss": 0.7236,
+      "step": 2758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.099538503722659e-06,
+      "loss": 0.8331,
+      "step": 2759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.093811681782161e-06,
+      "loss": 0.6051,
+      "step": 2760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0880917674840173e-06,
+      "loss": 0.7615,
+      "step": 2761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0823787658257132e-06,
+      "loss": 0.6616,
+      "step": 2762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0766726817987058e-06,
+      "loss": 0.6531,
+      "step": 2763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0709735203884085e-06,
+      "loss": 0.7214,
+      "step": 2764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0652812865741757e-06,
+      "loss": 0.7604,
+      "step": 2765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0595959853293233e-06,
+      "loss": 0.6373,
+      "step": 2766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0539176216210986e-06,
+      "loss": 0.7297,
+      "step": 2767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0482462004106895e-06,
+      "loss": 0.6823,
+      "step": 2768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0425817266532277e-06,
+      "loss": 0.7208,
+      "step": 2769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0369242052977613e-06,
+      "loss": 0.695,
+      "step": 2770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.031273641287276e-06,
+      "loss": 0.7415,
+      "step": 2771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.025630039558668e-06,
+      "loss": 0.7318,
+      "step": 2772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0199934050427627e-06,
+      "loss": 0.6973,
+      "step": 2773
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0143637426642857e-06,
+      "loss": 0.7363,
+      "step": 2774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.008741057341882e-06,
+      "loss": 0.8057,
+      "step": 2775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0031253539880935e-06,
+      "loss": 0.7348,
+      "step": 2776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9975166375093625e-06,
+      "loss": 0.7334,
+      "step": 2777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9919149128060313e-06,
+      "loss": 0.7157,
+      "step": 2778
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.986320184772328e-06,
+      "loss": 0.7532,
+      "step": 2779
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9807324582963715e-06,
+      "loss": 0.7263,
+      "step": 2780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.975151738260166e-06,
+      "loss": 0.8116,
+      "step": 2781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.969578029539585e-06,
+      "loss": 0.6697,
+      "step": 2782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.964011337004387e-06,
+      "loss": 0.7604,
+      "step": 2783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9584516655181896e-06,
+      "loss": 0.7377,
+      "step": 2784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9528990199384866e-06,
+      "loss": 0.8113,
+      "step": 2785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9473534051166276e-06,
+      "loss": 0.6903,
+      "step": 2786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9418148258978165e-06,
+      "loss": 0.7224,
+      "step": 2787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9362832871211177e-06,
+      "loss": 0.7775,
+      "step": 2788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9307587936194383e-06,
+      "loss": 0.7417,
+      "step": 2789
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9252413502195324e-06,
+      "loss": 0.6501,
+      "step": 2790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.919730961741998e-06,
+      "loss": 0.7544,
+      "step": 2791
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.91422763300126e-06,
+      "loss": 0.7447,
+      "step": 2792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9087313688055854e-06,
+      "loss": 0.7717,
+      "step": 2793
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9032421739570617e-06,
+      "loss": 0.6639,
+      "step": 2794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8977600532516004e-06,
+      "loss": 0.7515,
+      "step": 2795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.892285011478938e-06,
+      "loss": 0.7648,
+      "step": 2796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8868170534226193e-06,
+      "loss": 0.7496,
+      "step": 2797
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8813561838600058e-06,
+      "loss": 0.7677,
+      "step": 2798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.87590240756226e-06,
+      "loss": 0.7498,
+      "step": 2799
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.870455729294355e-06,
+      "loss": 0.7164,
+      "step": 2800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8650161538150513e-06,
+      "loss": 0.7216,
+      "step": 2801
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8595836858769135e-06,
+      "loss": 0.6237,
+      "step": 2802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8541583302262944e-06,
+      "loss": 0.7821,
+      "step": 2803
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8487400916033293e-06,
+      "loss": 0.7289,
+      "step": 2804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.843328974741936e-06,
+      "loss": 0.6963,
+      "step": 2805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.83792498436981e-06,
+      "loss": 0.7367,
+      "step": 2806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8325281252084238e-06,
+      "loss": 0.7442,
+      "step": 2807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8271384019730186e-06,
+      "loss": 0.6932,
+      "step": 2808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8217558193725948e-06,
+      "loss": 0.7311,
+      "step": 2809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8163803821099236e-06,
+      "loss": 0.8292,
+      "step": 2810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8110120948815247e-06,
+      "loss": 0.7699,
+      "step": 2811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8056509623776775e-06,
+      "loss": 0.7454,
+      "step": 2812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8002969892824062e-06,
+      "loss": 0.7209,
+      "step": 2813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.794950180273478e-06,
+      "loss": 0.7038,
+      "step": 2814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7896105400224073e-06,
+      "loss": 0.7268,
+      "step": 2815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7842780731944398e-06,
+      "loss": 0.6542,
+      "step": 2816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7789527844485544e-06,
+      "loss": 0.6643,
+      "step": 2817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7736346784374646e-06,
+      "loss": 0.7166,
+      "step": 2818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7683237598075976e-06,
+      "loss": 0.7562,
+      "step": 2819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.763020033199112e-06,
+      "loss": 0.766,
+      "step": 2820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.757723503245875e-06,
+      "loss": 0.6391,
+      "step": 2821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7524341745754657e-06,
+      "loss": 0.6777,
+      "step": 2822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.74715205180918e-06,
+      "loss": 0.786,
+      "step": 2823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.741877139562006e-06,
+      "loss": 0.6929,
+      "step": 2824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.736609442442645e-06,
+      "loss": 0.7842,
+      "step": 2825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.731348965053481e-06,
+      "loss": 0.7275,
+      "step": 2826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.726095711990604e-06,
+      "loss": 0.7315,
+      "step": 2827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7208496878437774e-06,
+      "loss": 0.7269,
+      "step": 2828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7156108971964602e-06,
+      "loss": 0.7173,
+      "step": 2829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7103793446257898e-06,
+      "loss": 0.6724,
+      "step": 2830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7051550347025759e-06,
+      "loss": 0.7666,
+      "step": 2831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6999379719913e-06,
+      "loss": 0.7378,
+      "step": 2832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.694728161050111e-06,
+      "loss": 0.7349,
+      "step": 2833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6895256064308285e-06,
+      "loss": 0.6879,
+      "step": 2834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6843303126789301e-06,
+      "loss": 0.6449,
+      "step": 2835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6791422843335415e-06,
+      "loss": 0.7565,
+      "step": 2836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6739615259274533e-06,
+      "loss": 0.7222,
+      "step": 2837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6687880419870917e-06,
+      "loss": 0.7716,
+      "step": 2838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6636218370325396e-06,
+      "loss": 0.8001,
+      "step": 2839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.65846291557751e-06,
+      "loss": 0.6717,
+      "step": 2840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.653311282129354e-06,
+      "loss": 0.7042,
+      "step": 2841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6481669411890632e-06,
+      "loss": 0.7408,
+      "step": 2842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.643029897251246e-06,
+      "loss": 0.7245,
+      "step": 2843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6379001548041474e-06,
+      "loss": 0.6846,
+      "step": 2844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.63277771832962e-06,
+      "loss": 0.674,
+      "step": 2845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6276625923031452e-06,
+      "loss": 0.7676,
+      "step": 2846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6225547811938135e-06,
+      "loss": 0.7043,
+      "step": 2847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6174542894643185e-06,
+      "loss": 0.8641,
+      "step": 2848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6123611215709679e-06,
+      "loss": 0.7332,
+      "step": 2849
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.607275281963664e-06,
+      "loss": 0.7754,
+      "step": 2850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6021967750859058e-06,
+      "loss": 0.7092,
+      "step": 2851
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.597125605374794e-06,
+      "loss": 0.7677,
+      "step": 2852
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.592061777261007e-06,
+      "loss": 0.7147,
+      "step": 2853
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587005295168821e-06,
+      "loss": 0.6778,
+      "step": 2854
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5819561635160819e-06,
+      "loss": 0.7257,
+      "step": 2855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5769143867142235e-06,
+      "loss": 0.785,
+      "step": 2856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5718799691682506e-06,
+      "loss": 0.7269,
+      "step": 2857
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5668529152767364e-06,
+      "loss": 0.7039,
+      "step": 2858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5618332294318216e-06,
+      "loss": 0.7229,
+      "step": 2859
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.556820916019206e-06,
+      "loss": 0.7657,
+      "step": 2860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.551815979418153e-06,
+      "loss": 0.7818,
+      "step": 2861
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5468184240014839e-06,
+      "loss": 0.7205,
+      "step": 2862
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.54182825413556e-06,
+      "loss": 0.7418,
+      "step": 2863
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5368454741803018e-06,
+      "loss": 0.7314,
+      "step": 2864
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5318700884891613e-06,
+      "loss": 0.747,
+      "step": 2865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5269021014091434e-06,
+      "loss": 0.6526,
+      "step": 2866
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5219415172807795e-06,
+      "loss": 0.7234,
+      "step": 2867
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5169883404381313e-06,
+      "loss": 0.7057,
+      "step": 2868
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5120425752087985e-06,
+      "loss": 0.7164,
+      "step": 2869
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5071042259138957e-06,
+      "loss": 0.6404,
+      "step": 2870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.502173296868067e-06,
+      "loss": 0.6617,
+      "step": 2871
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4972497923794627e-06,
+      "loss": 0.7018,
+      "step": 2872
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4923337167497542e-06,
+      "loss": 0.7373,
+      "step": 2873
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4874250742741247e-06,
+      "loss": 0.742,
+      "step": 2874
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4825238692412514e-06,
+      "loss": 0.6773,
+      "step": 2875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4776301059333275e-06,
+      "loss": 0.6523,
+      "step": 2876
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4727437886260343e-06,
+      "loss": 0.6997,
+      "step": 2877
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4678649215885476e-06,
+      "loss": 0.7178,
+      "step": 2878
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4629935090835434e-06,
+      "loss": 0.7999,
+      "step": 2879
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4581295553671725e-06,
+      "loss": 0.7176,
+      "step": 2880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4532730646890791e-06,
+      "loss": 0.7708,
+      "step": 2881
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4484240412923778e-06,
+      "loss": 0.7911,
+      "step": 2882
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4435824894136673e-06,
+      "loss": 0.7493,
+      "step": 2883
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4387484132830164e-06,
+      "loss": 0.7029,
+      "step": 2884
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4339218171239588e-06,
+      "loss": 0.736,
+      "step": 2885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4291027051534944e-06,
+      "loss": 0.6726,
+      "step": 2886
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4242910815820832e-06,
+      "loss": 0.7669,
+      "step": 2887
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4194869506136455e-06,
+      "loss": 0.5912,
+      "step": 2888
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4146903164455571e-06,
+      "loss": 0.804,
+      "step": 2889
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.409901183268636e-06,
+      "loss": 0.6633,
+      "step": 2890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4051195552671559e-06,
+      "loss": 0.7354,
+      "step": 2891
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.400345436618824e-06,
+      "loss": 0.6943,
+      "step": 2892
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.395578831494795e-06,
+      "loss": 0.7806,
+      "step": 2893
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3908197440596516e-06,
+      "loss": 0.7734,
+      "step": 2894
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.386068178471416e-06,
+      "loss": 0.7873,
+      "step": 2895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3813241388815312e-06,
+      "loss": 0.7703,
+      "step": 2896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.376587629434868e-06,
+      "loss": 0.7188,
+      "step": 2897
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.37185865426972e-06,
+      "loss": 0.674,
+      "step": 2898
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3671372175177934e-06,
+      "loss": 0.7213,
+      "step": 2899
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3624233233042128e-06,
+      "loss": 0.7352,
+      "step": 2900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3577169757475117e-06,
+      "loss": 0.7395,
+      "step": 2901
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3530181789596274e-06,
+      "loss": 0.691,
+      "step": 2902
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.348326937045905e-06,
+      "loss": 0.7303,
+      "step": 2903
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3436432541050837e-06,
+      "loss": 0.7617,
+      "step": 2904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3389671342292975e-06,
+      "loss": 0.7706,
+      "step": 2905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3342985815040821e-06,
+      "loss": 0.685,
+      "step": 2906
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.329637600008349e-06,
+      "loss": 0.7626,
+      "step": 2907
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.324984193814407e-06,
+      "loss": 0.7959,
+      "step": 2908
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3203383669879355e-06,
+      "loss": 0.753,
+      "step": 2909
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3157001235879985e-06,
+      "loss": 0.7025,
+      "step": 2910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3110694676670343e-06,
+      "loss": 0.7107,
+      "step": 2911
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3064464032708491e-06,
+      "loss": 0.7975,
+      "step": 2912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.301830934438616e-06,
+      "loss": 0.7802,
+      "step": 2913
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.297223065202875e-06,
+      "loss": 0.7688,
+      "step": 2914
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2926227995895234e-06,
+      "loss": 0.7785,
+      "step": 2915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2880301416178199e-06,
+      "loss": 0.7146,
+      "step": 2916
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2834450953003675e-06,
+      "loss": 0.6871,
+      "step": 2917
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2788676646431296e-06,
+      "loss": 0.6889,
+      "step": 2918
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2742978536454065e-06,
+      "loss": 0.7152,
+      "step": 2919
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2697356662998495e-06,
+      "loss": 0.7352,
+      "step": 2920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2651811065924414e-06,
+      "loss": 0.7467,
+      "step": 2921
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2606341785025067e-06,
+      "loss": 0.8094,
+      "step": 2922
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2560948860026989e-06,
+      "loss": 0.6955,
+      "step": 2923
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2515632330589987e-06,
+      "loss": 0.7445,
+      "step": 2924
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2470392236307193e-06,
+      "loss": 0.7601,
+      "step": 2925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2425228616704854e-06,
+      "loss": 0.6248,
+      "step": 2926
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2380141511242483e-06,
+      "loss": 0.7033,
+      "step": 2927
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2335130959312747e-06,
+      "loss": 0.7358,
+      "step": 2928
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2290197000241344e-06,
+      "loss": 0.7367,
+      "step": 2929
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2245339673287148e-06,
+      "loss": 0.7487,
+      "step": 2930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2200559017642022e-06,
+      "loss": 0.7417,
+      "step": 2931
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2155855072430833e-06,
+      "loss": 0.7092,
+      "step": 2932
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2111227876711484e-06,
+      "loss": 0.6969,
+      "step": 2933
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2066677469474764e-06,
+      "loss": 0.7366,
+      "step": 2934
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2022203889644425e-06,
+      "loss": 0.7123,
+      "step": 2935
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1977807176077028e-06,
+      "loss": 0.719,
+      "step": 2936
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1933487367562025e-06,
+      "loss": 0.7982,
+      "step": 2937
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1889244502821695e-06,
+      "loss": 0.6831,
+      "step": 2938
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1845078620511008e-06,
+      "loss": 0.8032,
+      "step": 2939
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1800989759217784e-06,
+      "loss": 0.6736,
+      "step": 2940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1756977957462457e-06,
+      "loss": 0.747,
+      "step": 2941
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1713043253698186e-06,
+      "loss": 0.689,
+      "step": 2942
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1669185686310713e-06,
+      "loss": 0.6898,
+      "step": 2943
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1625405293618453e-06,
+      "loss": 0.7173,
+      "step": 2944
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1581702113872396e-06,
+      "loss": 0.7548,
+      "step": 2945
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1538076185255987e-06,
+      "loss": 0.5945,
+      "step": 2946
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1494527545885281e-06,
+      "loss": 0.6461,
+      "step": 2947
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1451056233808711e-06,
+      "loss": 0.8052,
+      "step": 2948
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1407662287007237e-06,
+      "loss": 0.7178,
+      "step": 2949
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1364345743394157e-06,
+      "loss": 0.7203,
+      "step": 2950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.132110664081515e-06,
+      "loss": 0.7003,
+      "step": 2951
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1277945017048274e-06,
+      "loss": 0.7016,
+      "step": 2952
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1234860909803856e-06,
+      "loss": 0.717,
+      "step": 2953
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1191854356724518e-06,
+      "loss": 0.7718,
+      "step": 2954
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1148925395385135e-06,
+      "loss": 0.6972,
+      "step": 2955
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1106074063292726e-06,
+      "loss": 0.7966,
+      "step": 2956
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.106330039788659e-06,
+      "loss": 0.6712,
+      "step": 2957
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1020604436538052e-06,
+      "loss": 0.6541,
+      "step": 2958
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.097798621655064e-06,
+      "loss": 0.7491,
+      "step": 2959
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0935445775159926e-06,
+      "loss": 0.7678,
+      "step": 2960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0892983149533476e-06,
+      "loss": 0.8318,
+      "step": 2961
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.085059837677097e-06,
+      "loss": 0.739,
+      "step": 2962
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0808291493903978e-06,
+      "loss": 0.7282,
+      "step": 2963
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0766062537896072e-06,
+      "loss": 0.875,
+      "step": 2964
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0723911545642728e-06,
+      "loss": 0.7091,
+      "step": 2965
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.068183855397127e-06,
+      "loss": 0.7648,
+      "step": 2966
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0639843599640942e-06,
+      "loss": 0.7199,
+      "step": 2967
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0597926719342744e-06,
+      "loss": 0.7152,
+      "step": 2968
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0556087949699468e-06,
+      "loss": 0.7343,
+      "step": 2969
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0514327327265683e-06,
+      "loss": 0.7369,
+      "step": 2970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0472644888527683e-06,
+      "loss": 0.7518,
+      "step": 2971
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0431040669903458e-06,
+      "loss": 0.719,
+      "step": 2972
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.038951470774261e-06,
+      "loss": 0.8267,
+      "step": 2973
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0348067038326436e-06,
+      "loss": 0.7572,
+      "step": 2974
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0306697697867761e-06,
+      "loss": 0.6917,
+      "step": 2975
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.026540672251105e-06,
+      "loss": 0.6087,
+      "step": 2976
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0224194148332224e-06,
+      "loss": 0.7548,
+      "step": 2977
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.018306001133873e-06,
+      "loss": 0.7053,
+      "step": 2978
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0142004347469535e-06,
+      "loss": 0.7345,
+      "step": 2979
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0101027192594948e-06,
+      "loss": 0.692,
+      "step": 2980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.006012858251677e-06,
+      "loss": 0.7065,
+      "step": 2981
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0019308552968165e-06,
+      "loss": 0.6818,
+      "step": 2982
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.97856713961357e-07,
+      "loss": 0.7235,
+      "step": 2983
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.937904378048846e-07,
+      "loss": 0.785,
+      "step": 2984
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.897320303801018e-07,
+      "loss": 0.6751,
+      "step": 2985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.856814952328464e-07,
+      "loss": 0.7292,
+      "step": 2986
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.816388359020734e-07,
+      "loss": 0.7161,
+      "step": 2987
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.77604055919854e-07,
+      "loss": 0.7255,
+      "step": 2988
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.735771588113818e-07,
+      "loss": 0.7007,
+      "step": 2989
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.695581480949566e-07,
+      "loss": 0.8028,
+      "step": 2990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.655470272819946e-07,
+      "loss": 0.7381,
+      "step": 2991
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.615437998770105e-07,
+      "loss": 0.7164,
+      "step": 2992
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.575484693776294e-07,
+      "loss": 0.6808,
+      "step": 2993
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.535610392745764e-07,
+      "loss": 0.6888,
+      "step": 2994
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.495815130516695e-07,
+      "loss": 0.6979,
+      "step": 2995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.456098941858238e-07,
+      "loss": 0.7163,
+      "step": 2996
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.41646186147045e-07,
+      "loss": 0.7027,
+      "step": 2997
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.376903923984271e-07,
+      "loss": 0.703,
+      "step": 2998
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.337425163961533e-07,
+      "loss": 0.7022,
+      "step": 2999
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.298025615894824e-07,
+      "loss": 0.7286,
+      "step": 3000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.258705314207572e-07,
+      "loss": 0.7954,
+      "step": 3001
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.219464293253944e-07,
+      "loss": 0.7319,
+      "step": 3002
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.180302587318868e-07,
+      "loss": 0.7461,
+      "step": 3003
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.14122023061792e-07,
+      "loss": 0.6661,
+      "step": 3004
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.102217257297419e-07,
+      "loss": 0.7712,
+      "step": 3005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.063293701434273e-07,
+      "loss": 0.666,
+      "step": 3006
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.024449597036001e-07,
+      "loss": 0.7354,
+      "step": 3007
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.985684978040743e-07,
+      "loss": 0.7369,
+      "step": 3008
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.946999878317197e-07,
+      "loss": 0.6246,
+      "step": 3009
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.908394331664516e-07,
+      "loss": 0.6896,
+      "step": 3010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.869868371812451e-07,
+      "loss": 0.6939,
+      "step": 3011
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.8314220324211e-07,
+      "loss": 0.7446,
+      "step": 3012
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.793055347081103e-07,
+      "loss": 0.8044,
+      "step": 3013
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.754768349313436e-07,
+      "loss": 0.6939,
+      "step": 3014
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.716561072569474e-07,
+      "loss": 0.7522,
+      "step": 3015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.678433550230959e-07,
+      "loss": 0.7107,
+      "step": 3016
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.640385815609909e-07,
+      "loss": 0.6637,
+      "step": 3017
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.602417901948701e-07,
+      "loss": 0.6318,
+      "step": 3018
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.564529842419866e-07,
+      "loss": 0.7327,
+      "step": 3019
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.52672167012627e-07,
+      "loss": 0.7342,
+      "step": 3020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.488993418100955e-07,
+      "loss": 0.7536,
+      "step": 3021
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.451345119307097e-07,
+      "loss": 0.7222,
+      "step": 3022
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.413776806638041e-07,
+      "loss": 0.7468,
+      "step": 3023
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.37628851291723e-07,
+      "loss": 0.7296,
+      "step": 3024
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.338880270898242e-07,
+      "loss": 0.6954,
+      "step": 3025
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.301552113264688e-07,
+      "loss": 0.813,
+      "step": 3026
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.26430407263018e-07,
+      "loss": 0.7076,
+      "step": 3027
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.227136181538387e-07,
+      "loss": 0.8415,
+      "step": 3028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.190048472462897e-07,
+      "loss": 0.7034,
+      "step": 3029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.15304097780727e-07,
+      "loss": 0.755,
+      "step": 3030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.116113729905006e-07,
+      "loss": 0.7064,
+      "step": 3031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.079266761019455e-07,
+      "loss": 0.7281,
+      "step": 3032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.042500103343831e-07,
+      "loss": 0.697,
+      "step": 3033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.005813789001171e-07,
+      "loss": 0.7979,
+      "step": 3034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.969207850044359e-07,
+      "loss": 0.6905,
+      "step": 3035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.932682318456042e-07,
+      "loss": 0.726,
+      "step": 3036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.896237226148573e-07,
+      "loss": 0.6718,
+      "step": 3037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.85987260496408e-07,
+      "loss": 0.6968,
+      "step": 3038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.823588486674338e-07,
+      "loss": 0.7208,
+      "step": 3039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.787384902980821e-07,
+      "loss": 0.7552,
+      "step": 3040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.751261885514638e-07,
+      "loss": 0.749,
+      "step": 3041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.715219465836454e-07,
+      "loss": 0.7579,
+      "step": 3042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.679257675436591e-07,
+      "loss": 0.7269,
+      "step": 3043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.643376545734871e-07,
+      "loss": 0.6567,
+      "step": 3044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.607576108080694e-07,
+      "loss": 0.7264,
+      "step": 3045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.571856393752886e-07,
+      "loss": 0.7035,
+      "step": 3046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.536217433959802e-07,
+      "loss": 0.749,
+      "step": 3047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.500659259839249e-07,
+      "loss": 0.7598,
+      "step": 3048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.465181902458395e-07,
+      "loss": 0.6816,
+      "step": 3049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.429785392813871e-07,
+      "loss": 0.734,
+      "step": 3050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.394469761831602e-07,
+      "loss": 0.7459,
+      "step": 3051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.359235040366875e-07,
+      "loss": 0.7365,
+      "step": 3052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.324081259204319e-07,
+      "loss": 0.7213,
+      "step": 3053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.289008449057788e-07,
+      "loss": 0.6179,
+      "step": 3054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.25401664057046e-07,
+      "loss": 0.7638,
+      "step": 3055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.219105864314646e-07,
+      "loss": 0.7237,
+      "step": 3056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.184276150791969e-07,
+      "loss": 0.7929,
+      "step": 3057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.149527530433153e-07,
+      "loss": 0.725,
+      "step": 3058
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.114860033598103e-07,
+      "loss": 0.7133,
+      "step": 3059
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.080273690575823e-07,
+      "loss": 0.6568,
+      "step": 3060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.045768531584418e-07,
+      "loss": 0.6969,
+      "step": 3061
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.011344586771073e-07,
+      "loss": 0.6593,
+      "step": 3062
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.977001886212043e-07,
+      "loss": 0.7786,
+      "step": 3063
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.942740459912534e-07,
+      "loss": 0.6615,
+      "step": 3064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.908560337806802e-07,
+      "loss": 0.7288,
+      "step": 3065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.874461549758015e-07,
+      "loss": 0.7188,
+      "step": 3066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.840444125558354e-07,
+      "loss": 0.7956,
+      "step": 3067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.806508094928833e-07,
+      "loss": 0.7548,
+      "step": 3068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.772653487519377e-07,
+      "loss": 0.6847,
+      "step": 3069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.738880332908815e-07,
+      "loss": 0.6618,
+      "step": 3070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.705188660604745e-07,
+      "loss": 0.6652,
+      "step": 3071
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.67157850004363e-07,
+      "loss": 0.7081,
+      "step": 3072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.638049880590658e-07,
+      "loss": 0.7415,
+      "step": 3073
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.604602831539841e-07,
+      "loss": 0.7565,
+      "step": 3074
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.571237382113882e-07,
+      "loss": 0.7344,
+      "step": 3075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.537953561464183e-07,
+      "loss": 0.7258,
+      "step": 3076
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.504751398670872e-07,
+      "loss": 0.6662,
+      "step": 3077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.471630922742678e-07,
+      "loss": 0.7275,
+      "step": 3078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.438592162616963e-07,
+      "loss": 0.7462,
+      "step": 3079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.405635147159772e-07,
+      "loss": 0.6819,
+      "step": 3080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.372759905165615e-07,
+      "loss": 0.694,
+      "step": 3081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.339966465357661e-07,
+      "loss": 0.6766,
+      "step": 3082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.307254856387524e-07,
+      "loss": 0.6628,
+      "step": 3083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.274625106835375e-07,
+      "loss": 0.748,
+      "step": 3084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.242077245209877e-07,
+      "loss": 0.6659,
+      "step": 3085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.209611299948082e-07,
+      "loss": 0.7575,
+      "step": 3086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.177227299415534e-07,
+      "loss": 0.664,
+      "step": 3087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.144925271906122e-07,
+      "loss": 0.6558,
+      "step": 3088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.112705245642192e-07,
+      "loss": 0.7768,
+      "step": 3089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.080567248774372e-07,
+      "loss": 0.7421,
+      "step": 3090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.048511309381666e-07,
+      "loss": 0.6786,
+      "step": 3091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.016537455471383e-07,
+      "loss": 0.7237,
+      "step": 3092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.984645714979076e-07,
+      "loss": 0.7647,
+      "step": 3093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.952836115768612e-07,
+      "loss": 0.8151,
+      "step": 3094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.921108685632026e-07,
+      "loss": 0.7587,
+      "step": 3095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.889463452289623e-07,
+      "loss": 0.7258,
+      "step": 3096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.857900443389864e-07,
+      "loss": 0.7241,
+      "step": 3097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.826419686509354e-07,
+      "loss": 0.7583,
+      "step": 3098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.795021209152874e-07,
+      "loss": 0.6921,
+      "step": 3099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.763705038753276e-07,
+      "loss": 0.6854,
+      "step": 3100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.73247120267153e-07,
+      "loss": 0.7398,
+      "step": 3101
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.701319728196675e-07,
+      "loss": 0.7523,
+      "step": 3102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.670250642545739e-07,
+      "loss": 0.7321,
+      "step": 3103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.639263972863851e-07,
+      "loss": 0.7354,
+      "step": 3104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.608359746224057e-07,
+      "loss": 0.7031,
+      "step": 3105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.577537989627402e-07,
+      "loss": 0.6942,
+      "step": 3106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.54679873000289e-07,
+      "loss": 0.732,
+      "step": 3107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.516141994207414e-07,
+      "loss": 0.7727,
+      "step": 3108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.485567809025805e-07,
+      "loss": 0.6754,
+      "step": 3109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.455076201170739e-07,
+      "loss": 0.7694,
+      "step": 3110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.424667197282751e-07,
+      "loss": 0.793,
+      "step": 3111
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.394340823930244e-07,
+      "loss": 0.6128,
+      "step": 3112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.364097107609356e-07,
+      "loss": 0.64,
+      "step": 3113
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.333936074744062e-07,
+      "loss": 0.6684,
+      "step": 3114
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.303857751686048e-07,
+      "loss": 0.7183,
+      "step": 3115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.273862164714805e-07,
+      "loss": 0.6957,
+      "step": 3116
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.243949340037469e-07,
+      "loss": 0.6892,
+      "step": 3117
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.214119303788901e-07,
+      "loss": 0.7276,
+      "step": 3118
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.184372082031653e-07,
+      "loss": 0.7769,
+      "step": 3119
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.154707700755857e-07,
+      "loss": 0.6884,
+      "step": 3120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.125126185879336e-07,
+      "loss": 0.6848,
+      "step": 3121
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.095627563247452e-07,
+      "loss": 0.7395,
+      "step": 3122
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.066211858633196e-07,
+      "loss": 0.861,
+      "step": 3123
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.03687909773708e-07,
+      "loss": 0.7659,
+      "step": 3124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.007629306187145e-07,
+      "loss": 0.6177,
+      "step": 3125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.978462509538972e-07,
+      "loss": 0.7288,
+      "step": 3126
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9493787332756e-07,
+      "loss": 0.6441,
+      "step": 3127
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.920378002807558e-07,
+      "loss": 0.734,
+      "step": 3128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.891460343472798e-07,
+      "loss": 0.705,
+      "step": 3129
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8626257805367e-07,
+      "loss": 0.7399,
+      "step": 3130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.833874339192046e-07,
+      "loss": 0.6857,
+      "step": 3131
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.805206044558985e-07,
+      "loss": 0.782,
+      "step": 3132
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.776620921685027e-07,
+      "loss": 0.7473,
+      "step": 3133
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7481189955450256e-07,
+      "loss": 0.7506,
+      "step": 3134
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7197002910411073e-07,
+      "loss": 0.6841,
+      "step": 3135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6913648330027427e-07,
+      "loss": 0.6947,
+      "step": 3136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6631126461866183e-07,
+      "loss": 0.6832,
+      "step": 3137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6349437552767105e-07,
+      "loss": 0.6995,
+      "step": 3138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.606858184884178e-07,
+      "loss": 0.7468,
+      "step": 3139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5788559595474233e-07,
+      "loss": 0.7554,
+      "step": 3140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.550937103732012e-07,
+      "loss": 0.705,
+      "step": 3141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.523101641830663e-07,
+      "loss": 0.7221,
+      "step": 3142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4953495981632236e-07,
+      "loss": 0.72,
+      "step": 3143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4676809969766865e-07,
+      "loss": 0.6749,
+      "step": 3144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.440095862445115e-07,
+      "loss": 0.6887,
+      "step": 3145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.412594218669697e-07,
+      "loss": 0.7614,
+      "step": 3146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3851760896786e-07,
+      "loss": 0.7851,
+      "step": 3147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3578414994270846e-07,
+      "loss": 0.7216,
+      "step": 3148
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3305904717973803e-07,
+      "loss": 0.7164,
+      "step": 3149
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3034230305987635e-07,
+      "loss": 0.674,
+      "step": 3150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.276339199567425e-07,
+      "loss": 0.8094,
+      "step": 3151
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2493390023665147e-07,
+      "loss": 0.8124,
+      "step": 3152
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.222422462586151e-07,
+      "loss": 0.7194,
+      "step": 3153
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.195589603743311e-07,
+      "loss": 0.7282,
+      "step": 3154
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.168840449281897e-07,
+      "loss": 0.6682,
+      "step": 3155
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.142175022572659e-07,
+      "loss": 0.7323,
+      "step": 3156
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1155933469131937e-07,
+      "loss": 0.5992,
+      "step": 3157
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0890954455279574e-07,
+      "loss": 0.7395,
+      "step": 3158
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0626813415681534e-07,
+      "loss": 0.7736,
+      "step": 3159
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.036351058111809e-07,
+      "loss": 0.753,
+      "step": 3160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0101046181637236e-07,
+      "loss": 0.7383,
+      "step": 3161
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.98394204465542e-07,
+      "loss": 0.7368,
+      "step": 3162
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.957863360445158e-07,
+      "loss": 0.7629,
+      "step": 3163
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.931868588317911e-07,
+      "loss": 0.6751,
+      "step": 3164
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.905957750985323e-07,
+      "loss": 0.7128,
+      "step": 3165
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.880130871085708e-07,
+      "loss": 0.6879,
+      "step": 3166
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8543879711840257e-07,
+      "loss": 0.6471,
+      "step": 3167
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8287290737718975e-07,
+      "loss": 0.7536,
+      "step": 3168
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.803154201267489e-07,
+      "loss": 0.7224,
+      "step": 3169
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7776633760155944e-07,
+      "loss": 0.7931,
+      "step": 3170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.752256620287542e-07,
+      "loss": 0.7261,
+      "step": 3171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7269339562812646e-07,
+      "loss": 0.751,
+      "step": 3172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7016954061211975e-07,
+      "loss": 0.6552,
+      "step": 3173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.676540991858235e-07,
+      "loss": 0.7098,
+      "step": 3174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6514707354698644e-07,
+      "loss": 0.7125,
+      "step": 3175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.62648465885993e-07,
+      "loss": 0.763,
+      "step": 3176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6015827838588366e-07,
+      "loss": 0.6738,
+      "step": 3177
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5767651322233367e-07,
+      "loss": 0.7183,
+      "step": 3178
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.55203172563664e-07,
+      "loss": 0.7399,
+      "step": 3179
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.527382585708339e-07,
+      "loss": 0.7561,
+      "step": 3180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.502817733974395e-07,
+      "loss": 0.647,
+      "step": 3181
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.478337191897152e-07,
+      "loss": 0.7524,
+      "step": 3182
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.453940980865289e-07,
+      "loss": 0.7267,
+      "step": 3183
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4296291221937563e-07,
+      "loss": 0.7397,
+      "step": 3184
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.405401637123873e-07,
+      "loss": 0.7725,
+      "step": 3185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3812585468231963e-07,
+      "loss": 0.6713,
+      "step": 3186
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3571998723855634e-07,
+      "loss": 0.7221,
+      "step": 3187
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.333225634831072e-07,
+      "loss": 0.7691,
+      "step": 3188
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3093358551059887e-07,
+      "loss": 0.7187,
+      "step": 3189
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2855305540828717e-07,
+      "loss": 0.7186,
+      "step": 3190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2618097525604076e-07,
+      "loss": 0.704,
+      "step": 3191
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.238173471263495e-07,
+      "loss": 0.7421,
+      "step": 3192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2146217308431484e-07,
+      "loss": 0.7776,
+      "step": 3193
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1911545518765517e-07,
+      "loss": 0.7508,
+      "step": 3194
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1677719548670047e-07,
+      "loss": 0.656,
+      "step": 3195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1444739602438877e-07,
+      "loss": 0.6961,
+      "step": 3196
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1212605883626734e-07,
+      "loss": 0.8145,
+      "step": 3197
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0981318595049046e-07,
+      "loss": 0.7679,
+      "step": 3198
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.075087793878162e-07,
+      "loss": 0.7288,
+      "step": 3199
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.052128411616073e-07,
+      "loss": 0.7451,
+      "step": 3200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.029253732778259e-07,
+      "loss": 0.7459,
+      "step": 3201
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0064637773503547e-07,
+      "loss": 0.7393,
+      "step": 3202
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.983758565243944e-07,
+      "loss": 0.7645,
+      "step": 3203
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9611381162966024e-07,
+      "loss": 0.7862,
+      "step": 3204
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9386024502718214e-07,
+      "loss": 0.6242,
+      "step": 3205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9161515868590285e-07,
+      "loss": 0.7839,
+      "step": 3206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8937855456735665e-07,
+      "loss": 0.7129,
+      "step": 3207
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.871504346256637e-07,
+      "loss": 0.6867,
+      "step": 3208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.849308008075369e-07,
+      "loss": 0.7165,
+      "step": 3209
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.827196550522726e-07,
+      "loss": 0.7545,
+      "step": 3210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.805169992917467e-07,
+      "loss": 0.7004,
+      "step": 3211
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7832283545042415e-07,
+      "loss": 0.7181,
+      "step": 3212
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7613716544534596e-07,
+      "loss": 0.7006,
+      "step": 3213
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.739599911861346e-07,
+      "loss": 0.7536,
+      "step": 3214
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7179131457498844e-07,
+      "loss": 0.6969,
+      "step": 3215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6963113750668066e-07,
+      "loss": 0.7763,
+      "step": 3216
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.674794618685617e-07,
+      "loss": 0.6722,
+      "step": 3217
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.653362895405498e-07,
+      "loss": 0.7595,
+      "step": 3218
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6320162239513725e-07,
+      "loss": 0.7107,
+      "step": 3219
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6107546229738433e-07,
+      "loss": 0.7977,
+      "step": 3220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5895781110491735e-07,
+      "loss": 0.73,
+      "step": 3221
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.568486706679318e-07,
+      "loss": 0.7578,
+      "step": 3222
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5474804282918487e-07,
+      "loss": 0.7388,
+      "step": 3223
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.526559294239961e-07,
+      "loss": 0.7609,
+      "step": 3224
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.505723322802445e-07,
+      "loss": 0.7149,
+      "step": 3225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.484972532183716e-07,
+      "loss": 0.7521,
+      "step": 3226
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.464306940513772e-07,
+      "loss": 0.7354,
+      "step": 3227
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.443726565848137e-07,
+      "loss": 0.6773,
+      "step": 3228
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.423231426167916e-07,
+      "loss": 0.8461,
+      "step": 3229
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4028215393797073e-07,
+      "loss": 0.7245,
+      "step": 3230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.382496923315647e-07,
+      "loss": 0.7602,
+      "step": 3231
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3622575957333748e-07,
+      "loss": 0.7061,
+      "step": 3232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3421035743160125e-07,
+      "loss": 0.7192,
+      "step": 3233
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.32203487667213e-07,
+      "loss": 0.7636,
+      "step": 3234
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3020515203357574e-07,
+      "loss": 0.7253,
+      "step": 3235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2821535227663838e-07,
+      "loss": 0.7039,
+      "step": 3236
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2623409013488806e-07,
+      "loss": 0.7002,
+      "step": 3237
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2426136733935677e-07,
+      "loss": 0.7072,
+      "step": 3238
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2229718561361468e-07,
+      "loss": 0.7456,
+      "step": 3239
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2034154667376574e-07,
+      "loss": 0.7408,
+      "step": 3240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1839445222845645e-07,
+      "loss": 0.7329,
+      "step": 3241
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1645590397886273e-07,
+      "loss": 0.7985,
+      "step": 3242
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.145259036186953e-07,
+      "loss": 0.6585,
+      "step": 3243
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.126044528341975e-07,
+      "loss": 0.741,
+      "step": 3244
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1069155330414204e-07,
+      "loss": 0.6374,
+      "step": 3245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0878720669983088e-07,
+      "loss": 0.8122,
+      "step": 3246
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.06891414685092e-07,
+      "loss": 0.8156,
+      "step": 3247
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0500417891628155e-07,
+      "loss": 0.763,
+      "step": 3248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0312550104227947e-07,
+      "loss": 0.7731,
+      "step": 3249
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0125538270448608e-07,
+      "loss": 0.6481,
+      "step": 3250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.993938255368244e-07,
+      "loss": 0.712,
+      "step": 3251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.975408311657412e-07,
+      "loss": 0.6822,
+      "step": 3252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9569640121019695e-07,
+      "loss": 0.6864,
+      "step": 3253
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9386053728167152e-07,
+      "loss": 0.7133,
+      "step": 3254
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9203324098416076e-07,
+      "loss": 0.7228,
+      "step": 3255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9021451391417534e-07,
+      "loss": 0.7875,
+      "step": 3256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.884043576607375e-07,
+      "loss": 0.7189,
+      "step": 3257
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8660277380538327e-07,
+      "loss": 0.7324,
+      "step": 3258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8480976392215578e-07,
+      "loss": 0.6721,
+      "step": 3259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8302532957761076e-07,
+      "loss": 0.7057,
+      "step": 3260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8124947233081002e-07,
+      "loss": 0.74,
+      "step": 3261
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.794821937333191e-07,
+      "loss": 0.6698,
+      "step": 3262
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7772349532921286e-07,
+      "loss": 0.6468,
+      "step": 3263
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7597337865506548e-07,
+      "loss": 0.6939,
+      "step": 3264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7423184523995606e-07,
+      "loss": 0.6906,
+      "step": 3265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7249889660546416e-07,
+      "loss": 0.7493,
+      "step": 3266
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7077453426566749e-07,
+      "loss": 0.7629,
+      "step": 3267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6905875972714314e-07,
+      "loss": 0.7761,
+      "step": 3268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.673515744889642e-07,
+      "loss": 0.7263,
+      "step": 3269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.656529800426987e-07,
+      "loss": 0.7812,
+      "step": 3270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6396297787241166e-07,
+      "loss": 0.7374,
+      "step": 3271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6228156945465756e-07,
+      "loss": 0.742,
+      "step": 3272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606087562584857e-07,
+      "loss": 0.6797,
+      "step": 3273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5894453974543256e-07,
+      "loss": 0.7006,
+      "step": 3274
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5728892136952612e-07,
+      "loss": 0.7357,
+      "step": 3275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5564190257728151e-07,
+      "loss": 0.7527,
+      "step": 3276
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5400348480769988e-07,
+      "loss": 0.7132,
+      "step": 3277
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.523736694922684e-07,
+      "loss": 0.7309,
+      "step": 3278
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5075245805495798e-07,
+      "loss": 0.7147,
+      "step": 3279
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4913985191222115e-07,
+      "loss": 0.7177,
+      "step": 3280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4753585247299307e-07,
+      "loss": 0.6835,
+      "step": 3281
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4594046113869053e-07,
+      "loss": 0.6823,
+      "step": 3282
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4435367930320743e-07,
+      "loss": 0.7224,
+      "step": 3283
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4277550835291476e-07,
+      "loss": 0.712,
+      "step": 3284
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4120594966666512e-07,
+      "loss": 0.772,
+      "step": 3285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3964500461577934e-07,
+      "loss": 0.8012,
+      "step": 3286
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.380926745640576e-07,
+      "loss": 0.6669,
+      "step": 3287
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3654896086777281e-07,
+      "loss": 0.6657,
+      "step": 3288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3501386487566714e-07,
+      "loss": 0.7205,
+      "step": 3289
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3348738792895666e-07,
+      "loss": 0.7278,
+      "step": 3290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.319695313613234e-07,
+      "loss": 0.691,
+      "step": 3291
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3046029649891988e-07,
+      "loss": 0.7735,
+      "step": 3292
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2895968466036912e-07,
+      "loss": 0.7981,
+      "step": 3293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2746769715675234e-07,
+      "loss": 0.6819,
+      "step": 3294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2598433529162235e-07,
+      "loss": 0.7424,
+      "step": 3295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2450960036099247e-07,
+      "loss": 0.6638,
+      "step": 3296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.230434936533409e-07,
+      "loss": 0.7531,
+      "step": 3297
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2158601644960521e-07,
+      "loss": 0.6903,
+      "step": 3298
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2013717002318458e-07,
+      "loss": 0.705,
+      "step": 3299
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1869695563993755e-07,
+      "loss": 0.7124,
+      "step": 3300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1726537455817866e-07,
+      "loss": 0.7223,
+      "step": 3301
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1584242802868295e-07,
+      "loss": 0.7247,
+      "step": 3302
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1442811729468039e-07,
+      "loss": 0.7129,
+      "step": 3303
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1302244359185365e-07,
+      "loss": 0.8568,
+      "step": 3304
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1162540814834255e-07,
+      "loss": 0.7301,
+      "step": 3305
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1023701218473626e-07,
+      "loss": 0.7874,
+      "step": 3306
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0885725691407778e-07,
+      "loss": 0.7138,
+      "step": 3307
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.074861435418606e-07,
+      "loss": 0.7689,
+      "step": 3308
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0612367326602757e-07,
+      "loss": 0.6875,
+      "step": 3309
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0476984727696981e-07,
+      "loss": 0.7673,
+      "step": 3310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0342466675752561e-07,
+      "loss": 0.6618,
+      "step": 3311
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0208813288298036e-07,
+      "loss": 0.7621,
+      "step": 3312
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0076024682106444e-07,
+      "loss": 0.7813,
+      "step": 3313
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.944100973195425e-08,
+      "loss": 0.7714,
+      "step": 3314
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.813042276826779e-08,
+      "loss": 0.6475,
+      "step": 3315
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.682848707506354e-08,
+      "loss": 0.7269,
+      "step": 3316
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.553520378984604e-08,
+      "loss": 0.7451,
+      "step": 3317
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.425057404255589e-08,
+      "loss": 0.6779,
+      "step": 3318
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.297459895557747e-08,
+      "loss": 0.6669,
+      "step": 3319
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.170727964372905e-08,
+      "loss": 0.7448,
+      "step": 3320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.044861721426934e-08,
+      "loss": 0.8206,
+      "step": 3321
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.919861276689312e-08,
+      "loss": 0.622,
+      "step": 3322
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.79572673937279e-08,
+      "loss": 0.6485,
+      "step": 3323
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.672458217934166e-08,
+      "loss": 0.7592,
+      "step": 3324
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.550055820072955e-08,
+      "loss": 0.6937,
+      "step": 3325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.42851965273217e-08,
+      "loss": 0.7597,
+      "step": 3326
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.307849822098313e-08,
+      "loss": 0.6706,
+      "step": 3327
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.188046433600383e-08,
+      "loss": 0.718,
+      "step": 3328
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.069109591910984e-08,
+      "loss": 0.7451,
+      "step": 3329
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.951039400945215e-08,
+      "loss": 0.645,
+      "step": 3330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.833835963861002e-08,
+      "loss": 0.7818,
+      "step": 3331
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.717499383059213e-08,
+      "loss": 0.677,
+      "step": 3332
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.602029760183095e-08,
+      "loss": 0.7135,
+      "step": 3333
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.487427196118501e-08,
+      "loss": 0.6786,
+      "step": 3334
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.373691790993897e-08,
+      "loss": 0.6891,
+      "step": 3335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.260823644179793e-08,
+      "loss": 0.6447,
+      "step": 3336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.148822854289417e-08,
+      "loss": 0.7208,
+      "step": 3337
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.037689519177826e-08,
+      "loss": 0.727,
+      "step": 3338
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.927423735942352e-08,
+      "loss": 0.6814,
+      "step": 3339
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.818025600922151e-08,
+      "loss": 0.71,
+      "step": 3340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.709495209698546e-08,
+      "loss": 0.8063,
+      "step": 3341
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.601832657094575e-08,
+      "loss": 0.7662,
+      "step": 3342
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.495038037175106e-08,
+      "loss": 0.7395,
+      "step": 3343
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.38911144324672e-08,
+      "loss": 0.7401,
+      "step": 3344
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.284052967857501e-08,
+      "loss": 0.6827,
+      "step": 3345
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.179862702797135e-08,
+      "loss": 0.6728,
+      "step": 3346
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.076540739096804e-08,
+      "loss": 0.6628,
+      "step": 3347
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.974087167028741e-08,
+      "loss": 0.7525,
+      "step": 3348
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.872502076107012e-08,
+      "loss": 0.7325,
+      "step": 3349
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7717855550862846e-08,
+      "loss": 0.7424,
+      "step": 3350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6719376919629475e-08,
+      "loss": 0.6596,
+      "step": 3351
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5729585739739966e-08,
+      "loss": 0.7095,
+      "step": 3352
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.474848287597478e-08,
+      "loss": 0.6685,
+      "step": 3353
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.377606918552603e-08,
+      "loss": 0.7479,
+      "step": 3354
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.281234551799075e-08,
+      "loss": 0.6957,
+      "step": 3355
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.185731271537653e-08,
+      "loss": 0.7081,
+      "step": 3356
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.091097161209369e-08,
+      "loss": 0.7067,
+      "step": 3357
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.997332303496416e-08,
+      "loss": 0.7134,
+      "step": 3358
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.904436780321043e-08,
+      "loss": 0.6926,
+      "step": 3359
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8124106728461016e-08,
+      "loss": 0.7712,
+      "step": 3360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.721254061474945e-08,
+      "loss": 0.7371,
+      "step": 3361
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6309670258511964e-08,
+      "loss": 0.7305,
+      "step": 3362
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.541549644858645e-08,
+      "loss": 0.7221,
+      "step": 3363
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.453001996621242e-08,
+      "loss": 0.6983,
+      "step": 3364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.365324158503326e-08,
+      "loss": 0.6887,
+      "step": 3365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.278516207108952e-08,
+      "loss": 0.7409,
+      "step": 3366
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.192578218282339e-08,
+      "loss": 0.7321,
+      "step": 3367
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.107510267107429e-08,
+      "loss": 0.7089,
+      "step": 3368
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0233124279084327e-08,
+      "loss": 0.73,
+      "step": 3369
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.93998477424884e-08,
+      "loss": 0.7413,
+      "step": 3370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.857527378932191e-08,
+      "loss": 0.7239,
+      "step": 3371
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7759403140014136e-08,
+      "loss": 0.7006,
+      "step": 3372
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6952236507392655e-08,
+      "loss": 0.7979,
+      "step": 3373
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6153774596681125e-08,
+      "loss": 0.6666,
+      "step": 3374
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5364018105494834e-08,
+      "loss": 0.7485,
+      "step": 3375
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.458296772384406e-08,
+      "loss": 0.741,
+      "step": 3376
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.381062413413405e-08,
+      "loss": 0.7318,
+      "step": 3377
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3046988011161685e-08,
+      "loss": 0.7234,
+      "step": 3378
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.229206002211549e-08,
+      "loss": 0.8022,
+      "step": 3379
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1545840826577855e-08,
+      "loss": 0.639,
+      "step": 3380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.08083310765217e-08,
+      "loss": 0.7551,
+      "step": 3381
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.007953141630715e-08,
+      "loss": 0.7142,
+      "step": 3382
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9359442482689292e-08,
+      "loss": 0.7426,
+      "step": 3383
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8648064904809313e-08,
+      "loss": 0.7035,
+      "step": 3384
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7945399304198927e-08,
+      "loss": 0.73,
+      "step": 3385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.725144629477705e-08,
+      "loss": 0.799,
+      "step": 3386
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.656620648285091e-08,
+      "loss": 0.7377,
+      "step": 3387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.588968046711604e-08,
+      "loss": 0.6757,
+      "step": 3388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5221868838651854e-08,
+      "loss": 0.7212,
+      "step": 3389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4562772180927174e-08,
+      "loss": 0.7258,
+      "step": 3390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3912391069794706e-08,
+      "loss": 0.7406,
+      "step": 3391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3270726073493233e-08,
+      "loss": 0.7622,
+      "step": 3392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.263777775264653e-08,
+      "loss": 0.646,
+      "step": 3393
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.201354666026001e-08,
+      "loss": 0.7029,
+      "step": 3394
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.13980333417263e-08,
+      "loss": 0.7733,
+      "step": 3395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0791238334820772e-08,
+      "loss": 0.7394,
+      "step": 3396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0193162169698243e-08,
+      "loss": 0.8152,
+      "step": 3397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.960380536889961e-08,
+      "loss": 0.7035,
+      "step": 3398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9023168447346308e-08,
+      "loss": 0.8172,
+      "step": 3399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.845125191234143e-08,
+      "loss": 0.7224,
+      "step": 3400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7888056263568598e-08,
+      "loss": 0.751,
+      "step": 3401
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7333581993090875e-08,
+      "loss": 0.7464,
+      "step": 3402
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6787829585355187e-08,
+      "loss": 0.6889,
+      "step": 3403
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6250799517185668e-08,
+      "loss": 0.7437,
+      "step": 3404
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5722492257783663e-08,
+      "loss": 0.72,
+      "step": 3405
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5202908268733274e-08,
+      "loss": 0.7417,
+      "step": 3406
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4692048003994707e-08,
+      "loss": 0.712,
+      "step": 3407
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4189911909907594e-08,
+      "loss": 0.686,
+      "step": 3408
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3696500425188774e-08,
+      "loss": 0.6831,
+      "step": 3409
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3211813980931187e-08,
+      "loss": 0.725,
+      "step": 3410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.273585300060609e-08,
+      "loss": 0.7195,
+      "step": 3411
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2268617900060842e-08,
+      "loss": 0.731,
+      "step": 3412
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1810109087520005e-08,
+      "loss": 0.7512,
+      "step": 3413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1360326963582024e-08,
+      "loss": 0.7349,
+      "step": 3414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0919271921222552e-08,
+      "loss": 0.7635,
+      "step": 3415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0486944345791118e-08,
+      "loss": 0.6876,
+      "step": 3416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0063344615013349e-08,
+      "loss": 0.6837,
+      "step": 3417
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.648473098987643e-09,
+      "loss": 0.6999,
+      "step": 3418
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.242330160187386e-09,
+      "loss": 0.736,
+      "step": 3419
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.844916153462058e-09,
+      "loss": 0.7738,
+      "step": 3420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.45623142602947e-09,
+      "loss": 0.7326,
+      "step": 3421
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.076276317486864e-09,
+      "loss": 0.6245,
+      "step": 3422
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.705051159798693e-09,
+      "loss": 0.7724,
+      "step": 3423
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.3425562773066225e-09,
+      "loss": 0.7027,
+      "step": 3424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.988791986721755e-09,
+      "loss": 0.6808,
+      "step": 3425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.643758597127958e-09,
+      "loss": 0.7218,
+      "step": 3426
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.307456409980761e-09,
+      "loss": 0.6734,
+      "step": 3427
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.979885719109568e-09,
+      "loss": 0.7672,
+      "step": 3428
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6610468107110015e-09,
+      "loss": 0.7183,
+      "step": 3429
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3509399633566716e-09,
+      "loss": 0.7518,
+      "step": 3430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.049565447985405e-09,
+      "loss": 0.7186,
+      "step": 3431
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.756923527908796e-09,
+      "loss": 0.6553,
+      "step": 3432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.473014458808988e-09,
+      "loss": 0.7196,
+      "step": 3433
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.19783848873756e-09,
+      "loss": 0.7502,
+      "step": 3434
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.931395858115528e-09,
+      "loss": 0.7801,
+      "step": 3435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.673686799734455e-09,
+      "loss": 0.7738,
+      "step": 3436
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4247115387531225e-09,
+      "loss": 0.7413,
+      "step": 3437
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.184470292704189e-09,
+      "loss": 0.6728,
+      "step": 3438
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9529632714841994e-09,
+      "loss": 0.6881,
+      "step": 3439
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7301906773624653e-09,
+      "loss": 0.7015,
+      "step": 3440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5161527049755165e-09,
+      "loss": 0.6981,
+      "step": 3441
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.310849541327098e-09,
+      "loss": 0.7656,
+      "step": 3442
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.114281365792614e-09,
+      "loss": 0.7733,
+      "step": 3443
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9264483501124643e-09,
+      "loss": 0.7569,
+      "step": 3444
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.747350658396485e-09,
+      "loss": 0.7127,
+      "step": 3445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5769884471239506e-09,
+      "loss": 0.7679,
+      "step": 3446
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4153618651380208e-09,
+      "loss": 0.6973,
+      "step": 3447
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2624710536546235e-09,
+      "loss": 0.657,
+      "step": 3448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1183161462524627e-09,
+      "loss": 0.7399,
+      "step": 3449
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.82897268880789e-10,
+      "loss": 0.7254,
+      "step": 3450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.562145398549604e-10,
+      "loss": 0.7558,
+      "step": 3451
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.382680698586608e-10,
+      "loss": 0.7421,
+      "step": 3452
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.29057961939461e-10,
+      "loss": 0.7144,
+      "step": 3453
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.285843115165889e-10,
+      "loss": 0.6687,
+      "step": 3454
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.368472063731588e-10,
+      "loss": 0.6591,
+      "step": 3455
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.538467266595014e-10,
+      "loss": 0.7561,
+      "step": 3456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.795829448942744e-10,
+      "loss": 0.7579,
+      "step": 3457
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.140559259611319e-10,
+      "loss": 0.6991,
+      "step": 3458
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5726572710983435e-10,
+      "loss": 0.7396,
+      "step": 3459
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.092123979606896e-10,
+      "loss": 0.7027,
+      "step": 3460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.989598049567115e-11,
+      "loss": 0.7137,
+      "step": 3461
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.931650906729978e-11,
+      "loss": 0.7116,
+      "step": 3462
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7474010390872153e-11,
+      "loss": 0.6856,
+      "step": 3463
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.368503552232284e-12,
+      "loss": 0.724,
+      "step": 3464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.7079,
+      "step": 3465
+    },
+    {
+      "epoch": 1.0,
+      "step": 3465,
+      "total_flos": 4154536234188800.0,
+      "train_loss": 0.7781218005465222,
+      "train_runtime": 86131.649,
+      "train_samples_per_second": 7.724,
+      "train_steps_per_second": 0.04
+    }
+  ],
+  "max_steps": 3465,
+  "num_train_epochs": 1,
+  "total_flos": 4154536234188800.0,
+  "trial_name": null,
+  "trial_params": null
+}