diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,41448 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 6903,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8076923076923085e-08,
+      "loss": 1.2865,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.615384615384617e-08,
+      "loss": 0.6927,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4423076923076925e-07,
+      "loss": 1.2671,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769234e-07,
+      "loss": 1.2531,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.403846153846154e-07,
+      "loss": 1.2992,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.884615384615385e-07,
+      "loss": 0.6473,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3653846153846154e-07,
+      "loss": 0.6971,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 0.6821,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.326923076923077e-07,
+      "loss": 1.2583,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.807692307692308e-07,
+      "loss": 1.2588,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.288461538461539e-07,
+      "loss": 1.2489,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.76923076923077e-07,
+      "loss": 1.2285,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.25e-07,
+      "loss": 0.677,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.730769230769231e-07,
+      "loss": 1.2066,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.211538461538461e-07,
+      "loss": 1.1914,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 1.2039,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.173076923076924e-07,
+      "loss": 0.6431,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.653846153846154e-07,
+      "loss": 1.1703,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.134615384615385e-07,
+      "loss": 1.1931,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.615384615384617e-07,
+      "loss": 0.6705,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0096153846153846e-06,
+      "loss": 1.1293,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0576923076923078e-06,
+      "loss": 1.1565,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.105769230769231e-06,
+      "loss": 1.1435,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 1.1128,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.201923076923077e-06,
+      "loss": 1.097,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25e-06,
+      "loss": 1.1348,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2980769230769232e-06,
+      "loss": 0.6554,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3461538461538462e-06,
+      "loss": 0.6657,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3942307692307693e-06,
+      "loss": 1.0678,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4423076923076922e-06,
+      "loss": 1.1178,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4903846153846156e-06,
+      "loss": 0.6571,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 1.0499,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5865384615384616e-06,
+      "loss": 1.0547,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6346153846153848e-06,
+      "loss": 0.6306,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.682692307692308e-06,
+      "loss": 1.049,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7307692307692308e-06,
+      "loss": 1.0765,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.778846153846154e-06,
+      "loss": 1.0382,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.826923076923077e-06,
+      "loss": 0.6383,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8750000000000003e-06,
+      "loss": 1.0101,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 1.0182,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.971153846153846e-06,
+      "loss": 0.6319,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0192307692307692e-06,
+      "loss": 1.0144,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0673076923076924e-06,
+      "loss": 0.9988,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1153846153846155e-06,
+      "loss": 1.0072,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1634615384615387e-06,
+      "loss": 1.0102,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.211538461538462e-06,
+      "loss": 1.0067,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.259615384615385e-06,
+      "loss": 0.9774,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 0.9555,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.355769230769231e-06,
+      "loss": 1.0132,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.403846153846154e-06,
+      "loss": 0.9644,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.451923076923077e-06,
+      "loss": 1.0013,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 0.9386,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.548076923076923e-06,
+      "loss": 0.9642,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5961538461538465e-06,
+      "loss": 0.9723,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6442307692307696e-06,
+      "loss": 0.9605,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 0.9811,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.740384615384616e-06,
+      "loss": 0.9593,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7884615384615386e-06,
+      "loss": 0.9328,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8365384615384617e-06,
+      "loss": 0.9756,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8846153846153845e-06,
+      "loss": 0.9546,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.932692307692308e-06,
+      "loss": 0.6309,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.980769230769231e-06,
+      "loss": 0.6365,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.028846153846154e-06,
+      "loss": 0.8978,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 0.9226,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.125e-06,
+      "loss": 0.6108,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1730769230769233e-06,
+      "loss": 0.6,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2211538461538464e-06,
+      "loss": 0.9496,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2692307692307696e-06,
+      "loss": 0.8992,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3173076923076923e-06,
+      "loss": 0.9127,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.365384615384616e-06,
+      "loss": 0.9103,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.413461538461539e-06,
+      "loss": 0.8815,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 0.9337,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.509615384615385e-06,
+      "loss": 0.9139,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.557692307692308e-06,
+      "loss": 0.9142,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.605769230769231e-06,
+      "loss": 0.6139,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.653846153846154e-06,
+      "loss": 0.8844,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7019230769230774e-06,
+      "loss": 0.6127,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 0.9222,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7980769230769232e-06,
+      "loss": 0.8908,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 0.8996,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.89423076923077e-06,
+      "loss": 0.9007,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.942307692307692e-06,
+      "loss": 0.9165,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.990384615384616e-06,
+      "loss": 0.9027,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0384615384615385e-06,
+      "loss": 0.8982,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.086538461538462e-06,
+      "loss": 0.5874,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.134615384615385e-06,
+      "loss": 0.6027,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.182692307692308e-06,
+      "loss": 0.8775,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 0.6411,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.278846153846154e-06,
+      "loss": 0.8891,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.326923076923077e-06,
+      "loss": 0.8557,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3750000000000005e-06,
+      "loss": 0.8653,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.423076923076924e-06,
+      "loss": 0.8681,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.471153846153847e-06,
+      "loss": 0.8881,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.51923076923077e-06,
+      "loss": 0.8832,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.567307692307692e-06,
+      "loss": 0.8901,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 0.6348,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.663461538461539e-06,
+      "loss": 0.8811,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.711538461538462e-06,
+      "loss": 0.9004,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.759615384615385e-06,
+      "loss": 0.8993,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.807692307692308e-06,
+      "loss": 0.8805,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.855769230769231e-06,
+      "loss": 0.5888,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.903846153846154e-06,
+      "loss": 0.8876,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.951923076923077e-06,
+      "loss": 0.9055,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 0.8315,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0480769230769235e-06,
+      "loss": 0.5912,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.096153846153846e-06,
+      "loss": 0.8347,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.144230769230769e-06,
+      "loss": 0.8864,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.192307692307693e-06,
+      "loss": 0.8978,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.240384615384616e-06,
+      "loss": 0.5963,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.288461538461539e-06,
+      "loss": 0.8436,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3365384615384615e-06,
+      "loss": 0.8323,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 0.8537,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.432692307692308e-06,
+      "loss": 0.8301,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.480769230769232e-06,
+      "loss": 0.594,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.528846153846155e-06,
+      "loss": 0.8533,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.576923076923077e-06,
+      "loss": 0.8388,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.625e-06,
+      "loss": 0.8411,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6730769230769235e-06,
+      "loss": 0.8532,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.721153846153847e-06,
+      "loss": 0.8552,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.769230769230769e-06,
+      "loss": 0.8475,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.817307692307693e-06,
+      "loss": 0.859,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.865384615384616e-06,
+      "loss": 0.8465,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.913461538461539e-06,
+      "loss": 0.842,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.961538461538462e-06,
+      "loss": 0.6058,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.009615384615385e-06,
+      "loss": 0.5746,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.057692307692308e-06,
+      "loss": 0.8528,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.105769230769232e-06,
+      "loss": 0.8313,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 0.6043,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.201923076923078e-06,
+      "loss": 0.8382,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.25e-06,
+      "loss": 0.8577,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.2980769230769234e-06,
+      "loss": 0.8678,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.3461538461538466e-06,
+      "loss": 0.8565,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.394230769230769e-06,
+      "loss": 0.6022,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.442307692307693e-06,
+      "loss": 0.5738,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.490384615384616e-06,
+      "loss": 0.8663,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 0.8429,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.586538461538462e-06,
+      "loss": 0.8529,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.6346153846153846e-06,
+      "loss": 0.8653,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.682692307692308e-06,
+      "loss": 0.8183,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.730769230769232e-06,
+      "loss": 0.6083,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.778846153846155e-06,
+      "loss": 0.5997,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.826923076923078e-06,
+      "loss": 0.8031,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.875e-06,
+      "loss": 0.8786,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 0.8297,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.9711538461538465e-06,
+      "loss": 0.8473,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.01923076923077e-06,
+      "loss": 0.8333,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.067307692307694e-06,
+      "loss": 0.8148,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.115384615384616e-06,
+      "loss": 0.8482,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.163461538461539e-06,
+      "loss": 0.8428,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.211538461538462e-06,
+      "loss": 0.791,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.259615384615385e-06,
+      "loss": 0.8283,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 0.8189,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.355769230769232e-06,
+      "loss": 0.831,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.403846153846155e-06,
+      "loss": 0.6177,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.451923076923078e-06,
+      "loss": 0.8222,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.8311,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.548076923076923e-06,
+      "loss": 0.8183,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.5961538461538465e-06,
+      "loss": 0.8362,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.64423076923077e-06,
+      "loss": 0.8752,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.8119,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.740384615384616e-06,
+      "loss": 0.6069,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.78846153846154e-06,
+      "loss": 0.8509,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.836538461538462e-06,
+      "loss": 0.8141,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.884615384615384e-06,
+      "loss": 0.788,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.932692307692308e-06,
+      "loss": 0.6079,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.980769230769232e-06,
+      "loss": 0.8277,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.028846153846155e-06,
+      "loss": 0.8459,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 0.8203,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.125000000000001e-06,
+      "loss": 0.8206,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.173076923076923e-06,
+      "loss": 0.8185,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.221153846153847e-06,
+      "loss": 0.7894,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.26923076923077e-06,
+      "loss": 0.8119,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.317307692307694e-06,
+      "loss": 0.8108,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.365384615384616e-06,
+      "loss": 0.8162,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.41346153846154e-06,
+      "loss": 0.8229,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 0.6136,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.509615384615384e-06,
+      "loss": 0.6121,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.557692307692308e-06,
+      "loss": 0.5952,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.605769230769232e-06,
+      "loss": 0.8268,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.653846153846155e-06,
+      "loss": 0.6194,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.701923076923079e-06,
+      "loss": 0.8149,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 0.8253,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.798076923076923e-06,
+      "loss": 0.8295,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 0.8356,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.89423076923077e-06,
+      "loss": 0.7988,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.942307692307693e-06,
+      "loss": 0.8131,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.990384615384616e-06,
+      "loss": 0.7987,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.03846153846154e-06,
+      "loss": 0.6282,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.086538461538462e-06,
+      "loss": 0.5655,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.134615384615384e-06,
+      "loss": 0.8487,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.182692307692308e-06,
+      "loss": 0.8289,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.8199,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.278846153846155e-06,
+      "loss": 0.8172,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.326923076923079e-06,
+      "loss": 0.7943,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 0.8059,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.423076923076923e-06,
+      "loss": 0.8256,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.471153846153847e-06,
+      "loss": 0.7986,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.51923076923077e-06,
+      "loss": 0.7901,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.567307692307693e-06,
+      "loss": 0.8108,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 0.7864,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.66346153846154e-06,
+      "loss": 0.7955,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.711538461538462e-06,
+      "loss": 0.8364,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.759615384615386e-06,
+      "loss": 0.8078,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.807692307692308e-06,
+      "loss": 0.8215,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.855769230769232e-06,
+      "loss": 0.8271,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.903846153846155e-06,
+      "loss": 0.8157,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.951923076923079e-06,
+      "loss": 0.6192,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.8332,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999449523754e-06,
+      "loss": 0.7937,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999997798095137e-06,
+      "loss": 0.8266,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999995045714511e-06,
+      "loss": 0.6012,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999991192382485e-06,
+      "loss": 0.8494,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999986238099904e-06,
+      "loss": 0.8352,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999980182867862e-06,
+      "loss": 0.7904,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999973026687688e-06,
+      "loss": 0.8297,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999964769560963e-06,
+      "loss": 0.8076,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999955411489503e-06,
+      "loss": 0.8025,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999944952475369e-06,
+      "loss": 0.7836,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999933392520862e-06,
+      "loss": 0.8231,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999920731628529e-06,
+      "loss": 0.6259,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999906969801156e-06,
+      "loss": 0.8111,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999892107041778e-06,
+      "loss": 0.8168,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999876143353662e-06,
+      "loss": 0.8114,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999859078740325e-06,
+      "loss": 0.8054,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999840913205527e-06,
+      "loss": 0.8177,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999821646753265e-06,
+      "loss": 0.6086,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999801279387781e-06,
+      "loss": 0.7941,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999779811113564e-06,
+      "loss": 0.7854,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999757241935335e-06,
+      "loss": 0.817,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999733571858066e-06,
+      "loss": 0.7998,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999708800886972e-06,
+      "loss": 0.7968,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999682929027503e-06,
+      "loss": 0.8076,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999655956285356e-06,
+      "loss": 0.7767,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999627882666474e-06,
+      "loss": 0.7882,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999598708177035e-06,
+      "loss": 0.7973,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999568432823464e-06,
+      "loss": 0.7909,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999537056612427e-06,
+      "loss": 0.7837,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999504579550834e-06,
+      "loss": 0.7987,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999471001645835e-06,
+      "loss": 0.7915,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999436322904822e-06,
+      "loss": 0.7979,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999400543335436e-06,
+      "loss": 0.781,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99936366294555e-06,
+      "loss": 0.8253,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999325681743288e-06,
+      "loss": 0.7941,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99928659973701e-06,
+      "loss": 0.7805,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999246416935326e-06,
+      "loss": 0.7855,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999205133347078e-06,
+      "loss": 0.6155,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999162748981362e-06,
+      "loss": 0.7736,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999119263847508e-06,
+      "loss": 0.7829,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999074677955089e-06,
+      "loss": 0.7972,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999028991313926e-06,
+      "loss": 0.6072,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998982203934078e-06,
+      "loss": 0.5773,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998934315825846e-06,
+      "loss": 0.7955,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998885326999774e-06,
+      "loss": 0.7807,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998835237466651e-06,
+      "loss": 0.5545,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998784047237504e-06,
+      "loss": 0.8289,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998731756323606e-06,
+      "loss": 0.8253,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998678364736471e-06,
+      "loss": 0.8155,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998623872487853e-06,
+      "loss": 0.6112,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998568279589755e-06,
+      "loss": 0.845,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998511586054415e-06,
+      "loss": 0.7642,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998453791894317e-06,
+      "loss": 0.5943,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998394897122186e-06,
+      "loss": 0.8449,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998334901750991e-06,
+      "loss": 0.8065,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998273805793944e-06,
+      "loss": 0.5688,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998211609264494e-06,
+      "loss": 0.7829,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998148312176339e-06,
+      "loss": 0.6332,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998083914543413e-06,
+      "loss": 0.7895,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998018416379902e-06,
+      "loss": 0.7686,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997951817700222e-06,
+      "loss": 0.8055,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99788411851904e-06,
+      "loss": 0.7958,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997815318851262e-06,
+      "loss": 0.8242,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997745418712037e-06,
+      "loss": 0.7714,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997674418116759e-06,
+      "loss": 0.78,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997602317081057e-06,
+      "loss": 0.749,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99752911562081e-06,
+      "loss": 0.8043,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997454813752136e-06,
+      "loss": 0.5991,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997379411491397e-06,
+      "loss": 0.6054,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997302908855192e-06,
+      "loss": 0.7986,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99722530586037e-06,
+      "loss": 0.6007,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997146602524014e-06,
+      "loss": 0.5672,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997066798863457e-06,
+      "loss": 0.5866,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996985894896272e-06,
+      "loss": 0.7919,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99690389064027e-06,
+      "loss": 0.7932,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996820786113511e-06,
+      "loss": 0.8039,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996736581334293e-06,
+      "loss": 0.7878,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996651276321153e-06,
+      "loss": 0.6208,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996564871092879e-06,
+      "loss": 0.7921,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996477365668496e-06,
+      "loss": 0.5852,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996388760067268e-06,
+      "loss": 0.6007,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996299054308712e-06,
+      "loss": 0.7969,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996208248412574e-06,
+      "loss": 0.7964,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996116342398852e-06,
+      "loss": 0.7923,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99602333628778e-06,
+      "loss": 0.594,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99592923009984e-06,
+      "loss": 0.575,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995834023855754e-06,
+      "loss": 0.5853,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995737717576483e-06,
+      "loss": 0.5889,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995640311283233e-06,
+      "loss": 0.6088,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995541804997453e-06,
+      "loss": 0.5973,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995442198740832e-06,
+      "loss": 0.8554,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995341492535303e-06,
+      "loss": 0.7696,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99523968640304e-06,
+      "loss": 0.5865,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99513678036646e-06,
+      "loss": 0.8255,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995032774448223e-06,
+      "loss": 0.8261,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994927668671231e-06,
+      "loss": 0.7939,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994821463058623e-06,
+      "loss": 0.7634,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994714157633786e-06,
+      "loss": 0.7854,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994605752420352e-06,
+      "loss": 0.7781,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994496247442184e-06,
+      "loss": 0.7552,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9943856427234e-06,
+      "loss": 0.7602,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994273938288351e-06,
+      "loss": 0.5842,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994161134161635e-06,
+      "loss": 0.7897,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994047230368087e-06,
+      "loss": 0.7893,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99393222693279e-06,
+      "loss": 0.728,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993816123881067e-06,
+      "loss": 0.7448,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993698921238482e-06,
+      "loss": 0.8099,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993580619030842e-06,
+      "loss": 0.598,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993461217284195e-06,
+      "loss": 0.7673,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993340716024835e-06,
+      "loss": 0.8036,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993219115279292e-06,
+      "loss": 0.7739,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993096415074344e-06,
+      "loss": 0.5925,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992972615437006e-06,
+      "loss": 0.7693,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99284771639454e-06,
+      "loss": 0.7858,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992721717974444e-06,
+      "loss": 0.736,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992594620204468e-06,
+      "loss": 0.7525,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99246642311259e-06,
+      "loss": 0.757,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992337126727042e-06,
+      "loss": 0.7635,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992206731076295e-06,
+      "loss": 0.7746,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992075236189058e-06,
+      "loss": 0.7567,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991942642094286e-06,
+      "loss": 0.7851,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991808948821176e-06,
+      "loss": 0.7815,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991674156399165e-06,
+      "loss": 0.5923,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991538264857932e-06,
+      "loss": 0.7811,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9914012742274e-06,
+      "loss": 0.7619,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991263184537734e-06,
+      "loss": 0.7777,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991123995819339e-06,
+      "loss": 0.7849,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990983708102862e-06,
+      "loss": 0.794,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990842321419196e-06,
+      "loss": 0.789,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99069983579947e-06,
+      "loss": 0.8071,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990556251275058e-06,
+      "loss": 0.8002,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99041156787758e-06,
+      "loss": 0.5861,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99026578563889e-06,
+      "loss": 0.78,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990118904591088e-06,
+      "loss": 0.7988,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989970924766518e-06,
+      "loss": 0.5785,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989821846197762e-06,
+      "loss": 0.585,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989671668917645e-06,
+      "loss": 0.8159,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989520392959238e-06,
+      "loss": 0.7785,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989368018355846e-06,
+      "loss": 0.758,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989214545141024e-06,
+      "loss": 0.803,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989059973348562e-06,
+      "loss": 0.7769,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9889043030125e-06,
+      "loss": 0.7314,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988747534167112e-06,
+      "loss": 0.7561,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988589666846918e-06,
+      "loss": 0.7585,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988430701086676e-06,
+      "loss": 0.7795,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988270636921393e-06,
+      "loss": 0.7801,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988109474386311e-06,
+      "loss": 0.8177,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987947213516918e-06,
+      "loss": 0.7846,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98778385434894e-06,
+      "loss": 0.6023,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98761939691835e-06,
+      "loss": 0.5806,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987453841261357e-06,
+      "loss": 0.7896,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987287187414418e-06,
+      "loss": 0.7968,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987119435414225e-06,
+      "loss": 0.758,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98695058529772e-06,
+      "loss": 0.7555,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986780637102077e-06,
+      "loss": 0.7512,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98660959086472e-06,
+      "loss": 0.6194,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986437446623312e-06,
+      "loss": 0.743,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986264204415757e-06,
+      "loss": 0.8134,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986089864280199e-06,
+      "loss": 0.7951,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98591442625503e-06,
+      "loss": 0.764,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985737890378879e-06,
+      "loss": 0.7881,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985560256690614e-06,
+      "loss": 0.7596,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985381525229351e-06,
+      "loss": 0.7809,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985201696034445e-06,
+      "loss": 0.5608,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985020769145494e-06,
+      "loss": 0.7837,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984838744602332e-06,
+      "loss": 0.7713,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984655622445044e-06,
+      "loss": 0.7642,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984471402713949e-06,
+      "loss": 0.7891,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98428608544961e-06,
+      "loss": 0.7454,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984099670692835e-06,
+      "loss": 0.6012,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983912158484668e-06,
+      "loss": 0.7589,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983723548866398e-06,
+      "loss": 0.8058,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983533841879554e-06,
+      "loss": 0.7915,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983343037565909e-06,
+      "loss": 0.8119,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983151135967478e-06,
+      "loss": 0.7815,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982958137126512e-06,
+      "loss": 0.7817,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982764041085513e-06,
+      "loss": 0.7608,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982568847887213e-06,
+      "loss": 0.7457,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982372557574596e-06,
+      "loss": 0.777,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982175170190882e-06,
+      "loss": 0.7834,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98197668577953e-06,
+      "loss": 0.7476,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981777104384251e-06,
+      "loss": 0.7617,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981576426048989e-06,
+      "loss": 0.5841,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981374650817927e-06,
+      "loss": 0.7812,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981171778735501e-06,
+      "loss": 0.7699,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980967809846376e-06,
+      "loss": 0.7543,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980762744195465e-06,
+      "loss": 0.7183,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980556581827923e-06,
+      "loss": 0.7836,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980349322789144e-06,
+      "loss": 0.772,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980140967124766e-06,
+      "loss": 0.789,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979931514880665e-06,
+      "loss": 0.756,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979720966102962e-06,
+      "loss": 0.5828,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979509320838014e-06,
+      "loss": 0.7909,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979296579132428e-06,
+      "loss": 0.7291,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979082741033047e-06,
+      "loss": 0.7814,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978867806586955e-06,
+      "loss": 0.6055,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978651775841477e-06,
+      "loss": 0.7774,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978434648844184e-06,
+      "loss": 0.7443,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978216425642883e-06,
+      "loss": 0.7812,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977997106285627e-06,
+      "loss": 0.7484,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977776690820704e-06,
+      "loss": 0.6048,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977555179296652e-06,
+      "loss": 0.5896,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977332571762241e-06,
+      "loss": 0.6137,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977108868266493e-06,
+      "loss": 0.5953,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976884068858661e-06,
+      "loss": 0.6033,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976658173588244e-06,
+      "loss": 0.7489,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976431182504983e-06,
+      "loss": 0.7476,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976203095658859e-06,
+      "loss": 0.5825,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975973913100096e-06,
+      "loss": 0.6041,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975743634879156e-06,
+      "loss": 0.7823,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975512261046743e-06,
+      "loss": 0.7686,
+      "step": 419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975279791653804e-06,
+      "loss": 0.5874,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97504622675153e-06,
+      "loss": 0.6156,
+      "step": 421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974811566391347e-06,
+      "loss": 0.7516,
+      "step": 422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974575810624924e-06,
+      "loss": 0.7456,
+      "step": 423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974338959504173e-06,
+      "loss": 0.5787,
+      "step": 424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97410101308125e-06,
+      "loss": 0.5787,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97386197140854e-06,
+      "loss": 0.7451,
+      "step": 426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973621834538687e-06,
+      "loss": 0.8107,
+      "step": 427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97338060252456e-06,
+      "loss": 0.6122,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97313827541928e-06,
+      "loss": 0.7379,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972894853276204e-06,
+      "loss": 0.7829,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972650336148932e-06,
+      "loss": 0.7906,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972404724091303e-06,
+      "loss": 0.7564,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972158017157399e-06,
+      "loss": 0.7616,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971910215401543e-06,
+      "loss": 0.7602,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971661318878295e-06,
+      "loss": 0.7556,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971411327642467e-06,
+      "loss": 0.7768,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971160241749098e-06,
+      "loss": 0.7585,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970908061253479e-06,
+      "loss": 0.7221,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970654786211133e-06,
+      "loss": 0.5846,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970400416677833e-06,
+      "loss": 0.6063,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97014495270959e-06,
+      "loss": 0.7583,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969888394362648e-06,
+      "loss": 0.7544,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969630741693505e-06,
+      "loss": 0.7578,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96937199475889e-06,
+      "loss": 0.7621,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969112153615778e-06,
+      "loss": 0.7673,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968851218321387e-06,
+      "loss": 0.7538,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968589188933167e-06,
+      "loss": 0.6144,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968326065508815e-06,
+      "loss": 0.7339,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.968061848106272e-06,
+      "loss": 0.7908,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967796536783714e-06,
+      "loss": 0.7439,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96753013159956e-06,
+      "loss": 0.8115,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967262632612469e-06,
+      "loss": 0.7785,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966994039881342e-06,
+      "loss": 0.5834,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966724353465322e-06,
+      "loss": 0.7585,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96645357342379e-06,
+      "loss": 0.7282,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96618169981637e-06,
+      "loss": 0.7732,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965908732702928e-06,
+      "loss": 0.7471,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965634672143566e-06,
+      "loss": 0.7858,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965359518198627e-06,
+      "loss": 0.7306,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965083270928702e-06,
+      "loss": 0.593,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964805930394617e-06,
+      "loss": 0.7659,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96452749665744e-06,
+      "loss": 0.5948,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964247969778476e-06,
+      "loss": 0.7521,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963967349819277e-06,
+      "loss": 0.7618,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963685636841635e-06,
+      "loss": 0.7403,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963402830907575e-06,
+      "loss": 0.586,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963118932079374e-06,
+      "loss": 0.7587,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96283394041954e-06,
+      "loss": 0.7064,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962547855990827e-06,
+      "loss": 0.7862,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962260678856228e-06,
+      "loss": 0.721,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961972409078975e-06,
+      "loss": 0.7715,
+      "step": 471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961683046722544e-06,
+      "loss": 0.7619,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96139259185065e-06,
+      "loss": 0.73,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961101044527248e-06,
+      "loss": 0.7307,
+      "step": 474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960808404816534e-06,
+      "loss": 0.7427,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960514672782943e-06,
+      "loss": 0.7622,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960219848491154e-06,
+      "loss": 0.7337,
+      "step": 477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959923932006084e-06,
+      "loss": 0.7594,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95962692339289e-06,
+      "loss": 0.5991,
+      "step": 479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959328822716972e-06,
+      "loss": 0.7456,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959029630043969e-06,
+      "loss": 0.7156,
+      "step": 481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958729345439758e-06,
+      "loss": 0.7383,
+      "step": 482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958427968970462e-06,
+      "loss": 0.7423,
+      "step": 483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958125500702436e-06,
+      "loss": 0.7302,
+      "step": 484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957821940702288e-06,
+      "loss": 0.7271,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957517289036853e-06,
+      "loss": 0.7713,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957211545773217e-06,
+      "loss": 0.7704,
+      "step": 487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956904710978697e-06,
+      "loss": 0.5856,
+      "step": 488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95659678472086e-06,
+      "loss": 0.7504,
+      "step": 489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956287767067505e-06,
+      "loss": 0.5755,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955977658086674e-06,
+      "loss": 0.7771,
+      "step": 491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955666457846655e-06,
+      "loss": 0.7696,
+      "step": 492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955354166415966e-06,
+      "loss": 0.6007,
+      "step": 493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955040783863373e-06,
+      "loss": 0.7715,
+      "step": 494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954726310257881e-06,
+      "loss": 0.7473,
+      "step": 495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954410745668731e-06,
+      "loss": 0.7756,
+      "step": 496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954094090165409e-06,
+      "loss": 0.7581,
+      "step": 497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95377634381764e-06,
+      "loss": 0.7611,
+      "step": 498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95345750669539e-06,
+      "loss": 0.771,
+      "step": 499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95313757886886e-06,
+      "loss": 0.7387,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952816560408498e-06,
+      "loss": 0.5941,
+      "step": 501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952494451384988e-06,
+      "loss": 0.7558,
+      "step": 502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952171251869257e-06,
+      "loss": 0.7406,
+      "step": 503
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951846961932468e-06,
+      "loss": 0.681,
+      "step": 504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95152158164603e-06,
+      "loss": 0.7749,
+      "step": 505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951195111081585e-06,
+      "loss": 0.7443,
+      "step": 506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950867550311019e-06,
+      "loss": 0.7311,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95053889940646e-06,
+      "loss": 0.7231,
+      "step": 508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950209158440272e-06,
+      "loss": 0.7756,
+      "step": 509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949878327485063e-06,
+      "loss": 0.7706,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949546406613676e-06,
+      "loss": 0.5744,
+      "step": 511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9492133958992e-06,
+      "loss": 0.7962,
+      "step": 512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948879295414958e-06,
+      "loss": 0.7714,
+      "step": 513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948544105234518e-06,
+      "loss": 0.7033,
+      "step": 514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948207825431684e-06,
+      "loss": 0.7528,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947870456080502e-06,
+      "loss": 0.7257,
+      "step": 516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947531997255256e-06,
+      "loss": 0.5849,
+      "step": 517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.947192449030475e-06,
+      "loss": 0.5735,
+      "step": 518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946851811480922e-06,
+      "loss": 0.581,
+      "step": 519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946510084681602e-06,
+      "loss": 0.5837,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94616726870776e-06,
+      "loss": 0.5905,
+      "step": 521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945823363634883e-06,
+      "loss": 0.7646,
+      "step": 522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945478369538692e-06,
+      "loss": 0.7199,
+      "step": 523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945132286495155e-06,
+      "loss": 0.753,
+      "step": 524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944785114580475e-06,
+      "loss": 0.7967,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944436853871094e-06,
+      "loss": 0.748,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944087504443699e-06,
+      "loss": 0.5865,
+      "step": 527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94373706637521e-06,
+      "loss": 0.7674,
+      "step": 528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943385539742792e-06,
+      "loss": 0.7514,
+      "step": 529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943032924623847e-06,
+      "loss": 0.7548,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942679221096021e-06,
+      "loss": 0.6023,
+      "step": 531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94232442923719e-06,
+      "loss": 0.74,
+      "step": 532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941968549125481e-06,
+      "loss": 0.5804,
+      "step": 533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941611580839254e-06,
+      "loss": 0.7047,
+      "step": 534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941253524457108e-06,
+      "loss": 0.7486,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940894380057887e-06,
+      "loss": 0.7434,
+      "step": 536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940534147720666e-06,
+      "loss": 0.7366,
+      "step": 537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94017282752477e-06,
+      "loss": 0.7803,
+      "step": 538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939810419549757e-06,
+      "loss": 0.743,
+      "step": 539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939446923875425e-06,
+      "loss": 0.7265,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939082340581811e-06,
+      "loss": 0.7295,
+      "step": 541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938716669749196e-06,
+      "loss": 0.7499,
+      "step": 542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938349911458094e-06,
+      "loss": 0.7496,
+      "step": 543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937982065789262e-06,
+      "loss": 0.7474,
+      "step": 544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9376131328237e-06,
+      "loss": 0.7561,
+      "step": 545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937243112642639e-06,
+      "loss": 0.6031,
+      "step": 546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936872005327556e-06,
+      "loss": 0.7552,
+      "step": 547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936499810960165e-06,
+      "loss": 0.7532,
+      "step": 548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936126529622421e-06,
+      "loss": 0.6104,
+      "step": 549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935752161396515e-06,
+      "loss": 0.7688,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935376706364879e-06,
+      "loss": 0.7795,
+      "step": 551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935000164610186e-06,
+      "loss": 0.7574,
+      "step": 552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934622536215347e-06,
+      "loss": 0.7592,
+      "step": 553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934243821263511e-06,
+      "loss": 0.7399,
+      "step": 554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93386401983807e-06,
+      "loss": 0.7317,
+      "step": 555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93348313202265e-06,
+      "loss": 0.7276,
+      "step": 556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933101157901118e-06,
+      "loss": 0.7496,
+      "step": 557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932718097557584e-06,
+      "loss": 0.7157,
+      "step": 558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932333951076395e-06,
+      "loss": 0.5723,
+      "step": 559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93194871854213e-06,
+      "loss": 0.7519,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931562400039622e-06,
+      "loss": 0.7438,
+      "step": 561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931174995653929e-06,
+      "loss": 0.7609,
+      "step": 562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930786505470355e-06,
+      "loss": 0.7526,
+      "step": 563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930396929574444e-06,
+      "loss": 0.7615,
+      "step": 564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930006268051973e-06,
+      "loss": 0.727,
+      "step": 565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929614520988967e-06,
+      "loss": 0.7482,
+      "step": 566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929221688471678e-06,
+      "loss": 0.7447,
+      "step": 567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92882777058661e-06,
+      "loss": 0.7527,
+      "step": 568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928432767420499e-06,
+      "loss": 0.7428,
+      "step": 569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928036679060319e-06,
+      "loss": 0.7612,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927639505593287e-06,
+      "loss": 0.7426,
+      "step": 571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927241247106856e-06,
+      "loss": 0.6089,
+      "step": 572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926841903688716e-06,
+      "loss": 0.7445,
+      "step": 573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926441475426804e-06,
+      "loss": 0.6033,
+      "step": 574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926039962409285e-06,
+      "loss": 0.778,
+      "step": 575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925637364724572e-06,
+      "loss": 0.7345,
+      "step": 576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925233682461313e-06,
+      "loss": 0.7353,
+      "step": 577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924828915708393e-06,
+      "loss": 0.7604,
+      "step": 578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92442306455494e-06,
+      "loss": 0.7967,
+      "step": 579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924016129090317e-06,
+      "loss": 0.7763,
+      "step": 580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923608109404127e-06,
+      "loss": 0.7619,
+      "step": 581
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923199005586212e-06,
+      "loss": 0.7602,
+      "step": 582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922788817726656e-06,
+      "loss": 0.7559,
+      "step": 583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922377545915776e-06,
+      "loss": 0.749,
+      "step": 584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921965190244129e-06,
+      "loss": 0.7451,
+      "step": 585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921551750802514e-06,
+      "loss": 0.764,
+      "step": 586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.921137227681965e-06,
+      "loss": 0.7284,
+      "step": 587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920721620973757e-06,
+      "loss": 0.7429,
+      "step": 588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920304930769401e-06,
+      "loss": 0.7303,
+      "step": 589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91988715716065e-06,
+      "loss": 0.7619,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919468300239493e-06,
+      "loss": 0.7721,
+      "step": 591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919048360098159e-06,
+      "loss": 0.7258,
+      "step": 592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918627336829113e-06,
+      "loss": 0.7197,
+      "step": 593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918205230525065e-06,
+      "loss": 0.7869,
+      "step": 594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917782041278952e-06,
+      "loss": 0.7212,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917357769183962e-06,
+      "loss": 0.6009,
+      "step": 596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916932414333512e-06,
+      "loss": 0.7386,
+      "step": 597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916505976821262e-06,
+      "loss": 0.5792,
+      "step": 598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916078456741112e-06,
+      "loss": 0.7776,
+      "step": 599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915649854187195e-06,
+      "loss": 0.7391,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915220169253886e-06,
+      "loss": 0.5862,
+      "step": 601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914789402035796e-06,
+      "loss": 0.7704,
+      "step": 602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91435755262778e-06,
+      "loss": 0.7148,
+      "step": 603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913924621124922e-06,
+      "loss": 0.6054,
+      "step": 604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913490607622554e-06,
+      "loss": 0.7839,
+      "step": 605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91305551221624e-06,
+      "loss": 0.7418,
+      "step": 606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912619335001782e-06,
+      "loss": 0.7668,
+      "step": 607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912182076075225e-06,
+      "loss": 0.7442,
+      "step": 608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911743735532847e-06,
+      "loss": 0.7481,
+      "step": 609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911304313471166e-06,
+      "loss": 0.7562,
+      "step": 610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910863809986942e-06,
+      "loss": 0.7393,
+      "step": 611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910422225177166e-06,
+      "loss": 0.5914,
+      "step": 612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909979559139072e-06,
+      "loss": 0.7418,
+      "step": 613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909535811970132e-06,
+      "loss": 0.7595,
+      "step": 614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909090983768055e-06,
+      "loss": 0.7542,
+      "step": 615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908645074630786e-06,
+      "loss": 0.735,
+      "step": 616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90819808465651e-06,
+      "loss": 0.7393,
+      "step": 617
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907750013943652e-06,
+      "loss": 0.7354,
+      "step": 618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907300862590871e-06,
+      "loss": 0.7066,
+      "step": 619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906850630697068e-06,
+      "loss": 0.7567,
+      "step": 620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906399318361376e-06,
+      "loss": 0.7568,
+      "step": 621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905946925683177e-06,
+      "loss": 0.7556,
+      "step": 622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905493452762075e-06,
+      "loss": 0.7596,
+      "step": 623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905038899697924e-06,
+      "loss": 0.7282,
+      "step": 624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904583266590813e-06,
+      "loss": 0.5959,
+      "step": 625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904126553541068e-06,
+      "loss": 0.7486,
+      "step": 626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903668760649252e-06,
+      "loss": 0.7247,
+      "step": 627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903209888016167e-06,
+      "loss": 0.6168,
+      "step": 628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902749935742852e-06,
+      "loss": 0.7436,
+      "step": 629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902288903930585e-06,
+      "loss": 0.7407,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901826792680881e-06,
+      "loss": 0.5822,
+      "step": 631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90136360209549e-06,
+      "loss": 0.6361,
+      "step": 632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900899332276406e-06,
+      "loss": 0.5857,
+      "step": 633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900433983325854e-06,
+      "loss": 0.7625,
+      "step": 634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899967555346301e-06,
+      "loss": 0.7468,
+      "step": 635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899500048440448e-06,
+      "loss": 0.5804,
+      "step": 636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899031462711237e-06,
+      "loss": 0.7591,
+      "step": 637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898561798261846e-06,
+      "loss": 0.7342,
+      "step": 638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89809105519569e-06,
+      "loss": 0.5754,
+      "step": 639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897619233616424e-06,
+      "loss": 0.6199,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897146333627938e-06,
+      "loss": 0.7534,
+      "step": 641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896672355334358e-06,
+      "loss": 0.5762,
+      "step": 642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896197298840053e-06,
+      "loss": 0.7487,
+      "step": 643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89572116424962e-06,
+      "loss": 0.5895,
+      "step": 644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895243951667907e-06,
+      "loss": 0.7183,
+      "step": 645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894765661199987e-06,
+      "loss": 0.7356,
+      "step": 646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894286292951176e-06,
+      "loss": 0.7493,
+      "step": 647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893805847027025e-06,
+      "loss": 0.7458,
+      "step": 648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893324323533326e-06,
+      "loss": 0.5679,
+      "step": 649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892841722576103e-06,
+      "loss": 0.7078,
+      "step": 650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892358044261624e-06,
+      "loss": 0.7112,
+      "step": 651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891873288696388e-06,
+      "loss": 0.5737,
+      "step": 652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891387455987132e-06,
+      "loss": 0.5821,
+      "step": 653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890900546240837e-06,
+      "loss": 0.7535,
+      "step": 654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890412559564709e-06,
+      "loss": 0.7358,
+      "step": 655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889923496066202e-06,
+      "loss": 0.5946,
+      "step": 656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889433355853006e-06,
+      "loss": 0.7496,
+      "step": 657
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88894213903304e-06,
+      "loss": 0.7377,
+      "step": 658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888449845714465e-06,
+      "loss": 0.7556,
+      "step": 659
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887956476005683e-06,
+      "loss": 0.7401,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887462030015328e-06,
+      "loss": 0.7292,
+      "step": 661
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886966507852272e-06,
+      "loss": 0.7509,
+      "step": 662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886469909625624e-06,
+      "loss": 0.7389,
+      "step": 663
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885972235444732e-06,
+      "loss": 0.7576,
+      "step": 664
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885473485419176e-06,
+      "loss": 0.7595,
+      "step": 665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88497365965878e-06,
+      "loss": 0.7687,
+      "step": 666
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884472758273596e-06,
+      "loss": 0.745,
+      "step": 667
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883970781373921e-06,
+      "loss": 0.7222,
+      "step": 668
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883467729070286e-06,
+      "loss": 0.7264,
+      "step": 669
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882963601473458e-06,
+      "loss": 0.5795,
+      "step": 670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882458398694438e-06,
+      "loss": 0.5971,
+      "step": 671
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88195212084447e-06,
+      "loss": 0.7521,
+      "step": 672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881444768035031e-06,
+      "loss": 0.7214,
+      "step": 673
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880936340377836e-06,
+      "loss": 0.7256,
+      "step": 674
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880426837984833e-06,
+      "loss": 0.7665,
+      "step": 675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879916260968212e-06,
+      "loss": 0.7428,
+      "step": 676
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879404609440397e-06,
+      "loss": 0.7091,
+      "step": 677
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87889188351405e-06,
+      "loss": 0.7249,
+      "step": 678
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878378083302065e-06,
+      "loss": 0.734,
+      "step": 679
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87786320891758e-06,
+      "loss": 0.7082,
+      "step": 680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87734726047396e-06,
+      "loss": 0.7103,
+      "step": 681
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876830238084817e-06,
+      "loss": 0.7535,
+      "step": 682
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876312141863994e-06,
+      "loss": 0.7533,
+      "step": 683
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875792971925568e-06,
+      "loss": 0.7534,
+      "step": 684
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875272728383857e-06,
+      "loss": 0.7525,
+      "step": 685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874751411353414e-06,
+      "loss": 0.6029,
+      "step": 686
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874229020949026e-06,
+      "loss": 0.5794,
+      "step": 687
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873705557285723e-06,
+      "loss": 0.7724,
+      "step": 688
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87318102047876e-06,
+      "loss": 0.7429,
+      "step": 689
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872655410643641e-06,
+      "loss": 0.7255,
+      "step": 690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872128727896098e-06,
+      "loss": 0.7469,
+      "step": 691
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8716009723521e-06,
+      "loss": 0.7211,
+      "step": 692
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871072144127856e-06,
+      "loss": 0.7589,
+      "step": 693
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870542243339809e-06,
+      "loss": 0.6086,
+      "step": 694
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870011270104638e-06,
+      "loss": 0.5778,
+      "step": 695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869479224539256e-06,
+      "loss": 0.7398,
+      "step": 696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868946106760818e-06,
+      "loss": 0.7483,
+      "step": 697
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868411916886706e-06,
+      "loss": 0.7788,
+      "step": 698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867876655034551e-06,
+      "loss": 0.6124,
+      "step": 699
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867340321322207e-06,
+      "loss": 0.7434,
+      "step": 700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86680291586777e-06,
+      "loss": 0.7401,
+      "step": 701
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866264438789573e-06,
+      "loss": 0.6005,
+      "step": 702
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865724890206184e-06,
+      "loss": 0.7003,
+      "step": 703
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865184270236407e-06,
+      "loss": 0.7455,
+      "step": 704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864642578999278e-06,
+      "loss": 0.7408,
+      "step": 705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864099816614076e-06,
+      "loss": 0.7068,
+      "step": 706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863555983200309e-06,
+      "loss": 0.5792,
+      "step": 707
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863011078877726e-06,
+      "loss": 0.732,
+      "step": 708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86246510376631e-06,
+      "loss": 0.7143,
+      "step": 709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861918057986279e-06,
+      "loss": 0.7609,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861369941658085e-06,
+      "loss": 0.734,
+      "step": 711
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860820754902422e-06,
+      "loss": 0.6021,
+      "step": 712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860270497840214e-06,
+      "loss": 0.7251,
+      "step": 713
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85971917059262e-06,
+      "loss": 0.784,
+      "step": 714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85916677328104e-06,
+      "loss": 0.5721,
+      "step": 715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858613306027106e-06,
+      "loss": 0.6223,
+      "step": 716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858058768952688e-06,
+      "loss": 0.76,
+      "step": 717
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857503162179884e-06,
+      "loss": 0.783,
+      "step": 718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85694648583104e-06,
+      "loss": 0.7443,
+      "step": 719
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856388740028727e-06,
+      "loss": 0.7333,
+      "step": 720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855829924895757e-06,
+      "loss": 0.738,
+      "step": 721
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855270040555173e-06,
+      "loss": 0.7462,
+      "step": 722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854709087130261e-06,
+      "loss": 0.7224,
+      "step": 723
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854147064744533e-06,
+      "loss": 0.5744,
+      "step": 724
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.853583973521745e-06,
+      "loss": 0.7668,
+      "step": 725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85301981358588e-06,
+      "loss": 0.7166,
+      "step": 726
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852454585061164e-06,
+      "loss": 0.5875,
+      "step": 727
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851888288072053e-06,
+      "loss": 0.7561,
+      "step": 728
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851320922743242e-06,
+      "loss": 0.7708,
+      "step": 729
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85075248919966e-06,
+      "loss": 0.5748,
+      "step": 730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850182987566467e-06,
+      "loss": 0.5957,
+      "step": 731
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849612417969064e-06,
+      "loss": 0.7245,
+      "step": 732
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849040780533086e-06,
+      "loss": 0.5728,
+      "step": 733
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8484680753844e-06,
+      "loss": 0.5514,
+      "step": 734
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847894302649112e-06,
+      "loss": 0.719,
+      "step": 735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847319462453562e-06,
+      "loss": 0.7237,
+      "step": 736
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846743554924322e-06,
+      "loss": 0.5844,
+      "step": 737
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846166580188203e-06,
+      "loss": 0.7543,
+      "step": 738
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845588538372249e-06,
+      "loss": 0.5859,
+      "step": 739
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845009429603738e-06,
+      "loss": 0.7276,
+      "step": 740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844429254010184e-06,
+      "loss": 0.5921,
+      "step": 741
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84384801171934e-06,
+      "loss": 0.7273,
+      "step": 742
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843265702859187e-06,
+      "loss": 0.7278,
+      "step": 743
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842682327557945e-06,
+      "loss": 0.7612,
+      "step": 744
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842097885944066e-06,
+      "loss": 0.7067,
+      "step": 745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84151237814624e-06,
+      "loss": 0.5755,
+      "step": 746
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840925804293389e-06,
+      "loss": 0.7393,
+      "step": 747
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840338164514673e-06,
+      "loss": 0.7545,
+      "step": 748
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839749458939482e-06,
+      "loss": 0.7303,
+      "step": 749
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839159687697448e-06,
+      "loss": 0.5524,
+      "step": 750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838568850918428e-06,
+      "loss": 0.5855,
+      "step": 751
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83797694873252e-06,
+      "loss": 0.7254,
+      "step": 752
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837383981270056e-06,
+      "loss": 0.7433,
+      "step": 753
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836789948661602e-06,
+      "loss": 0.601,
+      "step": 754
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836194851037959e-06,
+      "loss": 0.5978,
+      "step": 755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83559868853016e-06,
+      "loss": 0.5956,
+      "step": 756
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835001461269473e-06,
+      "loss": 0.7148,
+      "step": 757
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834403169387408e-06,
+      "loss": 0.733,
+      "step": 758
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833803813015696e-06,
+      "loss": 0.7692,
+      "step": 759
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833203392286314e-06,
+      "loss": 0.6821,
+      "step": 760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832601907331467e-06,
+      "loss": 0.7144,
+      "step": 761
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831999358283597e-06,
+      "loss": 0.7467,
+      "step": 762
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831395745275378e-06,
+      "loss": 0.7258,
+      "step": 763
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830791068439725e-06,
+      "loss": 0.7404,
+      "step": 764
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830185327909775e-06,
+      "loss": 0.7263,
+      "step": 765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82957852381891e-06,
+      "loss": 0.7287,
+      "step": 766
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828970656300743e-06,
+      "loss": 0.5918,
+      "step": 767
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82836172548912e-06,
+      "loss": 0.6044,
+      "step": 768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827751731518123e-06,
+      "loss": 0.734,
+      "step": 769
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827140674522064e-06,
+      "loss": 0.7115,
+      "step": 770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826528554635493e-06,
+      "loss": 0.575,
+      "step": 771
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825915371993194e-06,
+      "loss": 0.5615,
+      "step": 772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825301126730183e-06,
+      "loss": 0.5657,
+      "step": 773
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824685818981711e-06,
+      "loss": 0.5854,
+      "step": 774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824069448883264e-06,
+      "loss": 0.7468,
+      "step": 775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82345201657056e-06,
+      "loss": 0.7165,
+      "step": 776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822833522179553e-06,
+      "loss": 0.7453,
+      "step": 777
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822213965846428e-06,
+      "loss": 0.7437,
+      "step": 778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821593347707606e-06,
+      "loss": 0.605,
+      "step": 779
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82097166789974e-06,
+      "loss": 0.7302,
+      "step": 780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820348926559718e-06,
+      "loss": 0.7533,
+      "step": 781
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819725123824665e-06,
+      "loss": 0.6978,
+      "step": 782
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819100259831933e-06,
+      "loss": 0.7469,
+      "step": 783
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818474334719114e-06,
+      "loss": 0.604,
+      "step": 784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817847348624027e-06,
+      "loss": 0.5696,
+      "step": 785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817219301684733e-06,
+      "loss": 0.7514,
+      "step": 786
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816590194039517e-06,
+      "loss": 0.7482,
+      "step": 787
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815960025826906e-06,
+      "loss": 0.7867,
+      "step": 788
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815328797185657e-06,
+      "loss": 0.5961,
+      "step": 789
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814696508254758e-06,
+      "loss": 0.732,
+      "step": 790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814063159173436e-06,
+      "loss": 0.749,
+      "step": 791
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813428750081146e-06,
+      "loss": 0.733,
+      "step": 792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81279328111758e-06,
+      "loss": 0.6136,
+      "step": 793
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.812156752422664e-06,
+      "loss": 0.7678,
+      "step": 794
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811519164136552e-06,
+      "loss": 0.7362,
+      "step": 795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810880516399636e-06,
+      "loss": 0.7288,
+      "step": 796
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810240809352542e-06,
+      "loss": 0.7328,
+      "step": 797
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809600043136125e-06,
+      "loss": 0.7267,
+      "step": 798
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808958217891478e-06,
+      "loss": 0.6013,
+      "step": 799
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808315333759924e-06,
+      "loss": 0.6106,
+      "step": 800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807671390883018e-06,
+      "loss": 0.7296,
+      "step": 801
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807026389402553e-06,
+      "loss": 0.5579,
+      "step": 802
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80638032946055e-06,
+      "loss": 0.7374,
+      "step": 803
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805733211199266e-06,
+      "loss": 0.5934,
+      "step": 804
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805085034761193e-06,
+      "loss": 0.7372,
+      "step": 805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804435800289047e-06,
+      "loss": 0.727,
+      "step": 806
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803785507925788e-06,
+      "loss": 0.7594,
+      "step": 807
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803134157814605e-06,
+      "loss": 0.5586,
+      "step": 808
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802481750098918e-06,
+      "loss": 0.7049,
+      "step": 809
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801828284922378e-06,
+      "loss": 0.7458,
+      "step": 810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801173762428875e-06,
+      "loss": 0.6909,
+      "step": 811
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800518182762527e-06,
+      "loss": 0.7486,
+      "step": 812
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799861546067688e-06,
+      "loss": 0.7296,
+      "step": 813
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799203852488941e-06,
+      "loss": 0.7342,
+      "step": 814
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798545102171107e-06,
+      "loss": 0.7192,
+      "step": 815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797885295259235e-06,
+      "loss": 0.5741,
+      "step": 816
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797224431898607e-06,
+      "loss": 0.7242,
+      "step": 817
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79656251223474e-06,
+      "loss": 0.7105,
+      "step": 818
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795899536413383e-06,
+      "loss": 0.5914,
+      "step": 819
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795235504580513e-06,
+      "loss": 0.7361,
+      "step": 820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79457041688235e-06,
+      "loss": 0.7108,
+      "step": 821
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793904273465336e-06,
+      "loss": 0.7113,
+      "step": 822
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79323707447615e-06,
+      "loss": 0.6911,
+      "step": 823
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792568820061703e-06,
+      "loss": 0.7645,
+      "step": 824
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791899510369139e-06,
+      "loss": 0.7535,
+      "step": 825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791229145545832e-06,
+      "loss": 0.5848,
+      "step": 826
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790557725739391e-06,
+      "loss": 0.6098,
+      "step": 827
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789885251097657e-06,
+      "loss": 0.7583,
+      "step": 828
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789211721768702e-06,
+      "loss": 0.5686,
+      "step": 829
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788537137900829e-06,
+      "loss": 0.5715,
+      "step": 830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787861499642575e-06,
+      "loss": 0.5559,
+      "step": 831
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787184807142713e-06,
+      "loss": 0.5879,
+      "step": 832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786507060550241e-06,
+      "loss": 0.7638,
+      "step": 833
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785828260014394e-06,
+      "loss": 0.7529,
+      "step": 834
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785148405684636e-06,
+      "loss": 0.7256,
+      "step": 835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784467497710665e-06,
+      "loss": 0.7391,
+      "step": 836
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78378553624241e-06,
+      "loss": 0.57,
+      "step": 837
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783102521430035e-06,
+      "loss": 0.743,
+      "step": 838
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78241845342393e-06,
+      "loss": 0.6967,
+      "step": 839
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78173333237472e-06,
+      "loss": 0.7355,
+      "step": 840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781047158433266e-06,
+      "loss": 0.5842,
+      "step": 841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780359931750653e-06,
+      "loss": 0.5682,
+      "step": 842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779671652478205e-06,
+      "loss": 0.7421,
+      "step": 843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778982320767474e-06,
+      "loss": 0.5714,
+      "step": 844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778291936770241e-06,
+      "loss": 0.7332,
+      "step": 845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777600500638525e-06,
+      "loss": 0.7395,
+      "step": 846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776908012524573e-06,
+      "loss": 0.7039,
+      "step": 847
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776214472580865e-06,
+      "loss": 0.7293,
+      "step": 848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775519880960112e-06,
+      "loss": 0.5987,
+      "step": 849
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774824237815253e-06,
+      "loss": 0.5675,
+      "step": 850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774127543299467e-06,
+      "loss": 0.7299,
+      "step": 851
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773429797566157e-06,
+      "loss": 0.5655,
+      "step": 852
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772731000768959e-06,
+      "loss": 0.757,
+      "step": 853
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772031153061744e-06,
+      "loss": 0.7281,
+      "step": 854
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77133025459861e-06,
+      "loss": 0.5742,
+      "step": 855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770628305533887e-06,
+      "loss": 0.7189,
+      "step": 856
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769925306022141e-06,
+      "loss": 0.5586,
+      "step": 857
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769221256218165e-06,
+      "loss": 0.7454,
+      "step": 858
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768516156276981e-06,
+      "loss": 0.5788,
+      "step": 859
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767810006353849e-06,
+      "loss": 0.6994,
+      "step": 860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767102806604253e-06,
+      "loss": 0.5909,
+      "step": 861
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766394557183914e-06,
+      "loss": 0.7894,
+      "step": 862
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.765685258248781e-06,
+      "loss": 0.7384,
+      "step": 863
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764974909955036e-06,
+      "loss": 0.5914,
+      "step": 864
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76426351245909e-06,
+      "loss": 0.5894,
+      "step": 865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763551065917586e-06,
+      "loss": 0.7311,
+      "step": 866
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762837570487398e-06,
+      "loss": 0.7232,
+      "step": 867
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762123026325632e-06,
+      "loss": 0.7384,
+      "step": 868
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761407433589622e-06,
+      "loss": 0.7249,
+      "step": 869
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760690792436936e-06,
+      "loss": 0.5997,
+      "step": 870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759973103025369e-06,
+      "loss": 0.7507,
+      "step": 871
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759254365512954e-06,
+      "loss": 0.7432,
+      "step": 872
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758534580057945e-06,
+      "loss": 0.7236,
+      "step": 873
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757813746818837e-06,
+      "loss": 0.7576,
+      "step": 874
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757091865954348e-06,
+      "loss": 0.7286,
+      "step": 875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756368937623429e-06,
+      "loss": 0.7675,
+      "step": 876
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755644961985263e-06,
+      "loss": 0.7182,
+      "step": 877
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75491993919926e-06,
+      "loss": 0.581,
+      "step": 878
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754193869425068e-06,
+      "loss": 0.5938,
+      "step": 879
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753466752822556e-06,
+      "loss": 0.7247,
+      "step": 880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752738589551831e-06,
+      "loss": 0.7351,
+      "step": 881
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752009379773226e-06,
+      "loss": 0.7536,
+      "step": 882
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751279123647307e-06,
+      "loss": 0.7056,
+      "step": 883
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750547821334868e-06,
+      "loss": 0.7122,
+      "step": 884
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749815472996937e-06,
+      "loss": 0.7329,
+      "step": 885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74908207879477e-06,
+      "loss": 0.5888,
+      "step": 886
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748347638889852e-06,
+      "loss": 0.5722,
+      "step": 887
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747612153443899e-06,
+      "loss": 0.7114,
+      "step": 888
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74687562261886e-06,
+      "loss": 0.5869,
+      "step": 889
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746138046576912e-06,
+      "loss": 0.7604,
+      "step": 890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745399425480459e-06,
+      "loss": 0.7282,
+      "step": 891
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744659759492143e-06,
+      "loss": 0.7073,
+      "step": 892
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743919048774829e-06,
+      "loss": 0.5789,
+      "step": 893
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743177293491614e-06,
+      "loss": 0.7348,
+      "step": 894
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742434493805827e-06,
+      "loss": 0.7405,
+      "step": 895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741690649881024e-06,
+      "loss": 0.7449,
+      "step": 896
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740945761880993e-06,
+      "loss": 0.742,
+      "step": 897
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740199829969752e-06,
+      "loss": 0.7197,
+      "step": 898
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739452854311546e-06,
+      "loss": 0.5944,
+      "step": 899
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738704835070855e-06,
+      "loss": 0.7447,
+      "step": 900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737955772412383e-06,
+      "loss": 0.686,
+      "step": 901
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737205666501067e-06,
+      "loss": 0.706,
+      "step": 902
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736454517502074e-06,
+      "loss": 0.5652,
+      "step": 903
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7357023255808e-06,
+      "loss": 0.7531,
+      "step": 904
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734949090902869e-06,
+      "loss": 0.7124,
+      "step": 905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73419481363414e-06,
+      "loss": 0.7055,
+      "step": 906
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733439493940693e-06,
+      "loss": 0.7126,
+      "step": 907
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732683131988844e-06,
+      "loss": 0.7238,
+      "step": 908
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731925727945137e-06,
+      "loss": 0.7177,
+      "step": 909
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731167281976346e-06,
+      "loss": 0.732,
+      "step": 910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73040779424947e-06,
+      "loss": 0.7149,
+      "step": 911
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729647264931746e-06,
+      "loss": 0.7456,
+      "step": 912
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72888569419063e-06,
+      "loss": 0.564,
+      "step": 913
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72812308219382e-06,
+      "loss": 0.7636,
+      "step": 914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727359429109228e-06,
+      "loss": 0.5775,
+      "step": 915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726594735105009e-06,
+      "loss": 0.7441,
+      "step": 916
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725829000349535e-06,
+      "loss": 0.7103,
+      "step": 917
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725062225011421e-06,
+      "loss": 0.7347,
+      "step": 918
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7242944092595e-06,
+      "loss": 0.7328,
+      "step": 919
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723525553262835e-06,
+      "loss": 0.7621,
+      "step": 920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722755657190727e-06,
+      "loss": 0.7063,
+      "step": 921
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721984721212695e-06,
+      "loss": 0.7397,
+      "step": 922
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721212745498493e-06,
+      "loss": 0.7257,
+      "step": 923
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720439730218104e-06,
+      "loss": 0.7074,
+      "step": 924
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719665675541737e-06,
+      "loss": 0.5758,
+      "step": 925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718890581639832e-06,
+      "loss": 0.7375,
+      "step": 926
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718114448683058e-06,
+      "loss": 0.7416,
+      "step": 927
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717337276842311e-06,
+      "loss": 0.5798,
+      "step": 928
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716559066288716e-06,
+      "loss": 0.7589,
+      "step": 929
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71577981719363e-06,
+      "loss": 0.707,
+      "step": 930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714999529728635e-06,
+      "loss": 0.5578,
+      "step": 931
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.714218204065545e-06,
+      "loss": 0.7268,
+      "step": 932
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713435840376395e-06,
+      "loss": 0.7257,
+      "step": 933
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71265243883346e-06,
+      "loss": 0.7183,
+      "step": 934
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711867999609233e-06,
+      "loss": 0.5553,
+      "step": 935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711082522876445e-06,
+      "loss": 0.7115,
+      "step": 936
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710296008808045e-06,
+      "loss": 0.7497,
+      "step": 937
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70950845757722e-06,
+      "loss": 0.7238,
+      "step": 938
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708719869357381e-06,
+      "loss": 0.5773,
+      "step": 939
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707930244322164e-06,
+      "loss": 0.7405,
+      "step": 940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707139582645442e-06,
+      "loss": 0.736,
+      "step": 941
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706347884501306e-06,
+      "loss": 0.7069,
+      "step": 942
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705555150064085e-06,
+      "loss": 0.7406,
+      "step": 943
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70476137950833e-06,
+      "loss": 0.7167,
+      "step": 944
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703966573008821e-06,
+      "loss": 0.5782,
+      "step": 945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703170730740567e-06,
+      "loss": 0.7559,
+      "step": 946
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702373852878806e-06,
+      "loss": 0.5752,
+      "step": 947
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701575939599e-06,
+      "loss": 0.5626,
+      "step": 948
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700776991076846e-06,
+      "loss": 0.7587,
+      "step": 949
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699977007488262e-06,
+      "loss": 0.7285,
+      "step": 950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699175989009398e-06,
+      "loss": 0.7338,
+      "step": 951
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69837393581663e-06,
+      "loss": 0.7062,
+      "step": 952
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697570848086564e-06,
+      "loss": 0.709,
+      "step": 953
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696766725996029e-06,
+      "loss": 0.5701,
+      "step": 954
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695961569722089e-06,
+      "loss": 0.7434,
+      "step": 955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69515537944203e-06,
+      "loss": 0.7124,
+      "step": 956
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694348155333367e-06,
+      "loss": 0.5846,
+      "step": 957
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693539897573843e-06,
+      "loss": 0.7418,
+      "step": 958
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.692730606341429e-06,
+      "loss": 0.5765,
+      "step": 959
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691920281814322e-06,
+      "loss": 0.7008,
+      "step": 960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691108924170952e-06,
+      "loss": 0.5654,
+      "step": 961
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690296533589967e-06,
+      "loss": 0.7265,
+      "step": 962
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689483110250252e-06,
+      "loss": 0.7313,
+      "step": 963
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688668654330912e-06,
+      "loss": 0.7202,
+      "step": 964
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687853166011284e-06,
+      "loss": 0.7323,
+      "step": 965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687036645470931e-06,
+      "loss": 0.5481,
+      "step": 966
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686219092889642e-06,
+      "loss": 0.5811,
+      "step": 967
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685400508447433e-06,
+      "loss": 0.7148,
+      "step": 968
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684580892324552e-06,
+      "loss": 0.7492,
+      "step": 969
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683760244701468e-06,
+      "loss": 0.7338,
+      "step": 970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682938565758883e-06,
+      "loss": 0.7129,
+      "step": 971
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682115855677719e-06,
+      "loss": 0.7333,
+      "step": 972
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68129211463913e-06,
+      "loss": 0.5504,
+      "step": 973
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680467342824498e-06,
+      "loss": 0.5838,
+      "step": 974
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679641540415428e-06,
+      "loss": 0.5682,
+      "step": 975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678814707593753e-06,
+      "loss": 0.5941,
+      "step": 976
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677986844541537e-06,
+      "loss": 0.7177,
+      "step": 977
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677157951441064e-06,
+      "loss": 0.7355,
+      "step": 978
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676328028474851e-06,
+      "loss": 0.7163,
+      "step": 979
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675497075825636e-06,
+      "loss": 0.7343,
+      "step": 980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674665093676392e-06,
+      "loss": 0.7245,
+      "step": 981
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67383208221031e-06,
+      "loss": 0.7263,
+      "step": 982
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672998041610812e-06,
+      "loss": 0.6066,
+      "step": 983
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672162972061546e-06,
+      "loss": 0.7128,
+      "step": 984
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671326873746382e-06,
+      "loss": 0.7437,
+      "step": 985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670489746849429e-06,
+      "loss": 0.6693,
+      "step": 986
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66965159155501e-06,
+      "loss": 0.5795,
+      "step": 987
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66881240804768e-06,
+      "loss": 0.6891,
+      "step": 988
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667972196512214e-06,
+      "loss": 0.7248,
+      "step": 989
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667130957133625e-06,
+      "loss": 0.6809,
+      "step": 990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666288690097146e-06,
+      "loss": 0.7122,
+      "step": 991
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665445395588232e-06,
+      "loss": 0.7064,
+      "step": 992
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664601073792572e-06,
+      "loss": 0.5613,
+      "step": 993
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663755724896076e-06,
+      "loss": 0.7103,
+      "step": 994
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66290934908488e-06,
+      "loss": 0.6849,
+      "step": 995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662061946545351e-06,
+      "loss": 0.7399,
+      "step": 996
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661213517464078e-06,
+      "loss": 0.5749,
+      "step": 997
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660364062027879e-06,
+      "loss": 0.7093,
+      "step": 998
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.65951358042379e-06,
+      "loss": 0.605,
+      "step": 999
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658662072839084e-06,
+      "loss": 0.5779,
+      "step": 1000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.657809539461256e-06,
+      "loss": 0.7384,
+      "step": 1001
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65695598047802e-06,
+      "loss": 0.6993,
+      "step": 1002
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65610139607733e-06,
+      "loss": 0.7509,
+      "step": 1003
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655245786447347e-06,
+      "loss": 0.5664,
+      "step": 1004
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654389151776475e-06,
+      "loss": 0.7195,
+      "step": 1005
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653531492253338e-06,
+      "loss": 0.7257,
+      "step": 1006
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652672808066779e-06,
+      "loss": 0.7101,
+      "step": 1007
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651813099405876e-06,
+      "loss": 0.5994,
+      "step": 1008
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650952366459928e-06,
+      "loss": 0.5841,
+      "step": 1009
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650090609418459e-06,
+      "loss": 0.5581,
+      "step": 1010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649227828471221e-06,
+      "loss": 0.701,
+      "step": 1011
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648364023808189e-06,
+      "loss": 0.5861,
+      "step": 1012
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647499195619567e-06,
+      "loss": 0.7201,
+      "step": 1013
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64663334409578e-06,
+      "loss": 0.7147,
+      "step": 1014
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64576646942748e-06,
+      "loss": 0.7152,
+      "step": 1015
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644898571805543e-06,
+      "loss": 0.7142,
+      "step": 1016
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644029651421077e-06,
+      "loss": 0.7428,
+      "step": 1017
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643159708465407e-06,
+      "loss": 0.7199,
+      "step": 1018
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642288743130083e-06,
+      "loss": 0.7088,
+      "step": 1019
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.641416755606889e-06,
+      "loss": 0.7367,
+      "step": 1020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640543746087824e-06,
+      "loss": 0.7026,
+      "step": 1021
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63966971476512e-06,
+      "loss": 0.5714,
+      "step": 1022
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638794661831226e-06,
+      "loss": 0.7201,
+      "step": 1023
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637918587478825e-06,
+      "loss": 0.7417,
+      "step": 1024
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637041491900816e-06,
+      "loss": 0.7444,
+      "step": 1025
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636163375290332e-06,
+      "loss": 0.588,
+      "step": 1026
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635284237840721e-06,
+      "loss": 0.5768,
+      "step": 1027
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634404079745565e-06,
+      "loss": 0.7086,
+      "step": 1028
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633522901198662e-06,
+      "loss": 0.7081,
+      "step": 1029
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632640702394043e-06,
+      "loss": 0.7203,
+      "step": 1030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631757483525958e-06,
+      "loss": 0.7038,
+      "step": 1031
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.630873244788884e-06,
+      "loss": 0.7234,
+      "step": 1032
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62998798637752e-06,
+      "loss": 0.7363,
+      "step": 1033
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629101708486794e-06,
+      "loss": 0.7188,
+      "step": 1034
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628214411311856e-06,
+      "loss": 0.7159,
+      "step": 1035
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627326095048079e-06,
+      "loss": 0.7211,
+      "step": 1036
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.626436759891062e-06,
+      "loss": 0.7055,
+      "step": 1037
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62554640603663e-06,
+      "loss": 0.7108,
+      "step": 1038
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624655033680827e-06,
+      "loss": 0.7439,
+      "step": 1039
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623762643019927e-06,
+      "loss": 0.7009,
+      "step": 1040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622869234250425e-06,
+      "loss": 0.704,
+      "step": 1041
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621974807569043e-06,
+      "loss": 0.746,
+      "step": 1042
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621079363172723e-06,
+      "loss": 0.7089,
+      "step": 1043
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620182901258635e-06,
+      "loss": 0.701,
+      "step": 1044
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619285422024171e-06,
+      "loss": 0.5795,
+      "step": 1045
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618386925666947e-06,
+      "loss": 0.5848,
+      "step": 1046
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617487412384802e-06,
+      "loss": 0.7333,
+      "step": 1047
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616586882375804e-06,
+      "loss": 0.7568,
+      "step": 1048
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615685335838237e-06,
+      "loss": 0.7414,
+      "step": 1049
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614782772970616e-06,
+      "loss": 0.5736,
+      "step": 1050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613879193971675e-06,
+      "loss": 0.7047,
+      "step": 1051
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612974599040374e-06,
+      "loss": 0.5905,
+      "step": 1052
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612068988375898e-06,
+      "loss": 0.5919,
+      "step": 1053
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61116236217765e-06,
+      "loss": 0.7373,
+      "step": 1054
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610254720645265e-06,
+      "loss": 0.7038,
+      "step": 1055
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609346063978593e-06,
+      "loss": 0.7387,
+      "step": 1056
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608436392377714e-06,
+      "loss": 0.5763,
+      "step": 1057
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607525706042928e-06,
+      "loss": 0.7268,
+      "step": 1058
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606614005174759e-06,
+      "loss": 0.7648,
+      "step": 1059
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605701289973957e-06,
+      "loss": 0.5773,
+      "step": 1060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604787560641492e-06,
+      "loss": 0.7294,
+      "step": 1061
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603872817378557e-06,
+      "loss": 0.747,
+      "step": 1062
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602957060386573e-06,
+      "loss": 0.7292,
+      "step": 1063
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602040289867178e-06,
+      "loss": 0.7357,
+      "step": 1064
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601122506022238e-06,
+      "loss": 0.7323,
+      "step": 1065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60020370905384e-06,
+      "loss": 0.5659,
+      "step": 1066
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599283899164293e-06,
+      "loss": 0.705,
+      "step": 1067
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598363076556132e-06,
+      "loss": 0.7586,
+      "step": 1068
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597441241432113e-06,
+      "loss": 0.7141,
+      "step": 1069
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.596518393995215e-06,
+      "loss": 0.7374,
+      "step": 1070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59559453444864e-06,
+      "loss": 0.7026,
+      "step": 1071
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594669662995815e-06,
+      "loss": 0.7517,
+      "step": 1072
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593743779840384e-06,
+      "loss": 0.6039,
+      "step": 1073
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592816885186221e-06,
+      "loss": 0.7197,
+      "step": 1074
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591888979237419e-06,
+      "loss": 0.6838,
+      "step": 1075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590960062198292e-06,
+      "loss": 0.5656,
+      "step": 1076
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59003013427338e-06,
+      "loss": 0.6967,
+      "step": 1077
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589099195667447e-06,
+      "loss": 0.5932,
+      "step": 1078
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588167246585474e-06,
+      "loss": 0.735,
+      "step": 1079
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587234287232665e-06,
+      "loss": 0.7,
+      "step": 1080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586300317814455e-06,
+      "loss": 0.7276,
+      "step": 1081
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58536533853649e-06,
+      "loss": 0.7264,
+      "step": 1082
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584429349604645e-06,
+      "loss": 0.6946,
+      "step": 1083
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583492351225014e-06,
+      "loss": 0.7449,
+      "step": 1084
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58255434360392e-06,
+      "loss": 0.7044,
+      "step": 1085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5816153269479e-06,
+      "loss": 0.5792,
+      "step": 1086
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580675301463718e-06,
+      "loss": 0.7206,
+      "step": 1087
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579734267358356e-06,
+      "loss": 0.7183,
+      "step": 1088
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578792224839025e-06,
+      "loss": 0.7215,
+      "step": 1089
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577849174113149e-06,
+      "loss": 0.7493,
+      "step": 1090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576905115388381e-06,
+      "loss": 0.7324,
+      "step": 1091
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575960048872595e-06,
+      "loss": 0.7001,
+      "step": 1092
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575013974773884e-06,
+      "loss": 0.6892,
+      "step": 1093
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.574066893300566e-06,
+      "loss": 0.5632,
+      "step": 1094
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573118804661179e-06,
+      "loss": 0.7076,
+      "step": 1095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572169709064482e-06,
+      "loss": 0.7128,
+      "step": 1096
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.571219606719457e-06,
+      "loss": 0.7102,
+      "step": 1097
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570268497835308e-06,
+      "loss": 0.7291,
+      "step": 1098
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56931638262146e-06,
+      "loss": 0.7299,
+      "step": 1099
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56836326128756e-06,
+      "loss": 0.7312,
+      "step": 1100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567409134043476e-06,
+      "loss": 0.728,
+      "step": 1101
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5664540010993e-06,
+      "loss": 0.5678,
+      "step": 1102
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565497862665337e-06,
+      "loss": 0.738,
+      "step": 1103
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.564540718952127e-06,
+      "loss": 0.5555,
+      "step": 1104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563582570170418e-06,
+      "loss": 0.7239,
+      "step": 1105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56262341653119e-06,
+      "loss": 0.716,
+      "step": 1106
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561663258245635e-06,
+      "loss": 0.5697,
+      "step": 1107
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.560702095525174e-06,
+      "loss": 0.5608,
+      "step": 1108
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559739928581447e-06,
+      "loss": 0.5835,
+      "step": 1109
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.558776757626309e-06,
+      "loss": 0.7347,
+      "step": 1110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557812582871846e-06,
+      "loss": 0.7288,
+      "step": 1111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556847404530359e-06,
+      "loss": 0.5884,
+      "step": 1112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555881222814369e-06,
+      "loss": 0.7083,
+      "step": 1113
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554914037936621e-06,
+      "loss": 0.5967,
+      "step": 1114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55394585011008e-06,
+      "loss": 0.6916,
+      "step": 1115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552976659547933e-06,
+      "loss": 0.7078,
+      "step": 1116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552006466463586e-06,
+      "loss": 0.7662,
+      "step": 1117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.551035271070665e-06,
+      "loss": 0.5975,
+      "step": 1118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.550063073583018e-06,
+      "loss": 0.7317,
+      "step": 1119
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549089874214717e-06,
+      "loss": 0.7218,
+      "step": 1120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548115673180047e-06,
+      "loss": 0.7041,
+      "step": 1121
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547140470693518e-06,
+      "loss": 0.5992,
+      "step": 1122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546164266969864e-06,
+      "loss": 0.7326,
+      "step": 1123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545187062224036e-06,
+      "loss": 0.5552,
+      "step": 1124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544208856671199e-06,
+      "loss": 0.7078,
+      "step": 1125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54322965052675e-06,
+      "loss": 0.7196,
+      "step": 1126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5422494440063e-06,
+      "loss": 0.706,
+      "step": 1127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.541268237325678e-06,
+      "loss": 0.7281,
+      "step": 1128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54028603070094e-06,
+      "loss": 0.7129,
+      "step": 1129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53930282434836e-06,
+      "loss": 0.7397,
+      "step": 1130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538318618484426e-06,
+      "loss": 0.7012,
+      "step": 1131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537333413325854e-06,
+      "loss": 0.7634,
+      "step": 1132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536347209089574e-06,
+      "loss": 0.727,
+      "step": 1133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53536000599274e-06,
+      "loss": 0.7211,
+      "step": 1134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534371804252727e-06,
+      "loss": 0.7271,
+      "step": 1135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533382604087127e-06,
+      "loss": 0.7496,
+      "step": 1136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53239240571375e-06,
+      "loss": 0.7441,
+      "step": 1137
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53140120935063e-06,
+      "loss": 0.5899,
+      "step": 1138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.530409015216018e-06,
+      "loss": 0.6834,
+      "step": 1139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.529415823528388e-06,
+      "loss": 0.5804,
+      "step": 1140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528421634506429e-06,
+      "loss": 0.591,
+      "step": 1141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527426448369053e-06,
+      "loss": 0.7413,
+      "step": 1142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52643026533539e-06,
+      "loss": 0.7054,
+      "step": 1143
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52543308562479e-06,
+      "loss": 0.6965,
+      "step": 1144
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524434909456822e-06,
+      "loss": 0.7133,
+      "step": 1145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.523435737051277e-06,
+      "loss": 0.5619,
+      "step": 1146
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.522435568628162e-06,
+      "loss": 0.5679,
+      "step": 1147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521434404407705e-06,
+      "loss": 0.5639,
+      "step": 1148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52043224461035e-06,
+      "loss": 0.5667,
+      "step": 1149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.519429089456769e-06,
+      "loss": 0.7325,
+      "step": 1150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518424939167842e-06,
+      "loss": 0.7016,
+      "step": 1151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.517419793964676e-06,
+      "loss": 0.6938,
+      "step": 1152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.516413654068593e-06,
+      "loss": 0.7085,
+      "step": 1153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515406519701136e-06,
+      "loss": 0.6894,
+      "step": 1154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.514398391084066e-06,
+      "loss": 0.7211,
+      "step": 1155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513389268439363e-06,
+      "loss": 0.6681,
+      "step": 1156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512379151989229e-06,
+      "loss": 0.6807,
+      "step": 1157
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.511368041956079e-06,
+      "loss": 0.7253,
+      "step": 1158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.510355938562552e-06,
+      "loss": 0.7049,
+      "step": 1159
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509342842031501e-06,
+      "loss": 0.5951,
+      "step": 1160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508328752586e-06,
+      "loss": 0.5698,
+      "step": 1161
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507313670449348e-06,
+      "loss": 0.7377,
+      "step": 1162
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.506297595845048e-06,
+      "loss": 0.7072,
+      "step": 1163
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.505280528996835e-06,
+      "loss": 0.7065,
+      "step": 1164
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.504262470128655e-06,
+      "loss": 0.6914,
+      "step": 1165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503243419464678e-06,
+      "loss": 0.7739,
+      "step": 1166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502223377229287e-06,
+      "loss": 0.7471,
+      "step": 1167
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.501202343647088e-06,
+      "loss": 0.7408,
+      "step": 1168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500180318942899e-06,
+      "loss": 0.7292,
+      "step": 1169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.499157303341761e-06,
+      "loss": 0.6942,
+      "step": 1170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498133297068937e-06,
+      "loss": 0.6084,
+      "step": 1171
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.497108300349898e-06,
+      "loss": 0.5962,
+      "step": 1172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49608231341034e-06,
+      "loss": 0.6759,
+      "step": 1173
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.495055336476176e-06,
+      "loss": 0.7051,
+      "step": 1174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494027369773536e-06,
+      "loss": 0.5718,
+      "step": 1175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49299841352877e-06,
+      "loss": 0.7273,
+      "step": 1176
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.491968467968445e-06,
+      "loss": 0.7177,
+      "step": 1177
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49093753331934e-06,
+      "loss": 0.5796,
+      "step": 1178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.489905609808462e-06,
+      "loss": 0.5692,
+      "step": 1179
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488872697663032e-06,
+      "loss": 0.6945,
+      "step": 1180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487838797110482e-06,
+      "loss": 0.699,
+      "step": 1181
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.486803908378472e-06,
+      "loss": 0.7243,
+      "step": 1182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.485768031694872e-06,
+      "loss": 0.5805,
+      "step": 1183
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.484731167287773e-06,
+      "loss": 0.717,
+      "step": 1184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483693315385482e-06,
+      "loss": 0.701,
+      "step": 1185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482654476216525e-06,
+      "loss": 0.5855,
+      "step": 1186
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.481614650009642e-06,
+      "loss": 0.5565,
+      "step": 1187
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480573836993799e-06,
+      "loss": 0.7559,
+      "step": 1188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479532037398167e-06,
+      "loss": 0.5507,
+      "step": 1189
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478489251452143e-06,
+      "loss": 0.7307,
+      "step": 1190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477445479385337e-06,
+      "loss": 0.57,
+      "step": 1191
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47640072142758e-06,
+      "loss": 0.6857,
+      "step": 1192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.475354977808917e-06,
+      "loss": 0.7361,
+      "step": 1193
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47430824875961e-06,
+      "loss": 0.564,
+      "step": 1194
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473260534510138e-06,
+      "loss": 0.6839,
+      "step": 1195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4722118352912e-06,
+      "loss": 0.6738,
+      "step": 1196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.471162151333708e-06,
+      "loss": 0.7236,
+      "step": 1197
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470111482868792e-06,
+      "loss": 0.7235,
+      "step": 1198
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.469059830127803e-06,
+      "loss": 0.5623,
+      "step": 1199
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4680071933423e-06,
+      "loss": 0.6898,
+      "step": 1200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466953572744064e-06,
+      "loss": 0.7102,
+      "step": 1201
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465898968565097e-06,
+      "loss": 0.7139,
+      "step": 1202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46484338103761e-06,
+      "loss": 0.7075,
+      "step": 1203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463786810394031e-06,
+      "loss": 0.7281,
+      "step": 1204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46272925686701e-06,
+      "loss": 0.6963,
+      "step": 1205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461670720689409e-06,
+      "loss": 0.7204,
+      "step": 1206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.460611202094308e-06,
+      "loss": 0.551,
+      "step": 1207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.459550701315001e-06,
+      "loss": 0.5696,
+      "step": 1208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458489218585003e-06,
+      "loss": 0.7124,
+      "step": 1209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.457426754138042e-06,
+      "loss": 0.7119,
+      "step": 1210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456363308208063e-06,
+      "loss": 0.7197,
+      "step": 1211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.455298881029223e-06,
+      "loss": 0.7019,
+      "step": 1212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.454233472835902e-06,
+      "loss": 0.6777,
+      "step": 1213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453167083862693e-06,
+      "loss": 0.6965,
+      "step": 1214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.452099714344402e-06,
+      "loss": 0.7318,
+      "step": 1215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451031364516056e-06,
+      "loss": 0.7088,
+      "step": 1216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.449962034612897e-06,
+      "loss": 0.7242,
+      "step": 1217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.448891724870378e-06,
+      "loss": 0.713,
+      "step": 1218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.447820435524173e-06,
+      "loss": 0.581,
+      "step": 1219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446748166810166e-06,
+      "loss": 0.5569,
+      "step": 1220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.445674918964466e-06,
+      "loss": 0.6642,
+      "step": 1221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44460069222339e-06,
+      "loss": 0.7241,
+      "step": 1222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.443525486823469e-06,
+      "loss": 0.7246,
+      "step": 1223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.442449303001458e-06,
+      "loss": 0.7209,
+      "step": 1224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.441372140994319e-06,
+      "loss": 0.6851,
+      "step": 1225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.440294001039235e-06,
+      "loss": 0.7047,
+      "step": 1226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439214883373603e-06,
+      "loss": 0.5602,
+      "step": 1227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.438134788235029e-06,
+      "loss": 0.6997,
+      "step": 1228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437053715861347e-06,
+      "loss": 0.7335,
+      "step": 1229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.435971666490593e-06,
+      "loss": 0.6472,
+      "step": 1230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434888640361027e-06,
+      "loss": 0.7033,
+      "step": 1231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43380463771112e-06,
+      "loss": 0.7172,
+      "step": 1232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43271965877956e-06,
+      "loss": 0.7325,
+      "step": 1233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43163370380525e-06,
+      "loss": 0.7217,
+      "step": 1234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430546773027302e-06,
+      "loss": 0.7184,
+      "step": 1235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.429458866685055e-06,
+      "loss": 0.6907,
+      "step": 1236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42836998501805e-06,
+      "loss": 0.7086,
+      "step": 1237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.427280128266049e-06,
+      "loss": 0.7173,
+      "step": 1238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.426189296669032e-06,
+      "loss": 0.7077,
+      "step": 1239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425097490467187e-06,
+      "loss": 0.7366,
+      "step": 1240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.424004709900918e-06,
+      "loss": 0.7644,
+      "step": 1241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.422910955210849e-06,
+      "loss": 0.7637,
+      "step": 1242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42181622663781e-06,
+      "loss": 0.5979,
+      "step": 1243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42072052442285e-06,
+      "loss": 0.6966,
+      "step": 1244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419623848807237e-06,
+      "loss": 0.7136,
+      "step": 1245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.418526200032445e-06,
+      "loss": 0.6913,
+      "step": 1246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.417427578340166e-06,
+      "loss": 0.7068,
+      "step": 1247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.416327983972304e-06,
+      "loss": 0.7194,
+      "step": 1248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.415227417170985e-06,
+      "loss": 0.7006,
+      "step": 1249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.414125878178537e-06,
+      "loss": 0.712,
+      "step": 1250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413023367237513e-06,
+      "loss": 0.7184,
+      "step": 1251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.411919884590675e-06,
+      "loss": 0.5703,
+      "step": 1252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410815430480994e-06,
+      "loss": 0.7109,
+      "step": 1253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40971000515167e-06,
+      "loss": 0.7219,
+      "step": 1254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408603608846098e-06,
+      "loss": 0.7056,
+      "step": 1255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.407496241807901e-06,
+      "loss": 0.6893,
+      "step": 1256
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.406387904280907e-06,
+      "loss": 0.7272,
+      "step": 1257
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.405278596509166e-06,
+      "loss": 0.7245,
+      "step": 1258
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.404168318736934e-06,
+      "loss": 0.7291,
+      "step": 1259
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403057071208685e-06,
+      "loss": 0.7132,
+      "step": 1260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.401944854169103e-06,
+      "loss": 0.7128,
+      "step": 1261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.400831667863091e-06,
+      "loss": 0.7258,
+      "step": 1262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.399717512535758e-06,
+      "loss": 0.6904,
+      "step": 1263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398602388432434e-06,
+      "loss": 0.591,
+      "step": 1264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.397486295798657e-06,
+      "loss": 0.7244,
+      "step": 1265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39636923488018e-06,
+      "loss": 0.6986,
+      "step": 1266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39525120592297e-06,
+      "loss": 0.6932,
+      "step": 1267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.394132209173205e-06,
+      "loss": 0.717,
+      "step": 1268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39301224487728e-06,
+      "loss": 0.5796,
+      "step": 1269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.391891313281798e-06,
+      "loss": 0.708,
+      "step": 1270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390769414633577e-06,
+      "loss": 0.7262,
+      "step": 1271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38964654917965e-06,
+      "loss": 0.7467,
+      "step": 1272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388522717167264e-06,
+      "loss": 0.7116,
+      "step": 1273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38739791884387e-06,
+      "loss": 0.7271,
+      "step": 1274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386272154457141e-06,
+      "loss": 0.7253,
+      "step": 1275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38514542425496e-06,
+      "loss": 0.743,
+      "step": 1276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.384017728485423e-06,
+      "loss": 0.7187,
+      "step": 1277
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.382889067396837e-06,
+      "loss": 0.7084,
+      "step": 1278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381759441237721e-06,
+      "loss": 0.6144,
+      "step": 1279
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.38062885025681e-06,
+      "loss": 0.7031,
+      "step": 1280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.379497294703048e-06,
+      "loss": 0.5899,
+      "step": 1281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.378364774825594e-06,
+      "loss": 0.5964,
+      "step": 1282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.377231290873817e-06,
+      "loss": 0.5807,
+      "step": 1283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376096843097299e-06,
+      "loss": 0.7153,
+      "step": 1284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.374961431745837e-06,
+      "loss": 0.7144,
+      "step": 1285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.373825057069436e-06,
+      "loss": 0.5625,
+      "step": 1286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372687719318316e-06,
+      "loss": 0.7032,
+      "step": 1287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371549418742907e-06,
+      "loss": 0.7124,
+      "step": 1288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.370410155593852e-06,
+      "loss": 0.7459,
+      "step": 1289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.369269930122004e-06,
+      "loss": 0.7112,
+      "step": 1290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.368128742578436e-06,
+      "loss": 0.7225,
+      "step": 1291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36698659321442e-06,
+      "loss": 0.5762,
+      "step": 1292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36584348228145e-06,
+      "loss": 0.695,
+      "step": 1293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.364699410031226e-06,
+      "loss": 0.7044,
+      "step": 1294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.363554376715662e-06,
+      "loss": 0.7397,
+      "step": 1295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.362408382586887e-06,
+      "loss": 0.5589,
+      "step": 1296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361261427897234e-06,
+      "loss": 0.6824,
+      "step": 1297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.360113512899254e-06,
+      "loss": 0.5799,
+      "step": 1298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358964637845704e-06,
+      "loss": 0.6847,
+      "step": 1299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35781480298956e-06,
+      "loss": 0.579,
+      "step": 1300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.356664008584e-06,
+      "loss": 0.6884,
+      "step": 1301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355512254882422e-06,
+      "loss": 0.6853,
+      "step": 1302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.354359542138428e-06,
+      "loss": 0.7185,
+      "step": 1303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.353205870605834e-06,
+      "loss": 0.6823,
+      "step": 1304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.352051240538674e-06,
+      "loss": 0.7149,
+      "step": 1305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.350895652191177e-06,
+      "loss": 0.6992,
+      "step": 1306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3497391058178e-06,
+      "loss": 0.7295,
+      "step": 1307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.348581601673198e-06,
+      "loss": 0.7249,
+      "step": 1308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347423140012248e-06,
+      "loss": 0.5896,
+      "step": 1309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34626372109003e-06,
+      "loss": 0.7165,
+      "step": 1310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.345103345161834e-06,
+      "loss": 0.5693,
+      "step": 1311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.343942012483169e-06,
+      "loss": 0.5703,
+      "step": 1312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.342779723309746e-06,
+      "loss": 0.5557,
+      "step": 1313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.341616477897489e-06,
+      "loss": 0.7222,
+      "step": 1314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340452276502537e-06,
+      "loss": 0.7155,
+      "step": 1315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.339287119381233e-06,
+      "loss": 0.5685,
+      "step": 1316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338121006790137e-06,
+      "loss": 0.5846,
+      "step": 1317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.336953938986013e-06,
+      "loss": 0.7098,
+      "step": 1318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335785916225837e-06,
+      "loss": 0.7239,
+      "step": 1319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334616938766801e-06,
+      "loss": 0.7318,
+      "step": 1320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3334470068663e-06,
+      "loss": 0.7348,
+      "step": 1321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.332276120781943e-06,
+      "loss": 0.5786,
+      "step": 1322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331104280771546e-06,
+      "loss": 0.5691,
+      "step": 1323
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.329931487093139e-06,
+      "loss": 0.7124,
+      "step": 1324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.328757740004956e-06,
+      "loss": 0.7303,
+      "step": 1325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.327583039765453e-06,
+      "loss": 0.7063,
+      "step": 1326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.326407386633281e-06,
+      "loss": 0.5542,
+      "step": 1327
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32523078086731e-06,
+      "loss": 0.7098,
+      "step": 1328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.324053222726616e-06,
+      "loss": 0.6766,
+      "step": 1329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.322874712470489e-06,
+      "loss": 0.6917,
+      "step": 1330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.321695250358422e-06,
+      "loss": 0.7103,
+      "step": 1331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.320514836650126e-06,
+      "loss": 0.6879,
+      "step": 1332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.319333471605513e-06,
+      "loss": 0.7274,
+      "step": 1333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31815115548471e-06,
+      "loss": 0.7097,
+      "step": 1334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31696788854805e-06,
+      "loss": 0.7052,
+      "step": 1335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31578367105608e-06,
+      "loss": 0.5693,
+      "step": 1336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31459850326955e-06,
+      "loss": 0.7069,
+      "step": 1337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.313412385449428e-06,
+      "loss": 0.7341,
+      "step": 1338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31222531785688e-06,
+      "loss": 0.7257,
+      "step": 1339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311037300753289e-06,
+      "loss": 0.6807,
+      "step": 1340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.309848334400247e-06,
+      "loss": 0.7096,
+      "step": 1341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30865841905955e-06,
+      "loss": 0.7303,
+      "step": 1342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30746755499321e-06,
+      "loss": 0.5848,
+      "step": 1343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.306275742463441e-06,
+      "loss": 0.7224,
+      "step": 1344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30508298173267e-06,
+      "loss": 0.6978,
+      "step": 1345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30388927306353e-06,
+      "loss": 0.5719,
+      "step": 1346
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.302694616718867e-06,
+      "loss": 0.7157,
+      "step": 1347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.30149901296173e-06,
+      "loss": 0.6926,
+      "step": 1348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.30030246205538e-06,
+      "loss": 0.683,
+      "step": 1349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.299104964263288e-06,
+      "loss": 0.6771,
+      "step": 1350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.297906519849132e-06,
+      "loss": 0.6782,
+      "step": 1351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.296707129076794e-06,
+      "loss": 0.7398,
+      "step": 1352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295506792210374e-06,
+      "loss": 0.7337,
+      "step": 1353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29430550951417e-06,
+      "loss": 0.6862,
+      "step": 1354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.293103281252697e-06,
+      "loss": 0.7232,
+      "step": 1355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29190010769067e-06,
+      "loss": 0.5616,
+      "step": 1356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29069598909302e-06,
+      "loss": 0.5578,
+      "step": 1357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.289490925724882e-06,
+      "loss": 0.6777,
+      "step": 1358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.288284917851598e-06,
+      "loss": 0.5507,
+      "step": 1359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28707796573872e-06,
+      "loss": 0.7142,
+      "step": 1360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.285870069652007e-06,
+      "loss": 0.5794,
+      "step": 1361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.284661229857428e-06,
+      "loss": 0.723,
+      "step": 1362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.283451446621155e-06,
+      "loss": 0.5624,
+      "step": 1363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.282240720209574e-06,
+      "loss": 0.6845,
+      "step": 1364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.281029050889274e-06,
+      "loss": 0.567,
+      "step": 1365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.279816438927052e-06,
+      "loss": 0.7003,
+      "step": 1366
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.278602884589916e-06,
+      "loss": 0.7273,
+      "step": 1367
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.277388388145077e-06,
+      "loss": 0.5507,
+      "step": 1368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.276172949859956e-06,
+      "loss": 0.7046,
+      "step": 1369
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.274956570002182e-06,
+      "loss": 0.7109,
+      "step": 1370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27373924883959e-06,
+      "loss": 0.7348,
+      "step": 1371
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.272520986640222e-06,
+      "loss": 0.69,
+      "step": 1372
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.271301783672328e-06,
+      "loss": 0.7169,
+      "step": 1373
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.270081640204367e-06,
+      "loss": 0.5841,
+      "step": 1374
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.268860556504998e-06,
+      "loss": 0.5576,
+      "step": 1375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.267638532843096e-06,
+      "loss": 0.7407,
+      "step": 1376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.266415569487738e-06,
+      "loss": 0.5633,
+      "step": 1377
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26519166670821e-06,
+      "loss": 0.7265,
+      "step": 1378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.263966824774e-06,
+      "loss": 0.7196,
+      "step": 1379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26274104395481e-06,
+      "loss": 0.5951,
+      "step": 1380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261514324520546e-06,
+      "loss": 0.7259,
+      "step": 1381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.260286666741315e-06,
+      "loss": 0.6991,
+      "step": 1382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.259058070887443e-06,
+      "loss": 0.701,
+      "step": 1383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25782853722945e-06,
+      "loss": 0.745,
+      "step": 1384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.256598066038067e-06,
+      "loss": 0.7132,
+      "step": 1385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.255366657584237e-06,
+      "loss": 0.7106,
+      "step": 1386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.254134312139098e-06,
+      "loss": 0.6395,
+      "step": 1387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.252901029974006e-06,
+      "loss": 0.7035,
+      "step": 1388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.251666811360516e-06,
+      "loss": 0.5553,
+      "step": 1389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25043165657039e-06,
+      "loss": 0.6904,
+      "step": 1390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2491955658756e-06,
+      "loss": 0.5724,
+      "step": 1391
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.247958539548319e-06,
+      "loss": 0.5698,
+      "step": 1392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24672057786093e-06,
+      "loss": 0.7243,
+      "step": 1393
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24548168108602e-06,
+      "loss": 0.6888,
+      "step": 1394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24424184949638e-06,
+      "loss": 0.6695,
+      "step": 1395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.243001083365015e-06,
+      "loss": 0.6906,
+      "step": 1396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241759382965123e-06,
+      "loss": 0.6809,
+      "step": 1397
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24051674857012e-06,
+      "loss": 0.6902,
+      "step": 1398
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.239273180453621e-06,
+      "loss": 0.7009,
+      "step": 1399
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.238028678889443e-06,
+      "loss": 0.7376,
+      "step": 1400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23678324415162e-06,
+      "loss": 0.5924,
+      "step": 1401
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.235536876514383e-06,
+      "loss": 0.6894,
+      "step": 1402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23428957625217e-06,
+      "loss": 0.5766,
+      "step": 1403
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.233041343639623e-06,
+      "loss": 0.7095,
+      "step": 1404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.231792178951594e-06,
+      "loss": 0.6076,
+      "step": 1405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.230542082463132e-06,
+      "loss": 0.6728,
+      "step": 1406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.229291054449502e-06,
+      "loss": 0.6798,
+      "step": 1407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.228039095186166e-06,
+      "loss": 0.732,
+      "step": 1408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226786204948794e-06,
+      "loss": 0.7073,
+      "step": 1409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.22553238401326e-06,
+      "loss": 0.5834,
+      "step": 1410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.224277632655645e-06,
+      "loss": 0.7267,
+      "step": 1411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.223021951152231e-06,
+      "loss": 0.579,
+      "step": 1412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.22176533977951e-06,
+      "loss": 0.6925,
+      "step": 1413
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.220507798814174e-06,
+      "loss": 0.698,
+      "step": 1414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.21924932853312e-06,
+      "loss": 0.7078,
+      "step": 1415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.217989929213453e-06,
+      "loss": 0.6995,
+      "step": 1416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.216729601132481e-06,
+      "loss": 0.5689,
+      "step": 1417
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.215468344567717e-06,
+      "loss": 0.6954,
+      "step": 1418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.214206159796876e-06,
+      "loss": 0.7215,
+      "step": 1419
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.21294304709788e-06,
+      "loss": 0.6769,
+      "step": 1420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.211679006748853e-06,
+      "loss": 0.6888,
+      "step": 1421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.210414039028128e-06,
+      "loss": 0.7148,
+      "step": 1422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.209148144214236e-06,
+      "loss": 0.7015,
+      "step": 1423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.207881322585915e-06,
+      "loss": 0.6734,
+      "step": 1424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.206613574422108e-06,
+      "loss": 0.5717,
+      "step": 1425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.205344900001963e-06,
+      "loss": 0.5909,
+      "step": 1426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.204075299604826e-06,
+      "loss": 0.5703,
+      "step": 1427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.202804773510254e-06,
+      "loss": 0.7195,
+      "step": 1428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.201533321998001e-06,
+      "loss": 0.6713,
+      "step": 1429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.200260945348034e-06,
+      "loss": 0.6974,
+      "step": 1430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.198987643840516e-06,
+      "loss": 0.6885,
+      "step": 1431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197713417755813e-06,
+      "loss": 0.7015,
+      "step": 1432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.196438267374501e-06,
+      "loss": 0.5744,
+      "step": 1433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.195162192977355e-06,
+      "loss": 0.7213,
+      "step": 1434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.193885194845356e-06,
+      "loss": 0.6714,
+      "step": 1435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.192607273259685e-06,
+      "loss": 0.5563,
+      "step": 1436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191328428501724e-06,
+      "loss": 0.7288,
+      "step": 1437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19004866085307e-06,
+      "loss": 0.7143,
+      "step": 1438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.188767970595514e-06,
+      "loss": 0.6817,
+      "step": 1439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.187486358011048e-06,
+      "loss": 0.6848,
+      "step": 1440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.186203823381873e-06,
+      "loss": 0.6998,
+      "step": 1441
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.184920366990393e-06,
+      "loss": 0.7284,
+      "step": 1442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.183635989119211e-06,
+      "loss": 0.6713,
+      "step": 1443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.182350690051134e-06,
+      "loss": 0.7018,
+      "step": 1444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181064470069175e-06,
+      "loss": 0.7021,
+      "step": 1445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.179777329456545e-06,
+      "loss": 0.6972,
+      "step": 1446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17848926849666e-06,
+      "loss": 0.7177,
+      "step": 1447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17720028747314e-06,
+      "loss": 0.6754,
+      "step": 1448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.175910386669809e-06,
+      "loss": 0.7295,
+      "step": 1449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.174619566370686e-06,
+      "loss": 0.6848,
+      "step": 1450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17332782686e-06,
+      "loss": 0.6935,
+      "step": 1451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17203516842218e-06,
+      "loss": 0.6926,
+      "step": 1452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.170741591341855e-06,
+      "loss": 0.722,
+      "step": 1453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.169447095903861e-06,
+      "loss": 0.581,
+      "step": 1454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.168151682393231e-06,
+      "loss": 0.568,
+      "step": 1455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.166855351095205e-06,
+      "loss": 0.694,
+      "step": 1456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.165558102295223e-06,
+      "loss": 0.6516,
+      "step": 1457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.164259936278923e-06,
+      "loss": 0.706,
+      "step": 1458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.162960853332154e-06,
+      "loss": 0.5735,
+      "step": 1459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.161660853740958e-06,
+      "loss": 0.7153,
+      "step": 1460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.160359937791585e-06,
+      "loss": 0.5897,
+      "step": 1461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159058105770485e-06,
+      "loss": 0.723,
+      "step": 1462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.157755357964303e-06,
+      "loss": 0.7377,
+      "step": 1463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.1564516946599e-06,
+      "loss": 0.5567,
+      "step": 1464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155147116144324e-06,
+      "loss": 0.6991,
+      "step": 1465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.153841622704834e-06,
+      "loss": 0.5357,
+      "step": 1466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.152535214628886e-06,
+      "loss": 0.6971,
+      "step": 1467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15122789220414e-06,
+      "loss": 0.7037,
+      "step": 1468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.149919655718453e-06,
+      "loss": 0.7284,
+      "step": 1469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14861050545989e-06,
+      "loss": 0.7246,
+      "step": 1470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14730044171671e-06,
+      "loss": 0.5995,
+      "step": 1471
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.145989464777378e-06,
+      "loss": 0.6981,
+      "step": 1472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14467757493056e-06,
+      "loss": 0.7348,
+      "step": 1473
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.143364772465121e-06,
+      "loss": 0.715,
+      "step": 1474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.142051057670127e-06,
+      "loss": 0.6993,
+      "step": 1475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.140736430834844e-06,
+      "loss": 0.5865,
+      "step": 1476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.139420892248744e-06,
+      "loss": 0.5721,
+      "step": 1477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.138104442201494e-06,
+      "loss": 0.6914,
+      "step": 1478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.136787080982963e-06,
+      "loss": 0.7221,
+      "step": 1479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.135468808883224e-06,
+      "loss": 0.7094,
+      "step": 1480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.134149626192544e-06,
+      "loss": 0.7171,
+      "step": 1481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.132829533201397e-06,
+      "loss": 0.6704,
+      "step": 1482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.131508530200457e-06,
+      "loss": 0.7214,
+      "step": 1483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.130186617480592e-06,
+      "loss": 0.7015,
+      "step": 1484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.128863795332879e-06,
+      "loss": 0.6971,
+      "step": 1485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.127540064048586e-06,
+      "loss": 0.6549,
+      "step": 1486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12621542391919e-06,
+      "loss": 0.5894,
+      "step": 1487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.124889875236362e-06,
+      "loss": 0.5609,
+      "step": 1488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.123563418291977e-06,
+      "loss": 0.7049,
+      "step": 1489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.122236053378108e-06,
+      "loss": 0.7023,
+      "step": 1490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.120907780787026e-06,
+      "loss": 0.6511,
+      "step": 1491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.119578600811204e-06,
+      "loss": 0.7052,
+      "step": 1492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11824851374332e-06,
+      "loss": 0.7063,
+      "step": 1493
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11691751987624e-06,
+      "loss": 0.6921,
+      "step": 1494
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.115585619503039e-06,
+      "loss": 0.5949,
+      "step": 1495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.114252812916988e-06,
+      "loss": 0.6624,
+      "step": 1496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.112919100411562e-06,
+      "loss": 0.5805,
+      "step": 1497
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.111584482280427e-06,
+      "loss": 0.5836,
+      "step": 1498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.110248958817455e-06,
+      "loss": 0.7115,
+      "step": 1499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.108912530316716e-06,
+      "loss": 0.705,
+      "step": 1500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.107575197072478e-06,
+      "loss": 0.7179,
+      "step": 1501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10623695937921e-06,
+      "loss": 0.7087,
+      "step": 1502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.104897817531579e-06,
+      "loss": 0.5668,
+      "step": 1503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10355777182445e-06,
+      "loss": 0.7338,
+      "step": 1504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.102216822552893e-06,
+      "loss": 0.585,
+      "step": 1505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.100874970012165e-06,
+      "loss": 0.7246,
+      "step": 1506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.099532214497735e-06,
+      "loss": 0.7328,
+      "step": 1507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.098188556305262e-06,
+      "loss": 0.7303,
+      "step": 1508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09684399573061e-06,
+      "loss": 0.7398,
+      "step": 1509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.095498533069834e-06,
+      "loss": 0.7112,
+      "step": 1510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.094152168619195e-06,
+      "loss": 0.6998,
+      "step": 1511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09280490267515e-06,
+      "loss": 0.6846,
+      "step": 1512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09145673553435e-06,
+      "loss": 0.6787,
+      "step": 1513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.090107667493655e-06,
+      "loss": 0.5778,
+      "step": 1514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.088757698850113e-06,
+      "loss": 0.6825,
+      "step": 1515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.087406829900973e-06,
+      "loss": 0.7278,
+      "step": 1516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.086055060943686e-06,
+      "loss": 0.6782,
+      "step": 1517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0847023922759e-06,
+      "loss": 0.7123,
+      "step": 1518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.083348824195456e-06,
+      "loss": 0.5692,
+      "step": 1519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.081994357000402e-06,
+      "loss": 0.6649,
+      "step": 1520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.080638990988971e-06,
+      "loss": 0.7119,
+      "step": 1521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.079282726459608e-06,
+      "loss": 0.7257,
+      "step": 1522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07792556371095e-06,
+      "loss": 0.5648,
+      "step": 1523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.076567503041827e-06,
+      "loss": 0.5835,
+      "step": 1524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.075208544751272e-06,
+      "loss": 0.6906,
+      "step": 1525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.073848689138519e-06,
+      "loss": 0.6749,
+      "step": 1526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072487936502992e-06,
+      "loss": 0.7098,
+      "step": 1527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.071126287144315e-06,
+      "loss": 0.6891,
+      "step": 1528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06976374136231e-06,
+      "loss": 0.656,
+      "step": 1529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.068400299457e-06,
+      "loss": 0.7076,
+      "step": 1530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.067035961728598e-06,
+      "loss": 0.7208,
+      "step": 1531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.065670728477523e-06,
+      "loss": 0.709,
+      "step": 1532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06430460000438e-06,
+      "loss": 0.6815,
+      "step": 1533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.062937576609983e-06,
+      "loss": 0.7112,
+      "step": 1534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.061569658595335e-06,
+      "loss": 0.6878,
+      "step": 1535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.060200846261639e-06,
+      "loss": 0.7118,
+      "step": 1536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.058831139910295e-06,
+      "loss": 0.7167,
+      "step": 1537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0574605398429e-06,
+      "loss": 0.7274,
+      "step": 1538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.056089046361243e-06,
+      "loss": 0.5717,
+      "step": 1539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.054716659767319e-06,
+      "loss": 0.6761,
+      "step": 1540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.053343380363312e-06,
+      "loss": 0.7164,
+      "step": 1541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.051969208451606e-06,
+      "loss": 0.6809,
+      "step": 1542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.050594144334778e-06,
+      "loss": 0.7006,
+      "step": 1543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.049218188315608e-06,
+      "loss": 0.7026,
+      "step": 1544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.047841340697065e-06,
+      "loss": 0.6734,
+      "step": 1545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.04646360178232e-06,
+      "loss": 0.5823,
+      "step": 1546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.045084971874738e-06,
+      "loss": 0.5716,
+      "step": 1547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.043705451277878e-06,
+      "loss": 0.5792,
+      "step": 1548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0423250402955e-06,
+      "loss": 0.701,
+      "step": 1549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.040943739231556e-06,
+      "loss": 0.6891,
+      "step": 1550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.039561548390195e-06,
+      "loss": 0.7018,
+      "step": 1551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.038178468075764e-06,
+      "loss": 0.6956,
+      "step": 1552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.036794498592802e-06,
+      "loss": 0.5801,
+      "step": 1553
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.035409640246046e-06,
+      "loss": 0.7202,
+      "step": 1554
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.034023893340431e-06,
+      "loss": 0.7037,
+      "step": 1555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.032637258181085e-06,
+      "loss": 0.7253,
+      "step": 1556
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.031249735073328e-06,
+      "loss": 0.6956,
+      "step": 1557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.029861324322684e-06,
+      "loss": 0.6947,
+      "step": 1558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.028472026234865e-06,
+      "loss": 0.6565,
+      "step": 1559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.027081841115784e-06,
+      "loss": 0.5709,
+      "step": 1560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.025690769271543e-06,
+      "loss": 0.7026,
+      "step": 1561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.024298811008445e-06,
+      "loss": 0.699,
+      "step": 1562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.022905966632986e-06,
+      "loss": 0.683,
+      "step": 1563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.021512236451857e-06,
+      "loss": 0.6908,
+      "step": 1564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.020117620771943e-06,
+      "loss": 0.712,
+      "step": 1565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.018722119900327e-06,
+      "loss": 0.7112,
+      "step": 1566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.017325734144283e-06,
+      "loss": 0.6668,
+      "step": 1567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.015928463811284e-06,
+      "loss": 0.6991,
+      "step": 1568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.014530309208993e-06,
+      "loss": 0.61,
+      "step": 1569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.013131270645272e-06,
+      "loss": 0.5404,
+      "step": 1570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.011731348428177e-06,
+      "loss": 0.7029,
+      "step": 1571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.010330542865956e-06,
+      "loss": 0.55,
+      "step": 1572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.008928854267054e-06,
+      "loss": 0.6875,
+      "step": 1573
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.007526282940108e-06,
+      "loss": 0.7087,
+      "step": 1574
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.006122829193953e-06,
+      "loss": 0.6595,
+      "step": 1575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.004718493337614e-06,
+      "loss": 0.7208,
+      "step": 1576
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.003313275680315e-06,
+      "loss": 0.5711,
+      "step": 1577
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.00190717653147e-06,
+      "loss": 0.7088,
+      "step": 1578
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.000500196200687e-06,
+      "loss": 0.7112,
+      "step": 1579
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.999092334997773e-06,
+      "loss": 0.5487,
+      "step": 1580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.997683593232722e-06,
+      "loss": 0.6629,
+      "step": 1581
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99627397121573e-06,
+      "loss": 0.7017,
+      "step": 1582
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99486346925718e-06,
+      "loss": 0.5861,
+      "step": 1583
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99345208766765e-06,
+      "loss": 0.7007,
+      "step": 1584
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.992039826757914e-06,
+      "loss": 0.6931,
+      "step": 1585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99062668683894e-06,
+      "loss": 0.6869,
+      "step": 1586
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.989212668221885e-06,
+      "loss": 0.6845,
+      "step": 1587
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987797771218104e-06,
+      "loss": 0.6919,
+      "step": 1588
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.986381996139144e-06,
+      "loss": 0.7153,
+      "step": 1589
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.984965343296745e-06,
+      "loss": 0.6586,
+      "step": 1590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.98354781300284e-06,
+      "loss": 0.5719,
+      "step": 1591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.982129405569556e-06,
+      "loss": 0.5824,
+      "step": 1592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.980710121309214e-06,
+      "loss": 0.5798,
+      "step": 1593
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.979289960534326e-06,
+      "loss": 0.6794,
+      "step": 1594
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.977868923557597e-06,
+      "loss": 0.5703,
+      "step": 1595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.976447010691927e-06,
+      "loss": 0.7159,
+      "step": 1596
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.975024222250407e-06,
+      "loss": 0.5837,
+      "step": 1597
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.973600558546322e-06,
+      "loss": 0.6898,
+      "step": 1598
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97217601989315e-06,
+      "loss": 0.5744,
+      "step": 1599
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97075060660456e-06,
+      "loss": 0.7241,
+      "step": 1600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969324318994412e-06,
+      "loss": 0.6867,
+      "step": 1601
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.967897157376766e-06,
+      "loss": 0.7226,
+      "step": 1602
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.966469122065865e-06,
+      "loss": 0.5528,
+      "step": 1603
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.965040213376152e-06,
+      "loss": 0.5743,
+      "step": 1604
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.963610431622258e-06,
+      "loss": 0.705,
+      "step": 1605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.962179777119007e-06,
+      "loss": 0.7077,
+      "step": 1606
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.960748250181415e-06,
+      "loss": 0.6909,
+      "step": 1607
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95931585112469e-06,
+      "loss": 0.7173,
+      "step": 1608
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957882580264235e-06,
+      "loss": 0.6735,
+      "step": 1609
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.956448437915643e-06,
+      "loss": 0.6834,
+      "step": 1610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.955013424394696e-06,
+      "loss": 0.6908,
+      "step": 1611
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95357754001737e-06,
+      "loss": 0.7138,
+      "step": 1612
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.952140785099836e-06,
+      "loss": 0.715,
+      "step": 1613
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.950703159958454e-06,
+      "loss": 0.713,
+      "step": 1614
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.949264664909772e-06,
+      "loss": 0.5634,
+      "step": 1615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.947825300270535e-06,
+      "loss": 0.6981,
+      "step": 1616
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.946385066357674e-06,
+      "loss": 0.7131,
+      "step": 1617
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.944943963488321e-06,
+      "loss": 0.69,
+      "step": 1618
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.943501991979791e-06,
+      "loss": 0.6792,
+      "step": 1619
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.942059152149589e-06,
+      "loss": 0.7024,
+      "step": 1620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.940615444315418e-06,
+      "loss": 0.6928,
+      "step": 1621
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.939170868795169e-06,
+      "loss": 0.7,
+      "step": 1622
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93772542590692e-06,
+      "loss": 0.7094,
+      "step": 1623
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.936279115968948e-06,
+      "loss": 0.5748,
+      "step": 1624
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.934831939299715e-06,
+      "loss": 0.6755,
+      "step": 1625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.933383896217877e-06,
+      "loss": 0.561,
+      "step": 1626
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.931934987042277e-06,
+      "loss": 0.7212,
+      "step": 1627
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93048521209195e-06,
+      "loss": 0.5651,
+      "step": 1628
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.929034571686128e-06,
+      "loss": 0.7294,
+      "step": 1629
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.927583066144223e-06,
+      "loss": 0.6737,
+      "step": 1630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.926130695785846e-06,
+      "loss": 0.6872,
+      "step": 1631
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.924677460930793e-06,
+      "loss": 0.5821,
+      "step": 1632
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.923223361899053e-06,
+      "loss": 0.6927,
+      "step": 1633
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.921768399010805e-06,
+      "loss": 0.6981,
+      "step": 1634
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.920312572586418e-06,
+      "loss": 0.6964,
+      "step": 1635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.918855882946454e-06,
+      "loss": 0.6902,
+      "step": 1636
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.917398330411656e-06,
+      "loss": 0.7265,
+      "step": 1637
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.91593991530297e-06,
+      "loss": 0.6584,
+      "step": 1638
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.914480637941519e-06,
+      "loss": 0.7194,
+      "step": 1639
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.913020498648625e-06,
+      "loss": 0.6748,
+      "step": 1640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.911559497745798e-06,
+      "loss": 0.6777,
+      "step": 1641
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.910097635554736e-06,
+      "loss": 0.705,
+      "step": 1642
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.908634912397324e-06,
+      "loss": 0.6845,
+      "step": 1643
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.907171328595643e-06,
+      "loss": 0.5913,
+      "step": 1644
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90570688447196e-06,
+      "loss": 0.5779,
+      "step": 1645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.904241580348731e-06,
+      "loss": 0.6648,
+      "step": 1646
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.902775416548602e-06,
+      "loss": 0.5633,
+      "step": 1647
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.901308393394407e-06,
+      "loss": 0.7079,
+      "step": 1648
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.899840511209173e-06,
+      "loss": 0.6696,
+      "step": 1649
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.898371770316113e-06,
+      "loss": 0.5771,
+      "step": 1650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.896902171038629e-06,
+      "loss": 0.6635,
+      "step": 1651
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.895431713700313e-06,
+      "loss": 0.6865,
+      "step": 1652
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.893960398624947e-06,
+      "loss": 0.6917,
+      "step": 1653
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.8924882261365e-06,
+      "loss": 0.6843,
+      "step": 1654
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.891015196559128e-06,
+      "loss": 0.6718,
+      "step": 1655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.889541310217181e-06,
+      "loss": 0.7095,
+      "step": 1656
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.888066567435194e-06,
+      "loss": 0.6859,
+      "step": 1657
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.886590968537892e-06,
+      "loss": 0.5786,
+      "step": 1658
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.885114513850187e-06,
+      "loss": 0.7081,
+      "step": 1659
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88363720369718e-06,
+      "loss": 0.7457,
+      "step": 1660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88215903840416e-06,
+      "loss": 0.6767,
+      "step": 1661
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.880680018296606e-06,
+      "loss": 0.5772,
+      "step": 1662
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.879200143700186e-06,
+      "loss": 0.6676,
+      "step": 1663
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.877719414940751e-06,
+      "loss": 0.566,
+      "step": 1664
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.876237832344345e-06,
+      "loss": 0.693,
+      "step": 1665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.8747553962372e-06,
+      "loss": 0.6905,
+      "step": 1666
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.873272106945731e-06,
+      "loss": 0.5677,
+      "step": 1667
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.871787964796546e-06,
+      "loss": 0.7101,
+      "step": 1668
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87030297011644e-06,
+      "loss": 0.5829,
+      "step": 1669
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.868817123232391e-06,
+      "loss": 0.5822,
+      "step": 1670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.867330424471574e-06,
+      "loss": 0.6814,
+      "step": 1671
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.865842874161341e-06,
+      "loss": 0.6957,
+      "step": 1672
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.864354472629238e-06,
+      "loss": 0.715,
+      "step": 1673
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.862865220202998e-06,
+      "loss": 0.5526,
+      "step": 1674
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86137511721054e-06,
+      "loss": 0.6914,
+      "step": 1675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.859884163979969e-06,
+      "loss": 0.6922,
+      "step": 1676
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85839236083958e-06,
+      "loss": 0.7054,
+      "step": 1677
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.856899708117854e-06,
+      "loss": 0.6962,
+      "step": 1678
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855406206143458e-06,
+      "loss": 0.7123,
+      "step": 1679
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.853911855245249e-06,
+      "loss": 0.6589,
+      "step": 1680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.852416655752264e-06,
+      "loss": 0.5632,
+      "step": 1681
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.850920607993738e-06,
+      "loss": 0.6907,
+      "step": 1682
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.849423712299084e-06,
+      "loss": 0.6955,
+      "step": 1683
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.847925968997901e-06,
+      "loss": 0.7036,
+      "step": 1684
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.846427378419983e-06,
+      "loss": 0.6904,
+      "step": 1685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.844927940895302e-06,
+      "loss": 0.7161,
+      "step": 1686
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84342765675402e-06,
+      "loss": 0.5823,
+      "step": 1687
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.841926526326487e-06,
+      "loss": 0.7005,
+      "step": 1688
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.840424549943239e-06,
+      "loss": 0.6964,
+      "step": 1689
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.838921727934992e-06,
+      "loss": 0.5815,
+      "step": 1690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.837418060632657e-06,
+      "loss": 0.6989,
+      "step": 1691
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.835913548367326e-06,
+      "loss": 0.7189,
+      "step": 1692
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.834408191470278e-06,
+      "loss": 0.7167,
+      "step": 1693
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.832901990272979e-06,
+      "loss": 0.732,
+      "step": 1694
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83139494510708e-06,
+      "loss": 0.7178,
+      "step": 1695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.829887056304419e-06,
+      "loss": 0.6949,
+      "step": 1696
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.828378324197017e-06,
+      "loss": 0.683,
+      "step": 1697
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.826868749117083e-06,
+      "loss": 0.7252,
+      "step": 1698
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.825358331397013e-06,
+      "loss": 0.6712,
+      "step": 1699
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.823847071369382e-06,
+      "loss": 0.6983,
+      "step": 1700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.822334969366961e-06,
+      "loss": 0.6709,
+      "step": 1701
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.820822025722695e-06,
+      "loss": 0.7298,
+      "step": 1702
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.819308240769726e-06,
+      "loss": 0.5895,
+      "step": 1703
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.817793614841368e-06,
+      "loss": 0.6878,
+      "step": 1704
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.816278148271132e-06,
+      "loss": 0.6915,
+      "step": 1705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.814761841392708e-06,
+      "loss": 0.5901,
+      "step": 1706
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.813244694539974e-06,
+      "loss": 0.6559,
+      "step": 1707
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.811726708046986e-06,
+      "loss": 0.6851,
+      "step": 1708
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.810207882247996e-06,
+      "loss": 0.5643,
+      "step": 1709
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.808688217477435e-06,
+      "loss": 0.569,
+      "step": 1710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.807167714069914e-06,
+      "loss": 0.7003,
+      "step": 1711
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.805646372360236e-06,
+      "loss": 0.702,
+      "step": 1712
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.804124192683387e-06,
+      "loss": 0.5863,
+      "step": 1713
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.802601175374535e-06,
+      "loss": 0.6976,
+      "step": 1714
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.801077320769034e-06,
+      "loss": 0.6,
+      "step": 1715
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.799552629202424e-06,
+      "loss": 0.5478,
+      "step": 1716
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.798027101010427e-06,
+      "loss": 0.6941,
+      "step": 1717
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.796500736528947e-06,
+      "loss": 0.6749,
+      "step": 1718
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79497353609408e-06,
+      "loss": 0.7192,
+      "step": 1719
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.793445500042097e-06,
+      "loss": 0.706,
+      "step": 1720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.791916628709456e-06,
+      "loss": 0.7136,
+      "step": 1721
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.790386922432805e-06,
+      "loss": 0.5586,
+      "step": 1722
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.788856381548969e-06,
+      "loss": 0.7082,
+      "step": 1723
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.787325006394953e-06,
+      "loss": 0.6735,
+      "step": 1724
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.785792797307959e-06,
+      "loss": 0.7094,
+      "step": 1725
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.784259754625362e-06,
+      "loss": 0.7117,
+      "step": 1726
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.782725878684722e-06,
+      "loss": 0.688,
+      "step": 1727
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.781191169823786e-06,
+      "loss": 0.6842,
+      "step": 1728
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.779655628380479e-06,
+      "loss": 0.6875,
+      "step": 1729
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.778119254692916e-06,
+      "loss": 0.5754,
+      "step": 1730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.776582049099392e-06,
+      "loss": 0.6822,
+      "step": 1731
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.775044011938382e-06,
+      "loss": 0.6969,
+      "step": 1732
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.773505143548551e-06,
+      "loss": 0.5919,
+      "step": 1733
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.771965444268741e-06,
+      "loss": 0.6601,
+      "step": 1734
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77042491443798e-06,
+      "loss": 0.5663,
+      "step": 1735
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.768883554395475e-06,
+      "loss": 0.7036,
+      "step": 1736
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.767341364480624e-06,
+      "loss": 0.7092,
+      "step": 1737
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.765798345032999e-06,
+      "loss": 0.5668,
+      "step": 1738
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.764254496392357e-06,
+      "loss": 0.7077,
+      "step": 1739
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.762709818898642e-06,
+      "loss": 0.5423,
+      "step": 1740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.761164312891978e-06,
+      "loss": 0.5672,
+      "step": 1741
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.759617978712667e-06,
+      "loss": 0.7005,
+      "step": 1742
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.7580708167012e-06,
+      "loss": 0.6977,
+      "step": 1743
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.756522827198246e-06,
+      "loss": 0.6824,
+      "step": 1744
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.754974010544658e-06,
+      "loss": 0.5883,
+      "step": 1745
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75342436708147e-06,
+      "loss": 0.6966,
+      "step": 1746
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.7518738971499e-06,
+      "loss": 0.7025,
+      "step": 1747
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.750322601091347e-06,
+      "loss": 0.6578,
+      "step": 1748
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.748770479247389e-06,
+      "loss": 0.6881,
+      "step": 1749
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.747217531959793e-06,
+      "loss": 0.5753,
+      "step": 1750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.745663759570499e-06,
+      "loss": 0.6724,
+      "step": 1751
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.744109162421633e-06,
+      "loss": 0.6966,
+      "step": 1752
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.742553740855507e-06,
+      "loss": 0.7144,
+      "step": 1753
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740997495214604e-06,
+      "loss": 0.7246,
+      "step": 1754
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.7394404258416e-06,
+      "loss": 0.6907,
+      "step": 1755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.737882533079344e-06,
+      "loss": 0.6917,
+      "step": 1756
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.736323817270868e-06,
+      "loss": 0.6583,
+      "step": 1757
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73476427875939e-06,
+      "loss": 0.7061,
+      "step": 1758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.733203917888301e-06,
+      "loss": 0.5933,
+      "step": 1759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.731642735001184e-06,
+      "loss": 0.7013,
+      "step": 1760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.730080730441791e-06,
+      "loss": 0.5747,
+      "step": 1761
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.728517904554063e-06,
+      "loss": 0.5577,
+      "step": 1762
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.726954257682119e-06,
+      "loss": 0.7208,
+      "step": 1763
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.725389790170256e-06,
+      "loss": 0.5932,
+      "step": 1764
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.72382450236296e-06,
+      "loss": 0.7288,
+      "step": 1765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.722258394604892e-06,
+      "loss": 0.6675,
+      "step": 1766
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.720691467240892e-06,
+      "loss": 0.5804,
+      "step": 1767
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71912372061598e-06,
+      "loss": 0.6854,
+      "step": 1768
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.717555155075366e-06,
+      "loss": 0.5649,
+      "step": 1769
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.715985770964427e-06,
+      "loss": 0.6845,
+      "step": 1770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.714415568628729e-06,
+      "loss": 0.5728,
+      "step": 1771
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.712844548414014e-06,
+      "loss": 0.6819,
+      "step": 1772
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.711272710666207e-06,
+      "loss": 0.7013,
+      "step": 1773
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.709700055731414e-06,
+      "loss": 0.5501,
+      "step": 1774
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.708126583955913e-06,
+      "loss": 0.6729,
+      "step": 1775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.706552295686176e-06,
+      "loss": 0.7022,
+      "step": 1776
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.704977191268836e-06,
+      "loss": 0.5598,
+      "step": 1777
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.703401271050722e-06,
+      "loss": 0.7039,
+      "step": 1778
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.701824535378839e-06,
+      "loss": 0.6753,
+      "step": 1779
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.700246984600364e-06,
+      "loss": 0.5634,
+      "step": 1780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69866861906266e-06,
+      "loss": 0.6766,
+      "step": 1781
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69708943911327e-06,
+      "loss": 0.5794,
+      "step": 1782
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.695509445099915e-06,
+      "loss": 0.702,
+      "step": 1783
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69392863737049e-06,
+      "loss": 0.5633,
+      "step": 1784
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69234701627308e-06,
+      "loss": 0.697,
+      "step": 1785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.690764582155937e-06,
+      "loss": 0.6945,
+      "step": 1786
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.6891813353675e-06,
+      "loss": 0.6902,
+      "step": 1787
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.687597276256389e-06,
+      "loss": 0.7075,
+      "step": 1788
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.686012405171394e-06,
+      "loss": 0.6514,
+      "step": 1789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.68442672246149e-06,
+      "loss": 0.6737,
+      "step": 1790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.682840228475828e-06,
+      "loss": 0.549,
+      "step": 1791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.681252923563741e-06,
+      "loss": 0.5835,
+      "step": 1792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.679664808074737e-06,
+      "loss": 0.6705,
+      "step": 1793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.678075882358506e-06,
+      "loss": 0.552,
+      "step": 1794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.676486146764912e-06,
+      "loss": 0.6919,
+      "step": 1795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.674895601644e-06,
+      "loss": 0.6826,
+      "step": 1796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.673304247345994e-06,
+      "loss": 0.5671,
+      "step": 1797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.671712084221294e-06,
+      "loss": 0.7338,
+      "step": 1798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.670119112620481e-06,
+      "loss": 0.5574,
+      "step": 1799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66852533289431e-06,
+      "loss": 0.5736,
+      "step": 1800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.666930745393719e-06,
+      "loss": 0.6693,
+      "step": 1801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.665335350469815e-06,
+      "loss": 0.721,
+      "step": 1802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.663739148473894e-06,
+      "loss": 0.5719,
+      "step": 1803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.662142139757427e-06,
+      "loss": 0.7186,
+      "step": 1804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.660544324672053e-06,
+      "loss": 0.706,
+      "step": 1805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.6589457035696e-06,
+      "loss": 0.6805,
+      "step": 1806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.657346276802071e-06,
+      "loss": 0.6538,
+      "step": 1807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65574604472164e-06,
+      "loss": 0.6548,
+      "step": 1808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.654145007680665e-06,
+      "loss": 0.6855,
+      "step": 1809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65254316603168e-06,
+      "loss": 0.7045,
+      "step": 1810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.650940520127393e-06,
+      "loss": 0.6985,
+      "step": 1811
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.649337070320694e-06,
+      "loss": 0.6947,
+      "step": 1812
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.647732816964644e-06,
+      "loss": 0.6978,
+      "step": 1813
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.646127760412489e-06,
+      "loss": 0.6883,
+      "step": 1814
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.644521901017644e-06,
+      "loss": 0.6721,
+      "step": 1815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.642915239133704e-06,
+      "loss": 0.5924,
+      "step": 1816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.641307775114443e-06,
+      "loss": 0.707,
+      "step": 1817
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.639699509313807e-06,
+      "loss": 0.6966,
+      "step": 1818
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.638090442085923e-06,
+      "loss": 0.5654,
+      "step": 1819
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.636480573785089e-06,
+      "loss": 0.5625,
+      "step": 1820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.634869904765784e-06,
+      "loss": 0.7063,
+      "step": 1821
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.633258435382662e-06,
+      "loss": 0.7127,
+      "step": 1822
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.631646165990558e-06,
+      "loss": 0.7013,
+      "step": 1823
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63003309694447e-06,
+      "loss": 0.6928,
+      "step": 1824
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.628419228599587e-06,
+      "loss": 0.6631,
+      "step": 1825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.626804561311264e-06,
+      "loss": 0.5755,
+      "step": 1826
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.625189095435038e-06,
+      "loss": 0.6795,
+      "step": 1827
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.623572831326617e-06,
+      "loss": 0.5801,
+      "step": 1828
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.621955769341888e-06,
+      "loss": 0.7001,
+      "step": 1829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.62033790983691e-06,
+      "loss": 0.6635,
+      "step": 1830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.618719253167927e-06,
+      "loss": 0.6868,
+      "step": 1831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.617099799691346e-06,
+      "loss": 0.685,
+      "step": 1832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.615479549763756e-06,
+      "loss": 0.6717,
+      "step": 1833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.613858503741923e-06,
+      "loss": 0.6776,
+      "step": 1834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.612236661982785e-06,
+      "loss": 0.6992,
+      "step": 1835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.610614024843453e-06,
+      "loss": 0.5609,
+      "step": 1836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.60899059268122e-06,
+      "loss": 0.6574,
+      "step": 1837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.60736636585355e-06,
+      "loss": 0.6883,
+      "step": 1838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.605741344718081e-06,
+      "loss": 0.6877,
+      "step": 1839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.604115529632628e-06,
+      "loss": 0.5662,
+      "step": 1840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.60248892095518e-06,
+      "loss": 0.6873,
+      "step": 1841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.6008615190439e-06,
+      "loss": 0.6753,
+      "step": 1842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.599233324257127e-06,
+      "loss": 0.5905,
+      "step": 1843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.597604336953374e-06,
+      "loss": 0.7127,
+      "step": 1844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.595974557491329e-06,
+      "loss": 0.6886,
+      "step": 1845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.594343986229854e-06,
+      "loss": 0.5856,
+      "step": 1846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.592712623527984e-06,
+      "loss": 0.5596,
+      "step": 1847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59108046974493e-06,
+      "loss": 0.6755,
+      "step": 1848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.589447525240079e-06,
+      "loss": 0.6889,
+      "step": 1849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.587813790372985e-06,
+      "loss": 0.538,
+      "step": 1850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.586179265503387e-06,
+      "loss": 0.6986,
+      "step": 1851
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.584543950991186e-06,
+      "loss": 0.6915,
+      "step": 1852
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.582907847196468e-06,
+      "loss": 0.7167,
+      "step": 1853
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.581270954479483e-06,
+      "loss": 0.6922,
+      "step": 1854
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.579633273200662e-06,
+      "loss": 0.6841,
+      "step": 1855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.577994803720605e-06,
+      "loss": 0.6612,
+      "step": 1856
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.57635554640009e-06,
+      "loss": 0.7108,
+      "step": 1857
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.574715501600064e-06,
+      "loss": 0.7286,
+      "step": 1858
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.57307466968165e-06,
+      "loss": 0.687,
+      "step": 1859
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.571433051006143e-06,
+      "loss": 0.6975,
+      "step": 1860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.569790645935013e-06,
+      "loss": 0.6104,
+      "step": 1861
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.568147454829898e-06,
+      "loss": 0.6834,
+      "step": 1862
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.566503478052619e-06,
+      "loss": 0.6965,
+      "step": 1863
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.564858715965158e-06,
+      "loss": 0.636,
+      "step": 1864
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.563213168929683e-06,
+      "loss": 0.6712,
+      "step": 1865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.561566837308521e-06,
+      "loss": 0.7058,
+      "step": 1866
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.559919721464183e-06,
+      "loss": 0.6976,
+      "step": 1867
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.558271821759347e-06,
+      "loss": 0.7005,
+      "step": 1868
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.556623138556865e-06,
+      "loss": 0.6846,
+      "step": 1869
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.55497367221976e-06,
+      "loss": 0.5473,
+      "step": 1870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.55332342311123e-06,
+      "loss": 0.6875,
+      "step": 1871
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.551672391594646e-06,
+      "loss": 0.6716,
+      "step": 1872
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.550020578033544e-06,
+      "loss": 0.7158,
+      "step": 1873
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.548367982791643e-06,
+      "loss": 0.6847,
+      "step": 1874
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.546714606232828e-06,
+      "loss": 0.7142,
+      "step": 1875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.545060448721155e-06,
+      "loss": 0.6873,
+      "step": 1876
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.543405510620855e-06,
+      "loss": 0.6847,
+      "step": 1877
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.541749792296327e-06,
+      "loss": 0.6696,
+      "step": 1878
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.540093294112149e-06,
+      "loss": 0.6865,
+      "step": 1879
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.538436016433062e-06,
+      "loss": 0.6728,
+      "step": 1880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.536777959623985e-06,
+      "loss": 0.6751,
+      "step": 1881
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.535119124050006e-06,
+      "loss": 0.6847,
+      "step": 1882
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.533459510076385e-06,
+      "loss": 0.7153,
+      "step": 1883
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.531799118068554e-06,
+      "loss": 0.581,
+      "step": 1884
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.530137948392113e-06,
+      "loss": 0.5866,
+      "step": 1885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.528476001412838e-06,
+      "loss": 0.7197,
+      "step": 1886
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.526813277496672e-06,
+      "loss": 0.663,
+      "step": 1887
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.525149777009733e-06,
+      "loss": 0.6945,
+      "step": 1888
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.523485500318307e-06,
+      "loss": 0.715,
+      "step": 1889
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.521820447788853e-06,
+      "loss": 0.655,
+      "step": 1890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.520154619788e-06,
+      "loss": 0.5794,
+      "step": 1891
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.518488016682545e-06,
+      "loss": 0.6584,
+      "step": 1892
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.516820638839458e-06,
+      "loss": 0.7299,
+      "step": 1893
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.515152486625883e-06,
+      "loss": 0.6899,
+      "step": 1894
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.51348356040913e-06,
+      "loss": 0.6304,
+      "step": 1895
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.51181386055668e-06,
+      "loss": 0.6791,
+      "step": 1896
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.510143387436183e-06,
+      "loss": 0.6712,
+      "step": 1897
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.508472141415468e-06,
+      "loss": 0.578,
+      "step": 1898
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.50680012286252e-06,
+      "loss": 0.6781,
+      "step": 1899
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.505127332145505e-06,
+      "loss": 0.7174,
+      "step": 1900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.503453769632756e-06,
+      "loss": 0.5811,
+      "step": 1901
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.501779435692774e-06,
+      "loss": 0.6764,
+      "step": 1902
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.500104330694234e-06,
+      "loss": 0.719,
+      "step": 1903
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.498428455005975e-06,
+      "loss": 0.57,
+      "step": 1904
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.496751808997011e-06,
+      "loss": 0.6982,
+      "step": 1905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.495074393036523e-06,
+      "loss": 0.6845,
+      "step": 1906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.493396207493863e-06,
+      "loss": 0.5712,
+      "step": 1907
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.49171725273855e-06,
+      "loss": 0.5635,
+      "step": 1908
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.490037529140274e-06,
+      "loss": 0.5622,
+      "step": 1909
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.488357037068893e-06,
+      "loss": 0.6734,
+      "step": 1910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.48667577689444e-06,
+      "loss": 0.5605,
+      "step": 1911
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.484993748987107e-06,
+      "loss": 0.5472,
+      "step": 1912
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.483310953717265e-06,
+      "loss": 0.5623,
+      "step": 1913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.481627391455447e-06,
+      "loss": 0.5524,
+      "step": 1914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.479943062572358e-06,
+      "loss": 0.6716,
+      "step": 1915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.47825796743887e-06,
+      "loss": 0.5638,
+      "step": 1916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.476572106426028e-06,
+      "loss": 0.6965,
+      "step": 1917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.474885479905039e-06,
+      "loss": 0.6648,
+      "step": 1918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.473198088247286e-06,
+      "loss": 0.698,
+      "step": 1919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.471509931824313e-06,
+      "loss": 0.6948,
+      "step": 1920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.469821011007837e-06,
+      "loss": 0.6983,
+      "step": 1921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.468131326169744e-06,
+      "loss": 0.6689,
+      "step": 1922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.466440877682084e-06,
+      "loss": 0.6882,
+      "step": 1923
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46474966591708e-06,
+      "loss": 0.7363,
+      "step": 1924
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.463057691247121e-06,
+      "loss": 0.6749,
+      "step": 1925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46136495404476e-06,
+      "loss": 0.6797,
+      "step": 1926
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.459671454682728e-06,
+      "loss": 0.7094,
+      "step": 1927
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.457977193533911e-06,
+      "loss": 0.6814,
+      "step": 1928
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.45628217097137e-06,
+      "loss": 0.7088,
+      "step": 1929
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.454586387368338e-06,
+      "loss": 0.6752,
+      "step": 1930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.452889843098206e-06,
+      "loss": 0.5606,
+      "step": 1931
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.451192538534538e-06,
+      "loss": 0.5673,
+      "step": 1932
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.449494474051065e-06,
+      "loss": 0.5461,
+      "step": 1933
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.447795650021685e-06,
+      "loss": 0.6821,
+      "step": 1934
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.44609606682046e-06,
+      "loss": 0.7021,
+      "step": 1935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.444395724821624e-06,
+      "loss": 0.7165,
+      "step": 1936
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.442694624399576e-06,
+      "loss": 0.5632,
+      "step": 1937
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.440992765928884e-06,
+      "loss": 0.703,
+      "step": 1938
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43929014978428e-06,
+      "loss": 0.715,
+      "step": 1939
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.437586776340662e-06,
+      "loss": 0.6585,
+      "step": 1940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.435882645973098e-06,
+      "loss": 0.7025,
+      "step": 1941
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.434177759056822e-06,
+      "loss": 0.7132,
+      "step": 1942
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.432472115967233e-06,
+      "loss": 0.6789,
+      "step": 1943
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.430765717079895e-06,
+      "loss": 0.6726,
+      "step": 1944
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.429058562770547e-06,
+      "loss": 0.668,
+      "step": 1945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.427350653415085e-06,
+      "loss": 0.6865,
+      "step": 1946
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.425641989389572e-06,
+      "loss": 0.663,
+      "step": 1947
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.423932571070243e-06,
+      "loss": 0.6698,
+      "step": 1948
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.422222398833492e-06,
+      "loss": 0.6826,
+      "step": 1949
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.420511473055887e-06,
+      "loss": 0.5746,
+      "step": 1950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.418799794114154e-06,
+      "loss": 0.5714,
+      "step": 1951
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.417087362385192e-06,
+      "loss": 0.6882,
+      "step": 1952
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.415374178246059e-06,
+      "loss": 0.5725,
+      "step": 1953
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.413660242073982e-06,
+      "loss": 0.5537,
+      "step": 1954
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.411945554246358e-06,
+      "loss": 0.6924,
+      "step": 1955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.410230115140738e-06,
+      "loss": 0.5618,
+      "step": 1956
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40851392513485e-06,
+      "loss": 0.6651,
+      "step": 1957
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.406796984606583e-06,
+      "loss": 0.6806,
+      "step": 1958
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.405079293933986e-06,
+      "loss": 0.6666,
+      "step": 1959
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.403360853495284e-06,
+      "loss": 0.6789,
+      "step": 1960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40164166366886e-06,
+      "loss": 0.5654,
+      "step": 1961
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.39992172483326e-06,
+      "loss": 0.6938,
+      "step": 1962
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.398201037367202e-06,
+      "loss": 0.5659,
+      "step": 1963
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.396479601649565e-06,
+      "loss": 0.6133,
+      "step": 1964
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.394757418059388e-06,
+      "loss": 0.7067,
+      "step": 1965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.393034486975884e-06,
+      "loss": 0.659,
+      "step": 1966
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.391310808778425e-06,
+      "loss": 0.6552,
+      "step": 1967
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.389586383846548e-06,
+      "loss": 0.6776,
+      "step": 1968
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.387861212559958e-06,
+      "loss": 0.695,
+      "step": 1969
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.386135295298516e-06,
+      "loss": 0.5733,
+      "step": 1970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.384408632442255e-06,
+      "loss": 0.6933,
+      "step": 1971
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.382681224371372e-06,
+      "loss": 0.6787,
+      "step": 1972
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.380953071466223e-06,
+      "loss": 0.5885,
+      "step": 1973
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.37922417410733e-06,
+      "loss": 0.6917,
+      "step": 1974
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.377494532675382e-06,
+      "loss": 0.6679,
+      "step": 1975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.37576414755123e-06,
+      "loss": 0.7003,
+      "step": 1976
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.374033019115889e-06,
+      "loss": 0.6524,
+      "step": 1977
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.372301147750534e-06,
+      "loss": 0.6591,
+      "step": 1978
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.370568533836508e-06,
+      "loss": 0.5523,
+      "step": 1979
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.368835177755316e-06,
+      "loss": 0.5462,
+      "step": 1980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.367101079888627e-06,
+      "loss": 0.7209,
+      "step": 1981
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.365366240618272e-06,
+      "loss": 0.5791,
+      "step": 1982
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.363630660326248e-06,
+      "loss": 0.5658,
+      "step": 1983
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.361894339394712e-06,
+      "loss": 0.6806,
+      "step": 1984
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.360157278205985e-06,
+      "loss": 0.5576,
+      "step": 1985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.358419477142553e-06,
+      "loss": 0.6822,
+      "step": 1986
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.35668093658706e-06,
+      "loss": 0.6706,
+      "step": 1987
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.354941656922317e-06,
+      "loss": 0.7153,
+      "step": 1988
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.3532016385313e-06,
+      "loss": 0.6782,
+      "step": 1989
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.351460881797145e-06,
+      "loss": 0.6759,
+      "step": 1990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.349719387103143e-06,
+      "loss": 0.6838,
+      "step": 1991
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.347977154832763e-06,
+      "loss": 0.6483,
+      "step": 1992
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.346234185369624e-06,
+      "loss": 0.6662,
+      "step": 1993
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.344490479097509e-06,
+      "loss": 0.726,
+      "step": 1994
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.34274603640037e-06,
+      "loss": 0.6733,
+      "step": 1995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.341000857662313e-06,
+      "loss": 0.5975,
+      "step": 1996
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.339254943267614e-06,
+      "loss": 0.7402,
+      "step": 1997
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.337508293600703e-06,
+      "loss": 0.6988,
+      "step": 1998
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.335760909046177e-06,
+      "loss": 0.672,
+      "step": 1999
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.334012789988795e-06,
+      "loss": 0.6797,
+      "step": 2000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.332263936813473e-06,
+      "loss": 0.7137,
+      "step": 2001
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.330514349905295e-06,
+      "loss": 0.6658,
+      "step": 2002
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.3287640296495e-06,
+      "loss": 0.7007,
+      "step": 2003
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.327012976431495e-06,
+      "loss": 0.6894,
+      "step": 2004
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.325261190636845e-06,
+      "loss": 0.5857,
+      "step": 2005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.323508672651272e-06,
+      "loss": 0.6662,
+      "step": 2006
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.321755422860671e-06,
+      "loss": 0.6627,
+      "step": 2007
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.320001441651086e-06,
+      "loss": 0.5856,
+      "step": 2008
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.31824672940873e-06,
+      "loss": 0.6841,
+      "step": 2009
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.316491286519971e-06,
+      "loss": 0.6794,
+      "step": 2010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.314735113371344e-06,
+      "loss": 0.6824,
+      "step": 2011
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.312978210349538e-06,
+      "loss": 0.6988,
+      "step": 2012
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.31122057784141e-06,
+      "loss": 0.6857,
+      "step": 2013
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.30946221623397e-06,
+      "loss": 0.7062,
+      "step": 2014
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.307703125914397e-06,
+      "loss": 0.7122,
+      "step": 2015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.305943307270023e-06,
+      "loss": 0.6652,
+      "step": 2016
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.304182760688345e-06,
+      "loss": 0.7102,
+      "step": 2017
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.302421486557015e-06,
+      "loss": 0.6726,
+      "step": 2018
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.300659485263854e-06,
+      "loss": 0.7064,
+      "step": 2019
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.298896757196835e-06,
+      "loss": 0.5737,
+      "step": 2020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.297133302744094e-06,
+      "loss": 0.7041,
+      "step": 2021
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.295369122293926e-06,
+      "loss": 0.7031,
+      "step": 2022
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29360421623479e-06,
+      "loss": 0.5638,
+      "step": 2023
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.291838584955298e-06,
+      "loss": 0.6413,
+      "step": 2024
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29007222884423e-06,
+      "loss": 0.6494,
+      "step": 2025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.288305148290516e-06,
+      "loss": 0.6939,
+      "step": 2026
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.286537343683251e-06,
+      "loss": 0.6878,
+      "step": 2027
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.284768815411693e-06,
+      "loss": 0.6693,
+      "step": 2028
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.282999563865248e-06,
+      "loss": 0.7057,
+      "step": 2029
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.281229589433495e-06,
+      "loss": 0.6719,
+      "step": 2030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.279458892506163e-06,
+      "loss": 0.687,
+      "step": 2031
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.277687473473142e-06,
+      "loss": 0.5822,
+      "step": 2032
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.275915332724482e-06,
+      "loss": 0.6902,
+      "step": 2033
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.274142470650392e-06,
+      "loss": 0.5708,
+      "step": 2034
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.272368887641238e-06,
+      "loss": 0.5695,
+      "step": 2035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.27059458408755e-06,
+      "loss": 0.6756,
+      "step": 2036
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.268819560380006e-06,
+      "loss": 0.6751,
+      "step": 2037
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.267043816909457e-06,
+      "loss": 0.6555,
+      "step": 2038
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.2652673540669e-06,
+      "loss": 0.6653,
+      "step": 2039
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.263490172243497e-06,
+      "loss": 0.5628,
+      "step": 2040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.261712271830564e-06,
+      "loss": 0.6912,
+      "step": 2041
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.259933653219583e-06,
+      "loss": 0.6978,
+      "step": 2042
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.258154316802185e-06,
+      "loss": 0.556,
+      "step": 2043
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.256374262970162e-06,
+      "loss": 0.6767,
+      "step": 2044
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.254593492115468e-06,
+      "loss": 0.6376,
+      "step": 2045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.25281200463021e-06,
+      "loss": 0.6998,
+      "step": 2046
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.251029800906656e-06,
+      "loss": 0.7231,
+      "step": 2047
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.249246881337227e-06,
+      "loss": 0.6647,
+      "step": 2048
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.24746324631451e-06,
+      "loss": 0.6763,
+      "step": 2049
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.245678896231241e-06,
+      "loss": 0.6464,
+      "step": 2050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.243893831480318e-06,
+      "loss": 0.7001,
+      "step": 2051
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.242108052454794e-06,
+      "loss": 0.5782,
+      "step": 2052
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.240321559547882e-06,
+      "loss": 0.7083,
+      "step": 2053
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.23853435315295e-06,
+      "loss": 0.6519,
+      "step": 2054
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.236746433663528e-06,
+      "loss": 0.6887,
+      "step": 2055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.234957801473292e-06,
+      "loss": 0.6584,
+      "step": 2056
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.233168456976084e-06,
+      "loss": 0.6741,
+      "step": 2057
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.231378400565904e-06,
+      "loss": 0.5605,
+      "step": 2058
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.229587632636902e-06,
+      "loss": 0.6687,
+      "step": 2059
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.227796153583389e-06,
+      "loss": 0.5697,
+      "step": 2060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.226003963799833e-06,
+      "loss": 0.5843,
+      "step": 2061
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.224211063680854e-06,
+      "loss": 0.6511,
+      "step": 2062
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.222417453621235e-06,
+      "loss": 0.5808,
+      "step": 2063
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.22062313401591e-06,
+      "loss": 0.5547,
+      "step": 2064
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.218828105259971e-06,
+      "loss": 0.6547,
+      "step": 2065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.217032367748667e-06,
+      "loss": 0.692,
+      "step": 2066
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.215235921877403e-06,
+      "loss": 0.7054,
+      "step": 2067
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.213438768041738e-06,
+      "loss": 0.6918,
+      "step": 2068
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.211640906637389e-06,
+      "loss": 0.6497,
+      "step": 2069
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.209842338060228e-06,
+      "loss": 0.7186,
+      "step": 2070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.208043062706282e-06,
+      "loss": 0.5766,
+      "step": 2071
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.206243080971734e-06,
+      "loss": 0.7057,
+      "step": 2072
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.204442393252925e-06,
+      "loss": 0.6772,
+      "step": 2073
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.202640999946348e-06,
+      "loss": 0.701,
+      "step": 2074
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.200838901448651e-06,
+      "loss": 0.5602,
+      "step": 2075
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.19903609815664e-06,
+      "loss": 0.5708,
+      "step": 2076
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.197232590467278e-06,
+      "loss": 0.5776,
+      "step": 2077
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.195428378777678e-06,
+      "loss": 0.6565,
+      "step": 2078
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.19362346348511e-06,
+      "loss": 0.5729,
+      "step": 2079
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.191817844986998e-06,
+      "loss": 0.6801,
+      "step": 2080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.190011523680925e-06,
+      "loss": 0.5736,
+      "step": 2081
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.188204499964622e-06,
+      "loss": 0.7172,
+      "step": 2082
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.186396774235986e-06,
+      "loss": 0.6625,
+      "step": 2083
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.184588346893052e-06,
+      "loss": 0.6688,
+      "step": 2084
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.182779218334021e-06,
+      "loss": 0.6708,
+      "step": 2085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.18096938895725e-06,
+      "loss": 0.7037,
+      "step": 2086
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.179158859161242e-06,
+      "loss": 0.6538,
+      "step": 2087
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.17734762934466e-06,
+      "loss": 0.5812,
+      "step": 2088
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.17553569990632e-06,
+      "loss": 0.6816,
+      "step": 2089
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.173723071245192e-06,
+      "loss": 0.6646,
+      "step": 2090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.171909743760397e-06,
+      "loss": 0.6743,
+      "step": 2091
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.170095717851215e-06,
+      "loss": 0.6974,
+      "step": 2092
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.168280993917078e-06,
+      "loss": 0.5741,
+      "step": 2093
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.166465572357567e-06,
+      "loss": 0.6811,
+      "step": 2094
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.164649453572423e-06,
+      "loss": 0.5617,
+      "step": 2095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.162832637961541e-06,
+      "loss": 0.6703,
+      "step": 2096
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.16101512592496e-06,
+      "loss": 0.6792,
+      "step": 2097
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.159196917862886e-06,
+      "loss": 0.6769,
+      "step": 2098
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.157378014175666e-06,
+      "loss": 0.6884,
+      "step": 2099
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.155558415263807e-06,
+      "loss": 0.6782,
+      "step": 2100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.153738121527968e-06,
+      "loss": 0.6704,
+      "step": 2101
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.15191713336896e-06,
+      "loss": 0.5801,
+      "step": 2102
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.150095451187747e-06,
+      "loss": 0.6822,
+      "step": 2103
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.148273075385445e-06,
+      "loss": 0.7054,
+      "step": 2104
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.146450006363327e-06,
+      "loss": 0.5767,
+      "step": 2105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.144626244522812e-06,
+      "loss": 0.6862,
+      "step": 2106
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.142801790265477e-06,
+      "loss": 0.5918,
+      "step": 2107
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.140976643993051e-06,
+      "loss": 0.6874,
+      "step": 2108
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.139150806107408e-06,
+      "loss": 0.6763,
+      "step": 2109
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.137324277010587e-06,
+      "loss": 0.6844,
+      "step": 2110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.135497057104768e-06,
+      "loss": 0.6758,
+      "step": 2111
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.133669146792289e-06,
+      "loss": 0.6912,
+      "step": 2112
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13184054647564e-06,
+      "loss": 0.6731,
+      "step": 2113
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.130011256557459e-06,
+      "loss": 0.5952,
+      "step": 2114
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.128181277440539e-06,
+      "loss": 0.695,
+      "step": 2115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.126350609527821e-06,
+      "loss": 0.5631,
+      "step": 2116
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.124519253222409e-06,
+      "loss": 0.6833,
+      "step": 2117
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.122687208927541e-06,
+      "loss": 0.6826,
+      "step": 2118
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.120854477046621e-06,
+      "loss": 0.6959,
+      "step": 2119
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.119021057983199e-06,
+      "loss": 0.5545,
+      "step": 2120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.117186952140973e-06,
+      "loss": 0.6737,
+      "step": 2121
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.115352159923799e-06,
+      "loss": 0.6655,
+      "step": 2122
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.113516681735682e-06,
+      "loss": 0.6766,
+      "step": 2123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.111680517980772e-06,
+      "loss": 0.697,
+      "step": 2124
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.109843669063379e-06,
+      "loss": 0.6935,
+      "step": 2125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.108006135387958e-06,
+      "loss": 0.7145,
+      "step": 2126
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.106167917359115e-06,
+      "loss": 0.732,
+      "step": 2127
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10432901538161e-06,
+      "loss": 0.6738,
+      "step": 2128
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.102489429860355e-06,
+      "loss": 0.658,
+      "step": 2129
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.100649161200403e-06,
+      "loss": 0.7098,
+      "step": 2130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.098808209806966e-06,
+      "loss": 0.6784,
+      "step": 2131
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.096966576085406e-06,
+      "loss": 0.687,
+      "step": 2132
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.09512426044123e-06,
+      "loss": 0.6942,
+      "step": 2133
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.093281263280102e-06,
+      "loss": 0.5902,
+      "step": 2134
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.091437585007828e-06,
+      "loss": 0.6658,
+      "step": 2135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.089593226030375e-06,
+      "loss": 0.5605,
+      "step": 2136
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.087748186753845e-06,
+      "loss": 0.5779,
+      "step": 2137
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.085902467584505e-06,
+      "loss": 0.6945,
+      "step": 2138
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.084056068928762e-06,
+      "loss": 0.6856,
+      "step": 2139
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.082208991193175e-06,
+      "loss": 0.7075,
+      "step": 2140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.080361234784455e-06,
+      "loss": 0.7072,
+      "step": 2141
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.078512800109457e-06,
+      "loss": 0.6656,
+      "step": 2142
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.076663687575192e-06,
+      "loss": 0.571,
+      "step": 2143
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.074813897588815e-06,
+      "loss": 0.7043,
+      "step": 2144
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.072963430557636e-06,
+      "loss": 0.667,
+      "step": 2145
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.071112286889105e-06,
+      "loss": 0.5849,
+      "step": 2146
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.069260466990827e-06,
+      "loss": 0.7098,
+      "step": 2147
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.06740797127056e-06,
+      "loss": 0.6973,
+      "step": 2148
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.065554800136201e-06,
+      "loss": 0.6674,
+      "step": 2149
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.063700953995802e-06,
+      "loss": 0.6715,
+      "step": 2150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.061846433257563e-06,
+      "loss": 0.6843,
+      "step": 2151
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.059991238329833e-06,
+      "loss": 0.6736,
+      "step": 2152
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.058135369621105e-06,
+      "loss": 0.566,
+      "step": 2153
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.056278827540025e-06,
+      "loss": 0.5558,
+      "step": 2154
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.054421612495387e-06,
+      "loss": 0.6617,
+      "step": 2155
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.052563724896131e-06,
+      "loss": 0.5882,
+      "step": 2156
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.050705165151346e-06,
+      "loss": 0.6911,
+      "step": 2157
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.048845933670274e-06,
+      "loss": 0.67,
+      "step": 2158
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.04698603086229e-06,
+      "loss": 0.5799,
+      "step": 2159
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.045125457136936e-06,
+      "loss": 0.5627,
+      "step": 2160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.043264212903889e-06,
+      "loss": 0.6756,
+      "step": 2161
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.041402298572979e-06,
+      "loss": 0.591,
+      "step": 2162
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.039539714554179e-06,
+      "loss": 0.683,
+      "step": 2163
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.037676461257616e-06,
+      "loss": 0.5386,
+      "step": 2164
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.035812539093557e-06,
+      "loss": 0.6886,
+      "step": 2165
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.033947948472422e-06,
+      "loss": 0.6675,
+      "step": 2166
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.032082689804777e-06,
+      "loss": 0.693,
+      "step": 2167
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.030216763501331e-06,
+      "loss": 0.7012,
+      "step": 2168
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.028350169972945e-06,
+      "loss": 0.7029,
+      "step": 2169
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.026482909630626e-06,
+      "loss": 0.6877,
+      "step": 2170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.024614982885527e-06,
+      "loss": 0.6781,
+      "step": 2171
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.022746390148948e-06,
+      "loss": 0.7066,
+      "step": 2172
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.020877131832332e-06,
+      "loss": 0.7066,
+      "step": 2173
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.019007208347276e-06,
+      "loss": 0.6829,
+      "step": 2174
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.017136620105518e-06,
+      "loss": 0.6386,
+      "step": 2175
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.015265367518941e-06,
+      "loss": 0.6482,
+      "step": 2176
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.013393450999582e-06,
+      "loss": 0.6507,
+      "step": 2177
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.011520870959616e-06,
+      "loss": 0.6859,
+      "step": 2178
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.009647627811368e-06,
+      "loss": 0.5651,
+      "step": 2179
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.007773721967309e-06,
+      "loss": 0.5628,
+      "step": 2180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.005899153840055e-06,
+      "loss": 0.6978,
+      "step": 2181
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.004023923842367e-06,
+      "loss": 0.6719,
+      "step": 2182
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.002148032387154e-06,
+      "loss": 0.7314,
+      "step": 2183
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.00027147988747e-06,
+      "loss": 0.6491,
+      "step": 2184
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.998394266756511e-06,
+      "loss": 0.7139,
+      "step": 2185
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.996516393407625e-06,
+      "loss": 0.5835,
+      "step": 2186
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.9946378602543e-06,
+      "loss": 0.6932,
+      "step": 2187
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.992758667710173e-06,
+      "loss": 0.5659,
+      "step": 2188
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99087881618902e-06,
+      "loss": 0.6785,
+      "step": 2189
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.988998306104773e-06,
+      "loss": 0.6712,
+      "step": 2190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.987117137871497e-06,
+      "loss": 0.6851,
+      "step": 2191
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.98523531190341e-06,
+      "loss": 0.5653,
+      "step": 2192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.98335282861487e-06,
+      "loss": 0.6736,
+      "step": 2193
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.981469688420385e-06,
+      "loss": 0.6892,
+      "step": 2194
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.979585891734602e-06,
+      "loss": 0.563,
+      "step": 2195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.977701438972317e-06,
+      "loss": 0.576,
+      "step": 2196
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.975816330548466e-06,
+      "loss": 0.688,
+      "step": 2197
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.973930566878137e-06,
+      "loss": 0.6598,
+      "step": 2198
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.972044148376552e-06,
+      "loss": 0.6527,
+      "step": 2199
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.970157075459083e-06,
+      "loss": 0.6751,
+      "step": 2200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.968269348541249e-06,
+      "loss": 0.5732,
+      "step": 2201
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.966380968038706e-06,
+      "loss": 0.7034,
+      "step": 2202
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.964491934367259e-06,
+      "loss": 0.685,
+      "step": 2203
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.962602247942855e-06,
+      "loss": 0.6732,
+      "step": 2204
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.960711909181586e-06,
+      "loss": 0.6897,
+      "step": 2205
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.958820918499686e-06,
+      "loss": 0.6932,
+      "step": 2206
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95692927631353e-06,
+      "loss": 0.5642,
+      "step": 2207
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.955036983039646e-06,
+      "loss": 0.5379,
+      "step": 2208
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.953144039094693e-06,
+      "loss": 0.5873,
+      "step": 2209
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.951250444895485e-06,
+      "loss": 0.6803,
+      "step": 2210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.94935620085897e-06,
+      "loss": 0.7019,
+      "step": 2211
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.947461307402239e-06,
+      "loss": 0.6907,
+      "step": 2212
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.945565764942536e-06,
+      "loss": 0.6533,
+      "step": 2213
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.943669573897238e-06,
+      "loss": 0.6941,
+      "step": 2214
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.941772734683868e-06,
+      "loss": 0.675,
+      "step": 2215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.939875247720096e-06,
+      "loss": 0.7177,
+      "step": 2216
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.937977113423725e-06,
+      "loss": 0.6619,
+      "step": 2217
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.93607833221271e-06,
+      "loss": 0.6874,
+      "step": 2218
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.934178904505143e-06,
+      "loss": 0.6878,
+      "step": 2219
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.93227883071926e-06,
+      "loss": 0.5937,
+      "step": 2220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.93037811127344e-06,
+      "loss": 0.6911,
+      "step": 2221
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.928476746586203e-06,
+      "loss": 0.6764,
+      "step": 2222
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92657473707621e-06,
+      "loss": 0.65,
+      "step": 2223
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.924672083162268e-06,
+      "loss": 0.6771,
+      "step": 2224
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.922768785263321e-06,
+      "loss": 0.6766,
+      "step": 2225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92086484379846e-06,
+      "loss": 0.7188,
+      "step": 2226
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.918960259186912e-06,
+      "loss": 0.5677,
+      "step": 2227
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.917055031848049e-06,
+      "loss": 0.714,
+      "step": 2228
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.915149162201384e-06,
+      "loss": 0.6745,
+      "step": 2229
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.913242650666572e-06,
+      "loss": 0.6667,
+      "step": 2230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.911335497663408e-06,
+      "loss": 0.6926,
+      "step": 2231
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.909427703611829e-06,
+      "loss": 0.5636,
+      "step": 2232
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.907519268931916e-06,
+      "loss": 0.6308,
+      "step": 2233
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.905610194043883e-06,
+      "loss": 0.6993,
+      "step": 2234
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.903700479368095e-06,
+      "loss": 0.5573,
+      "step": 2235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.901790125325049e-06,
+      "loss": 0.5617,
+      "step": 2236
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.89987913233539e-06,
+      "loss": 0.6739,
+      "step": 2237
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.897967500819898e-06,
+      "loss": 0.5796,
+      "step": 2238
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.896055231199497e-06,
+      "loss": 0.6699,
+      "step": 2239
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.894142323895253e-06,
+      "loss": 0.6852,
+      "step": 2240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.892228779328367e-06,
+      "loss": 0.6765,
+      "step": 2241
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.890314597920183e-06,
+      "loss": 0.6909,
+      "step": 2242
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.88839978009219e-06,
+      "loss": 0.7076,
+      "step": 2243
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.886484326266006e-06,
+      "loss": 0.6874,
+      "step": 2244
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.884568236863402e-06,
+      "loss": 0.6863,
+      "step": 2245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.882651512306278e-06,
+      "loss": 0.6689,
+      "step": 2246
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.88073415301668e-06,
+      "loss": 0.6787,
+      "step": 2247
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.878816159416793e-06,
+      "loss": 0.6388,
+      "step": 2248
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.876897531928943e-06,
+      "loss": 0.6963,
+      "step": 2249
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.874978270975589e-06,
+      "loss": 0.6827,
+      "step": 2250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.873058376979336e-06,
+      "loss": 0.6849,
+      "step": 2251
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.871137850362928e-06,
+      "loss": 0.6628,
+      "step": 2252
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.869216691549246e-06,
+      "loss": 0.6634,
+      "step": 2253
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.86729490096131e-06,
+      "loss": 0.6652,
+      "step": 2254
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.86537247902228e-06,
+      "loss": 0.7106,
+      "step": 2255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.863449426155455e-06,
+      "loss": 0.6575,
+      "step": 2256
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.861525742784273e-06,
+      "loss": 0.6514,
+      "step": 2257
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.859601429332311e-06,
+      "loss": 0.6842,
+      "step": 2258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.857676486223285e-06,
+      "loss": 0.6648,
+      "step": 2259
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.85575091388105e-06,
+      "loss": 0.6902,
+      "step": 2260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.853824712729597e-06,
+      "loss": 0.6273,
+      "step": 2261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.851897883193057e-06,
+      "loss": 0.6945,
+      "step": 2262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.8499704256957e-06,
+      "loss": 0.5877,
+      "step": 2263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.848042340661937e-06,
+      "loss": 0.6912,
+      "step": 2264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.846113628516309e-06,
+      "loss": 0.5925,
+      "step": 2265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.844184289683503e-06,
+      "loss": 0.7141,
+      "step": 2266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.84225432458834e-06,
+      "loss": 0.6539,
+      "step": 2267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.84032373365578e-06,
+      "loss": 0.5587,
+      "step": 2268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.838392517310921e-06,
+      "loss": 0.6394,
+      "step": 2269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.836460675979e-06,
+      "loss": 0.6499,
+      "step": 2270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.834528210085387e-06,
+      "loss": 0.6768,
+      "step": 2271
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.832595120055595e-06,
+      "loss": 0.6951,
+      "step": 2272
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.830661406315272e-06,
+      "loss": 0.5652,
+      "step": 2273
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.828727069290203e-06,
+      "loss": 0.6793,
+      "step": 2274
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.82679210940631e-06,
+      "loss": 0.6761,
+      "step": 2275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.824856527089653e-06,
+      "loss": 0.6801,
+      "step": 2276
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.822920322766427e-06,
+      "loss": 0.6877,
+      "step": 2277
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.820983496862971e-06,
+      "loss": 0.6542,
+      "step": 2278
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.819046049805752e-06,
+      "loss": 0.6673,
+      "step": 2279
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.817107982021377e-06,
+      "loss": 0.6749,
+      "step": 2280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.815169293936592e-06,
+      "loss": 0.6966,
+      "step": 2281
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.813229985978276e-06,
+      "loss": 0.6788,
+      "step": 2282
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.811290058573447e-06,
+      "loss": 0.7198,
+      "step": 2283
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.809349512149258e-06,
+      "loss": 0.6985,
+      "step": 2284
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.807408347133e-06,
+      "loss": 0.7024,
+      "step": 2285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.805466563952099e-06,
+      "loss": 0.6599,
+      "step": 2286
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.803524163034116e-06,
+      "loss": 0.6763,
+      "step": 2287
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.801581144806752e-06,
+      "loss": 0.5523,
+      "step": 2288
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.799637509697834e-06,
+      "loss": 0.6579,
+      "step": 2289
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.797693258135341e-06,
+      "loss": 0.6754,
+      "step": 2290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.795748390547375e-06,
+      "loss": 0.6721,
+      "step": 2291
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.793802907362176e-06,
+      "loss": 0.6829,
+      "step": 2292
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.791856809008122e-06,
+      "loss": 0.6704,
+      "step": 2293
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.789910095913727e-06,
+      "loss": 0.7302,
+      "step": 2294
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.787962768507635e-06,
+      "loss": 0.68,
+      "step": 2295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.786014827218632e-06,
+      "loss": 0.6612,
+      "step": 2296
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.784066272475635e-06,
+      "loss": 0.654,
+      "step": 2297
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.782117104707698e-06,
+      "loss": 0.6606,
+      "step": 2298
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.78016732434401e-06,
+      "loss": 0.6909,
+      "step": 2299
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.77821693181389e-06,
+      "loss": 0.658,
+      "step": 2300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.7762659275468e-06,
+      "loss": 0.67,
+      "step": 2301
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.774314311972332e-06,
+      "loss": 0.6532,
+      "step": 2302
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.772362085520212e-06,
+      "loss": 0.7011,
+      "step": 2303
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.770409248620303e-06,
+      "loss": 0.6605,
+      "step": 2304
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.7684558017026e-06,
+      "loss": 0.5619,
+      "step": 2305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.766501745197234e-06,
+      "loss": 0.6444,
+      "step": 2306
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.76454707953447e-06,
+      "loss": 0.6745,
+      "step": 2307
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.762591805144706e-06,
+      "loss": 0.6831,
+      "step": 2308
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.760635922458474e-06,
+      "loss": 0.5925,
+      "step": 2309
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.758679431906443e-06,
+      "loss": 0.6687,
+      "step": 2310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.756722333919414e-06,
+      "loss": 0.6956,
+      "step": 2311
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.754764628928317e-06,
+      "loss": 0.6688,
+      "step": 2312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.752806317364226e-06,
+      "loss": 0.6529,
+      "step": 2313
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.750847399658336e-06,
+      "loss": 0.6752,
+      "step": 2314
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.748887876241988e-06,
+      "loss": 0.636,
+      "step": 2315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.74692774754665e-06,
+      "loss": 0.6789,
+      "step": 2316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.744967014003919e-06,
+      "loss": 0.6948,
+      "step": 2317
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.743005676045534e-06,
+      "loss": 0.6673,
+      "step": 2318
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.741043734103362e-06,
+      "loss": 0.6656,
+      "step": 2319
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.739081188609404e-06,
+      "loss": 0.6855,
+      "step": 2320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.737118039995793e-06,
+      "loss": 0.6696,
+      "step": 2321
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.735154288694794e-06,
+      "loss": 0.684,
+      "step": 2322
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.733189935138811e-06,
+      "loss": 0.6774,
+      "step": 2323
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.731224979760375e-06,
+      "loss": 0.6867,
+      "step": 2324
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.729259422992149e-06,
+      "loss": 0.6558,
+      "step": 2325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.727293265266926e-06,
+      "loss": 0.6824,
+      "step": 2326
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.725326507017644e-06,
+      "loss": 0.7119,
+      "step": 2327
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.723359148677356e-06,
+      "loss": 0.5702,
+      "step": 2328
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.721391190679262e-06,
+      "loss": 0.5583,
+      "step": 2329
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.719422633456684e-06,
+      "loss": 0.5944,
+      "step": 2330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.71745347744308e-06,
+      "loss": 0.6932,
+      "step": 2331
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.715483723072041e-06,
+      "loss": 0.6886,
+      "step": 2332
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.713513370777287e-06,
+      "loss": 0.5603,
+      "step": 2333
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.711542420992673e-06,
+      "loss": 0.5742,
+      "step": 2334
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.709570874152179e-06,
+      "loss": 0.6705,
+      "step": 2335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.707598730689924e-06,
+      "loss": 0.7262,
+      "step": 2336
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.705625991040156e-06,
+      "loss": 0.6603,
+      "step": 2337
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.703652655637251e-06,
+      "loss": 0.6585,
+      "step": 2338
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70167872491572e-06,
+      "loss": 0.6617,
+      "step": 2339
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.699704199310204e-06,
+      "loss": 0.5531,
+      "step": 2340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.697729079255474e-06,
+      "loss": 0.5721,
+      "step": 2341
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.695753365186434e-06,
+      "loss": 0.6628,
+      "step": 2342
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.693777057538116e-06,
+      "loss": 0.6804,
+      "step": 2343
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.691800156745683e-06,
+      "loss": 0.581,
+      "step": 2344
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.689822663244435e-06,
+      "loss": 0.6936,
+      "step": 2345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.687844577469792e-06,
+      "loss": 0.6672,
+      "step": 2346
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.685865899857311e-06,
+      "loss": 0.6619,
+      "step": 2347
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.683886630842681e-06,
+      "loss": 0.6489,
+      "step": 2348
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.681906770861714e-06,
+      "loss": 0.6878,
+      "step": 2349
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.679926320350358e-06,
+      "loss": 0.7303,
+      "step": 2350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.67794527974469e-06,
+      "loss": 0.6755,
+      "step": 2351
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.675963649480916e-06,
+      "loss": 0.6562,
+      "step": 2352
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.673981429995372e-06,
+      "loss": 0.6663,
+      "step": 2353
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.671998621724523e-06,
+      "loss": 0.653,
+      "step": 2354
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.670015225104966e-06,
+      "loss": 0.6377,
+      "step": 2355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.668031240573425e-06,
+      "loss": 0.6922,
+      "step": 2356
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.666046668566756e-06,
+      "loss": 0.6915,
+      "step": 2357
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.664061509521943e-06,
+      "loss": 0.6629,
+      "step": 2358
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.662075763876095e-06,
+      "loss": 0.6674,
+      "step": 2359
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66008943206646e-06,
+      "loss": 0.6818,
+      "step": 2360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.658102514530405e-06,
+      "loss": 0.6993,
+      "step": 2361
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.656115011705433e-06,
+      "loss": 0.6672,
+      "step": 2362
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.654126924029175e-06,
+      "loss": 0.6708,
+      "step": 2363
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.652138251939384e-06,
+      "loss": 0.6784,
+      "step": 2364
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.650148995873949e-06,
+      "loss": 0.5895,
+      "step": 2365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.648159156270884e-06,
+      "loss": 0.683,
+      "step": 2366
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.646168733568338e-06,
+      "loss": 0.6973,
+      "step": 2367
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.644177728204576e-06,
+      "loss": 0.6864,
+      "step": 2368
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.642186140618004e-06,
+      "loss": 0.632,
+      "step": 2369
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.640193971247147e-06,
+      "loss": 0.6936,
+      "step": 2370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.638201220530664e-06,
+      "loss": 0.65,
+      "step": 2371
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.636207888907338e-06,
+      "loss": 0.6558,
+      "step": 2372
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.634213976816085e-06,
+      "loss": 0.6597,
+      "step": 2373
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.632219484695942e-06,
+      "loss": 0.6542,
+      "step": 2374
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.630224412986078e-06,
+      "loss": 0.6848,
+      "step": 2375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.628228762125791e-06,
+      "loss": 0.6711,
+      "step": 2376
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.626232532554501e-06,
+      "loss": 0.6564,
+      "step": 2377
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.62423572471176e-06,
+      "loss": 0.6616,
+      "step": 2378
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.622238339037248e-06,
+      "loss": 0.6765,
+      "step": 2379
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.6202403759707675e-06,
+      "loss": 0.6711,
+      "step": 2380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.6182418359522535e-06,
+      "loss": 0.5692,
+      "step": 2381
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6162427194217645e-06,
+      "loss": 0.7184,
+      "step": 2382
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.614243026819486e-06,
+      "loss": 0.5598,
+      "step": 2383
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.612242758585733e-06,
+      "loss": 0.5761,
+      "step": 2384
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.610241915160943e-06,
+      "loss": 0.6687,
+      "step": 2385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.608240496985686e-06,
+      "loss": 0.6611,
+      "step": 2386
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.606238504500654e-06,
+      "loss": 0.6604,
+      "step": 2387
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.604235938146667e-06,
+      "loss": 0.5502,
+      "step": 2388
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6022327983646684e-06,
+      "loss": 0.6927,
+      "step": 2389
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6002290855957336e-06,
+      "loss": 0.6407,
+      "step": 2390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.59822480028106e-06,
+      "loss": 0.6815,
+      "step": 2391
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.596219942861971e-06,
+      "loss": 0.6725,
+      "step": 2392
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5942145137799205e-06,
+      "loss": 0.6757,
+      "step": 2393
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.592208513476484e-06,
+      "loss": 0.6675,
+      "step": 2394
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.590201942393361e-06,
+      "loss": 0.5784,
+      "step": 2395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.588194800972381e-06,
+      "loss": 0.6704,
+      "step": 2396
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.586187089655497e-06,
+      "loss": 0.6879,
+      "step": 2397
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.584178808884788e-06,
+      "loss": 0.5587,
+      "step": 2398
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.582169959102461e-06,
+      "loss": 0.6866,
+      "step": 2399
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.580160540750842e-06,
+      "loss": 0.6794,
+      "step": 2400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.578150554272388e-06,
+      "loss": 0.67,
+      "step": 2401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.576140000109678e-06,
+      "loss": 0.6973,
+      "step": 2402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.574128878705417e-06,
+      "loss": 0.5282,
+      "step": 2403
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5721171905024335e-06,
+      "loss": 0.6823,
+      "step": 2404
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.570104935943685e-06,
+      "loss": 0.6391,
+      "step": 2405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5680921154722496e-06,
+      "loss": 0.6793,
+      "step": 2406
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5660787295313324e-06,
+      "loss": 0.5538,
+      "step": 2407
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.564064778564257e-06,
+      "loss": 0.6924,
+      "step": 2408
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.562050263014482e-06,
+      "loss": 0.6588,
+      "step": 2409
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.560035183325582e-06,
+      "loss": 0.6773,
+      "step": 2410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.55801953994126e-06,
+      "loss": 0.6874,
+      "step": 2411
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.556003333305337e-06,
+      "loss": 0.7189,
+      "step": 2412
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5539865638617684e-06,
+      "loss": 0.6847,
+      "step": 2413
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.551969232054625e-06,
+      "loss": 0.6611,
+      "step": 2414
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.549951338328103e-06,
+      "loss": 0.7029,
+      "step": 2415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.547932883126526e-06,
+      "loss": 0.6623,
+      "step": 2416
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5459138668943365e-06,
+      "loss": 0.6429,
+      "step": 2417
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5438942900761035e-06,
+      "loss": 0.6692,
+      "step": 2418
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.541874153116518e-06,
+      "loss": 0.645,
+      "step": 2419
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.539853456460397e-06,
+      "loss": 0.6703,
+      "step": 2420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.537832200552676e-06,
+      "loss": 0.7306,
+      "step": 2421
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5358103858384176e-06,
+      "loss": 0.6875,
+      "step": 2422
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.533788012762807e-06,
+      "loss": 0.6688,
+      "step": 2423
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.531765081771149e-06,
+      "loss": 0.6752,
+      "step": 2424
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.529741593308878e-06,
+      "loss": 0.6672,
+      "step": 2425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.527717547821543e-06,
+      "loss": 0.6888,
+      "step": 2426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5256929457548215e-06,
+      "loss": 0.6691,
+      "step": 2427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.523667787554512e-06,
+      "loss": 0.5673,
+      "step": 2428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.521642073666533e-06,
+      "loss": 0.6885,
+      "step": 2429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.519615804536928e-06,
+      "loss": 0.548,
+      "step": 2430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.517588980611864e-06,
+      "loss": 0.6796,
+      "step": 2431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.515561602337627e-06,
+      "loss": 0.7071,
+      "step": 2432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.513533670160627e-06,
+      "loss": 0.6818,
+      "step": 2433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.511505184527394e-06,
+      "loss": 0.6806,
+      "step": 2434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5094761458845845e-06,
+      "loss": 0.6656,
+      "step": 2435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.50744655467897e-06,
+      "loss": 0.6826,
+      "step": 2436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.505416411357449e-06,
+      "loss": 0.6212,
+      "step": 2437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.50338571636704e-06,
+      "loss": 0.6349,
+      "step": 2438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.50135447015488e-06,
+      "loss": 0.662,
+      "step": 2439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.499322673168234e-06,
+      "loss": 0.7046,
+      "step": 2440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.4972903258544815e-06,
+      "loss": 0.5664,
+      "step": 2441
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.495257428661128e-06,
+      "loss": 0.663,
+      "step": 2442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.493223982035798e-06,
+      "loss": 0.6677,
+      "step": 2443
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.491189986426236e-06,
+      "loss": 0.5794,
+      "step": 2444
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.489155442280308e-06,
+      "loss": 0.6583,
+      "step": 2445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.487120350046003e-06,
+      "loss": 0.6529,
+      "step": 2446
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.485084710171429e-06,
+      "loss": 0.6882,
+      "step": 2447
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.483048523104812e-06,
+      "loss": 0.6548,
+      "step": 2448
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.4810117892945035e-06,
+      "loss": 0.543,
+      "step": 2449
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.478974509188972e-06,
+      "loss": 0.6806,
+      "step": 2450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.476936683236808e-06,
+      "loss": 0.6911,
+      "step": 2451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.474898311886722e-06,
+      "loss": 0.6601,
+      "step": 2452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.472859395587542e-06,
+      "loss": 0.6835,
+      "step": 2453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.470819934788219e-06,
+      "loss": 0.684,
+      "step": 2454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.468779929937823e-06,
+      "loss": 0.6736,
+      "step": 2455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.466739381485542e-06,
+      "loss": 0.6706,
+      "step": 2456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.464698289880689e-06,
+      "loss": 0.6272,
+      "step": 2457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.46265665557269e-06,
+      "loss": 0.563,
+      "step": 2458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4606144790110935e-06,
+      "loss": 0.6495,
+      "step": 2459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.458571760645569e-06,
+      "loss": 0.5614,
+      "step": 2460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.456528500925903e-06,
+      "loss": 0.6843,
+      "step": 2461
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.454484700302001e-06,
+      "loss": 0.6621,
+      "step": 2462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.452440359223891e-06,
+      "loss": 0.686,
+      "step": 2463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.450395478141714e-06,
+      "loss": 0.5551,
+      "step": 2464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4483500575057355e-06,
+      "loss": 0.6855,
+      "step": 2465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.446304097766337e-06,
+      "loss": 0.6864,
+      "step": 2466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.444257599374019e-06,
+      "loss": 0.6802,
+      "step": 2467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.442210562779403e-06,
+      "loss": 0.5607,
+      "step": 2468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.440162988433226e-06,
+      "loss": 0.6859,
+      "step": 2469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.438114876786344e-06,
+      "loss": 0.562,
+      "step": 2470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.436066228289731e-06,
+      "loss": 0.6745,
+      "step": 2471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4340170433944814e-06,
+      "loss": 0.6743,
+      "step": 2472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.431967322551804e-06,
+      "loss": 0.6841,
+      "step": 2473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.42991706621303e-06,
+      "loss": 0.5543,
+      "step": 2474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.427866274829607e-06,
+      "loss": 0.668,
+      "step": 2475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.425814948853099e-06,
+      "loss": 0.5679,
+      "step": 2476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.423763088735187e-06,
+      "loss": 0.6715,
+      "step": 2477
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.421710694927673e-06,
+      "loss": 0.5635,
+      "step": 2478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.419657767882473e-06,
+      "loss": 0.6593,
+      "step": 2479
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.417604308051622e-06,
+      "loss": 0.6887,
+      "step": 2480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415550315887275e-06,
+      "loss": 0.6781,
+      "step": 2481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.413495791841699e-06,
+      "loss": 0.6877,
+      "step": 2482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.411440736367281e-06,
+      "loss": 0.648,
+      "step": 2483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.409385149916526e-06,
+      "loss": 0.5642,
+      "step": 2484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.407329032942054e-06,
+      "loss": 0.5431,
+      "step": 2485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.405272385896601e-06,
+      "loss": 0.6757,
+      "step": 2486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.403215209233022e-06,
+      "loss": 0.6684,
+      "step": 2487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.401157503404289e-06,
+      "loss": 0.6806,
+      "step": 2488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.399099268863488e-06,
+      "loss": 0.6705,
+      "step": 2489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.397040506063823e-06,
+      "loss": 0.6653,
+      "step": 2490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.394981215458615e-06,
+      "loss": 0.6617,
+      "step": 2491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.392921397501296e-06,
+      "loss": 0.6696,
+      "step": 2492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.390861052645424e-06,
+      "loss": 0.5645,
+      "step": 2493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.388800181344665e-06,
+      "loss": 0.6471,
+      "step": 2494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.386738784052802e-06,
+      "loss": 0.6753,
+      "step": 2495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.384676861223738e-06,
+      "loss": 0.5278,
+      "step": 2496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.382614413311486e-06,
+      "loss": 0.6272,
+      "step": 2497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.380551440770178e-06,
+      "loss": 0.6779,
+      "step": 2498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.378487944054064e-06,
+      "loss": 0.5445,
+      "step": 2499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.376423923617501e-06,
+      "loss": 0.6982,
+      "step": 2500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.374359379914971e-06,
+      "loss": 0.6982,
+      "step": 2501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3722943134010664e-06,
+      "loss": 0.6626,
+      "step": 2502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.370228724530492e-06,
+      "loss": 0.5894,
+      "step": 2503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.368162613758074e-06,
+      "loss": 0.7226,
+      "step": 2504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3660959815387504e-06,
+      "loss": 0.6772,
+      "step": 2505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3640288283275715e-06,
+      "loss": 0.5958,
+      "step": 2506
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.361961154579708e-06,
+      "loss": 0.6421,
+      "step": 2507
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.35989296075044e-06,
+      "loss": 0.6997,
+      "step": 2508
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3578242472951635e-06,
+      "loss": 0.6898,
+      "step": 2509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.35575501466939e-06,
+      "loss": 0.6826,
+      "step": 2510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.353685263328748e-06,
+      "loss": 0.6816,
+      "step": 2511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.351614993728972e-06,
+      "loss": 0.6736,
+      "step": 2512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.349544206325921e-06,
+      "loss": 0.6618,
+      "step": 2513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.347472901575558e-06,
+      "loss": 0.647,
+      "step": 2514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.345401079933967e-06,
+      "loss": 0.6832,
+      "step": 2515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.343328741857343e-06,
+      "loss": 0.6684,
+      "step": 2516
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.341255887801995e-06,
+      "loss": 0.6681,
+      "step": 2517
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.339182518224345e-06,
+      "loss": 0.6649,
+      "step": 2518
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3371086335809315e-06,
+      "loss": 0.6749,
+      "step": 2519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.335034234328402e-06,
+      "loss": 0.666,
+      "step": 2520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.332959320923521e-06,
+      "loss": 0.6514,
+      "step": 2521
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.330883893823164e-06,
+      "loss": 0.5331,
+      "step": 2522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3288079534843194e-06,
+      "loss": 0.7129,
+      "step": 2523
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.326731500364091e-06,
+      "loss": 0.663,
+      "step": 2524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.324654534919695e-06,
+      "loss": 0.5659,
+      "step": 2525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.322577057608458e-06,
+      "loss": 0.5794,
+      "step": 2526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3204990688878185e-06,
+      "loss": 0.6654,
+      "step": 2527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.318420569215334e-06,
+      "loss": 0.6512,
+      "step": 2528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.316341559048667e-06,
+      "loss": 0.7015,
+      "step": 2529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.314262038845599e-06,
+      "loss": 0.548,
+      "step": 2530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.312182009064019e-06,
+      "loss": 0.6595,
+      "step": 2531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.31010147016193e-06,
+      "loss": 0.6644,
+      "step": 2532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.308020422597445e-06,
+      "loss": 0.641,
+      "step": 2533
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.305938866828795e-06,
+      "loss": 0.7013,
+      "step": 2534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.303856803314313e-06,
+      "loss": 0.6603,
+      "step": 2535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.301774232512456e-06,
+      "loss": 0.5357,
+      "step": 2536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.299691154881782e-06,
+      "loss": 0.6776,
+      "step": 2537
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.297607570880967e-06,
+      "loss": 0.5735,
+      "step": 2538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.295523480968793e-06,
+      "loss": 0.543,
+      "step": 2539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.293438885604163e-06,
+      "loss": 0.5465,
+      "step": 2540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.291353785246078e-06,
+      "loss": 0.6626,
+      "step": 2541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2892681803536625e-06,
+      "loss": 0.6444,
+      "step": 2542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2871820713861455e-06,
+      "loss": 0.6916,
+      "step": 2543
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.285095458802868e-06,
+      "loss": 0.667,
+      "step": 2544
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.28300834306328e-06,
+      "loss": 0.7105,
+      "step": 2545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2809207246269495e-06,
+      "loss": 0.6999,
+      "step": 2546
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.278832603953546e-06,
+      "loss": 0.6518,
+      "step": 2547
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.276743981502856e-06,
+      "loss": 0.6755,
+      "step": 2548
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.274654857734773e-06,
+      "loss": 0.5325,
+      "step": 2549
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.272565233109305e-06,
+      "loss": 0.6659,
+      "step": 2550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.270475108086563e-06,
+      "loss": 0.6706,
+      "step": 2551
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2683844831267766e-06,
+      "loss": 0.6851,
+      "step": 2552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.266293358690279e-06,
+      "loss": 0.6275,
+      "step": 2553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.264201735237517e-06,
+      "loss": 0.6723,
+      "step": 2554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2621096132290465e-06,
+      "loss": 0.6558,
+      "step": 2555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.260016993125533e-06,
+      "loss": 0.6669,
+      "step": 2556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.257923875387751e-06,
+      "loss": 0.6306,
+      "step": 2557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.255830260476585e-06,
+      "loss": 0.6942,
+      "step": 2558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.25373614885303e-06,
+      "loss": 0.6545,
+      "step": 2559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.251641540978188e-06,
+      "loss": 0.66,
+      "step": 2560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.249546437313273e-06,
+      "loss": 0.6442,
+      "step": 2561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.247450838319605e-06,
+      "loss": 0.6679,
+      "step": 2562
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2453547444586195e-06,
+      "loss": 0.6636,
+      "step": 2563
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.243258156191851e-06,
+      "loss": 0.6454,
+      "step": 2564
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.241161073980951e-06,
+      "loss": 0.5706,
+      "step": 2565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.239063498287678e-06,
+      "loss": 0.543,
+      "step": 2566
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.236965429573896e-06,
+      "loss": 0.6902,
+      "step": 2567
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.234866868301581e-06,
+      "loss": 0.6532,
+      "step": 2568
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.232767814932817e-06,
+      "loss": 0.6828,
+      "step": 2569
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.230668269929795e-06,
+      "loss": 0.6936,
+      "step": 2570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.228568233754814e-06,
+      "loss": 0.6705,
+      "step": 2571
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.226467706870283e-06,
+      "loss": 0.635,
+      "step": 2572
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2243666897387165e-06,
+      "loss": 0.6654,
+      "step": 2573
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.22226518282274e-06,
+      "loss": 0.673,
+      "step": 2574
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2201631865850855e-06,
+      "loss": 0.565,
+      "step": 2575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.218060701488593e-06,
+      "loss": 0.5425,
+      "step": 2576
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.215957727996208e-06,
+      "loss": 0.6549,
+      "step": 2577
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.213854266570985e-06,
+      "loss": 0.5707,
+      "step": 2578
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.211750317676089e-06,
+      "loss": 0.5644,
+      "step": 2579
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.209645881774787e-06,
+      "loss": 0.6683,
+      "step": 2580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.207540959330458e-06,
+      "loss": 0.6633,
+      "step": 2581
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.205435550806585e-06,
+      "loss": 0.6784,
+      "step": 2582
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.203329656666756e-06,
+      "loss": 0.5642,
+      "step": 2583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.201223277374673e-06,
+      "loss": 0.6595,
+      "step": 2584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.199116413394139e-06,
+      "loss": 0.6673,
+      "step": 2585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.197009065189065e-06,
+      "loss": 0.6424,
+      "step": 2586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.194901233223471e-06,
+      "loss": 0.6648,
+      "step": 2587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.192792917961481e-06,
+      "loss": 0.687,
+      "step": 2588
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.190684119867324e-06,
+      "loss": 0.6794,
+      "step": 2589
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1885748394053374e-06,
+      "loss": 0.5722,
+      "step": 2590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.186465077039968e-06,
+      "loss": 0.6562,
+      "step": 2591
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.184354833235761e-06,
+      "loss": 0.6568,
+      "step": 2592
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1822441084573766e-06,
+      "loss": 0.5675,
+      "step": 2593
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.180132903169573e-06,
+      "loss": 0.6835,
+      "step": 2594
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.178021217837219e-06,
+      "loss": 0.6651,
+      "step": 2595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.175909052925287e-06,
+      "loss": 0.6752,
+      "step": 2596
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1737964088988565e-06,
+      "loss": 0.6604,
+      "step": 2597
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.17168328622311e-06,
+      "loss": 0.6887,
+      "step": 2598
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.169569685363338e-06,
+      "loss": 0.7005,
+      "step": 2599
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.167455606784936e-06,
+      "loss": 0.6631,
+      "step": 2600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.165341050953403e-06,
+      "loss": 0.6716,
+      "step": 2601
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.163226018334345e-06,
+      "loss": 0.5623,
+      "step": 2602
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.161110509393472e-06,
+      "loss": 0.6747,
+      "step": 2603
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.158994524596596e-06,
+      "loss": 0.68,
+      "step": 2604
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.156878064409642e-06,
+      "loss": 0.652,
+      "step": 2605
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.15476112929863e-06,
+      "loss": 0.6812,
+      "step": 2606
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1526437197296924e-06,
+      "loss": 0.6851,
+      "step": 2607
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.150525836169061e-06,
+      "loss": 0.6564,
+      "step": 2608
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1484074790830725e-06,
+      "loss": 0.5628,
+      "step": 2609
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.146288648938171e-06,
+      "loss": 0.6513,
+      "step": 2610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.144169346200902e-06,
+      "loss": 0.665,
+      "step": 2611
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.142049571337914e-06,
+      "loss": 0.5834,
+      "step": 2612
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.139929324815965e-06,
+      "loss": 0.6758,
+      "step": 2613
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.137808607101912e-06,
+      "loss": 0.6686,
+      "step": 2614
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1356874186627135e-06,
+      "loss": 0.6586,
+      "step": 2615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1335657599654394e-06,
+      "loss": 0.6779,
+      "step": 2616
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.131443631477255e-06,
+      "loss": 0.5551,
+      "step": 2617
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.129321033665436e-06,
+      "loss": 0.6808,
+      "step": 2618
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.127197966997358e-06,
+      "loss": 0.695,
+      "step": 2619
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.125074431940499e-06,
+      "loss": 0.6681,
+      "step": 2620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.122950428962441e-06,
+      "loss": 0.6722,
+      "step": 2621
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1208259585308694e-06,
+      "loss": 0.6655,
+      "step": 2622
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.118701021113572e-06,
+      "loss": 0.6808,
+      "step": 2623
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.116575617178441e-06,
+      "loss": 0.6925,
+      "step": 2624
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.11444974719347e-06,
+      "loss": 0.5564,
+      "step": 2625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.112323411626756e-06,
+      "loss": 0.6881,
+      "step": 2626
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.110196610946496e-06,
+      "loss": 0.6674,
+      "step": 2627
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.108069345620993e-06,
+      "loss": 0.673,
+      "step": 2628
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1059416161186475e-06,
+      "loss": 0.6795,
+      "step": 2629
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1038134229079705e-06,
+      "loss": 0.6828,
+      "step": 2630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1016847664575665e-06,
+      "loss": 0.6564,
+      "step": 2631
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.099555647236145e-06,
+      "loss": 0.6608,
+      "step": 2632
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.097426065712519e-06,
+      "loss": 0.6322,
+      "step": 2633
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.095296022355604e-06,
+      "loss": 0.6757,
+      "step": 2634
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.09316551763441e-06,
+      "loss": 0.7056,
+      "step": 2635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.091034552018059e-06,
+      "loss": 0.6788,
+      "step": 2636
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.088903125975768e-06,
+      "loss": 0.672,
+      "step": 2637
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.086771239976856e-06,
+      "loss": 0.6452,
+      "step": 2638
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.084638894490744e-06,
+      "loss": 0.6717,
+      "step": 2639
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.082506089986956e-06,
+      "loss": 0.6586,
+      "step": 2640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.080372826935112e-06,
+      "loss": 0.5597,
+      "step": 2641
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.078239105804938e-06,
+      "loss": 0.6443,
+      "step": 2642
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.07610492706626e-06,
+      "loss": 0.5836,
+      "step": 2643
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.073970291189003e-06,
+      "loss": 0.6851,
+      "step": 2644
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.071835198643193e-06,
+      "loss": 0.6086,
+      "step": 2645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0696996498989605e-06,
+      "loss": 0.5483,
+      "step": 2646
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0675636454265286e-06,
+      "loss": 0.6571,
+      "step": 2647
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0654271856962255e-06,
+      "loss": 0.6668,
+      "step": 2648
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0632902711784814e-06,
+      "loss": 0.6986,
+      "step": 2649
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.061152902343825e-06,
+      "loss": 0.676,
+      "step": 2650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.059015079662882e-06,
+      "loss": 0.5494,
+      "step": 2651
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.056876803606384e-06,
+      "loss": 0.6724,
+      "step": 2652
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0547380746451564e-06,
+      "loss": 0.6783,
+      "step": 2653
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.052598893250127e-06,
+      "loss": 0.6625,
+      "step": 2654
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.050459259892324e-06,
+      "loss": 0.6373,
+      "step": 2655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.048319175042876e-06,
+      "loss": 0.67,
+      "step": 2656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.046178639173006e-06,
+      "loss": 0.6915,
+      "step": 2657
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0440376527540436e-06,
+      "loss": 0.6516,
+      "step": 2658
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.04189621625741e-06,
+      "loss": 0.6751,
+      "step": 2659
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.03975433015463e-06,
+      "loss": 0.6829,
+      "step": 2660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.037611994917327e-06,
+      "loss": 0.5785,
+      "step": 2661
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.035469211017223e-06,
+      "loss": 0.6393,
+      "step": 2662
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.03332597892614e-06,
+      "loss": 0.6423,
+      "step": 2663
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.031182299115996e-06,
+      "loss": 0.5619,
+      "step": 2664
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.029038172058809e-06,
+      "loss": 0.6983,
+      "step": 2665
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.026893598226695e-06,
+      "loss": 0.6517,
+      "step": 2666
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.024748578091868e-06,
+      "loss": 0.6665,
+      "step": 2667
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.022603112126642e-06,
+      "loss": 0.669,
+      "step": 2668
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.020457200803431e-06,
+      "loss": 0.6657,
+      "step": 2669
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.018310844594741e-06,
+      "loss": 0.591,
+      "step": 2670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.016164043973179e-06,
+      "loss": 0.6662,
+      "step": 2671
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.014016799411452e-06,
+      "loss": 0.574,
+      "step": 2672
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.011869111382361e-06,
+      "loss": 0.651,
+      "step": 2673
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.009720980358808e-06,
+      "loss": 0.6601,
+      "step": 2674
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.00757240681379e-06,
+      "loss": 0.6671,
+      "step": 2675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.005423391220403e-06,
+      "loss": 0.668,
+      "step": 2676
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.003273934051839e-06,
+      "loss": 0.7229,
+      "step": 2677
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.00112403578139e-06,
+      "loss": 0.6517,
+      "step": 2678
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.998973696882442e-06,
+      "loss": 0.6703,
+      "step": 2679
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9968229178284775e-06,
+      "loss": 0.6632,
+      "step": 2680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9946716990930805e-06,
+      "loss": 0.6872,
+      "step": 2681
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.992520041149928e-06,
+      "loss": 0.5475,
+      "step": 2682
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.990367944472795e-06,
+      "loss": 0.6792,
+      "step": 2683
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9882154095355516e-06,
+      "loss": 0.6812,
+      "step": 2684
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.986062436812165e-06,
+      "loss": 0.6573,
+      "step": 2685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.983909026776701e-06,
+      "loss": 0.6751,
+      "step": 2686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.981755179903318e-06,
+      "loss": 0.6529,
+      "step": 2687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.979600896666277e-06,
+      "loss": 0.6868,
+      "step": 2688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.977446177539924e-06,
+      "loss": 0.6783,
+      "step": 2689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9752910229987135e-06,
+      "loss": 0.6808,
+      "step": 2690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9731354335171885e-06,
+      "loss": 0.5635,
+      "step": 2691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.970979409569988e-06,
+      "loss": 0.5798,
+      "step": 2692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9688229516318485e-06,
+      "loss": 0.689,
+      "step": 2693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9666660601776015e-06,
+      "loss": 0.6682,
+      "step": 2694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.964508735682174e-06,
+      "loss": 0.6776,
+      "step": 2695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.96235097862059e-06,
+      "loss": 0.6748,
+      "step": 2696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.960192789467965e-06,
+      "loss": 0.6666,
+      "step": 2697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.958034168699513e-06,
+      "loss": 0.6531,
+      "step": 2698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.955875116790541e-06,
+      "loss": 0.5704,
+      "step": 2699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.953715634216451e-06,
+      "loss": 0.6694,
+      "step": 2700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.951555721452743e-06,
+      "loss": 0.6834,
+      "step": 2701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.949395378975008e-06,
+      "loss": 0.6611,
+      "step": 2702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.947234607258931e-06,
+      "loss": 0.6599,
+      "step": 2703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.945073406780296e-06,
+      "loss": 0.6749,
+      "step": 2704
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.942911778014978e-06,
+      "loss": 0.6786,
+      "step": 2705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.940749721438947e-06,
+      "loss": 0.5786,
+      "step": 2706
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.938587237528267e-06,
+      "loss": 0.6861,
+      "step": 2707
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.936424326759097e-06,
+      "loss": 0.6337,
+      "step": 2708
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.93426098960769e-06,
+      "loss": 0.6425,
+      "step": 2709
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.932097226550389e-06,
+      "loss": 0.6404,
+      "step": 2710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.929933038063638e-06,
+      "loss": 0.574,
+      "step": 2711
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.92776842462397e-06,
+      "loss": 0.6347,
+      "step": 2712
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.92560338670801e-06,
+      "loss": 0.6692,
+      "step": 2713
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.923437924792482e-06,
+      "loss": 0.5732,
+      "step": 2714
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.921272039354196e-06,
+      "loss": 0.5668,
+      "step": 2715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.919105730870065e-06,
+      "loss": 0.6697,
+      "step": 2716
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.916938999817085e-06,
+      "loss": 0.5259,
+      "step": 2717
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.914771846672349e-06,
+      "loss": 0.5644,
+      "step": 2718
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.912604271913048e-06,
+      "loss": 0.6427,
+      "step": 2719
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.910436276016459e-06,
+      "loss": 0.6722,
+      "step": 2720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.908267859459952e-06,
+      "loss": 0.6613,
+      "step": 2721
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.906099022720996e-06,
+      "loss": 0.6699,
+      "step": 2722
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.903929766277146e-06,
+      "loss": 0.6772,
+      "step": 2723
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.901760090606049e-06,
+      "loss": 0.6546,
+      "step": 2724
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.899589996185451e-06,
+      "loss": 0.6652,
+      "step": 2725
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.897419483493185e-06,
+      "loss": 0.6316,
+      "step": 2726
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.895248553007177e-06,
+      "loss": 0.6819,
+      "step": 2727
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.893077205205445e-06,
+      "loss": 0.6471,
+      "step": 2728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8909054405661e-06,
+      "loss": 0.7095,
+      "step": 2729
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.888733259567343e-06,
+      "loss": 0.6512,
+      "step": 2730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.886560662687467e-06,
+      "loss": 0.668,
+      "step": 2731
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.88438765040486e-06,
+      "loss": 0.6647,
+      "step": 2732
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.882214223197996e-06,
+      "loss": 0.6779,
+      "step": 2733
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.880040381545444e-06,
+      "loss": 0.6668,
+      "step": 2734
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.877866125925864e-06,
+      "loss": 0.5685,
+      "step": 2735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.875691456818003e-06,
+      "loss": 0.6968,
+      "step": 2736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.873516374700707e-06,
+      "loss": 0.6734,
+      "step": 2737
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.871340880052907e-06,
+      "loss": 0.6529,
+      "step": 2738
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.869164973353624e-06,
+      "loss": 0.6696,
+      "step": 2739
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8669886550819755e-06,
+      "loss": 0.6589,
+      "step": 2740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.864811925717163e-06,
+      "loss": 0.6766,
+      "step": 2741
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.862634785738484e-06,
+      "loss": 0.6656,
+      "step": 2742
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.860457235625322e-06,
+      "loss": 0.6773,
+      "step": 2743
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.858279275857154e-06,
+      "loss": 0.5724,
+      "step": 2744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.856100906913545e-06,
+      "loss": 0.6348,
+      "step": 2745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.853922129274153e-06,
+      "loss": 0.5403,
+      "step": 2746
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.851742943418724e-06,
+      "loss": 0.6645,
+      "step": 2747
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.849563349827092e-06,
+      "loss": 0.6415,
+      "step": 2748
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8473833489791845e-06,
+      "loss": 0.5721,
+      "step": 2749
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.845202941355016e-06,
+      "loss": 0.6627,
+      "step": 2750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.843022127434691e-06,
+      "loss": 0.6683,
+      "step": 2751
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8408409076984075e-06,
+      "loss": 0.5413,
+      "step": 2752
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.838659282626445e-06,
+      "loss": 0.5685,
+      "step": 2753
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.836477252699179e-06,
+      "loss": 0.6975,
+      "step": 2754
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.83429481839707e-06,
+      "loss": 0.6435,
+      "step": 2755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.832111980200672e-06,
+      "loss": 0.5425,
+      "step": 2756
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.829928738590623e-06,
+      "loss": 0.574,
+      "step": 2757
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.827745094047653e-06,
+      "loss": 0.5414,
+      "step": 2758
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.82556104705258e-06,
+      "loss": 0.6839,
+      "step": 2759
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.82337659808631e-06,
+      "loss": 0.6744,
+      "step": 2760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.821191747629836e-06,
+      "loss": 0.6482,
+      "step": 2761
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.819006496164247e-06,
+      "loss": 0.6713,
+      "step": 2762
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.816820844170706e-06,
+      "loss": 0.617,
+      "step": 2763
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.814634792130481e-06,
+      "loss": 0.6543,
+      "step": 2764
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8124483405249154e-06,
+      "loss": 0.5843,
+      "step": 2765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.810261489835447e-06,
+      "loss": 0.7009,
+      "step": 2766
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8080742405435985e-06,
+      "loss": 0.6773,
+      "step": 2767
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8058865931309796e-06,
+      "loss": 0.6588,
+      "step": 2768
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.803698548079294e-06,
+      "loss": 0.6168,
+      "step": 2769
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.801510105870324e-06,
+      "loss": 0.5396,
+      "step": 2770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.799321266985947e-06,
+      "loss": 0.6457,
+      "step": 2771
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.797132031908124e-06,
+      "loss": 0.6728,
+      "step": 2772
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.794942401118901e-06,
+      "loss": 0.5677,
+      "step": 2773
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.792752375100417e-06,
+      "loss": 0.5654,
+      "step": 2774
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.790561954334894e-06,
+      "loss": 0.6064,
+      "step": 2775
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.78837113930464e-06,
+      "loss": 0.6562,
+      "step": 2776
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.786179930492054e-06,
+      "loss": 0.5549,
+      "step": 2777
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7839883283796205e-06,
+      "loss": 0.6546,
+      "step": 2778
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.781796333449906e-06,
+      "loss": 0.6736,
+      "step": 2779
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.779603946185569e-06,
+      "loss": 0.6625,
+      "step": 2780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.777411167069351e-06,
+      "loss": 0.6259,
+      "step": 2781
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.775217996584082e-06,
+      "loss": 0.6514,
+      "step": 2782
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.773024435212678e-06,
+      "loss": 0.5791,
+      "step": 2783
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.770830483438141e-06,
+      "loss": 0.5609,
+      "step": 2784
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7686361417435564e-06,
+      "loss": 0.6509,
+      "step": 2785
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.766441410612097e-06,
+      "loss": 0.6631,
+      "step": 2786
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.764246290527023e-06,
+      "loss": 0.6474,
+      "step": 2787
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.762050781971678e-06,
+      "loss": 0.7032,
+      "step": 2788
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.759854885429494e-06,
+      "loss": 0.6575,
+      "step": 2789
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.757658601383986e-06,
+      "loss": 0.6484,
+      "step": 2790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.755461930318755e-06,
+      "loss": 0.6571,
+      "step": 2791
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.753264872717484e-06,
+      "loss": 0.6595,
+      "step": 2792
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7510674290639475e-06,
+      "loss": 0.5503,
+      "step": 2793
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.748869599842002e-06,
+      "loss": 0.5793,
+      "step": 2794
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.746671385535586e-06,
+      "loss": 0.6402,
+      "step": 2795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7444727866287286e-06,
+      "loss": 0.67,
+      "step": 2796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.742273803605538e-06,
+      "loss": 0.6702,
+      "step": 2797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.740074436950209e-06,
+      "loss": 0.6453,
+      "step": 2798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7378746871470226e-06,
+      "loss": 0.5594,
+      "step": 2799
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.735674554680342e-06,
+      "loss": 0.682,
+      "step": 2800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.733474040034616e-06,
+      "loss": 0.6458,
+      "step": 2801
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.731273143694379e-06,
+      "loss": 0.6763,
+      "step": 2802
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7290718661442445e-06,
+      "loss": 0.6982,
+      "step": 2803
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.726870207868913e-06,
+      "loss": 0.6473,
+      "step": 2804
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.724668169353169e-06,
+      "loss": 0.686,
+      "step": 2805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.722465751081882e-06,
+      "loss": 0.6004,
+      "step": 2806
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.720262953540002e-06,
+      "loss": 0.6676,
+      "step": 2807
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7180597772125665e-06,
+      "loss": 0.6569,
+      "step": 2808
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.715856222584691e-06,
+      "loss": 0.5699,
+      "step": 2809
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.713652290141578e-06,
+      "loss": 0.6702,
+      "step": 2810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7114479803685125e-06,
+      "loss": 0.6807,
+      "step": 2811
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7092432937508645e-06,
+      "loss": 0.6627,
+      "step": 2812
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.707038230774081e-06,
+      "loss": 0.5403,
+      "step": 2813
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.704832791923701e-06,
+      "loss": 0.5577,
+      "step": 2814
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.702626977685339e-06,
+      "loss": 0.5444,
+      "step": 2815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.700420788544693e-06,
+      "loss": 0.5563,
+      "step": 2816
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.698214224987546e-06,
+      "loss": 0.6532,
+      "step": 2817
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.69600728749976e-06,
+      "loss": 0.6542,
+      "step": 2818
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.693799976567286e-06,
+      "loss": 0.6562,
+      "step": 2819
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.691592292676151e-06,
+      "loss": 0.685,
+      "step": 2820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.689384236312465e-06,
+      "loss": 0.6399,
+      "step": 2821
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.687175807962423e-06,
+      "loss": 0.6393,
+      "step": 2822
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.684967008112299e-06,
+      "loss": 0.6579,
+      "step": 2823
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.682757837248447e-06,
+      "loss": 0.666,
+      "step": 2824
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6805482958573085e-06,
+      "loss": 0.5894,
+      "step": 2825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.678338384425405e-06,
+      "loss": 0.6856,
+      "step": 2826
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.676128103439334e-06,
+      "loss": 0.6952,
+      "step": 2827
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.673917453385782e-06,
+      "loss": 0.6704,
+      "step": 2828
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.671706434751511e-06,
+      "loss": 0.6842,
+      "step": 2829
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6694950480233655e-06,
+      "loss": 0.6319,
+      "step": 2830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6672832936882735e-06,
+      "loss": 0.6776,
+      "step": 2831
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.665071172233243e-06,
+      "loss": 0.6615,
+      "step": 2832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.662858684145359e-06,
+      "loss": 0.6827,
+      "step": 2833
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.660645829911794e-06,
+      "loss": 0.6922,
+      "step": 2834
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.658432610019796e-06,
+      "loss": 0.6778,
+      "step": 2835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6562190249566935e-06,
+      "loss": 0.6773,
+      "step": 2836
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6540050752099e-06,
+      "loss": 0.6735,
+      "step": 2837
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.651790761266903e-06,
+      "loss": 0.6626,
+      "step": 2838
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.649576083615274e-06,
+      "loss": 0.5779,
+      "step": 2839
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6473610427426675e-06,
+      "loss": 0.6664,
+      "step": 2840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.64514563913681e-06,
+      "loss": 0.6598,
+      "step": 2841
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.642929873285515e-06,
+      "loss": 0.6834,
+      "step": 2842
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.64071374567667e-06,
+      "loss": 0.6708,
+      "step": 2843
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.638497256798251e-06,
+      "loss": 0.6542,
+      "step": 2844
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.636280407138301e-06,
+      "loss": 0.6872,
+      "step": 2845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.634063197184955e-06,
+      "loss": 0.6609,
+      "step": 2846
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.631845627426418e-06,
+      "loss": 0.6675,
+      "step": 2847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.629627698350978e-06,
+      "loss": 0.6309,
+      "step": 2848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6274094104470025e-06,
+      "loss": 0.5894,
+      "step": 2849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.625190764202938e-06,
+      "loss": 0.6846,
+      "step": 2850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.622971760107308e-06,
+      "loss": 0.6618,
+      "step": 2851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.62075239864872e-06,
+      "loss": 0.6813,
+      "step": 2852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.618532680315851e-06,
+      "loss": 0.6636,
+      "step": 2853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.616312605597464e-06,
+      "loss": 0.5551,
+      "step": 2854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6140921749824e-06,
+      "loss": 0.664,
+      "step": 2855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.611871388959575e-06,
+      "loss": 0.6669,
+      "step": 2856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.609650248017985e-06,
+      "loss": 0.6709,
+      "step": 2857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.607428752646705e-06,
+      "loss": 0.6717,
+      "step": 2858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.605206903334886e-06,
+      "loss": 0.6592,
+      "step": 2859
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.602984700571758e-06,
+      "loss": 0.6577,
+      "step": 2860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.600762144846632e-06,
+      "loss": 0.5835,
+      "step": 2861
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.598539236648889e-06,
+      "loss": 0.5676,
+      "step": 2862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.596315976467995e-06,
+      "loss": 0.6579,
+      "step": 2863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.594092364793491e-06,
+      "loss": 0.6456,
+      "step": 2864
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.591868402114995e-06,
+      "loss": 0.646,
+      "step": 2865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.589644088922201e-06,
+      "loss": 0.6668,
+      "step": 2866
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.587419425704883e-06,
+      "loss": 0.6954,
+      "step": 2867
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.585194412952889e-06,
+      "loss": 0.5582,
+      "step": 2868
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.582969051156147e-06,
+      "loss": 0.6613,
+      "step": 2869
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.580743340804663e-06,
+      "loss": 0.6611,
+      "step": 2870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.578517282388512e-06,
+      "loss": 0.6854,
+      "step": 2871
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.576290876397856e-06,
+      "loss": 0.6229,
+      "step": 2872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.574064123322925e-06,
+      "loss": 0.5585,
+      "step": 2873
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.571837023654031e-06,
+      "loss": 0.6564,
+      "step": 2874
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.569609577881557e-06,
+      "loss": 0.6294,
+      "step": 2875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.56738178649597e-06,
+      "loss": 0.6583,
+      "step": 2876
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.565153649987803e-06,
+      "loss": 0.666,
+      "step": 2877
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.562925168847677e-06,
+      "loss": 0.6714,
+      "step": 2878
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.560696343566278e-06,
+      "loss": 0.6841,
+      "step": 2879
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.558467174634371e-06,
+      "loss": 0.6525,
+      "step": 2880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.556237662542801e-06,
+      "loss": 0.6562,
+      "step": 2881
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.554007807782485e-06,
+      "loss": 0.5659,
+      "step": 2882
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.551777610844414e-06,
+      "loss": 0.6655,
+      "step": 2883
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.549547072219659e-06,
+      "loss": 0.5372,
+      "step": 2884
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.54731619239936e-06,
+      "loss": 0.6425,
+      "step": 2885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.545084971874738e-06,
+      "loss": 0.5473,
+      "step": 2886
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.542853411137085e-06,
+      "loss": 0.592,
+      "step": 2887
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.540621510677771e-06,
+      "loss": 0.6223,
+      "step": 2888
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.538389270988238e-06,
+      "loss": 0.6399,
+      "step": 2889
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.536156692560005e-06,
+      "loss": 0.5477,
+      "step": 2890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5339237758846655e-06,
+      "loss": 0.6477,
+      "step": 2891
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.531690521453883e-06,
+      "loss": 0.5302,
+      "step": 2892
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5294569297594005e-06,
+      "loss": 0.6715,
+      "step": 2893
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.527223001293036e-06,
+      "loss": 0.5772,
+      "step": 2894
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.524988736546676e-06,
+      "loss": 0.6829,
+      "step": 2895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.522754136012287e-06,
+      "loss": 0.5369,
+      "step": 2896
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.520519200181905e-06,
+      "loss": 0.6772,
+      "step": 2897
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.518283929547643e-06,
+      "loss": 0.6688,
+      "step": 2898
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.516048324601685e-06,
+      "loss": 0.6636,
+      "step": 2899
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5138123858362914e-06,
+      "loss": 0.6711,
+      "step": 2900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.511576113743792e-06,
+      "loss": 0.5577,
+      "step": 2901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.509339508816597e-06,
+      "loss": 0.5655,
+      "step": 2902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.507102571547182e-06,
+      "loss": 0.6322,
+      "step": 2903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5048653024281005e-06,
+      "loss": 0.6499,
+      "step": 2904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.502627701951978e-06,
+      "loss": 0.6306,
+      "step": 2905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.500389770611512e-06,
+      "loss": 0.6361,
+      "step": 2906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.498151508899474e-06,
+      "loss": 0.651,
+      "step": 2907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.49591291730871e-06,
+      "loss": 0.6702,
+      "step": 2908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.493673996332135e-06,
+      "loss": 0.6584,
+      "step": 2909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.491434746462737e-06,
+      "loss": 0.6724,
+      "step": 2910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.489195168193579e-06,
+      "loss": 0.5496,
+      "step": 2911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.486955262017794e-06,
+      "loss": 0.5433,
+      "step": 2912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.484715028428588e-06,
+      "loss": 0.6379,
+      "step": 2913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.482474467919242e-06,
+      "loss": 0.6457,
+      "step": 2914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.480233580983104e-06,
+      "loss": 0.5994,
+      "step": 2915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.477992368113594e-06,
+      "loss": 0.6221,
+      "step": 2916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.475750829804208e-06,
+      "loss": 0.672,
+      "step": 2917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.473508966548512e-06,
+      "loss": 0.6816,
+      "step": 2918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.47126677884014e-06,
+      "loss": 0.5515,
+      "step": 2919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.469024267172805e-06,
+      "loss": 0.6708,
+      "step": 2920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.466781432040283e-06,
+      "loss": 0.6393,
+      "step": 2921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.464538273936427e-06,
+      "loss": 0.6895,
+      "step": 2922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.462294793355158e-06,
+      "loss": 0.6704,
+      "step": 2923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4600509907904694e-06,
+      "loss": 0.6589,
+      "step": 2924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.457806866736424e-06,
+      "loss": 0.6781,
+      "step": 2925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.45556242168716e-06,
+      "loss": 0.6279,
+      "step": 2926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.453317656136879e-06,
+      "loss": 0.6641,
+      "step": 2927
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.45107257057986e-06,
+      "loss": 0.6599,
+      "step": 2928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4488271655104485e-06,
+      "loss": 0.5696,
+      "step": 2929
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.44658144142306e-06,
+      "loss": 0.5512,
+      "step": 2930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.444335398812183e-06,
+      "loss": 0.646,
+      "step": 2931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.442089038172375e-06,
+      "loss": 0.6265,
+      "step": 2932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.439842359998263e-06,
+      "loss": 0.6617,
+      "step": 2933
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.437595364784544e-06,
+      "loss": 0.6715,
+      "step": 2934
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.435348053025985e-06,
+      "loss": 0.6697,
+      "step": 2935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.433100425217421e-06,
+      "loss": 0.6692,
+      "step": 2936
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.430852481853761e-06,
+      "loss": 0.6566,
+      "step": 2937
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.42860422342998e-06,
+      "loss": 0.6763,
+      "step": 2938
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.426355650441122e-06,
+      "loss": 0.5607,
+      "step": 2939
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4241067633823025e-06,
+      "loss": 0.662,
+      "step": 2940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.421857562748707e-06,
+      "loss": 0.67,
+      "step": 2941
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.419608049035584e-06,
+      "loss": 0.6611,
+      "step": 2942
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.417358222738257e-06,
+      "loss": 0.6703,
+      "step": 2943
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4151080843521166e-06,
+      "loss": 0.6665,
+      "step": 2944
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.41285763437262e-06,
+      "loss": 0.6393,
+      "step": 2945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.410606873295299e-06,
+      "loss": 0.544,
+      "step": 2946
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.408355801615747e-06,
+      "loss": 0.6652,
+      "step": 2947
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4061044198296285e-06,
+      "loss": 0.6717,
+      "step": 2948
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4038527284326765e-06,
+      "loss": 0.6896,
+      "step": 2949
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.401600727920692e-06,
+      "loss": 0.6681,
+      "step": 2950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.399348418789545e-06,
+      "loss": 0.6328,
+      "step": 2951
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.397095801535173e-06,
+      "loss": 0.6572,
+      "step": 2952
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.394842876653579e-06,
+      "loss": 0.5554,
+      "step": 2953
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.392589644640836e-06,
+      "loss": 0.5738,
+      "step": 2954
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3903361059930865e-06,
+      "loss": 0.5443,
+      "step": 2955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.388082261206535e-06,
+      "loss": 0.6569,
+      "step": 2956
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.385828110777459e-06,
+      "loss": 0.5468,
+      "step": 2957
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.383573655202202e-06,
+      "loss": 0.5655,
+      "step": 2958
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.381318894977171e-06,
+      "loss": 0.6621,
+      "step": 2959
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.379063830598843e-06,
+      "loss": 0.6563,
+      "step": 2960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.376808462563764e-06,
+      "loss": 0.6733,
+      "step": 2961
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.374552791368543e-06,
+      "loss": 0.6993,
+      "step": 2962
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.372296817509855e-06,
+      "loss": 0.5376,
+      "step": 2963
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.37004054148445e-06,
+      "loss": 0.5753,
+      "step": 2964
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.367783963789135e-06,
+      "loss": 0.6195,
+      "step": 2965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.365527084920785e-06,
+      "loss": 0.6536,
+      "step": 2966
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.363269905376348e-06,
+      "loss": 0.6628,
+      "step": 2967
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.361012425652828e-06,
+      "loss": 0.6702,
+      "step": 2968
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.358754646247305e-06,
+      "loss": 0.6556,
+      "step": 2969
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.35649656765692e-06,
+      "loss": 0.5693,
+      "step": 2970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.354238190378879e-06,
+      "loss": 0.6783,
+      "step": 2971
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.351979514910455e-06,
+      "loss": 0.6347,
+      "step": 2972
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.349720541748988e-06,
+      "loss": 0.705,
+      "step": 2973
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.347461271391882e-06,
+      "loss": 0.6598,
+      "step": 2974
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.345201704336607e-06,
+      "loss": 0.6382,
+      "step": 2975
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.342941841080699e-06,
+      "loss": 0.5482,
+      "step": 2976
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3406816821217554e-06,
+      "loss": 0.6168,
+      "step": 2977
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.338421227957446e-06,
+      "loss": 0.5497,
+      "step": 2978
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.336160479085498e-06,
+      "loss": 0.6584,
+      "step": 2979
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.333899436003708e-06,
+      "loss": 0.5487,
+      "step": 2980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.331638099209937e-06,
+      "loss": 0.6998,
+      "step": 2981
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.329376469202108e-06,
+      "loss": 0.6505,
+      "step": 2982
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.327114546478211e-06,
+      "loss": 0.6732,
+      "step": 2983
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.324852331536302e-06,
+      "loss": 0.6431,
+      "step": 2984
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.322589824874498e-06,
+      "loss": 0.5334,
+      "step": 2985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.32032702699098e-06,
+      "loss": 0.6551,
+      "step": 2986
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3180639383839956e-06,
+      "loss": 0.5653,
+      "step": 2987
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.315800559551857e-06,
+      "loss": 0.6663,
+      "step": 2988
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.313536890992935e-06,
+      "loss": 0.638,
+      "step": 2989
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.311272933205672e-06,
+      "loss": 0.5497,
+      "step": 2990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.309008686688567e-06,
+      "loss": 0.6236,
+      "step": 2991
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.306744151940187e-06,
+      "loss": 0.6447,
+      "step": 2992
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.304479329459161e-06,
+      "loss": 0.6431,
+      "step": 2993
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.302214219744181e-06,
+      "loss": 0.5211,
+      "step": 2994
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.299948823294002e-06,
+      "loss": 0.5407,
+      "step": 2995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.297683140607445e-06,
+      "loss": 0.5427,
+      "step": 2996
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.29541717218339e-06,
+      "loss": 0.6696,
+      "step": 2997
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2931509185207815e-06,
+      "loss": 0.6495,
+      "step": 2998
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2908843801186285e-06,
+      "loss": 0.6571,
+      "step": 2999
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.288617557475999e-06,
+      "loss": 0.6733,
+      "step": 3000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.286350451092027e-06,
+      "loss": 0.6584,
+      "step": 3001
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.28408306146591e-06,
+      "loss": 0.6468,
+      "step": 3002
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.281815389096903e-06,
+      "loss": 0.6215,
+      "step": 3003
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.279547434484325e-06,
+      "loss": 0.5602,
+      "step": 3004
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.277279198127561e-06,
+      "loss": 0.6585,
+      "step": 3005
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.275010680526051e-06,
+      "loss": 0.6629,
+      "step": 3006
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.272741882179305e-06,
+      "loss": 0.7036,
+      "step": 3007
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.270472803586891e-06,
+      "loss": 0.5551,
+      "step": 3008
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2682034452484365e-06,
+      "loss": 0.5381,
+      "step": 3009
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2659338076636325e-06,
+      "loss": 0.5629,
+      "step": 3010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.263663891332232e-06,
+      "loss": 0.66,
+      "step": 3011
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.26139369675405e-06,
+      "loss": 0.6292,
+      "step": 3012
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.25912322442896e-06,
+      "loss": 0.6927,
+      "step": 3013
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2568524748569e-06,
+      "loss": 0.5641,
+      "step": 3014
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.254581448537868e-06,
+      "loss": 0.6343,
+      "step": 3015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2523101459719204e-06,
+      "loss": 0.5592,
+      "step": 3016
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.250038567659178e-06,
+      "loss": 0.6543,
+      "step": 3017
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.24776671409982e-06,
+      "loss": 0.6652,
+      "step": 3018
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2454945857940865e-06,
+      "loss": 0.6793,
+      "step": 3019
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2432221832422805e-06,
+      "loss": 0.6419,
+      "step": 3020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.240949506944762e-06,
+      "loss": 0.671,
+      "step": 3021
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.238676557401952e-06,
+      "loss": 0.5967,
+      "step": 3022
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.236403335114333e-06,
+      "loss": 0.558,
+      "step": 3023
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.23412984058245e-06,
+      "loss": 0.6495,
+      "step": 3024
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2318560743069005e-06,
+      "loss": 0.5779,
+      "step": 3025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.229582036788346e-06,
+      "loss": 0.6437,
+      "step": 3026
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2273077285275115e-06,
+      "loss": 0.5713,
+      "step": 3027
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.225033150025177e-06,
+      "loss": 0.6643,
+      "step": 3028
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.222758301782183e-06,
+      "loss": 0.6399,
+      "step": 3029
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.220483184299426e-06,
+      "loss": 0.6773,
+      "step": 3030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.218207798077871e-06,
+      "loss": 0.5314,
+      "step": 3031
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2159321436185326e-06,
+      "loss": 0.6688,
+      "step": 3032
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.213656221422488e-06,
+      "loss": 0.6632,
+      "step": 3033
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.211380031990878e-06,
+      "loss": 0.5547,
+      "step": 3034
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.209103575824894e-06,
+      "loss": 0.6707,
+      "step": 3035
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.206826853425789e-06,
+      "loss": 0.6539,
+      "step": 3036
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.20454986529488e-06,
+      "loss": 0.5343,
+      "step": 3037
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.202272611933533e-06,
+      "loss": 0.6382,
+      "step": 3038
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1999950938431805e-06,
+      "loss": 0.6725,
+      "step": 3039
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.197717311525312e-06,
+      "loss": 0.6627,
+      "step": 3040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.195439265481472e-06,
+      "loss": 0.6541,
+      "step": 3041
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.193160956213262e-06,
+      "loss": 0.5381,
+      "step": 3042
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.190882384222348e-06,
+      "loss": 0.7027,
+      "step": 3043
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.188603550010447e-06,
+      "loss": 0.6153,
+      "step": 3044
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.186324454079337e-06,
+      "loss": 0.6322,
+      "step": 3045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.184045096930855e-06,
+      "loss": 0.6849,
+      "step": 3046
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.181765479066895e-06,
+      "loss": 0.6481,
+      "step": 3047
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.179485600989401e-06,
+      "loss": 0.6841,
+      "step": 3048
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.177205463200388e-06,
+      "loss": 0.5638,
+      "step": 3049
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.174925066201914e-06,
+      "loss": 0.6789,
+      "step": 3050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1726444104961054e-06,
+      "loss": 0.6054,
+      "step": 3051
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.170363496585141e-06,
+      "loss": 0.631,
+      "step": 3052
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.168082324971254e-06,
+      "loss": 0.6568,
+      "step": 3053
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.165800896156736e-06,
+      "loss": 0.6509,
+      "step": 3054
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.163519210643939e-06,
+      "loss": 0.6785,
+      "step": 3055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.161237268935266e-06,
+      "loss": 0.5545,
+      "step": 3056
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.15895507153318e-06,
+      "loss": 0.6622,
+      "step": 3057
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1566726189402e-06,
+      "loss": 0.6353,
+      "step": 3058
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1543899116589e-06,
+      "loss": 0.6276,
+      "step": 3059
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.152106950191907e-06,
+      "loss": 0.6546,
+      "step": 3060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.149823735041913e-06,
+      "loss": 0.6347,
+      "step": 3061
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.147540266711656e-06,
+      "loss": 0.6899,
+      "step": 3062
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.145256545703936e-06,
+      "loss": 0.6289,
+      "step": 3063
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1429725725216075e-06,
+      "loss": 0.6683,
+      "step": 3064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.140688347667577e-06,
+      "loss": 0.6591,
+      "step": 3065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.13840387164481e-06,
+      "loss": 0.6398,
+      "step": 3066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.136119144956329e-06,
+      "loss": 0.6292,
+      "step": 3067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.133834168105206e-06,
+      "loss": 0.6495,
+      "step": 3068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.131548941594573e-06,
+      "loss": 0.6601,
+      "step": 3069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.129263465927613e-06,
+      "loss": 0.6679,
+      "step": 3070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.126977741607568e-06,
+      "loss": 0.6751,
+      "step": 3071
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.124691769137733e-06,
+      "loss": 0.5635,
+      "step": 3072
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.122405549021456e-06,
+      "loss": 0.5331,
+      "step": 3073
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.120119081762143e-06,
+      "loss": 0.5534,
+      "step": 3074
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.117832367863249e-06,
+      "loss": 0.6109,
+      "step": 3075
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.115545407828288e-06,
+      "loss": 0.7301,
+      "step": 3076
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.11325820216083e-06,
+      "loss": 0.6587,
+      "step": 3077
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.110970751364493e-06,
+      "loss": 0.648,
+      "step": 3078
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1086830559429515e-06,
+      "loss": 0.5568,
+      "step": 3079
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.106395116399936e-06,
+      "loss": 0.575,
+      "step": 3080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.104106933239227e-06,
+      "loss": 0.6658,
+      "step": 3081
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.101818506964663e-06,
+      "loss": 0.6466,
+      "step": 3082
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.099529838080132e-06,
+      "loss": 0.6708,
+      "step": 3083
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0972409270895786e-06,
+      "loss": 0.6782,
+      "step": 3084
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.094951774496998e-06,
+      "loss": 0.6648,
+      "step": 3085
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.09266238080644e-06,
+      "loss": 0.6509,
+      "step": 3086
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.090372746522009e-06,
+      "loss": 0.5758,
+      "step": 3087
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.088082872147858e-06,
+      "loss": 0.6561,
+      "step": 3088
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.085792758188196e-06,
+      "loss": 0.5621,
+      "step": 3089
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.083502405147288e-06,
+      "loss": 0.6736,
+      "step": 3090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.081211813529444e-06,
+      "loss": 0.6744,
+      "step": 3091
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.078920983839032e-06,
+      "loss": 0.6495,
+      "step": 3092
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.07662991658047e-06,
+      "loss": 0.6687,
+      "step": 3093
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.07433861225823e-06,
+      "loss": 0.659,
+      "step": 3094
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0720470713768355e-06,
+      "loss": 0.6945,
+      "step": 3095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0697552944408626e-06,
+      "loss": 0.6631,
+      "step": 3096
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.067463281954938e-06,
+      "loss": 0.54,
+      "step": 3097
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.06517103442374e-06,
+      "loss": 0.642,
+      "step": 3098
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0628785523520025e-06,
+      "loss": 0.6696,
+      "step": 3099
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.060585836244506e-06,
+      "loss": 0.5586,
+      "step": 3100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.058292886606084e-06,
+      "loss": 0.6527,
+      "step": 3101
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.055999703941627e-06,
+      "loss": 0.6324,
+      "step": 3102
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.053706288756067e-06,
+      "loss": 0.6374,
+      "step": 3103
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.051412641554393e-06,
+      "loss": 0.674,
+      "step": 3104
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.049118762841647e-06,
+      "loss": 0.6793,
+      "step": 3105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.046824653122918e-06,
+      "loss": 0.6606,
+      "step": 3106
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.044530312903343e-06,
+      "loss": 0.6466,
+      "step": 3107
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.042235742688121e-06,
+      "loss": 0.5786,
+      "step": 3108
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.039940942982492e-06,
+      "loss": 0.6754,
+      "step": 3109
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.037645914291745e-06,
+      "loss": 0.561,
+      "step": 3110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.03535065712123e-06,
+      "loss": 0.5463,
+      "step": 3111
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.033055171976337e-06,
+      "loss": 0.6541,
+      "step": 3112
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.030759459362509e-06,
+      "loss": 0.6878,
+      "step": 3113
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.028463519785242e-06,
+      "loss": 0.6609,
+      "step": 3114
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.026167353750082e-06,
+      "loss": 0.6273,
+      "step": 3115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.023870961762619e-06,
+      "loss": 0.6597,
+      "step": 3116
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.021574344328499e-06,
+      "loss": 0.675,
+      "step": 3117
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.019277501953416e-06,
+      "loss": 0.6697,
+      "step": 3118
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.016980435143109e-06,
+      "loss": 0.6672,
+      "step": 3119
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.014683144403375e-06,
+      "loss": 0.6437,
+      "step": 3120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.012385630240054e-06,
+      "loss": 0.6283,
+      "step": 3121
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.010087893159034e-06,
+      "loss": 0.6421,
+      "step": 3122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.007789933666259e-06,
+      "loss": 0.619,
+      "step": 3123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.005491752267717e-06,
+      "loss": 0.6745,
+      "step": 3124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.003193349469444e-06,
+      "loss": 0.5812,
+      "step": 3125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.000894725777526e-06,
+      "loss": 0.6723,
+      "step": 3126
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9985958816981e-06,
+      "loss": 0.6481,
+      "step": 3127
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.996296817737349e-06,
+      "loss": 0.6654,
+      "step": 3128
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.993997534401507e-06,
+      "loss": 0.6489,
+      "step": 3129
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.991698032196851e-06,
+      "loss": 0.5611,
+      "step": 3130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.98939831162971e-06,
+      "loss": 0.6693,
+      "step": 3131
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9870983732064615e-06,
+      "loss": 0.6576,
+      "step": 3132
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9847982174335314e-06,
+      "loss": 0.6484,
+      "step": 3133
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9824978448173895e-06,
+      "loss": 0.6536,
+      "step": 3134
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.98019725586456e-06,
+      "loss": 0.6428,
+      "step": 3135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.977896451081604e-06,
+      "loss": 0.6591,
+      "step": 3136
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.975595430975144e-06,
+      "loss": 0.5338,
+      "step": 3137
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.973294196051839e-06,
+      "loss": 0.663,
+      "step": 3138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.970992746818398e-06,
+      "loss": 0.6439,
+      "step": 3139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.968691083781583e-06,
+      "loss": 0.6764,
+      "step": 3140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9663892074481944e-06,
+      "loss": 0.5524,
+      "step": 3141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.964087118325082e-06,
+      "loss": 0.6672,
+      "step": 3142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9617848169191496e-06,
+      "loss": 0.6714,
+      "step": 3143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.959482303737337e-06,
+      "loss": 0.6373,
+      "step": 3144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9571795792866385e-06,
+      "loss": 0.6283,
+      "step": 3145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.954876644074092e-06,
+      "loss": 0.6293,
+      "step": 3146
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9525734986067794e-06,
+      "loss": 0.5637,
+      "step": 3147
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.950270143391834e-06,
+      "loss": 0.543,
+      "step": 3148
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.947966578936432e-06,
+      "loss": 0.6643,
+      "step": 3149
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.945662805747798e-06,
+      "loss": 0.6543,
+      "step": 3150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.943358824333196e-06,
+      "loss": 0.6635,
+      "step": 3151
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.941054635199947e-06,
+      "loss": 0.6647,
+      "step": 3152
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.938750238855408e-06,
+      "loss": 0.6237,
+      "step": 3153
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.936445635806985e-06,
+      "loss": 0.6497,
+      "step": 3154
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.934140826562133e-06,
+      "loss": 0.6309,
+      "step": 3155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.931835811628344e-06,
+      "loss": 0.6722,
+      "step": 3156
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.929530591513165e-06,
+      "loss": 0.5745,
+      "step": 3157
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.92722516672418e-06,
+      "loss": 0.6415,
+      "step": 3158
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.924919537769025e-06,
+      "loss": 0.5374,
+      "step": 3159
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9226137051553745e-06,
+      "loss": 0.5786,
+      "step": 3160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.920307669390954e-06,
+      "loss": 0.6645,
+      "step": 3161
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9180014309835275e-06,
+      "loss": 0.652,
+      "step": 3162
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9156949904409075e-06,
+      "loss": 0.5733,
+      "step": 3163
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.913388348270952e-06,
+      "loss": 0.6555,
+      "step": 3164
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.911081504981562e-06,
+      "loss": 0.639,
+      "step": 3165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.908774461080679e-06,
+      "loss": 0.6594,
+      "step": 3166
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.906467217076296e-06,
+      "loss": 0.6859,
+      "step": 3167
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.904159773476445e-06,
+      "loss": 0.6254,
+      "step": 3168
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.901852130789201e-06,
+      "loss": 0.6265,
+      "step": 3169
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.899544289522687e-06,
+      "loss": 0.6488,
+      "step": 3170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.897236250185069e-06,
+      "loss": 0.5685,
+      "step": 3171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.894928013284551e-06,
+      "loss": 0.5531,
+      "step": 3172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.892619579329391e-06,
+      "loss": 0.6637,
+      "step": 3173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8903109488278795e-06,
+      "loss": 0.6552,
+      "step": 3174
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8880021222883555e-06,
+      "loss": 0.5517,
+      "step": 3175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8856931002192005e-06,
+      "loss": 0.659,
+      "step": 3176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.883383883128843e-06,
+      "loss": 0.6178,
+      "step": 3177
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.881074471525746e-06,
+      "loss": 0.6622,
+      "step": 3178
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.878764865918424e-06,
+      "loss": 0.663,
+      "step": 3179
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.876455066815427e-06,
+      "loss": 0.5974,
+      "step": 3180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.87414507472535e-06,
+      "loss": 0.6767,
+      "step": 3181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8718348901568345e-06,
+      "loss": 0.6542,
+      "step": 3182
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.869524513618561e-06,
+      "loss": 0.6248,
+      "step": 3183
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.86721394561925e-06,
+      "loss": 0.6278,
+      "step": 3184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8649031866676685e-06,
+      "loss": 0.6704,
+      "step": 3185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8625922372726244e-06,
+      "loss": 0.6578,
+      "step": 3186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.860281097942964e-06,
+      "loss": 0.6789,
+      "step": 3187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.857969769187579e-06,
+      "loss": 0.5386,
+      "step": 3188
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.855658251515403e-06,
+      "loss": 0.6612,
+      "step": 3189
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.853346545435412e-06,
+      "loss": 0.6352,
+      "step": 3190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.851034651456619e-06,
+      "loss": 0.6466,
+      "step": 3191
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.848722570088081e-06,
+      "loss": 0.632,
+      "step": 3192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8464103018388985e-06,
+      "loss": 0.6424,
+      "step": 3193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.84409784721821e-06,
+      "loss": 0.6338,
+      "step": 3194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.841785206735192e-06,
+      "loss": 0.6522,
+      "step": 3195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.839472380899074e-06,
+      "loss": 0.6866,
+      "step": 3196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.837159370219112e-06,
+      "loss": 0.6691,
+      "step": 3197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.834846175204612e-06,
+      "loss": 0.5773,
+      "step": 3198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.832532796364915e-06,
+      "loss": 0.6496,
+      "step": 3199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8302192342094075e-06,
+      "loss": 0.6702,
+      "step": 3200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.827905489247513e-06,
+      "loss": 0.6266,
+      "step": 3201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.825591561988695e-06,
+      "loss": 0.5555,
+      "step": 3202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.823277452942461e-06,
+      "loss": 0.656,
+      "step": 3203
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8209631626183525e-06,
+      "loss": 0.6571,
+      "step": 3204
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.818648691525957e-06,
+      "loss": 0.6673,
+      "step": 3205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.816334040174898e-06,
+      "loss": 0.5757,
+      "step": 3206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.814019209074839e-06,
+      "loss": 0.6591,
+      "step": 3207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.811704198735485e-06,
+      "loss": 0.6575,
+      "step": 3208
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809389009666579e-06,
+      "loss": 0.6801,
+      "step": 3209
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.807073642377903e-06,
+      "loss": 0.6472,
+      "step": 3210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.804758097379281e-06,
+      "loss": 0.642,
+      "step": 3211
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802442375180571e-06,
+      "loss": 0.6389,
+      "step": 3212
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800126476291675e-06,
+      "loss": 0.6779,
+      "step": 3213
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.797810401222529e-06,
+      "loss": 0.573,
+      "step": 3214
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.795494150483115e-06,
+      "loss": 0.6754,
+      "step": 3215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.793177724583447e-06,
+      "loss": 0.6416,
+      "step": 3216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.790861124033582e-06,
+      "loss": 0.6175,
+      "step": 3217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.78854434934361e-06,
+      "loss": 0.5678,
+      "step": 3218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.786227401023666e-06,
+      "loss": 0.6239,
+      "step": 3219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.783910279583916e-06,
+      "loss": 0.645,
+      "step": 3220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.781592985534573e-06,
+      "loss": 0.6521,
+      "step": 3221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.77927551938588e-06,
+      "loss": 0.6433,
+      "step": 3222
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.776957881648123e-06,
+      "loss": 0.6607,
+      "step": 3223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.774640072831622e-06,
+      "loss": 0.6613,
+      "step": 3224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.772322093446737e-06,
+      "loss": 0.6564,
+      "step": 3225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.770003944003865e-06,
+      "loss": 0.6609,
+      "step": 3226
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.767685625013441e-06,
+      "loss": 0.7103,
+      "step": 3227
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.765367136985936e-06,
+      "loss": 0.6684,
+      "step": 3228
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.763048480431861e-06,
+      "loss": 0.6583,
+      "step": 3229
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.76072965586176e-06,
+      "loss": 0.674,
+      "step": 3230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.758410663786216e-06,
+      "loss": 0.6609,
+      "step": 3231
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.756091504715851e-06,
+      "loss": 0.6282,
+      "step": 3232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.75377217916132e-06,
+      "loss": 0.5659,
+      "step": 3233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.751452687633316e-06,
+      "loss": 0.681,
+      "step": 3234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.749133030642573e-06,
+      "loss": 0.6762,
+      "step": 3235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.746813208699853e-06,
+      "loss": 0.5612,
+      "step": 3236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.74449322231596e-06,
+      "loss": 0.6632,
+      "step": 3237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7421730720017334e-06,
+      "loss": 0.5626,
+      "step": 3238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.739852758268048e-06,
+      "loss": 0.6431,
+      "step": 3239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.737532281625814e-06,
+      "loss": 0.555,
+      "step": 3240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.73521164258598e-06,
+      "loss": 0.5451,
+      "step": 3241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7328908416595286e-06,
+      "loss": 0.5579,
+      "step": 3242
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.730569879357477e-06,
+      "loss": 0.6532,
+      "step": 3243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7282487561908805e-06,
+      "loss": 0.5578,
+      "step": 3244
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.725927472670826e-06,
+      "loss": 0.6686,
+      "step": 3245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.723606029308438e-06,
+      "loss": 0.6706,
+      "step": 3246
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7212844266148805e-06,
+      "loss": 0.6675,
+      "step": 3247
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.718962665101344e-06,
+      "loss": 0.6417,
+      "step": 3248
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7166407452790604e-06,
+      "loss": 0.6484,
+      "step": 3249
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.714318667659295e-06,
+      "loss": 0.5376,
+      "step": 3250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.711996432753344e-06,
+      "loss": 0.6897,
+      "step": 3251
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.709674041072542e-06,
+      "loss": 0.6435,
+      "step": 3252
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.707351493128263e-06,
+      "loss": 0.6309,
+      "step": 3253
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7050287894319035e-06,
+      "loss": 0.5731,
+      "step": 3254
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.702705930494905e-06,
+      "loss": 0.6484,
+      "step": 3255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.700382916828737e-06,
+      "loss": 0.5623,
+      "step": 3256
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.698059748944905e-06,
+      "loss": 0.6346,
+      "step": 3257
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.695736427354948e-06,
+      "loss": 0.5843,
+      "step": 3258
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.693412952570442e-06,
+      "loss": 0.5716,
+      "step": 3259
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6910893251029905e-06,
+      "loss": 0.6687,
+      "step": 3260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.688765545464238e-06,
+      "loss": 0.6798,
+      "step": 3261
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6864416141658565e-06,
+      "loss": 0.6929,
+      "step": 3262
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.684117531719552e-06,
+      "loss": 0.6691,
+      "step": 3263
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6817932986370675e-06,
+      "loss": 0.6437,
+      "step": 3264
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.679468915430177e-06,
+      "loss": 0.6819,
+      "step": 3265
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6771443826106876e-06,
+      "loss": 0.6653,
+      "step": 3266
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.67481970069044e-06,
+      "loss": 0.6496,
+      "step": 3267
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6724948701813046e-06,
+      "loss": 0.6411,
+      "step": 3268
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6701698915951885e-06,
+      "loss": 0.6607,
+      "step": 3269
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.66784476544403e-06,
+      "loss": 0.5712,
+      "step": 3270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6655194922398e-06,
+      "loss": 0.6638,
+      "step": 3271
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6631940724945e-06,
+      "loss": 0.6185,
+      "step": 3272
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.660868506720169e-06,
+      "loss": 0.6573,
+      "step": 3273
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.658542795428871e-06,
+      "loss": 0.6536,
+      "step": 3274
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.656216939132706e-06,
+      "loss": 0.6487,
+      "step": 3275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.653890938343806e-06,
+      "loss": 0.6485,
+      "step": 3276
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.651564793574335e-06,
+      "loss": 0.6696,
+      "step": 3277
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.649238505336486e-06,
+      "loss": 0.6278,
+      "step": 3278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.646912074142488e-06,
+      "loss": 0.6252,
+      "step": 3279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.644585500504599e-06,
+      "loss": 0.6188,
+      "step": 3280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6422587849351055e-06,
+      "loss": 0.6931,
+      "step": 3281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.63993192794633e-06,
+      "loss": 0.56,
+      "step": 3282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.637604930050625e-06,
+      "loss": 0.6299,
+      "step": 3283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.635277791760372e-06,
+      "loss": 0.6634,
+      "step": 3284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.632950513587986e-06,
+      "loss": 0.524,
+      "step": 3285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.630623096045911e-06,
+      "loss": 0.6521,
+      "step": 3286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.628295539646621e-06,
+      "loss": 0.6695,
+      "step": 3287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.625967844902624e-06,
+      "loss": 0.6833,
+      "step": 3288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.623640012326455e-06,
+      "loss": 0.652,
+      "step": 3289
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.62131204243068e-06,
+      "loss": 0.6643,
+      "step": 3290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6189839357278975e-06,
+      "loss": 0.6393,
+      "step": 3291
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6166556927307325e-06,
+      "loss": 0.6593,
+      "step": 3292
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.614327313951844e-06,
+      "loss": 0.6289,
+      "step": 3293
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.611998799903918e-06,
+      "loss": 0.559,
+      "step": 3294
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.609670151099671e-06,
+      "loss": 0.6375,
+      "step": 3295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.607341368051848e-06,
+      "loss": 0.5462,
+      "step": 3296
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.605012451273229e-06,
+      "loss": 0.565,
+      "step": 3297
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6026834012766155e-06,
+      "loss": 0.6715,
+      "step": 3298
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6003542185748435e-06,
+      "loss": 0.624,
+      "step": 3299
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.598024903680778e-06,
+      "loss": 0.6532,
+      "step": 3300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.595695457107309e-06,
+      "loss": 0.6489,
+      "step": 3301
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.593365879367361e-06,
+      "loss": 0.5431,
+      "step": 3302
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.591036170973885e-06,
+      "loss": 0.5726,
+      "step": 3303
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.588706332439858e-06,
+      "loss": 0.6446,
+      "step": 3304
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.586376364278292e-06,
+      "loss": 0.6359,
+      "step": 3305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.584046267002222e-06,
+      "loss": 0.6847,
+      "step": 3306
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5817160411247115e-06,
+      "loss": 0.6524,
+      "step": 3307
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.579385687158857e-06,
+      "loss": 0.6555,
+      "step": 3308
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.57705520561778e-06,
+      "loss": 0.6582,
+      "step": 3309
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.574724597014628e-06,
+      "loss": 0.6342,
+      "step": 3310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.572393861862582e-06,
+      "loss": 0.659,
+      "step": 3311
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.570063000674845e-06,
+      "loss": 0.6477,
+      "step": 3312
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.567732013964653e-06,
+      "loss": 0.6434,
+      "step": 3313
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.565400902245264e-06,
+      "loss": 0.6349,
+      "step": 3314
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.56306966602997e-06,
+      "loss": 0.642,
+      "step": 3315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.560738305832085e-06,
+      "loss": 0.5688,
+      "step": 3316
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.558406822164953e-06,
+      "loss": 0.6437,
+      "step": 3317
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.556075215541945e-06,
+      "loss": 0.6453,
+      "step": 3318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.553743486476457e-06,
+      "loss": 0.6602,
+      "step": 3319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.551411635481915e-06,
+      "loss": 0.5553,
+      "step": 3320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.54907966307177e-06,
+      "loss": 0.6721,
+      "step": 3321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5467475697595e-06,
+      "loss": 0.555,
+      "step": 3322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.544415356058611e-06,
+      "loss": 0.6357,
+      "step": 3323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.542083022482634e-06,
+      "loss": 0.6659,
+      "step": 3324
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.539750569545124e-06,
+      "loss": 0.6632,
+      "step": 3325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.537417997759669e-06,
+      "loss": 0.6558,
+      "step": 3326
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.535085307639875e-06,
+      "loss": 0.567,
+      "step": 3327
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.532752499699381e-06,
+      "loss": 0.5742,
+      "step": 3328
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.530419574451849e-06,
+      "loss": 0.6619,
+      "step": 3329
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.528086532410968e-06,
+      "loss": 0.5473,
+      "step": 3330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.525753374090448e-06,
+      "loss": 0.6475,
+      "step": 3331
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.523420100004031e-06,
+      "loss": 0.5649,
+      "step": 3332
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5210867106654806e-06,
+      "loss": 0.6525,
+      "step": 3333
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.518753206588588e-06,
+      "loss": 0.6633,
+      "step": 3334
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.516419588287169e-06,
+      "loss": 0.6786,
+      "step": 3335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.514085856275064e-06,
+      "loss": 0.6334,
+      "step": 3336
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.511752011066136e-06,
+      "loss": 0.6381,
+      "step": 3337
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.509418053174279e-06,
+      "loss": 0.6634,
+      "step": 3338
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.507083983113406e-06,
+      "loss": 0.5591,
+      "step": 3339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.504749801397457e-06,
+      "loss": 0.5565,
+      "step": 3340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.502415508540401e-06,
+      "loss": 0.5476,
+      "step": 3341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.500081105056222e-06,
+      "loss": 0.6332,
+      "step": 3342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.497746591458934e-06,
+      "loss": 0.6565,
+      "step": 3343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.495411968262577e-06,
+      "loss": 0.638,
+      "step": 3344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.493077235981211e-06,
+      "loss": 0.656,
+      "step": 3345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.490742395128923e-06,
+      "loss": 0.5699,
+      "step": 3346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.488407446219824e-06,
+      "loss": 0.6546,
+      "step": 3347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4860723897680424e-06,
+      "loss": 0.6646,
+      "step": 3348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.483737226287742e-06,
+      "loss": 0.7015,
+      "step": 3349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4814019562931e-06,
+      "loss": 0.6436,
+      "step": 3350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.47906658029832e-06,
+      "loss": 0.6574,
+      "step": 3351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.476731098817631e-06,
+      "loss": 0.6376,
+      "step": 3352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.474395512365286e-06,
+      "loss": 0.6919,
+      "step": 3353
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.472059821455554e-06,
+      "loss": 0.6635,
+      "step": 3354
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.469724026602736e-06,
+      "loss": 0.6623,
+      "step": 3355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4673881283211505e-06,
+      "loss": 0.6727,
+      "step": 3356
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.465052127125139e-06,
+      "loss": 0.6699,
+      "step": 3357
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.462716023529067e-06,
+      "loss": 0.6866,
+      "step": 3358
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.460379818047327e-06,
+      "loss": 0.6444,
+      "step": 3359
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.458043511194322e-06,
+      "loss": 0.646,
+      "step": 3360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.455707103484489e-06,
+      "loss": 0.661,
+      "step": 3361
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.453370595432281e-06,
+      "loss": 0.6381,
+      "step": 3362
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.451033987552177e-06,
+      "loss": 0.6595,
+      "step": 3363
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.448697280358674e-06,
+      "loss": 0.6195,
+      "step": 3364
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.446360474366291e-06,
+      "loss": 0.6297,
+      "step": 3365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.444023570089574e-06,
+      "loss": 0.6851,
+      "step": 3366
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.441686568043086e-06,
+      "loss": 0.5743,
+      "step": 3367
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.439349468741414e-06,
+      "loss": 0.6337,
+      "step": 3368
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.437012272699161e-06,
+      "loss": 0.6708,
+      "step": 3369
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.434674980430959e-06,
+      "loss": 0.6686,
+      "step": 3370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4323375924514555e-06,
+      "loss": 0.6664,
+      "step": 3371
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.430000109275322e-06,
+      "loss": 0.6523,
+      "step": 3372
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.42766253141725e-06,
+      "loss": 0.6517,
+      "step": 3373
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4253248593919535e-06,
+      "loss": 0.6352,
+      "step": 3374
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.422987093714162e-06,
+      "loss": 0.6692,
+      "step": 3375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.420649234898634e-06,
+      "loss": 0.5597,
+      "step": 3376
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.41831128346014e-06,
+      "loss": 0.6787,
+      "step": 3377
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.415973239913475e-06,
+      "loss": 0.5729,
+      "step": 3378
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.413635104773457e-06,
+      "loss": 0.6825,
+      "step": 3379
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.411296878554918e-06,
+      "loss": 0.6183,
+      "step": 3380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4089585617727136e-06,
+      "loss": 0.6914,
+      "step": 3381
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.406620154941721e-06,
+      "loss": 0.6518,
+      "step": 3382
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.404281658576832e-06,
+      "loss": 0.6877,
+      "step": 3383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.401943073192963e-06,
+      "loss": 0.5425,
+      "step": 3384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.399604399305051e-06,
+      "loss": 0.6588,
+      "step": 3385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.397265637428046e-06,
+      "loss": 0.5643,
+      "step": 3386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.394926788076921e-06,
+      "loss": 0.6368,
+      "step": 3387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.392587851766671e-06,
+      "loss": 0.6655,
+      "step": 3388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.390248829012305e-06,
+      "loss": 0.553,
+      "step": 3389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.387909720328854e-06,
+      "loss": 0.5695,
+      "step": 3390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.385570526231372e-06,
+      "loss": 0.6353,
+      "step": 3391
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.383231247234921e-06,
+      "loss": 0.6687,
+      "step": 3392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.380891883854591e-06,
+      "loss": 0.6683,
+      "step": 3393
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.378552436605487e-06,
+      "loss": 0.5531,
+      "step": 3394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.376212906002734e-06,
+      "loss": 0.6582,
+      "step": 3395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3738732925614715e-06,
+      "loss": 0.6618,
+      "step": 3396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.371533596796864e-06,
+      "loss": 0.661,
+      "step": 3397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.369193819224089e-06,
+      "loss": 0.6646,
+      "step": 3398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3668539603583425e-06,
+      "loss": 0.6531,
+      "step": 3399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3645140207148396e-06,
+      "loss": 0.6455,
+      "step": 3400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.362174000808813e-06,
+      "loss": 0.6574,
+      "step": 3401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.359833901155511e-06,
+      "loss": 0.6266,
+      "step": 3402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.357493722270205e-06,
+      "loss": 0.6557,
+      "step": 3403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.355153464668177e-06,
+      "loss": 0.6436,
+      "step": 3404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.352813128864731e-06,
+      "loss": 0.6423,
+      "step": 3405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3504727153751865e-06,
+      "loss": 0.5681,
+      "step": 3406
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.348132224714879e-06,
+      "loss": 0.7051,
+      "step": 3407
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.345791657399165e-06,
+      "loss": 0.6323,
+      "step": 3408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3434510139434116e-06,
+      "loss": 0.6559,
+      "step": 3409
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.341110294863009e-06,
+      "loss": 0.5732,
+      "step": 3410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.338769500673362e-06,
+      "loss": 0.6587,
+      "step": 3411
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.336428631889888e-06,
+      "loss": 0.6919,
+      "step": 3412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.334087689028027e-06,
+      "loss": 0.6274,
+      "step": 3413
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.331746672603231e-06,
+      "loss": 0.5629,
+      "step": 3414
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.329405583130969e-06,
+      "loss": 0.6536,
+      "step": 3415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.327064421126726e-06,
+      "loss": 0.6738,
+      "step": 3416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.324723187106008e-06,
+      "loss": 0.6221,
+      "step": 3417
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.322381881584329e-06,
+      "loss": 0.6464,
+      "step": 3418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.320040505077222e-06,
+      "loss": 0.5544,
+      "step": 3419
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.317699058100237e-06,
+      "loss": 0.6672,
+      "step": 3420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.315357541168937e-06,
+      "loss": 0.5271,
+      "step": 3421
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3130159547989035e-06,
+      "loss": 0.6585,
+      "step": 3422
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.310674299505731e-06,
+      "loss": 0.6346,
+      "step": 3423
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.308332575805031e-06,
+      "loss": 0.6808,
+      "step": 3424
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.305990784212425e-06,
+      "loss": 0.6313,
+      "step": 3425
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.303648925243557e-06,
+      "loss": 0.6591,
+      "step": 3426
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.30130699941408e-06,
+      "loss": 0.5727,
+      "step": 3427
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.298965007239665e-06,
+      "loss": 0.5502,
+      "step": 3428
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2966229492359965e-06,
+      "loss": 0.6727,
+      "step": 3429
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.294280825918773e-06,
+      "loss": 0.6375,
+      "step": 3430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.291938637803706e-06,
+      "loss": 0.5582,
+      "step": 3431
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.289596385406527e-06,
+      "loss": 0.6568,
+      "step": 3432
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.287254069242974e-06,
+      "loss": 0.6456,
+      "step": 3433
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.284911689828804e-06,
+      "loss": 0.658,
+      "step": 3434
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.28256924767979e-06,
+      "loss": 0.6376,
+      "step": 3435
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.280226743311711e-06,
+      "loss": 0.5658,
+      "step": 3436
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.277884177240363e-06,
+      "loss": 0.6473,
+      "step": 3437
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2755415499815625e-06,
+      "loss": 0.5849,
+      "step": 3438
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2731988620511295e-06,
+      "loss": 0.6028,
+      "step": 3439
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.270856113964903e-06,
+      "loss": 0.5521,
+      "step": 3440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.268513306238734e-06,
+      "loss": 0.6783,
+      "step": 3441
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.266170439388487e-06,
+      "loss": 0.5676,
+      "step": 3442
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.263827513930037e-06,
+      "loss": 0.5422,
+      "step": 3443
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.261484530379276e-06,
+      "loss": 0.6908,
+      "step": 3444
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.259141489252104e-06,
+      "loss": 0.6517,
+      "step": 3445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.25679839106444e-06,
+      "loss": 0.6289,
+      "step": 3446
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.25445523633221e-06,
+      "loss": 0.6371,
+      "step": 3447
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.252112025571355e-06,
+      "loss": 0.6258,
+      "step": 3448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.249768759297828e-06,
+      "loss": 0.6475,
+      "step": 3449
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.247425438027592e-06,
+      "loss": 0.573,
+      "step": 3450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.245082062276625e-06,
+      "loss": 0.6232,
+      "step": 3451
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.242738632560919e-06,
+      "loss": 0.5661,
+      "step": 3452
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.24039514939647e-06,
+      "loss": 0.5532,
+      "step": 3453
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.238051613299293e-06,
+      "loss": 0.6298,
+      "step": 3454
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2357080247854155e-06,
+      "loss": 0.6248,
+      "step": 3455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2333643843708684e-06,
+      "loss": 0.5616,
+      "step": 3456
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.231020692571701e-06,
+      "loss": 0.5324,
+      "step": 3457
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.228676949903974e-06,
+      "loss": 0.6651,
+      "step": 3458
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2263331568837525e-06,
+      "loss": 0.6602,
+      "step": 3459
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2239893140271214e-06,
+      "loss": 0.609,
+      "step": 3460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.221645421850173e-06,
+      "loss": 0.6276,
+      "step": 3461
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.219301480869008e-06,
+      "loss": 0.6604,
+      "step": 3462
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.21695749159974e-06,
+      "loss": 0.6343,
+      "step": 3463
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2146134545584945e-06,
+      "loss": 0.5655,
+      "step": 3464
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.212269370261405e-06,
+      "loss": 0.5779,
+      "step": 3465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.209925239224616e-06,
+      "loss": 0.6475,
+      "step": 3466
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.207581061964287e-06,
+      "loss": 0.6274,
+      "step": 3467
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.205236838996578e-06,
+      "loss": 0.5721,
+      "step": 3468
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.202892570837668e-06,
+      "loss": 0.6375,
+      "step": 3469
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.200548258003741e-06,
+      "loss": 0.6149,
+      "step": 3470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.198203901010993e-06,
+      "loss": 0.6817,
+      "step": 3471
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.195859500375628e-06,
+      "loss": 0.675,
+      "step": 3472
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.193515056613865e-06,
+      "loss": 0.6809,
+      "step": 3473
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.191170570241925e-06,
+      "loss": 0.5753,
+      "step": 3474
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.18882604177604e-06,
+      "loss": 0.6209,
+      "step": 3475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.186481471732455e-06,
+      "loss": 0.6586,
+      "step": 3476
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.184136860627422e-06,
+      "loss": 0.6111,
+      "step": 3477
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.181792208977201e-06,
+      "loss": 0.6762,
+      "step": 3478
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.179447517298065e-06,
+      "loss": 0.6231,
+      "step": 3479
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1771027861062905e-06,
+      "loss": 0.5621,
+      "step": 3480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.174758015918163e-06,
+      "loss": 0.6167,
+      "step": 3481
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.172413207249983e-06,
+      "loss": 0.5585,
+      "step": 3482
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.170068360618051e-06,
+      "loss": 0.6545,
+      "step": 3483
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.167723476538683e-06,
+      "loss": 0.6671,
+      "step": 3484
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1653785555281995e-06,
+      "loss": 0.564,
+      "step": 3485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.16303359810293e-06,
+      "loss": 0.662,
+      "step": 3486
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1606886047792095e-06,
+      "loss": 0.6416,
+      "step": 3487
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.158343576073387e-06,
+      "loss": 0.6388,
+      "step": 3488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.155998512501811e-06,
+      "loss": 0.6689,
+      "step": 3489
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.153653414580846e-06,
+      "loss": 0.5456,
+      "step": 3490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.15130828282686e-06,
+      "loss": 0.6094,
+      "step": 3491
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.148963117756227e-06,
+      "loss": 0.6558,
+      "step": 3492
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.14661791988533e-06,
+      "loss": 0.6641,
+      "step": 3493
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.144272689730562e-06,
+      "loss": 0.5512,
+      "step": 3494
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.141927427808318e-06,
+      "loss": 0.6426,
+      "step": 3495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.139582134635002e-06,
+      "loss": 0.6494,
+      "step": 3496
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.137236810727025e-06,
+      "loss": 0.6584,
+      "step": 3497
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.134891456600809e-06,
+      "loss": 0.6304,
+      "step": 3498
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.132546072772774e-06,
+      "loss": 0.681,
+      "step": 3499
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1302006597593544e-06,
+      "loss": 0.6516,
+      "step": 3500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.127855218076987e-06,
+      "loss": 0.594,
+      "step": 3501
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.125509748242113e-06,
+      "loss": 0.5503,
+      "step": 3502
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.123164250771186e-06,
+      "loss": 0.6632,
+      "step": 3503
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.120818726180662e-06,
+      "loss": 0.6328,
+      "step": 3504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.118473174987002e-06,
+      "loss": 0.6433,
+      "step": 3505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.116127597706675e-06,
+      "loss": 0.6525,
+      "step": 3506
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.113781994856154e-06,
+      "loss": 0.5107,
+      "step": 3507
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.111436366951919e-06,
+      "loss": 0.5604,
+      "step": 3508
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.109090714510454e-06,
+      "loss": 0.6367,
+      "step": 3509
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.106745038048251e-06,
+      "loss": 0.6959,
+      "step": 3510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1043993380818045e-06,
+      "loss": 0.6462,
+      "step": 3511
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.102053615127617e-06,
+      "loss": 0.6167,
+      "step": 3512
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0997078697021904e-06,
+      "loss": 0.6391,
+      "step": 3513
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.097362102322041e-06,
+      "loss": 0.674,
+      "step": 3514
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.09501631350368e-06,
+      "loss": 0.6498,
+      "step": 3515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.09267050376363e-06,
+      "loss": 0.6675,
+      "step": 3516
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0903246736184165e-06,
+      "loss": 0.6372,
+      "step": 3517
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.087978823584567e-06,
+      "loss": 0.6549,
+      "step": 3518
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.085632954178616e-06,
+      "loss": 0.6339,
+      "step": 3519
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.083287065917104e-06,
+      "loss": 0.6382,
+      "step": 3520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.08094115931657e-06,
+      "loss": 0.5732,
+      "step": 3521
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0785952348935616e-06,
+      "loss": 0.6709,
+      "step": 3522
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.07624929316463e-06,
+      "loss": 0.5464,
+      "step": 3523
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.07390333464633e-06,
+      "loss": 0.6341,
+      "step": 3524
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.071557359855215e-06,
+      "loss": 0.5442,
+      "step": 3525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.069211369307851e-06,
+      "loss": 0.6536,
+      "step": 3526
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0668653635207994e-06,
+      "loss": 0.6273,
+      "step": 3527
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.064519343010631e-06,
+      "loss": 0.6469,
+      "step": 3528
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.062173308293918e-06,
+      "loss": 0.6614,
+      "step": 3529
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.059827259887232e-06,
+      "loss": 0.5219,
+      "step": 3530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.05748119830715e-06,
+      "loss": 0.6556,
+      "step": 3531
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.055135124070257e-06,
+      "loss": 0.6223,
+      "step": 3532
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.052789037693133e-06,
+      "loss": 0.5734,
+      "step": 3533
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.050442939692363e-06,
+      "loss": 0.645,
+      "step": 3534
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.048096830584538e-06,
+      "loss": 0.6362,
+      "step": 3535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.045750710886248e-06,
+      "loss": 0.6429,
+      "step": 3536
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0434045811140854e-06,
+      "loss": 0.6239,
+      "step": 3537
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0410584417846475e-06,
+      "loss": 0.6402,
+      "step": 3538
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.038712293414531e-06,
+      "loss": 0.6451,
+      "step": 3539
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.036366136520335e-06,
+      "loss": 0.5625,
+      "step": 3540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.03401997161866e-06,
+      "loss": 0.6532,
+      "step": 3541
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.031673799226112e-06,
+      "loss": 0.5496,
+      "step": 3542
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0293276198592935e-06,
+      "loss": 0.6812,
+      "step": 3543
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.026981434034812e-06,
+      "loss": 0.6163,
+      "step": 3544
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.024635242269276e-06,
+      "loss": 0.6694,
+      "step": 3545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.022289045079292e-06,
+      "loss": 0.6096,
+      "step": 3546
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0199428429814725e-06,
+      "loss": 0.6329,
+      "step": 3547
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.017596636492429e-06,
+      "loss": 0.6402,
+      "step": 3548
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.015250426128772e-06,
+      "loss": 0.6278,
+      "step": 3549
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.012904212407119e-06,
+      "loss": 0.553,
+      "step": 3550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0105579958440795e-06,
+      "loss": 0.6296,
+      "step": 3551
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.008211776956269e-06,
+      "loss": 0.5543,
+      "step": 3552
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.005865556260303e-06,
+      "loss": 0.5754,
+      "step": 3553
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0035193342727985e-06,
+      "loss": 0.6512,
+      "step": 3554
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.001173111510369e-06,
+      "loss": 0.6515,
+      "step": 3555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.998826888489631e-06,
+      "loss": 0.6427,
+      "step": 3556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.996480665727203e-06,
+      "loss": 0.6429,
+      "step": 3557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9941344437396985e-06,
+      "loss": 0.6346,
+      "step": 3558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.991788223043733e-06,
+      "loss": 0.6716,
+      "step": 3559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.989442004155923e-06,
+      "loss": 0.5478,
+      "step": 3560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.987095787592883e-06,
+      "loss": 0.6448,
+      "step": 3561
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.984749573871228e-06,
+      "loss": 0.6442,
+      "step": 3562
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.982403363507572e-06,
+      "loss": 0.5449,
+      "step": 3563
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.980057157018529e-06,
+      "loss": 0.5527,
+      "step": 3564
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.977710954920709e-06,
+      "loss": 0.6513,
+      "step": 3565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.975364757730727e-06,
+      "loss": 0.63,
+      "step": 3566
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.973018565965191e-06,
+      "loss": 0.6397,
+      "step": 3567
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.970672380140707e-06,
+      "loss": 0.6413,
+      "step": 3568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.96832620077389e-06,
+      "loss": 0.672,
+      "step": 3569
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.965980028381341e-06,
+      "loss": 0.6426,
+      "step": 3570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.963633863479667e-06,
+      "loss": 0.5846,
+      "step": 3571
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.961287706585471e-06,
+      "loss": 0.5397,
+      "step": 3572
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.958941558215354e-06,
+      "loss": 0.6444,
+      "step": 3573
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9565954188859145e-06,
+      "loss": 0.6418,
+      "step": 3574
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.954249289113753e-06,
+      "loss": 0.68,
+      "step": 3575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.951903169415464e-06,
+      "loss": 0.649,
+      "step": 3576
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.949557060307639e-06,
+      "loss": 0.6665,
+      "step": 3577
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9472109623068694e-06,
+      "loss": 0.639,
+      "step": 3578
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.944864875929745e-06,
+      "loss": 0.6199,
+      "step": 3579
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.94251880169285e-06,
+      "loss": 0.5416,
+      "step": 3580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.94017274011277e-06,
+      "loss": 0.6572,
+      "step": 3581
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.937826691706084e-06,
+      "loss": 0.6213,
+      "step": 3582
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.93548065698937e-06,
+      "loss": 0.6491,
+      "step": 3583
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.933134636479201e-06,
+      "loss": 0.5727,
+      "step": 3584
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.93078863069215e-06,
+      "loss": 0.6299,
+      "step": 3585
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928442640144786e-06,
+      "loss": 0.5633,
+      "step": 3586
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.926096665353672e-06,
+      "loss": 0.612,
+      "step": 3587
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.923750706835371e-06,
+      "loss": 0.6108,
+      "step": 3588
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.92140476510644e-06,
+      "loss": 0.6368,
+      "step": 3589
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.919058840683432e-06,
+      "loss": 0.6574,
+      "step": 3590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.916712934082897e-06,
+      "loss": 0.629,
+      "step": 3591
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.914367045821384e-06,
+      "loss": 0.6236,
+      "step": 3592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.912021176415434e-06,
+      "loss": 0.6158,
+      "step": 3593
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.909675326381585e-06,
+      "loss": 0.5431,
+      "step": 3594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.907329496236371e-06,
+      "loss": 0.6344,
+      "step": 3595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9049836864963214e-06,
+      "loss": 0.6879,
+      "step": 3596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.90263789767796e-06,
+      "loss": 0.5335,
+      "step": 3597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.900292130297809e-06,
+      "loss": 0.5452,
+      "step": 3598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8979463848723845e-06,
+      "loss": 0.6394,
+      "step": 3599
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.895600661918196e-06,
+      "loss": 0.629,
+      "step": 3600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8932549619517514e-06,
+      "loss": 0.5756,
+      "step": 3601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8909092854895475e-06,
+      "loss": 0.6458,
+      "step": 3602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8885636330480824e-06,
+      "loss": 0.6012,
+      "step": 3603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.886218005143846e-06,
+      "loss": 0.6211,
+      "step": 3604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.883872402293326e-06,
+      "loss": 0.5607,
+      "step": 3605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.881526825012999e-06,
+      "loss": 0.637,
+      "step": 3606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.87918127381934e-06,
+      "loss": 0.6598,
+      "step": 3607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8768357492288145e-06,
+      "loss": 0.5651,
+      "step": 3608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.874490251757888e-06,
+      "loss": 0.6553,
+      "step": 3609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.872144781923015e-06,
+      "loss": 0.6385,
+      "step": 3610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.869799340240647e-06,
+      "loss": 0.6441,
+      "step": 3611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8674539272272274e-06,
+      "loss": 0.5518,
+      "step": 3612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8651085433991935e-06,
+      "loss": 0.6205,
+      "step": 3613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.862763189272976e-06,
+      "loss": 0.5519,
+      "step": 3614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.860417865365e-06,
+      "loss": 0.635,
+      "step": 3615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.858072572191684e-06,
+      "loss": 0.6715,
+      "step": 3616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.85572731026944e-06,
+      "loss": 0.6536,
+      "step": 3617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8533820801146715e-06,
+      "loss": 0.6605,
+      "step": 3618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.851036882243776e-06,
+      "loss": 0.5391,
+      "step": 3619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.848691717173143e-06,
+      "loss": 0.6205,
+      "step": 3620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.846346585419155e-06,
+      "loss": 0.6477,
+      "step": 3621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.84400148749819e-06,
+      "loss": 0.538,
+      "step": 3622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.841656423926615e-06,
+      "loss": 0.683,
+      "step": 3623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.839311395220791e-06,
+      "loss": 0.6628,
+      "step": 3624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.836966401897072e-06,
+      "loss": 0.6575,
+      "step": 3625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.834621444471803e-06,
+      "loss": 0.6385,
+      "step": 3626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.832276523461317e-06,
+      "loss": 0.6378,
+      "step": 3627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.829931639381949e-06,
+      "loss": 0.6385,
+      "step": 3628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.827586792750018e-06,
+      "loss": 0.6462,
+      "step": 3629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8252419840818384e-06,
+      "loss": 0.6572,
+      "step": 3630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.822897213893712e-06,
+      "loss": 0.5383,
+      "step": 3631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.820552482701938e-06,
+      "loss": 0.6307,
+      "step": 3632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.818207791022799e-06,
+      "loss": 0.6782,
+      "step": 3633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8158631393725794e-06,
+      "loss": 0.6618,
+      "step": 3634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.813518528267546e-06,
+      "loss": 0.6738,
+      "step": 3635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.811173958223962e-06,
+      "loss": 0.6295,
+      "step": 3636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.808829429758079e-06,
+      "loss": 0.6278,
+      "step": 3637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.806484943386136e-06,
+      "loss": 0.5438,
+      "step": 3638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.804140499624372e-06,
+      "loss": 0.6368,
+      "step": 3639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.801796098989009e-06,
+      "loss": 0.6773,
+      "step": 3640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.79945174199626e-06,
+      "loss": 0.6375,
+      "step": 3641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.797107429162335e-06,
+      "loss": 0.5518,
+      "step": 3642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7947631610034245e-06,
+      "loss": 0.6648,
+      "step": 3643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.792418938035714e-06,
+      "loss": 0.6461,
+      "step": 3644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.790074760775384e-06,
+      "loss": 0.6425,
+      "step": 3645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.787730629738597e-06,
+      "loss": 0.6354,
+      "step": 3646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.785386545441506e-06,
+      "loss": 0.5387,
+      "step": 3647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.783042508400261e-06,
+      "loss": 0.6548,
+      "step": 3648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.780698519130994e-06,
+      "loss": 0.6498,
+      "step": 3649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7783545781498275e-06,
+      "loss": 0.5542,
+      "step": 3650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7760106859728785e-06,
+      "loss": 0.6227,
+      "step": 3651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.773666843116248e-06,
+      "loss": 0.6434,
+      "step": 3652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.771323050096028e-06,
+      "loss": 0.662,
+      "step": 3653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7689793074283005e-06,
+      "loss": 0.6548,
+      "step": 3654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.766635615629134e-06,
+      "loss": 0.6227,
+      "step": 3655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.764291975214585e-06,
+      "loss": 0.6546,
+      "step": 3656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.761948386700707e-06,
+      "loss": 0.5493,
+      "step": 3657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.759604850603531e-06,
+      "loss": 0.6353,
+      "step": 3658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.757261367439083e-06,
+      "loss": 0.6288,
+      "step": 3659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7549179377233765e-06,
+      "loss": 0.5734,
+      "step": 3660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.75257456197241e-06,
+      "loss": 0.6496,
+      "step": 3661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.750231240702173e-06,
+      "loss": 0.6653,
+      "step": 3662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.747887974428646e-06,
+      "loss": 0.5678,
+      "step": 3663
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.745544763667791e-06,
+      "loss": 0.6386,
+      "step": 3664
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.743201608935561e-06,
+      "loss": 0.6384,
+      "step": 3665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.740858510747897e-06,
+      "loss": 0.6279,
+      "step": 3666
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.738515469620727e-06,
+      "loss": 0.6581,
+      "step": 3667
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.736172486069964e-06,
+      "loss": 0.6443,
+      "step": 3668
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7338295606115146e-06,
+      "loss": 0.5606,
+      "step": 3669
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.731486693761268e-06,
+      "loss": 0.6646,
+      "step": 3670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.729143886035099e-06,
+      "loss": 0.6287,
+      "step": 3671
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.726801137948872e-06,
+      "loss": 0.6541,
+      "step": 3672
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.72445845001844e-06,
+      "loss": 0.6166,
+      "step": 3673
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.722115822759637e-06,
+      "loss": 0.6654,
+      "step": 3674
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.719773256688292e-06,
+      "loss": 0.6368,
+      "step": 3675
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.717430752320212e-06,
+      "loss": 0.5461,
+      "step": 3676
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.715088310171197e-06,
+      "loss": 0.6384,
+      "step": 3677
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.712745930757027e-06,
+      "loss": 0.5558,
+      "step": 3678
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.710403614593475e-06,
+      "loss": 0.6223,
+      "step": 3679
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.708061362196293e-06,
+      "loss": 0.6206,
+      "step": 3680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.705719174081229e-06,
+      "loss": 0.6206,
+      "step": 3681
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.703377050764005e-06,
+      "loss": 0.6532,
+      "step": 3682
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.701034992760337e-06,
+      "loss": 0.6282,
+      "step": 3683
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.698693000585921e-06,
+      "loss": 0.6408,
+      "step": 3684
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.696351074756446e-06,
+      "loss": 0.624,
+      "step": 3685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.694009215787576e-06,
+      "loss": 0.6599,
+      "step": 3686
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.691667424194971e-06,
+      "loss": 0.649,
+      "step": 3687
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.68932570049427e-06,
+      "loss": 0.6248,
+      "step": 3688
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.686984045201098e-06,
+      "loss": 0.5514,
+      "step": 3689
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.684642458831064e-06,
+      "loss": 0.5596,
+      "step": 3690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.682300941899764e-06,
+      "loss": 0.5401,
+      "step": 3691
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.679959494922779e-06,
+      "loss": 0.5455,
+      "step": 3692
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.677618118415673e-06,
+      "loss": 0.6488,
+      "step": 3693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.675276812893993e-06,
+      "loss": 0.6574,
+      "step": 3694
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.672935578873275e-06,
+      "loss": 0.6429,
+      "step": 3695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.670594416869033e-06,
+      "loss": 0.5245,
+      "step": 3696
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.668253327396771e-06,
+      "loss": 0.6334,
+      "step": 3697
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.665912310971974e-06,
+      "loss": 0.6273,
+      "step": 3698
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.663571368110113e-06,
+      "loss": 0.6467,
+      "step": 3699
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.66123049932664e-06,
+      "loss": 0.6195,
+      "step": 3700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.658889705136992e-06,
+      "loss": 0.5589,
+      "step": 3701
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.65654898605659e-06,
+      "loss": 0.5677,
+      "step": 3702
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.654208342600836e-06,
+      "loss": 0.5526,
+      "step": 3703
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6518677752851214e-06,
+      "loss": 0.6329,
+      "step": 3704
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.649527284624814e-06,
+      "loss": 0.6563,
+      "step": 3705
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.64718687113527e-06,
+      "loss": 0.6411,
+      "step": 3706
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.644846535331825e-06,
+      "loss": 0.6483,
+      "step": 3707
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6425062777297975e-06,
+      "loss": 0.6302,
+      "step": 3708
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.640166098844489e-06,
+      "loss": 0.5525,
+      "step": 3709
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.637825999191189e-06,
+      "loss": 0.6272,
+      "step": 3710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.635485979285161e-06,
+      "loss": 0.6379,
+      "step": 3711
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.633146039641659e-06,
+      "loss": 0.5423,
+      "step": 3712
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.630806180775913e-06,
+      "loss": 0.6616,
+      "step": 3713
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.628466403203138e-06,
+      "loss": 0.6762,
+      "step": 3714
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6261267074385285e-06,
+      "loss": 0.6412,
+      "step": 3715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.623787093997269e-06,
+      "loss": 0.5394,
+      "step": 3716
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.621447563394514e-06,
+      "loss": 0.6496,
+      "step": 3717
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.619108116145411e-06,
+      "loss": 0.5482,
+      "step": 3718
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.616768752765081e-06,
+      "loss": 0.632,
+      "step": 3719
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6144294737686315e-06,
+      "loss": 0.6635,
+      "step": 3720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.612090279671145e-06,
+      "loss": 0.6428,
+      "step": 3721
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.609751170987696e-06,
+      "loss": 0.6453,
+      "step": 3722
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.60741214823333e-06,
+      "loss": 0.6602,
+      "step": 3723
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.60507321192308e-06,
+      "loss": 0.5574,
+      "step": 3724
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.602734362571956e-06,
+      "loss": 0.632,
+      "step": 3725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.600395600694952e-06,
+      "loss": 0.6186,
+      "step": 3726
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.598056926807036e-06,
+      "loss": 0.54,
+      "step": 3727
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.595718341423169e-06,
+      "loss": 0.554,
+      "step": 3728
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5933798450582805e-06,
+      "loss": 0.6522,
+      "step": 3729
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.591041438227288e-06,
+      "loss": 0.5446,
+      "step": 3730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.588703121445084e-06,
+      "loss": 0.6366,
+      "step": 3731
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.586364895226546e-06,
+      "loss": 0.6765,
+      "step": 3732
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.584026760086525e-06,
+      "loss": 0.5298,
+      "step": 3733
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.581688716539861e-06,
+      "loss": 0.5708,
+      "step": 3734
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.579350765101367e-06,
+      "loss": 0.6491,
+      "step": 3735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5770129062858385e-06,
+      "loss": 0.6237,
+      "step": 3736
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.574675140608049e-06,
+      "loss": 0.5485,
+      "step": 3737
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.572337468582751e-06,
+      "loss": 0.5658,
+      "step": 3738
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.569999890724679e-06,
+      "loss": 0.5435,
+      "step": 3739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.567662407548546e-06,
+      "loss": 0.6535,
+      "step": 3740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.565325019569042e-06,
+      "loss": 0.6136,
+      "step": 3741
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.56298772730084e-06,
+      "loss": 0.6344,
+      "step": 3742
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5606505312585895e-06,
+      "loss": 0.638,
+      "step": 3743
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.558313431956914e-06,
+      "loss": 0.6281,
+      "step": 3744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.555976429910427e-06,
+      "loss": 0.6529,
+      "step": 3745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.55363952563371e-06,
+      "loss": 0.6102,
+      "step": 3746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.551302719641329e-06,
+      "loss": 0.5251,
+      "step": 3747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.548966012447825e-06,
+      "loss": 0.6496,
+      "step": 3748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.546629404567721e-06,
+      "loss": 0.633,
+      "step": 3749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.544292896515511e-06,
+      "loss": 0.565,
+      "step": 3750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.541956488805679e-06,
+      "loss": 0.6346,
+      "step": 3751
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.539620181952676e-06,
+      "loss": 0.5516,
+      "step": 3752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5372839764709334e-06,
+      "loss": 0.6487,
+      "step": 3753
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.534947872874862e-06,
+      "loss": 0.6534,
+      "step": 3754
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.532611871678851e-06,
+      "loss": 0.6309,
+      "step": 3755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.530275973397264e-06,
+      "loss": 0.6412,
+      "step": 3756
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.527940178544446e-06,
+      "loss": 0.6189,
+      "step": 3757
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.525604487634716e-06,
+      "loss": 0.6319,
+      "step": 3758
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.52326890118237e-06,
+      "loss": 0.6387,
+      "step": 3759
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.520933419701681e-06,
+      "loss": 0.6303,
+      "step": 3760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.518598043706902e-06,
+      "loss": 0.6496,
+      "step": 3761
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.516262773712258e-06,
+      "loss": 0.6111,
+      "step": 3762
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5139276102319575e-06,
+      "loss": 0.6538,
+      "step": 3763
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.511592553780179e-06,
+      "loss": 0.6344,
+      "step": 3764
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.509257604871079e-06,
+      "loss": 0.6179,
+      "step": 3765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.50692276401879e-06,
+      "loss": 0.6513,
+      "step": 3766
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5045880317374255e-06,
+      "loss": 0.52,
+      "step": 3767
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5022534085410665e-06,
+      "loss": 0.6463,
+      "step": 3768
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.499918894943779e-06,
+      "loss": 0.5535,
+      "step": 3769
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4975844914596015e-06,
+      "loss": 0.6372,
+      "step": 3770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.495250198602544e-06,
+      "loss": 0.6338,
+      "step": 3771
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.492916016886596e-06,
+      "loss": 0.5657,
+      "step": 3772
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4905819468257235e-06,
+      "loss": 0.6299,
+      "step": 3773
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4882479889338645e-06,
+      "loss": 0.5615,
+      "step": 3774
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.485914143724937e-06,
+      "loss": 0.6204,
+      "step": 3775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.483580411712832e-06,
+      "loss": 0.549,
+      "step": 3776
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.481246793411413e-06,
+      "loss": 0.6538,
+      "step": 3777
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.47891328933452e-06,
+      "loss": 0.6332,
+      "step": 3778
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4765798999959715e-06,
+      "loss": 0.6184,
+      "step": 3779
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.474246625909553e-06,
+      "loss": 0.6557,
+      "step": 3780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4719134675890345e-06,
+      "loss": 0.5523,
+      "step": 3781
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4695804255481514e-06,
+      "loss": 0.5323,
+      "step": 3782
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.467247500300621e-06,
+      "loss": 0.5912,
+      "step": 3783
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.464914692360126e-06,
+      "loss": 0.676,
+      "step": 3784
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.462582002240334e-06,
+      "loss": 0.6429,
+      "step": 3785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.460249430454876e-06,
+      "loss": 0.6634,
+      "step": 3786
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.457916977517368e-06,
+      "loss": 0.6308,
+      "step": 3787
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.45558464394139e-06,
+      "loss": 0.5445,
+      "step": 3788
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.453252430240501e-06,
+      "loss": 0.6187,
+      "step": 3789
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.450920336928232e-06,
+      "loss": 0.6378,
+      "step": 3790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4485883645180875e-06,
+      "loss": 0.646,
+      "step": 3791
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.446256513523544e-06,
+      "loss": 0.6552,
+      "step": 3792
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.443924784458056e-06,
+      "loss": 0.6206,
+      "step": 3793
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.441593177835048e-06,
+      "loss": 0.6536,
+      "step": 3794
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.439261694167917e-06,
+      "loss": 0.6487,
+      "step": 3795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.436930333970033e-06,
+      "loss": 0.6625,
+      "step": 3796
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.434599097754736e-06,
+      "loss": 0.6441,
+      "step": 3797
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.432267986035349e-06,
+      "loss": 0.6373,
+      "step": 3798
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.429936999325155e-06,
+      "loss": 0.6442,
+      "step": 3799
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.42760613813742e-06,
+      "loss": 0.5491,
+      "step": 3800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.425275402985373e-06,
+      "loss": 0.654,
+      "step": 3801
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.422944794382223e-06,
+      "loss": 0.6412,
+      "step": 3802
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4206143128411434e-06,
+      "loss": 0.6445,
+      "step": 3803
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.418283958875289e-06,
+      "loss": 0.6416,
+      "step": 3804
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4159537329977794e-06,
+      "loss": 0.6644,
+      "step": 3805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.413623635721709e-06,
+      "loss": 0.63,
+      "step": 3806
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.411293667560143e-06,
+      "loss": 0.61,
+      "step": 3807
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.408963829026119e-06,
+      "loss": 0.6513,
+      "step": 3808
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.40663412063264e-06,
+      "loss": 0.6369,
+      "step": 3809
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.404304542892692e-06,
+      "loss": 0.6576,
+      "step": 3810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.401975096319223e-06,
+      "loss": 0.5672,
+      "step": 3811
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.399645781425157e-06,
+      "loss": 0.5293,
+      "step": 3812
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.397316598723385e-06,
+      "loss": 0.6242,
+      "step": 3813
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.394987548726773e-06,
+      "loss": 0.6227,
+      "step": 3814
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.392658631948152e-06,
+      "loss": 0.6559,
+      "step": 3815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.390329848900331e-06,
+      "loss": 0.6283,
+      "step": 3816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.388001200096083e-06,
+      "loss": 0.5533,
+      "step": 3817
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.385672686048158e-06,
+      "loss": 0.6319,
+      "step": 3818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.383344307269269e-06,
+      "loss": 0.6655,
+      "step": 3819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.381016064272106e-06,
+      "loss": 0.6241,
+      "step": 3820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.378687957569321e-06,
+      "loss": 0.6501,
+      "step": 3821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.376359987673547e-06,
+      "loss": 0.6023,
+      "step": 3822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.374032155097377e-06,
+      "loss": 0.6448,
+      "step": 3823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.37170446035338e-06,
+      "loss": 0.5668,
+      "step": 3824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.369376903954091e-06,
+      "loss": 0.6356,
+      "step": 3825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.367049486412016e-06,
+      "loss": 0.6166,
+      "step": 3826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.364722208239628e-06,
+      "loss": 0.555,
+      "step": 3827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.362395069949376e-06,
+      "loss": 0.5621,
+      "step": 3828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.360068072053671e-06,
+      "loss": 0.531,
+      "step": 3829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.357741215064896e-06,
+      "loss": 0.633,
+      "step": 3830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.355414499495403e-06,
+      "loss": 0.5726,
+      "step": 3831
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.353087925857513e-06,
+      "loss": 0.5568,
+      "step": 3832
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.350761494663514e-06,
+      "loss": 0.5622,
+      "step": 3833
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.348435206425666e-06,
+      "loss": 0.6176,
+      "step": 3834
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.346109061656196e-06,
+      "loss": 0.6367,
+      "step": 3835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3437830608672954e-06,
+      "loss": 0.5687,
+      "step": 3836
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3414572045711315e-06,
+      "loss": 0.6783,
+      "step": 3837
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.339131493279834e-06,
+      "loss": 0.6578,
+      "step": 3838
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3368059275055e-06,
+      "loss": 0.5516,
+      "step": 3839
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.334480507760201e-06,
+      "loss": 0.5658,
+      "step": 3840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.332155234555972e-06,
+      "loss": 0.5641,
+      "step": 3841
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.329830108404813e-06,
+      "loss": 0.581,
+      "step": 3842
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.327505129818698e-06,
+      "loss": 0.6283,
+      "step": 3843
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.325180299309564e-06,
+      "loss": 0.621,
+      "step": 3844
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.322855617389313e-06,
+      "loss": 0.6349,
+      "step": 3845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.320531084569824e-06,
+      "loss": 0.5482,
+      "step": 3846
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.318206701362934e-06,
+      "loss": 0.6502,
+      "step": 3847
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.31588246828045e-06,
+      "loss": 0.5569,
+      "step": 3848
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.313558385834147e-06,
+      "loss": 0.6635,
+      "step": 3849
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.311234454535763e-06,
+      "loss": 0.6401,
+      "step": 3850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.30891067489701e-06,
+      "loss": 0.6698,
+      "step": 3851
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.30658704742956e-06,
+      "loss": 0.5475,
+      "step": 3852
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.304263572645053e-06,
+      "loss": 0.6379,
+      "step": 3853
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.301940251055097e-06,
+      "loss": 0.6518,
+      "step": 3854
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2996170831712655e-06,
+      "loss": 0.6195,
+      "step": 3855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.297294069505095e-06,
+      "loss": 0.617,
+      "step": 3856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2949712105680964e-06,
+      "loss": 0.6719,
+      "step": 3857
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.292648506871738e-06,
+      "loss": 0.6318,
+      "step": 3858
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2903259589274585e-06,
+      "loss": 0.6635,
+      "step": 3859
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.288003567246658e-06,
+      "loss": 0.5726,
+      "step": 3860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.285681332340708e-06,
+      "loss": 0.653,
+      "step": 3861
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.28335925472094e-06,
+      "loss": 0.6599,
+      "step": 3862
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2810373348986565e-06,
+      "loss": 0.6682,
+      "step": 3863
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.278715573385121e-06,
+      "loss": 0.6273,
+      "step": 3864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.276393970691563e-06,
+      "loss": 0.6254,
+      "step": 3865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.274072527329176e-06,
+      "loss": 0.5695,
+      "step": 3866
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.271751243809122e-06,
+      "loss": 0.6492,
+      "step": 3867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.269430120642523e-06,
+      "loss": 0.6435,
+      "step": 3868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.267109158340471e-06,
+      "loss": 0.6413,
+      "step": 3869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.264788357414021e-06,
+      "loss": 0.6464,
+      "step": 3870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.262467718374187e-06,
+      "loss": 0.6663,
+      "step": 3871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.260147241731954e-06,
+      "loss": 0.6896,
+      "step": 3872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.257826927998269e-06,
+      "loss": 0.5211,
+      "step": 3873
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.255506777684041e-06,
+      "loss": 0.6144,
+      "step": 3874
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.253186791300148e-06,
+      "loss": 0.6618,
+      "step": 3875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.250866969357429e-06,
+      "loss": 0.6224,
+      "step": 3876
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2485473123666845e-06,
+      "loss": 0.6366,
+      "step": 3877
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.246227820838683e-06,
+      "loss": 0.6256,
+      "step": 3878
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.243908495284152e-06,
+      "loss": 0.6399,
+      "step": 3879
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.241589336213784e-06,
+      "loss": 0.6082,
+      "step": 3880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.239270344138241e-06,
+      "loss": 0.5474,
+      "step": 3881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.236951519568141e-06,
+      "loss": 0.6132,
+      "step": 3882
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.234632863014065e-06,
+      "loss": 0.5467,
+      "step": 3883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.232314374986561e-06,
+      "loss": 0.6669,
+      "step": 3884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.229996055996137e-06,
+      "loss": 0.6556,
+      "step": 3885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2276779065532644e-06,
+      "loss": 0.6083,
+      "step": 3886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.225359927168379e-06,
+      "loss": 0.6464,
+      "step": 3887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.223042118351879e-06,
+      "loss": 0.6399,
+      "step": 3888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.220724480614121e-06,
+      "loss": 0.6215,
+      "step": 3889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.21840701446543e-06,
+      "loss": 0.5885,
+      "step": 3890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.216089720416085e-06,
+      "loss": 0.6712,
+      "step": 3891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.213772598976336e-06,
+      "loss": 0.6661,
+      "step": 3892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2114556506563905e-06,
+      "loss": 0.6309,
+      "step": 3893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2091388759664194e-06,
+      "loss": 0.6554,
+      "step": 3894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2068222754165536e-06,
+      "loss": 0.6295,
+      "step": 3895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.204505849516886e-06,
+      "loss": 0.6334,
+      "step": 3896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.202189598777472e-06,
+      "loss": 0.6143,
+      "step": 3897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.199873523708327e-06,
+      "loss": 0.6269,
+      "step": 3898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1975576248194305e-06,
+      "loss": 0.5728,
+      "step": 3899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.195241902620721e-06,
+      "loss": 0.5655,
+      "step": 3900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.192926357622098e-06,
+      "loss": 0.6389,
+      "step": 3901
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.190610990333423e-06,
+      "loss": 0.6179,
+      "step": 3902
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.188295801264515e-06,
+      "loss": 0.665,
+      "step": 3903
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.185980790925162e-06,
+      "loss": 0.6575,
+      "step": 3904
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1836659598251026e-06,
+      "loss": 0.6353,
+      "step": 3905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.181351308474044e-06,
+      "loss": 0.5678,
+      "step": 3906
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.179036837381649e-06,
+      "loss": 0.6397,
+      "step": 3907
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.176722547057542e-06,
+      "loss": 0.6734,
+      "step": 3908
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.174408438011306e-06,
+      "loss": 0.6275,
+      "step": 3909
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.172094510752489e-06,
+      "loss": 0.6553,
+      "step": 3910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.169780765790593e-06,
+      "loss": 0.5554,
+      "step": 3911
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.167467203635086e-06,
+      "loss": 0.6642,
+      "step": 3912
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.165153824795391e-06,
+      "loss": 0.5541,
+      "step": 3913
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.162840629780891e-06,
+      "loss": 0.6138,
+      "step": 3914
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.160527619100927e-06,
+      "loss": 0.6351,
+      "step": 3915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.158214793264808e-06,
+      "loss": 0.6169,
+      "step": 3916
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1559021527817916e-06,
+      "loss": 0.6045,
+      "step": 3917
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.153589698161103e-06,
+      "loss": 0.6265,
+      "step": 3918
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1512774299119205e-06,
+      "loss": 0.5523,
+      "step": 3919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.148965348543383e-06,
+      "loss": 0.5904,
+      "step": 3920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1466534545645885e-06,
+      "loss": 0.5562,
+      "step": 3921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1443417484845965e-06,
+      "loss": 0.6398,
+      "step": 3922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.142030230812422e-06,
+      "loss": 0.5887,
+      "step": 3923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.139718902057038e-06,
+      "loss": 0.5663,
+      "step": 3924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.137407762727378e-06,
+      "loss": 0.6196,
+      "step": 3925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.135096813332333e-06,
+      "loss": 0.5649,
+      "step": 3926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.13278605438075e-06,
+      "loss": 0.563,
+      "step": 3927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.130475486381441e-06,
+      "loss": 0.627,
+      "step": 3928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.128165109843167e-06,
+      "loss": 0.6292,
+      "step": 3929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.125854925274651e-06,
+      "loss": 0.6659,
+      "step": 3930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.123544933184576e-06,
+      "loss": 0.617,
+      "step": 3931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.12123513408158e-06,
+      "loss": 0.6079,
+      "step": 3932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.118925528474254e-06,
+      "loss": 0.6679,
+      "step": 3933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.116616116871158e-06,
+      "loss": 0.6582,
+      "step": 3934
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1143068997808e-06,
+      "loss": 0.6026,
+      "step": 3935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.111997877711646e-06,
+      "loss": 0.6454,
+      "step": 3936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.109689051172123e-06,
+      "loss": 0.6237,
+      "step": 3937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.107380420670612e-06,
+      "loss": 0.6438,
+      "step": 3938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.105071986715449e-06,
+      "loss": 0.6218,
+      "step": 3939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.102763749814933e-06,
+      "loss": 0.5882,
+      "step": 3940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1004557104773145e-06,
+      "loss": 0.6207,
+      "step": 3941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.098147869210801e-06,
+      "loss": 0.6301,
+      "step": 3942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.095840226523558e-06,
+      "loss": 0.554,
+      "step": 3943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.093532782923707e-06,
+      "loss": 0.6107,
+      "step": 3944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.091225538919321e-06,
+      "loss": 0.6385,
+      "step": 3945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0889184950184405e-06,
+      "loss": 0.5436,
+      "step": 3946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.08661165172905e-06,
+      "loss": 0.5438,
+      "step": 3947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.084305009559093e-06,
+      "loss": 0.6258,
+      "step": 3948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.081998569016475e-06,
+      "loss": 0.5267,
+      "step": 3949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.07969233060905e-06,
+      "loss": 0.5443,
+      "step": 3950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.077386294844626e-06,
+      "loss": 0.6428,
+      "step": 3951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.075080462230977e-06,
+      "loss": 0.5407,
+      "step": 3952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.072774833275822e-06,
+      "loss": 0.552,
+      "step": 3953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.070469408486837e-06,
+      "loss": 0.6264,
+      "step": 3954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.068164188371657e-06,
+      "loss": 0.625,
+      "step": 3955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.065859173437868e-06,
+      "loss": 0.641,
+      "step": 3956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.063554364193015e-06,
+      "loss": 0.5586,
+      "step": 3957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.061249761144593e-06,
+      "loss": 0.646,
+      "step": 3958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058945364800054e-06,
+      "loss": 0.5455,
+      "step": 3959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056641175666805e-06,
+      "loss": 0.5512,
+      "step": 3960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.054337194252204e-06,
+      "loss": 0.6224,
+      "step": 3961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052033421063568e-06,
+      "loss": 0.6163,
+      "step": 3962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049729856608166e-06,
+      "loss": 0.6278,
+      "step": 3963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047426501393221e-06,
+      "loss": 0.6583,
+      "step": 3964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.04512335592591e-06,
+      "loss": 0.5536,
+      "step": 3965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.042820420713363e-06,
+      "loss": 0.5525,
+      "step": 3966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.040517696262664e-06,
+      "loss": 0.6651,
+      "step": 3967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.038215183080851e-06,
+      "loss": 0.6494,
+      "step": 3968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.035912881674917e-06,
+      "loss": 0.6379,
+      "step": 3969
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.033610792551807e-06,
+      "loss": 0.6551,
+      "step": 3970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.031308916218419e-06,
+      "loss": 0.6591,
+      "step": 3971
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.029007253181603e-06,
+      "loss": 0.6107,
+      "step": 3972
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0267058039481635e-06,
+      "loss": 0.6276,
+      "step": 3973
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.024404569024857e-06,
+      "loss": 0.6704,
+      "step": 3974
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.022103548918396e-06,
+      "loss": 0.5558,
+      "step": 3975
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.019802744135443e-06,
+      "loss": 0.654,
+      "step": 3976
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.017502155182612e-06,
+      "loss": 0.6198,
+      "step": 3977
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.015201782566471e-06,
+      "loss": 0.6305,
+      "step": 3978
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.01290162679354e-06,
+      "loss": 0.6232,
+      "step": 3979
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.010601688370291e-06,
+      "loss": 0.6537,
+      "step": 3980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.008301967803151e-06,
+      "loss": 0.5598,
+      "step": 3981
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.006002465598495e-06,
+      "loss": 0.6618,
+      "step": 3982
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0037031822626516e-06,
+      "loss": 0.555,
+      "step": 3983
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0014041183019014e-06,
+      "loss": 0.6491,
+      "step": 3984
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9991052742224756e-06,
+      "loss": 0.6272,
+      "step": 3985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.996806650530558e-06,
+      "loss": 0.5523,
+      "step": 3986
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.994508247732284e-06,
+      "loss": 0.6134,
+      "step": 3987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9922100663337415e-06,
+      "loss": 0.55,
+      "step": 3988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.989912106840967e-06,
+      "loss": 0.6124,
+      "step": 3989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.987614369759949e-06,
+      "loss": 0.6497,
+      "step": 3990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.985316855596627e-06,
+      "loss": 0.6595,
+      "step": 3991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.983019564856892e-06,
+      "loss": 0.5696,
+      "step": 3992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.980722498046586e-06,
+      "loss": 0.6383,
+      "step": 3993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.978425655671503e-06,
+      "loss": 0.6576,
+      "step": 3994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.976129038237383e-06,
+      "loss": 0.6447,
+      "step": 3995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.973832646249921e-06,
+      "loss": 0.6234,
+      "step": 3996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.971536480214759e-06,
+      "loss": 0.6001,
+      "step": 3997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.969240540637492e-06,
+      "loss": 0.5653,
+      "step": 3998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.966944828023664e-06,
+      "loss": 0.5689,
+      "step": 3999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.964649342878771e-06,
+      "loss": 0.5986,
+      "step": 4000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.962354085708256e-06,
+      "loss": 0.6476,
+      "step": 4001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.960059057017511e-06,
+      "loss": 0.6193,
+      "step": 4002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.957764257311882e-06,
+      "loss": 0.6305,
+      "step": 4003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.955469687096657e-06,
+      "loss": 0.65,
+      "step": 4004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.953175346877084e-06,
+      "loss": 0.5226,
+      "step": 4005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.950881237158355e-06,
+      "loss": 0.5355,
+      "step": 4006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.948587358445608e-06,
+      "loss": 0.6416,
+      "step": 4007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.946293711243936e-06,
+      "loss": 0.6234,
+      "step": 4008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.944000296058374e-06,
+      "loss": 0.6357,
+      "step": 4009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.941707113393916e-06,
+      "loss": 0.6201,
+      "step": 4010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9394141637554955e-06,
+      "loss": 0.6595,
+      "step": 4011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.937121447647998e-06,
+      "loss": 0.6471,
+      "step": 4012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.934828965576261e-06,
+      "loss": 0.6101,
+      "step": 4013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.932536718045064e-06,
+      "loss": 0.6601,
+      "step": 4014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.930244705559138e-06,
+      "loss": 0.6238,
+      "step": 4015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.927952928623165e-06,
+      "loss": 0.6356,
+      "step": 4016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9256613877417715e-06,
+      "loss": 0.6271,
+      "step": 4017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.923370083419531e-06,
+      "loss": 0.5681,
+      "step": 4018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.92107901616097e-06,
+      "loss": 0.5264,
+      "step": 4019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.918788186470559e-06,
+      "loss": 0.5486,
+      "step": 4020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.916497594852713e-06,
+      "loss": 0.6372,
+      "step": 4021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.914207241811804e-06,
+      "loss": 0.5298,
+      "step": 4022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.911917127852144e-06,
+      "loss": 0.6136,
+      "step": 4023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.909627253477993e-06,
+      "loss": 0.54,
+      "step": 4024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9073376191935615e-06,
+      "loss": 0.5132,
+      "step": 4025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.905048225503004e-06,
+      "loss": 0.655,
+      "step": 4026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.902759072910422e-06,
+      "loss": 0.6426,
+      "step": 4027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.900470161919869e-06,
+      "loss": 0.5274,
+      "step": 4028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.898181493035339e-06,
+      "loss": 0.6497,
+      "step": 4029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.895893066760774e-06,
+      "loss": 0.5431,
+      "step": 4030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.893604883600066e-06,
+      "loss": 0.6315,
+      "step": 4031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.891316944057051e-06,
+      "loss": 0.621,
+      "step": 4032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.889029248635508e-06,
+      "loss": 0.5567,
+      "step": 4033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.886741797839171e-06,
+      "loss": 0.6752,
+      "step": 4034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.884454592171713e-06,
+      "loss": 0.6222,
+      "step": 4035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.882167632136753e-06,
+      "loss": 0.6195,
+      "step": 4036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.87988091823786e-06,
+      "loss": 0.6452,
+      "step": 4037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.877594450978547e-06,
+      "loss": 0.6662,
+      "step": 4038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.875308230862268e-06,
+      "loss": 0.6189,
+      "step": 4039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.873022258392433e-06,
+      "loss": 0.618,
+      "step": 4040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.870736534072389e-06,
+      "loss": 0.6181,
+      "step": 4041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.868451058405429e-06,
+      "loss": 0.6488,
+      "step": 4042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.866165831894796e-06,
+      "loss": 0.6357,
+      "step": 4043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.863880855043673e-06,
+      "loss": 0.6308,
+      "step": 4044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.86159612835519e-06,
+      "loss": 0.6256,
+      "step": 4045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.859311652332424e-06,
+      "loss": 0.5545,
+      "step": 4046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.857027427478395e-06,
+      "loss": 0.6251,
+      "step": 4047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.854743454296065e-06,
+      "loss": 0.6196,
+      "step": 4048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.852459733288345e-06,
+      "loss": 0.6221,
+      "step": 4049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8501762649580894e-06,
+      "loss": 0.6124,
+      "step": 4050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.847893049808092e-06,
+      "loss": 0.6307,
+      "step": 4051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.845610088341102e-06,
+      "loss": 0.5565,
+      "step": 4052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.843327381059802e-06,
+      "loss": 0.5473,
+      "step": 4053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.841044928466821e-06,
+      "loss": 0.6542,
+      "step": 4054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.838762731064735e-06,
+      "loss": 0.6766,
+      "step": 4055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.836480789356063e-06,
+      "loss": 0.6311,
+      "step": 4056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.834199103843264e-06,
+      "loss": 0.647,
+      "step": 4057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.831917675028748e-06,
+      "loss": 0.5694,
+      "step": 4058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.829636503414861e-06,
+      "loss": 0.6527,
+      "step": 4059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.827355589503895e-06,
+      "loss": 0.5496,
+      "step": 4060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.825074933798087e-06,
+      "loss": 0.6281,
+      "step": 4061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.822794536799614e-06,
+      "loss": 0.6187,
+      "step": 4062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.820514399010598e-06,
+      "loss": 0.6315,
+      "step": 4063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.818234520933108e-06,
+      "loss": 0.6271,
+      "step": 4064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.815954903069146e-06,
+      "loss": 0.652,
+      "step": 4065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8136755459206644e-06,
+      "loss": 0.5686,
+      "step": 4066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.811396449989555e-06,
+      "loss": 0.6412,
+      "step": 4067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.809117615777654e-06,
+      "loss": 0.6647,
+      "step": 4068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.806839043786738e-06,
+      "loss": 0.6117,
+      "step": 4069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.80456073451853e-06,
+      "loss": 0.5545,
+      "step": 4070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.802282688474689e-06,
+      "loss": 0.6277,
+      "step": 4071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.80000490615682e-06,
+      "loss": 0.6424,
+      "step": 4072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7977273880664683e-06,
+      "loss": 0.6457,
+      "step": 4073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.795450134705122e-06,
+      "loss": 0.5471,
+      "step": 4074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.793173146574211e-06,
+      "loss": 0.6033,
+      "step": 4075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7908964241751077e-06,
+      "loss": 0.6317,
+      "step": 4076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7886199680091234e-06,
+      "loss": 0.6354,
+      "step": 4077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7863437785775124e-06,
+      "loss": 0.6476,
+      "step": 4078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.784067856381469e-06,
+      "loss": 0.6493,
+      "step": 4079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.78179220192213e-06,
+      "loss": 0.6441,
+      "step": 4080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7795168157005734e-06,
+      "loss": 0.5452,
+      "step": 4081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.777241698217819e-06,
+      "loss": 0.6373,
+      "step": 4082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.774966849974825e-06,
+      "loss": 0.663,
+      "step": 4083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.77269227147249e-06,
+      "loss": 0.6352,
+      "step": 4084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.770417963211656e-06,
+      "loss": 0.5656,
+      "step": 4085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.768143925693102e-06,
+      "loss": 0.6623,
+      "step": 4086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7658701594175513e-06,
+      "loss": 0.6012,
+      "step": 4087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.763596664885667e-06,
+      "loss": 0.6268,
+      "step": 4088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7613234425980493e-06,
+      "loss": 0.5537,
+      "step": 4089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7590504930552405e-06,
+      "loss": 0.6534,
+      "step": 4090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.756777816757722e-06,
+      "loss": 0.6031,
+      "step": 4091
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.754505414205914e-06,
+      "loss": 0.6871,
+      "step": 4092
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7522332859001815e-06,
+      "loss": 0.6306,
+      "step": 4093
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7499614323408227e-06,
+      "loss": 0.5318,
+      "step": 4094
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.747689854028081e-06,
+      "loss": 0.6244,
+      "step": 4095
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.745418551462134e-06,
+      "loss": 0.6698,
+      "step": 4096
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.743147525143102e-06,
+      "loss": 0.5457,
+      "step": 4097
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7408767755710406e-06,
+      "loss": 0.651,
+      "step": 4098
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.738606303245952e-06,
+      "loss": 0.6246,
+      "step": 4099
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7363361086677693e-06,
+      "loss": 0.6384,
+      "step": 4100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7340661923363696e-06,
+      "loss": 0.6412,
+      "step": 4101
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.731796554751566e-06,
+      "loss": 0.6256,
+      "step": 4102
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.729527196413112e-06,
+      "loss": 0.6304,
+      "step": 4103
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.727258117820695e-06,
+      "loss": 0.6079,
+      "step": 4104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.72498931947395e-06,
+      "loss": 0.6362,
+      "step": 4105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7227208018724413e-06,
+      "loss": 0.633,
+      "step": 4106
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.720452565515677e-06,
+      "loss": 0.6193,
+      "step": 4107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7181846109031007e-06,
+      "loss": 0.6277,
+      "step": 4108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7159169385340933e-06,
+      "loss": 0.607,
+      "step": 4109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.713649548907973e-06,
+      "loss": 0.567,
+      "step": 4110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.711382442524002e-06,
+      "loss": 0.5515,
+      "step": 4111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7091156198813727e-06,
+      "loss": 0.5577,
+      "step": 4112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.70684908147922e-06,
+      "loss": 0.6529,
+      "step": 4113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.704582827816612e-06,
+      "loss": 0.653,
+      "step": 4114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7023168593925553e-06,
+      "loss": 0.6278,
+      "step": 4115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7000511767059976e-06,
+      "loss": 0.5486,
+      "step": 4116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6977857802558205e-06,
+      "loss": 0.6345,
+      "step": 4117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6955206705408397e-06,
+      "loss": 0.6338,
+      "step": 4118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6932558480598147e-06,
+      "loss": 0.6405,
+      "step": 4119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.690991313311435e-06,
+      "loss": 0.6691,
+      "step": 4120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6887270667943285e-06,
+      "loss": 0.6138,
+      "step": 4121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6864631090070656e-06,
+      "loss": 0.6648,
+      "step": 4122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.684199440448145e-06,
+      "loss": 0.6518,
+      "step": 4123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.681936061616005e-06,
+      "loss": 0.6123,
+      "step": 4124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.679672973009022e-06,
+      "loss": 0.636,
+      "step": 4125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6774101751255043e-06,
+      "loss": 0.6253,
+      "step": 4126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.675147668463698e-06,
+      "loss": 0.6215,
+      "step": 4127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.672885453521789e-06,
+      "loss": 0.6473,
+      "step": 4128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6706235307978933e-06,
+      "loss": 0.6346,
+      "step": 4129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6683619007900647e-06,
+      "loss": 0.6475,
+      "step": 4130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6661005639962932e-06,
+      "loss": 0.6396,
+      "step": 4131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6638395209145044e-06,
+      "loss": 0.6722,
+      "step": 4132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.661578772042555e-06,
+      "loss": 0.5569,
+      "step": 4133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6593183178782454e-06,
+      "loss": 0.6057,
+      "step": 4134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6570581589193032e-06,
+      "loss": 0.5567,
+      "step": 4135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6547982956633953e-06,
+      "loss": 0.5395,
+      "step": 4136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6525387286081193e-06,
+      "loss": 0.6268,
+      "step": 4137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.650279458251014e-06,
+      "loss": 0.6402,
+      "step": 4138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6480204850895455e-06,
+      "loss": 0.6308,
+      "step": 4139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6457618096211222e-06,
+      "loss": 0.6564,
+      "step": 4140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.643503432343082e-06,
+      "loss": 0.6444,
+      "step": 4141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.641245353752696e-06,
+      "loss": 0.5557,
+      "step": 4142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6389875743471724e-06,
+      "loss": 0.6196,
+      "step": 4143
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.636730094623655e-06,
+      "loss": 0.6173,
+      "step": 4144
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6344729150792147e-06,
+      "loss": 0.6146,
+      "step": 4145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6322160362108666e-06,
+      "loss": 0.627,
+      "step": 4146
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6299594585155513e-06,
+      "loss": 0.6673,
+      "step": 4147
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6277031824901453e-06,
+      "loss": 0.557,
+      "step": 4148
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6254472086314594e-06,
+      "loss": 0.6244,
+      "step": 4149
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6231915374362385e-06,
+      "loss": 0.6367,
+      "step": 4150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6209361694011575e-06,
+      "loss": 0.5941,
+      "step": 4151
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.618681105022831e-06,
+      "loss": 0.6836,
+      "step": 4152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6164263447977997e-06,
+      "loss": 0.6418,
+      "step": 4153
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.614171889222542e-06,
+      "loss": 0.6236,
+      "step": 4154
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6119177387934656e-06,
+      "loss": 0.5421,
+      "step": 4155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.609663894006916e-06,
+      "loss": 0.6119,
+      "step": 4156
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6074103553591644e-06,
+      "loss": 0.5402,
+      "step": 4157
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6051571233464223e-06,
+      "loss": 0.6248,
+      "step": 4158
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.602904198464829e-06,
+      "loss": 0.6077,
+      "step": 4159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.600651581210457e-06,
+      "loss": 0.6338,
+      "step": 4160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5983992720793094e-06,
+      "loss": 0.5388,
+      "step": 4161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5961472715673247e-06,
+      "loss": 0.5401,
+      "step": 4162
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5938955801703723e-06,
+      "loss": 0.6436,
+      "step": 4163
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5916441983842545e-06,
+      "loss": 0.5393,
+      "step": 4164
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5893931267047023e-06,
+      "loss": 0.5495,
+      "step": 4165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5871423656273806e-06,
+      "loss": 0.6294,
+      "step": 4166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5848919156478855e-06,
+      "loss": 0.654,
+      "step": 4167
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.582641777261744e-06,
+      "loss": 0.6387,
+      "step": 4168
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5803919509644168e-06,
+      "loss": 0.6104,
+      "step": 4169
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5781424372512945e-06,
+      "loss": 0.633,
+      "step": 4170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.575893236617698e-06,
+      "loss": 0.6116,
+      "step": 4171
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.573644349558879e-06,
+      "loss": 0.5386,
+      "step": 4172
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5713957765700224e-06,
+      "loss": 0.5921,
+      "step": 4173
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.56914751814624e-06,
+      "loss": 0.6431,
+      "step": 4174
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5668995747825797e-06,
+      "loss": 0.5474,
+      "step": 4175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.564651946974017e-06,
+      "loss": 0.6111,
+      "step": 4176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.562404635215458e-06,
+      "loss": 0.6295,
+      "step": 4177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.560157640001739e-06,
+      "loss": 0.6245,
+      "step": 4178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5579109618276275e-06,
+      "loss": 0.6102,
+      "step": 4179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.555664601187817e-06,
+      "loss": 0.6448,
+      "step": 4180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.553418558576941e-06,
+      "loss": 0.6251,
+      "step": 4181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5511728344895523e-06,
+      "loss": 0.6211,
+      "step": 4182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5489274294201414e-06,
+      "loss": 0.6149,
+      "step": 4183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5466823438631226e-06,
+      "loss": 0.6526,
+      "step": 4184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5444375783128428e-06,
+      "loss": 0.555,
+      "step": 4185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.542193133263576e-06,
+      "loss": 0.515,
+      "step": 4186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5399490092095322e-06,
+      "loss": 0.6286,
+      "step": 4187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5377052066448437e-06,
+      "loss": 0.6257,
+      "step": 4188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.535461726063575e-06,
+      "loss": 0.6327,
+      "step": 4189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5332185679597184e-06,
+      "loss": 0.6258,
+      "step": 4190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5309757328271978e-06,
+      "loss": 0.6438,
+      "step": 4191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.52873322115986e-06,
+      "loss": 0.64,
+      "step": 4192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.52649103345149e-06,
+      "loss": 0.6329,
+      "step": 4193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5242491701957924e-06,
+      "loss": 0.6473,
+      "step": 4194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.522007631886408e-06,
+      "loss": 0.6258,
+      "step": 4195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.519766419016899e-06,
+      "loss": 0.5389,
+      "step": 4196
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.51752553208076e-06,
+      "loss": 0.5289,
+      "step": 4197
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5152849715714112e-06,
+      "loss": 0.6551,
+      "step": 4198
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5130447379822076e-06,
+      "loss": 0.6059,
+      "step": 4199
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5108048318064226e-06,
+      "loss": 0.6243,
+      "step": 4200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5085652535372648e-06,
+      "loss": 0.6544,
+      "step": 4201
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5063260036678683e-06,
+      "loss": 0.6003,
+      "step": 4202
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.504087082691293e-06,
+      "loss": 0.6347,
+      "step": 4203
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.501848491100526e-06,
+      "loss": 0.6196,
+      "step": 4204
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4996102293884896e-06,
+      "loss": 0.6231,
+      "step": 4205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.497372298048024e-06,
+      "loss": 0.6453,
+      "step": 4206
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4951346975719016e-06,
+      "loss": 0.6297,
+      "step": 4207
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4928974284528205e-06,
+      "loss": 0.5626,
+      "step": 4208
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.490660491183406e-06,
+      "loss": 0.643,
+      "step": 4209
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.488423886256208e-06,
+      "loss": 0.5697,
+      "step": 4210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4861876141637106e-06,
+      "loss": 0.6209,
+      "step": 4211
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.483951675398316e-06,
+      "loss": 0.5568,
+      "step": 4212
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.481716070452359e-06,
+      "loss": 0.6336,
+      "step": 4213
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.479480799818097e-06,
+      "loss": 0.6521,
+      "step": 4214
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4772458639877132e-06,
+      "loss": 0.6266,
+      "step": 4215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4750112634533245e-06,
+      "loss": 0.632,
+      "step": 4216
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.472776998706966e-06,
+      "loss": 0.5714,
+      "step": 4217
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4705430702406008e-06,
+      "loss": 0.5456,
+      "step": 4218
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.468309478546119e-06,
+      "loss": 0.6479,
+      "step": 4219
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4660762241153374e-06,
+      "loss": 0.6349,
+      "step": 4220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4638433074399945e-06,
+      "loss": 0.6621,
+      "step": 4221
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4616107290117618e-06,
+      "loss": 0.634,
+      "step": 4222
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4593784893222305e-06,
+      "loss": 0.6379,
+      "step": 4223
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4571465888629164e-06,
+      "loss": 0.6498,
+      "step": 4224
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4549150281252635e-06,
+      "loss": 0.5976,
+      "step": 4225
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.452683807600642e-06,
+      "loss": 0.6735,
+      "step": 4226
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.450452927780342e-06,
+      "loss": 0.6248,
+      "step": 4227
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4482223891555865e-06,
+      "loss": 0.6149,
+      "step": 4228
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4459921922175165e-06,
+      "loss": 0.6285,
+      "step": 4229
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4437623374572005e-06,
+      "loss": 0.5457,
+      "step": 4230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4415328253656298e-06,
+      "loss": 0.5695,
+      "step": 4231
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4393036564337253e-06,
+      "loss": 0.5896,
+      "step": 4232
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.437074831152324e-06,
+      "loss": 0.5468,
+      "step": 4233
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4348463500121963e-06,
+      "loss": 0.6473,
+      "step": 4234
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.432618213504032e-06,
+      "loss": 0.6391,
+      "step": 4235
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.430390422118445e-06,
+      "loss": 0.6421,
+      "step": 4236
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.428162976345971e-06,
+      "loss": 0.6282,
+      "step": 4237
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.425935876677077e-06,
+      "loss": 0.6554,
+      "step": 4238
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.423709123602144e-06,
+      "loss": 0.5814,
+      "step": 4239
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.421482717611487e-06,
+      "loss": 0.5424,
+      "step": 4240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.419256659195338e-06,
+      "loss": 0.6343,
+      "step": 4241
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4170309488438537e-06,
+      "loss": 0.5517,
+      "step": 4242
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.414805587047112e-06,
+      "loss": 0.5508,
+      "step": 4243
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4125805742951196e-06,
+      "loss": 0.5389,
+      "step": 4244
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4103559110777994e-06,
+      "loss": 0.6399,
+      "step": 4245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4081315978850065e-06,
+      "loss": 0.6475,
+      "step": 4246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.40590763520651e-06,
+      "loss": 0.6433,
+      "step": 4247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.403684023532007e-06,
+      "loss": 0.6432,
+      "step": 4248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4014607633511126e-06,
+      "loss": 0.5467,
+      "step": 4249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.399237855153371e-06,
+      "loss": 0.6437,
+      "step": 4250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.397015299428242e-06,
+      "loss": 0.6307,
+      "step": 4251
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3947930966651153e-06,
+      "loss": 0.653,
+      "step": 4252
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.392571247353297e-06,
+      "loss": 0.6678,
+      "step": 4253
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.390349751982017e-06,
+      "loss": 0.6449,
+      "step": 4254
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3881286110404267e-06,
+      "loss": 0.6695,
+      "step": 4255
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3859078250176013e-06,
+      "loss": 0.6436,
+      "step": 4256
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3836873944025356e-06,
+      "loss": 0.5666,
+      "step": 4257
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3814673196841496e-06,
+      "loss": 0.6062,
+      "step": 4258
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.379247601351282e-06,
+      "loss": 0.6238,
+      "step": 4259
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3770282398926923e-06,
+      "loss": 0.6344,
+      "step": 4260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.374809235797064e-06,
+      "loss": 0.6515,
+      "step": 4261
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3725905895529996e-06,
+      "loss": 0.6323,
+      "step": 4262
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.370372301649023e-06,
+      "loss": 0.6361,
+      "step": 4263
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3681543725735843e-06,
+      "loss": 0.6404,
+      "step": 4264
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3659368028150474e-06,
+      "loss": 0.633,
+      "step": 4265
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3637195928617006e-06,
+      "loss": 0.6209,
+      "step": 4266
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.361502743201752e-06,
+      "loss": 0.5529,
+      "step": 4267
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3592862543233295e-06,
+      "loss": 0.6165,
+      "step": 4268
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3570701267144865e-06,
+      "loss": 0.5442,
+      "step": 4269
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3548543608631904e-06,
+      "loss": 0.6065,
+      "step": 4270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3526389572573338e-06,
+      "loss": 0.5886,
+      "step": 4271
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3504239163847263e-06,
+      "loss": 0.6441,
+      "step": 4272
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3482092387330996e-06,
+      "loss": 0.629,
+      "step": 4273
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3459949247901014e-06,
+      "loss": 0.5513,
+      "step": 4274
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.343780975043307e-06,
+      "loss": 0.6275,
+      "step": 4275
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.341567389980205e-06,
+      "loss": 0.6301,
+      "step": 4276
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.339354170088207e-06,
+      "loss": 0.6176,
+      "step": 4277
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.337141315854642e-06,
+      "loss": 0.5308,
+      "step": 4278
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.33492882776676e-06,
+      "loss": 0.5261,
+      "step": 4279
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3327167063117273e-06,
+      "loss": 0.5502,
+      "step": 4280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.330504951976636e-06,
+      "loss": 0.6032,
+      "step": 4281
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3282935652484905e-06,
+      "loss": 0.5439,
+      "step": 4282
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.32608254661422e-06,
+      "loss": 0.5298,
+      "step": 4283
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3238718965606674e-06,
+      "loss": 0.5552,
+      "step": 4284
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.321661615574597e-06,
+      "loss": 0.6141,
+      "step": 4285
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.319451704142691e-06,
+      "loss": 0.6109,
+      "step": 4286
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3172421627515537e-06,
+      "loss": 0.622,
+      "step": 4287
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.315032991887703e-06,
+      "loss": 0.6232,
+      "step": 4288
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.312824192037578e-06,
+      "loss": 0.603,
+      "step": 4289
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3106157636875356e-06,
+      "loss": 0.5399,
+      "step": 4290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.308407707323851e-06,
+      "loss": 0.6013,
+      "step": 4291
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.306200023432714e-06,
+      "loss": 0.6478,
+      "step": 4292
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.30399271250024e-06,
+      "loss": 0.5729,
+      "step": 4293
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.301785775012456e-06,
+      "loss": 0.6158,
+      "step": 4294
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2995792114553092e-06,
+      "loss": 0.6039,
+      "step": 4295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.297373022314664e-06,
+      "loss": 0.6313,
+      "step": 4296
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.295167208076301e-06,
+      "loss": 0.5847,
+      "step": 4297
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2929617692259187e-06,
+      "loss": 0.6351,
+      "step": 4298
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2907567062491376e-06,
+      "loss": 0.6771,
+      "step": 4299
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.288552019631488e-06,
+      "loss": 0.6367,
+      "step": 4300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2863477098584238e-06,
+      "loss": 0.561,
+      "step": 4301
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2841437774153116e-06,
+      "loss": 0.6337,
+      "step": 4302
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2819402227874364e-06,
+      "loss": 0.6075,
+      "step": 4303
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2797370464599974e-06,
+      "loss": 0.6517,
+      "step": 4304
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.277534248918119e-06,
+      "loss": 0.6514,
+      "step": 4305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2753318306468323e-06,
+      "loss": 0.6346,
+      "step": 4306
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.273129792131089e-06,
+      "loss": 0.6239,
+      "step": 4307
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.270928133855758e-06,
+      "loss": 0.6288,
+      "step": 4308
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2687268563056236e-06,
+      "loss": 0.6208,
+      "step": 4309
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2665259599653836e-06,
+      "loss": 0.6183,
+      "step": 4310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.264325445319659e-06,
+      "loss": 0.6061,
+      "step": 4311
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2621253128529795e-06,
+      "loss": 0.624,
+      "step": 4312
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.259925563049793e-06,
+      "loss": 0.5494,
+      "step": 4313
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2577261963944653e-06,
+      "loss": 0.6153,
+      "step": 4314
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2555272133712744e-06,
+      "loss": 0.5459,
+      "step": 4315
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.253328614464414e-06,
+      "loss": 0.6211,
+      "step": 4316
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.251130400157999e-06,
+      "loss": 0.6481,
+      "step": 4317
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2489325709360537e-06,
+      "loss": 0.6281,
+      "step": 4318
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2467351272825166e-06,
+      "loss": 0.5577,
+      "step": 4319
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.244538069681248e-06,
+      "loss": 0.545,
+      "step": 4320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2423413986160135e-06,
+      "loss": 0.6284,
+      "step": 4321
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2401451145705056e-06,
+      "loss": 0.6554,
+      "step": 4322
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2379492180283222e-06,
+      "loss": 0.5723,
+      "step": 4323
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2357537094729787e-06,
+      "loss": 0.6347,
+      "step": 4324
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2335585893879045e-06,
+      "loss": 0.5581,
+      "step": 4325
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2313638582564465e-06,
+      "loss": 0.6398,
+      "step": 4326
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2291695165618597e-06,
+      "loss": 0.6123,
+      "step": 4327
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.226975564787322e-06,
+      "loss": 0.6,
+      "step": 4328
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2247820034159182e-06,
+      "loss": 0.5082,
+      "step": 4329
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2225888329306508e-06,
+      "loss": 0.5529,
+      "step": 4330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.220396053814433e-06,
+      "loss": 0.6431,
+      "step": 4331
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.218203666550096e-06,
+      "loss": 0.5865,
+      "step": 4332
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2160116716203804e-06,
+      "loss": 0.6187,
+      "step": 4333
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2138200695079456e-06,
+      "loss": 0.6409,
+      "step": 4334
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.211628860695361e-06,
+      "loss": 0.6014,
+      "step": 4335
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2094380456651085e-06,
+      "loss": 0.637,
+      "step": 4336
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2072476248995845e-06,
+      "loss": 0.6266,
+      "step": 4337
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2050575988811008e-06,
+      "loss": 0.6337,
+      "step": 4338
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2028679680918774e-06,
+      "loss": 0.5334,
+      "step": 4339
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.200678733014053e-06,
+      "loss": 0.6417,
+      "step": 4340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1984898941296764e-06,
+      "loss": 0.6237,
+      "step": 4341
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.196301451920708e-06,
+      "loss": 0.6211,
+      "step": 4342
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1941134068690217e-06,
+      "loss": 0.5552,
+      "step": 4343
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.191925759456404e-06,
+      "loss": 0.6057,
+      "step": 4344
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1897385101645534e-06,
+      "loss": 0.5213,
+      "step": 4345
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.187551659475085e-06,
+      "loss": 0.6349,
+      "step": 4346
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1853652078695197e-06,
+      "loss": 0.6113,
+      "step": 4347
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1831791558292945e-06,
+      "loss": 0.5945,
+      "step": 4348
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1809935038357564e-06,
+      "loss": 0.6288,
+      "step": 4349
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1788082523701647e-06,
+      "loss": 0.6527,
+      "step": 4350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1766234019136918e-06,
+      "loss": 0.5541,
+      "step": 4351
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1744389529474206e-06,
+      "loss": 0.5495,
+      "step": 4352
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.172254905952348e-06,
+      "loss": 0.6453,
+      "step": 4353
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.170071261409379e-06,
+      "loss": 0.5739,
+      "step": 4354
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.16788801979933e-06,
+      "loss": 0.6053,
+      "step": 4355
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1657051816029317e-06,
+      "loss": 0.6649,
+      "step": 4356
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.163522747300823e-06,
+      "loss": 0.6532,
+      "step": 4357
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1613407173735563e-06,
+      "loss": 0.6275,
+      "step": 4358
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1591590923015946e-06,
+      "loss": 0.6172,
+      "step": 4359
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.15697787256531e-06,
+      "loss": 0.5379,
+      "step": 4360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1547970586449868e-06,
+      "loss": 0.5564,
+      "step": 4361
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.152616651020818e-06,
+      "loss": 0.55,
+      "step": 4362
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1504366501729096e-06,
+      "loss": 0.6385,
+      "step": 4363
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.148257056581277e-06,
+      "loss": 0.637,
+      "step": 4364
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1460778707258478e-06,
+      "loss": 0.6486,
+      "step": 4365
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.143899093086456e-06,
+      "loss": 0.6531,
+      "step": 4366
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1417207241428484e-06,
+      "loss": 0.6085,
+      "step": 4367
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1395427643746802e-06,
+      "loss": 0.6251,
+      "step": 4368
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.137365214261518e-06,
+      "loss": 0.6218,
+      "step": 4369
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1351880742828376e-06,
+      "loss": 0.6222,
+      "step": 4370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.133011344918026e-06,
+      "loss": 0.5606,
+      "step": 4371
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.130835026646377e-06,
+      "loss": 0.6139,
+      "step": 4372
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1286591199470957e-06,
+      "loss": 0.6174,
+      "step": 4373
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.126483625299293e-06,
+      "loss": 0.5643,
+      "step": 4374
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1243085431819974e-06,
+      "loss": 0.6269,
+      "step": 4375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1221338740741374e-06,
+      "loss": 0.5399,
+      "step": 4376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1199596184545573e-06,
+      "loss": 0.5445,
+      "step": 4377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.117785776802006e-06,
+      "loss": 0.6457,
+      "step": 4378
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.115612349595142e-06,
+      "loss": 0.6483,
+      "step": 4379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1134393373125327e-06,
+      "loss": 0.6002,
+      "step": 4380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.111266740432658e-06,
+      "loss": 0.5545,
+      "step": 4381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1090945594339005e-06,
+      "loss": 0.6322,
+      "step": 4382
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1069227947945556e-06,
+      "loss": 0.6578,
+      "step": 4383
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.104751446992824e-06,
+      "loss": 0.635,
+      "step": 4384
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.102580516506817e-06,
+      "loss": 0.6116,
+      "step": 4385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.100410003814549e-06,
+      "loss": 0.6348,
+      "step": 4386
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0982399093939517e-06,
+      "loss": 0.6294,
+      "step": 4387
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.096070233722856e-06,
+      "loss": 0.6336,
+      "step": 4388
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0939009772790052e-06,
+      "loss": 0.6081,
+      "step": 4389
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0917321405400492e-06,
+      "loss": 0.6098,
+      "step": 4390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0895637239835436e-06,
+      "loss": 0.6123,
+      "step": 4391
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0873957280869526e-06,
+      "loss": 0.5452,
+      "step": 4392
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.085228153327651e-06,
+      "loss": 0.6243,
+      "step": 4393
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0830610001829173e-06,
+      "loss": 0.6531,
+      "step": 4394
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0808942691299373e-06,
+      "loss": 0.6418,
+      "step": 4395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0787279606458047e-06,
+      "loss": 0.538,
+      "step": 4396
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.076562075207521e-06,
+      "loss": 0.5165,
+      "step": 4397
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.07439661329199e-06,
+      "loss": 0.6411,
+      "step": 4398
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0722315753760317e-06,
+      "loss": 0.6209,
+      "step": 4399
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.070066961936363e-06,
+      "loss": 0.6444,
+      "step": 4400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.067902773449612e-06,
+      "loss": 0.6433,
+      "step": 4401
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.065739010392313e-06,
+      "loss": 0.6445,
+      "step": 4402
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0635756732409054e-06,
+      "loss": 0.6263,
+      "step": 4403
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0614127624717338e-06,
+      "loss": 0.5303,
+      "step": 4404
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0592502785610546e-06,
+      "loss": 0.557,
+      "step": 4405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.057088221985024e-06,
+      "loss": 0.6337,
+      "step": 4406
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0549265932197055e-06,
+      "loss": 0.5321,
+      "step": 4407
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0527653927410704e-06,
+      "loss": 0.5376,
+      "step": 4408
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0506046210249953e-06,
+      "loss": 0.6183,
+      "step": 4409
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.048444278547258e-06,
+      "loss": 0.5466,
+      "step": 4410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0462843657835494e-06,
+      "loss": 0.6374,
+      "step": 4411
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.044124883209461e-06,
+      "loss": 0.6024,
+      "step": 4412
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0419658313004884e-06,
+      "loss": 0.532,
+      "step": 4413
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0398072105320364e-06,
+      "loss": 0.6335,
+      "step": 4414
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0376490213794117e-06,
+      "loss": 0.634,
+      "step": 4415
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.035491264317826e-06,
+      "loss": 0.6491,
+      "step": 4416
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0333339398223993e-06,
+      "loss": 0.6642,
+      "step": 4417
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0311770483681536e-06,
+      "loss": 0.5368,
+      "step": 4418
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0290205904300137e-06,
+      "loss": 0.6135,
+      "step": 4419
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.026864566482813e-06,
+      "loss": 0.5495,
+      "step": 4420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.024708977001288e-06,
+      "loss": 0.6428,
+      "step": 4421
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0225538224600758e-06,
+      "loss": 0.6305,
+      "step": 4422
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0203991033337256e-06,
+      "loss": 0.5191,
+      "step": 4423
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.018244820096683e-06,
+      "loss": 0.6289,
+      "step": 4424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0160909732233007e-06,
+      "loss": 0.5666,
+      "step": 4425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0139375631878372e-06,
+      "loss": 0.6284,
+      "step": 4426
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0117845904644492e-06,
+      "loss": 0.6195,
+      "step": 4427
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0096320555272063e-06,
+      "loss": 0.5357,
+      "step": 4428
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0074799588500725e-06,
+      "loss": 0.6333,
+      "step": 4429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.00532830090692e-06,
+      "loss": 0.6251,
+      "step": 4430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0031770821715233e-06,
+      "loss": 0.5278,
+      "step": 4431
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.00102630311756e-06,
+      "loss": 0.5305,
+      "step": 4432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.99887596421861e-06,
+      "loss": 0.6425,
+      "step": 4433
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.996726065948161e-06,
+      "loss": 0.6206,
+      "step": 4434
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.994576608779598e-06,
+      "loss": 0.6376,
+      "step": 4435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9924275931862117e-06,
+      "loss": 0.5991,
+      "step": 4436
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9902790196411944e-06,
+      "loss": 0.6133,
+      "step": 4437
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.988130888617641e-06,
+      "loss": 0.6594,
+      "step": 4438
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9859832005885493e-06,
+      "loss": 0.5317,
+      "step": 4439
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9838359560268214e-06,
+      "loss": 0.6296,
+      "step": 4440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9816891554052607e-06,
+      "loss": 0.5375,
+      "step": 4441
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9795427991965705e-06,
+      "loss": 0.6174,
+      "step": 4442
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.977396887873358e-06,
+      "loss": 0.5068,
+      "step": 4443
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.975251421908133e-06,
+      "loss": 0.617,
+      "step": 4444
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.973106401773307e-06,
+      "loss": 0.6553,
+      "step": 4445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9709618279411922e-06,
+      "loss": 0.6441,
+      "step": 4446
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.968817700884005e-06,
+      "loss": 0.6393,
+      "step": 4447
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.966674021073861e-06,
+      "loss": 0.6303,
+      "step": 4448
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.964530788982778e-06,
+      "loss": 0.6152,
+      "step": 4449
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.962388005082675e-06,
+      "loss": 0.6676,
+      "step": 4450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.960245669845372e-06,
+      "loss": 0.5577,
+      "step": 4451
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.958103783742592e-06,
+      "loss": 0.6402,
+      "step": 4452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9559623472459585e-06,
+      "loss": 0.6217,
+      "step": 4453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.953821360826995e-06,
+      "loss": 0.6514,
+      "step": 4454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.951680824957126e-06,
+      "loss": 0.6193,
+      "step": 4455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.949540740107677e-06,
+      "loss": 0.6295,
+      "step": 4456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9474011067498744e-06,
+      "loss": 0.5494,
+      "step": 4457
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.945261925354845e-06,
+      "loss": 0.6275,
+      "step": 4458
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9431231963936176e-06,
+      "loss": 0.5233,
+      "step": 4459
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.940984920337119e-06,
+      "loss": 0.637,
+      "step": 4460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.938847097656177e-06,
+      "loss": 0.6261,
+      "step": 4461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.93670972882152e-06,
+      "loss": 0.6422,
+      "step": 4462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9345728143037753e-06,
+      "loss": 0.6257,
+      "step": 4463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9324363545734735e-06,
+      "loss": 0.5514,
+      "step": 4464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.930300350101042e-06,
+      "loss": 0.6152,
+      "step": 4465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9281648013568074e-06,
+      "loss": 0.6165,
+      "step": 4466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.926029708810999e-06,
+      "loss": 0.5544,
+      "step": 4467
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9238950729337424e-06,
+      "loss": 0.6296,
+      "step": 4468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9217608941950627e-06,
+      "loss": 0.6165,
+      "step": 4469
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9196271730648897e-06,
+      "loss": 0.632,
+      "step": 4470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9174939100130474e-06,
+      "loss": 0.6285,
+      "step": 4471
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9153611055092586e-06,
+      "loss": 0.6562,
+      "step": 4472
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9132287600231452e-06,
+      "loss": 0.62,
+      "step": 4473
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9110968740242335e-06,
+      "loss": 0.532,
+      "step": 4474
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.908965447981941e-06,
+      "loss": 0.6254,
+      "step": 4475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.90683448236559e-06,
+      "loss": 0.6502,
+      "step": 4476
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.904703977644398e-06,
+      "loss": 0.6136,
+      "step": 4477
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.902573934287482e-06,
+      "loss": 0.6362,
+      "step": 4478
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.900444352763857e-06,
+      "loss": 0.624,
+      "step": 4479
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8983152335424348e-06,
+      "loss": 0.5995,
+      "step": 4480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8961865770920304e-06,
+      "loss": 0.6092,
+      "step": 4481
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.894058383881353e-06,
+      "loss": 0.5311,
+      "step": 4482
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8919306543790094e-06,
+      "loss": 0.658,
+      "step": 4483
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8898033890535064e-06,
+      "loss": 0.6124,
+      "step": 4484
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8876765883732447e-06,
+      "loss": 0.5359,
+      "step": 4485
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8855502528065303e-06,
+      "loss": 0.5706,
+      "step": 4486
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.883424382821558e-06,
+      "loss": 0.5739,
+      "step": 4487
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.881298978886428e-06,
+      "loss": 0.6188,
+      "step": 4488
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8791740414691322e-06,
+      "loss": 0.6328,
+      "step": 4489
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8770495710375613e-06,
+      "loss": 0.5351,
+      "step": 4490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8749255680595034e-06,
+      "loss": 0.6173,
+      "step": 4491
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.872802033002643e-06,
+      "loss": 0.6585,
+      "step": 4492
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.870678966334564e-06,
+      "loss": 0.546,
+      "step": 4493
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8685563685227456e-06,
+      "loss": 0.625,
+      "step": 4494
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8664342400345635e-06,
+      "loss": 0.6408,
+      "step": 4495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.864312581337289e-06,
+      "loss": 0.6481,
+      "step": 4496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8621913928980904e-06,
+      "loss": 0.6421,
+      "step": 4497
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.860070675184036e-06,
+      "loss": 0.6294,
+      "step": 4498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.857950428662085e-06,
+      "loss": 0.6433,
+      "step": 4499
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.855830653799099e-06,
+      "loss": 0.6417,
+      "step": 4500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8537113510618297e-06,
+      "loss": 0.6521,
+      "step": 4501
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8515925209169287e-06,
+      "loss": 0.6173,
+      "step": 4502
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.849474163830941e-06,
+      "loss": 0.5709,
+      "step": 4503
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.847356280270308e-06,
+      "loss": 0.5779,
+      "step": 4504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.84523887070137e-06,
+      "loss": 0.614,
+      "step": 4505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8431219355903595e-06,
+      "loss": 0.6555,
+      "step": 4506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8410054754034054e-06,
+      "loss": 0.6645,
+      "step": 4507
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8388894906065317e-06,
+      "loss": 0.6136,
+      "step": 4508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.836773981665657e-06,
+      "loss": 0.5309,
+      "step": 4509
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8346589490465993e-06,
+      "loss": 0.6171,
+      "step": 4510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8325443932150646e-06,
+      "loss": 0.648,
+      "step": 4511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.830430314636663e-06,
+      "loss": 0.5503,
+      "step": 4512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.828316713776892e-06,
+      "loss": 0.6162,
+      "step": 4513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8262035911011465e-06,
+      "loss": 0.5989,
+      "step": 4514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8240909470747156e-06,
+      "loss": 0.6435,
+      "step": 4515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.821978782162782e-06,
+      "loss": 0.5502,
+      "step": 4516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.819867096830428e-06,
+      "loss": 0.6397,
+      "step": 4517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.817755891542625e-06,
+      "loss": 0.6366,
+      "step": 4518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8156451667642405e-06,
+      "loss": 0.5247,
+      "step": 4519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.813534922960035e-06,
+      "loss": 0.6278,
+      "step": 4520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.811425160594665e-06,
+      "loss": 0.6163,
+      "step": 4521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.809315880132678e-06,
+      "loss": 0.6157,
+      "step": 4522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.807207082038522e-06,
+      "loss": 0.6435,
+      "step": 4523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.805098766776529e-06,
+      "loss": 0.609,
+      "step": 4524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8029909348109354e-06,
+      "loss": 0.6391,
+      "step": 4525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8008835866058625e-06,
+      "loss": 0.6387,
+      "step": 4526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7987767226253287e-06,
+      "loss": 0.5449,
+      "step": 4527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7966703433332443e-06,
+      "loss": 0.6477,
+      "step": 4528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7945644491934174e-06,
+      "loss": 0.6051,
+      "step": 4529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.792459040669544e-06,
+      "loss": 0.6581,
+      "step": 4530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.790354118225215e-06,
+      "loss": 0.5941,
+      "step": 4531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7882496823239137e-06,
+      "loss": 0.6493,
+      "step": 4532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7861457334290155e-06,
+      "loss": 0.5271,
+      "step": 4533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7840422720037943e-06,
+      "loss": 0.6554,
+      "step": 4534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7819392985114097e-06,
+      "loss": 0.6216,
+      "step": 4535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7798368134149145e-06,
+      "loss": 0.5686,
+      "step": 4536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7777348171772604e-06,
+      "loss": 0.5946,
+      "step": 4537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.775633310261285e-06,
+      "loss": 0.6274,
+      "step": 4538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7735322931297184e-06,
+      "loss": 0.6182,
+      "step": 4539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.771431766245187e-06,
+      "loss": 0.617,
+      "step": 4540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.769331730070206e-06,
+      "loss": 0.6414,
+      "step": 4541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7672321850671842e-06,
+      "loss": 0.5224,
+      "step": 4542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7651331316984198e-06,
+      "loss": 0.6076,
+      "step": 4543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7630345704261057e-06,
+      "loss": 0.6361,
+      "step": 4544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.760936501712323e-06,
+      "loss": 0.6271,
+      "step": 4545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7588389260190495e-06,
+      "loss": 0.5753,
+      "step": 4546
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7567418438081504e-06,
+      "loss": 0.6115,
+      "step": 4547
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7546452555413817e-06,
+      "loss": 0.5556,
+      "step": 4548
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.752549161680395e-06,
+      "loss": 0.6505,
+      "step": 4549
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.750453562686729e-06,
+      "loss": 0.6116,
+      "step": 4550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.748358459021813e-06,
+      "loss": 0.6533,
+      "step": 4551
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.746263851146971e-06,
+      "loss": 0.5314,
+      "step": 4552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.744169739523416e-06,
+      "loss": 0.6287,
+      "step": 4553
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.742076124612251e-06,
+      "loss": 0.6262,
+      "step": 4554
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7399830068744683e-06,
+      "loss": 0.5622,
+      "step": 4555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.737890386770955e-06,
+      "loss": 0.5328,
+      "step": 4556
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7357982647624836e-06,
+      "loss": 0.6269,
+      "step": 4557
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7337066413097223e-06,
+      "loss": 0.5803,
+      "step": 4558
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7316155168732255e-06,
+      "loss": 0.6573,
+      "step": 4559
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7295248919134387e-06,
+      "loss": 0.5966,
+      "step": 4560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7274347668906963e-06,
+      "loss": 0.6099,
+      "step": 4561
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7253451422652277e-06,
+      "loss": 0.5471,
+      "step": 4562
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7232560184971437e-06,
+      "loss": 0.6466,
+      "step": 4563
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.721167396046454e-06,
+      "loss": 0.6256,
+      "step": 4564
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.719079275373052e-06,
+      "loss": 0.6293,
+      "step": 4565
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7169916569367204e-06,
+      "loss": 0.6286,
+      "step": 4566
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7149045411971354e-06,
+      "loss": 0.6443,
+      "step": 4567
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7128179286138574e-06,
+      "loss": 0.6341,
+      "step": 4568
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7107318196463383e-06,
+      "loss": 0.5948,
+      "step": 4569
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7086462147539228e-06,
+      "loss": 0.5922,
+      "step": 4570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7065611143958403e-06,
+      "loss": 0.6155,
+      "step": 4571
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.704476519031209e-06,
+      "loss": 0.6376,
+      "step": 4572
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.702392429119035e-06,
+      "loss": 0.5478,
+      "step": 4573
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7003088451182204e-06,
+      "loss": 0.6632,
+      "step": 4574
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6982257674875444e-06,
+      "loss": 0.6134,
+      "step": 4575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6961431966856866e-06,
+      "loss": 0.5489,
+      "step": 4576
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6940611331712075e-06,
+      "loss": 0.5976,
+      "step": 4577
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.691979577402556e-06,
+      "loss": 0.5657,
+      "step": 4578
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6898985298380725e-06,
+      "loss": 0.5469,
+      "step": 4579
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6878179909359834e-06,
+      "loss": 0.5637,
+      "step": 4580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.685737961154401e-06,
+      "loss": 0.6229,
+      "step": 4581
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6836584409513335e-06,
+      "loss": 0.6142,
+      "step": 4582
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6815794307846676e-06,
+      "loss": 0.6037,
+      "step": 4583
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.679500931112183e-06,
+      "loss": 0.6364,
+      "step": 4584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6774229423915443e-06,
+      "loss": 0.6227,
+      "step": 4585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.675345465080307e-06,
+      "loss": 0.5555,
+      "step": 4586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6732684996359084e-06,
+      "loss": 0.6336,
+      "step": 4587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6711920465156805e-06,
+      "loss": 0.6217,
+      "step": 4588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.669116106176838e-06,
+      "loss": 0.6243,
+      "step": 4589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6670406790764806e-06,
+      "loss": 0.5951,
+      "step": 4590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6649657656716e-06,
+      "loss": 0.5034,
+      "step": 4591
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6628913664190693e-06,
+      "loss": 0.6277,
+      "step": 4592
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6608174817756562e-06,
+      "loss": 0.6504,
+      "step": 4593
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6587441121980075e-06,
+      "loss": 0.6309,
+      "step": 4594
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6566712581426594e-06,
+      "loss": 0.6572,
+      "step": 4595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6545989200660362e-06,
+      "loss": 0.6283,
+      "step": 4596
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6525270984244455e-06,
+      "loss": 0.5272,
+      "step": 4597
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.650455793674081e-06,
+      "loss": 0.6302,
+      "step": 4598
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.648385006271027e-06,
+      "loss": 0.517,
+      "step": 4599
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.646314736671253e-06,
+      "loss": 0.6445,
+      "step": 4600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6442449853306097e-06,
+      "loss": 0.6336,
+      "step": 4601
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6421757527048373e-06,
+      "loss": 0.6184,
+      "step": 4602
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6401070392495627e-06,
+      "loss": 0.6594,
+      "step": 4603
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.638038845420293e-06,
+      "loss": 0.6114,
+      "step": 4604
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.635971171672429e-06,
+      "loss": 0.6317,
+      "step": 4605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6339040184612512e-06,
+      "loss": 0.6157,
+      "step": 4606
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6318373862419277e-06,
+      "loss": 0.5649,
+      "step": 4607
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6297712754695102e-06,
+      "loss": 0.5324,
+      "step": 4608
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6277056865989374e-06,
+      "loss": 0.6172,
+      "step": 4609
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6256406200850305e-06,
+      "loss": 0.5456,
+      "step": 4610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6235760763825004e-06,
+      "loss": 0.5824,
+      "step": 4611
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.621512055945937e-06,
+      "loss": 0.6154,
+      "step": 4612
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.619448559229822e-06,
+      "loss": 0.6397,
+      "step": 4613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.617385586688515e-06,
+      "loss": 0.6405,
+      "step": 4614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.615323138776264e-06,
+      "loss": 0.5408,
+      "step": 4615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6132612159471976e-06,
+      "loss": 0.5675,
+      "step": 4616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6111998186553355e-06,
+      "loss": 0.6231,
+      "step": 4617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.609138947354577e-06,
+      "loss": 0.5429,
+      "step": 4618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.607078602498705e-06,
+      "loss": 0.6161,
+      "step": 4619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6050187845413887e-06,
+      "loss": 0.5411,
+      "step": 4620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.60295949393618e-06,
+      "loss": 0.6327,
+      "step": 4621
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6009007311365135e-06,
+      "loss": 0.6345,
+      "step": 4622
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.598842496595713e-06,
+      "loss": 0.6156,
+      "step": 4623
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5967847907669785e-06,
+      "loss": 0.6231,
+      "step": 4624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5947276141034006e-06,
+      "loss": 0.5936,
+      "step": 4625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.592670967057948e-06,
+      "loss": 0.5469,
+      "step": 4626
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.590614850083476e-06,
+      "loss": 0.6202,
+      "step": 4627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.588559263632719e-06,
+      "loss": 0.6435,
+      "step": 4628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.586504208158302e-06,
+      "loss": 0.6246,
+      "step": 4629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5844496841127265e-06,
+      "loss": 0.6294,
+      "step": 4630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5823956919483794e-06,
+      "loss": 0.5526,
+      "step": 4631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5803422321175296e-06,
+      "loss": 0.6023,
+      "step": 4632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5782893050723303e-06,
+      "loss": 0.6617,
+      "step": 4633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5762369112648144e-06,
+      "loss": 0.5458,
+      "step": 4634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.574185051146903e-06,
+      "loss": 0.5612,
+      "step": 4635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5721337251703927e-06,
+      "loss": 0.6584,
+      "step": 4636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.57008293378697e-06,
+      "loss": 0.5508,
+      "step": 4637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5680326774481965e-06,
+      "loss": 0.6272,
+      "step": 4638
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5659829566055194e-06,
+      "loss": 0.6592,
+      "step": 4639
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5639337717102697e-06,
+      "loss": 0.6315,
+      "step": 4640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5618851232136576e-06,
+      "loss": 0.6208,
+      "step": 4641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5598370115667757e-06,
+      "loss": 0.6338,
+      "step": 4642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.557789437220598e-06,
+      "loss": 0.6014,
+      "step": 4643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5557424006259823e-06,
+      "loss": 0.5489,
+      "step": 4644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5536959022336637e-06,
+      "loss": 0.6669,
+      "step": 4645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5516499424942666e-06,
+      "loss": 0.6205,
+      "step": 4646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5496045218582882e-06,
+      "loss": 0.6414,
+      "step": 4647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5475596407761122e-06,
+      "loss": 0.636,
+      "step": 4648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5455152996979993e-06,
+      "loss": 0.6374,
+      "step": 4649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5434714990740984e-06,
+      "loss": 0.5914,
+      "step": 4650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5414282393544308e-06,
+      "loss": 0.6108,
+      "step": 4651
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.539385520988907e-06,
+      "loss": 0.5507,
+      "step": 4652
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.537343344427311e-06,
+      "loss": 0.6277,
+      "step": 4653
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5353017101193124e-06,
+      "loss": 0.6453,
+      "step": 4654
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5332606185144595e-06,
+      "loss": 0.6465,
+      "step": 4655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5312200700621796e-06,
+      "loss": 0.6082,
+      "step": 4656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5291800652117822e-06,
+      "loss": 0.5442,
+      "step": 4657
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.527140604412459e-06,
+      "loss": 0.625,
+      "step": 4658
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5251016881132796e-06,
+      "loss": 0.6431,
+      "step": 4659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5230633167631936e-06,
+      "loss": 0.6032,
+      "step": 4660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5210254908110277e-06,
+      "loss": 0.6008,
+      "step": 4661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5189882107054973e-06,
+      "loss": 0.6294,
+      "step": 4662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5169514768951876e-06,
+      "loss": 0.5753,
+      "step": 4663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5149152898285724e-06,
+      "loss": 0.5765,
+      "step": 4664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5128796499539983e-06,
+      "loss": 0.6296,
+      "step": 4665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5108445577196937e-06,
+      "loss": 0.5437,
+      "step": 4666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.508810013573767e-06,
+      "loss": 0.5325,
+      "step": 4667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5067760179642047e-06,
+      "loss": 0.6385,
+      "step": 4668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5047425713388728e-06,
+      "loss": 0.6053,
+      "step": 4669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5027096741455193e-06,
+      "loss": 0.5959,
+      "step": 4670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.500677326831768e-06,
+      "loss": 0.5987,
+      "step": 4671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4986455298451224e-06,
+      "loss": 0.5978,
+      "step": 4672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.496614283632962e-06,
+      "loss": 0.6366,
+      "step": 4673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.494583588642553e-06,
+      "loss": 0.5332,
+      "step": 4674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4925534453210302e-06,
+      "loss": 0.6124,
+      "step": 4675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4905238541154163e-06,
+      "loss": 0.6356,
+      "step": 4676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.488494815472606e-06,
+      "loss": 0.6187,
+      "step": 4677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4864663298393745e-06,
+      "loss": 0.5499,
+      "step": 4678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.484438397662375e-06,
+      "loss": 0.6222,
+      "step": 4679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4824110193881384e-06,
+      "loss": 0.6719,
+      "step": 4680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4803841954630724e-06,
+      "loss": 0.5707,
+      "step": 4681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.478357926333469e-06,
+      "loss": 0.6535,
+      "step": 4682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.476332212445491e-06,
+      "loss": 0.5359,
+      "step": 4683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4743070542451806e-06,
+      "loss": 0.6142,
+      "step": 4684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4722824521784596e-06,
+      "loss": 0.5338,
+      "step": 4685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4702584066911234e-06,
+      "loss": 0.6276,
+      "step": 4686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4682349182288494e-06,
+      "loss": 0.601,
+      "step": 4687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.466211987237193e-06,
+      "loss": 0.6049,
+      "step": 4688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.464189614161583e-06,
+      "loss": 0.6214,
+      "step": 4689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4621677994473254e-06,
+      "loss": 0.6328,
+      "step": 4690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.460146543539605e-06,
+      "loss": 0.5289,
+      "step": 4691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.458125846883482e-06,
+      "loss": 0.6227,
+      "step": 4692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4561057099238973e-06,
+      "loss": 0.5978,
+      "step": 4693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4540861331056647e-06,
+      "loss": 0.6065,
+      "step": 4694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.452067116873476e-06,
+      "loss": 0.6693,
+      "step": 4695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4500486616718988e-06,
+      "loss": 0.613,
+      "step": 4696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.448030767945378e-06,
+      "loss": 0.625,
+      "step": 4697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.446013436138233e-06,
+      "loss": 0.6028,
+      "step": 4698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.443996666694664e-06,
+      "loss": 0.6306,
+      "step": 4699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4419804600587417e-06,
+      "loss": 0.6123,
+      "step": 4700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4399648166744184e-06,
+      "loss": 0.6229,
+      "step": 4701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4379497369855186e-06,
+      "loss": 0.6271,
+      "step": 4702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.435935221435744e-06,
+      "loss": 0.5394,
+      "step": 4703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4339212704686692e-06,
+      "loss": 0.6323,
+      "step": 4704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4319078845277504e-06,
+      "loss": 0.6,
+      "step": 4705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4298950640563155e-06,
+      "loss": 0.628,
+      "step": 4706
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4278828094975678e-06,
+      "loss": 0.6298,
+      "step": 4707
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.425871121294586e-06,
+      "loss": 0.6086,
+      "step": 4708
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.423859999890325e-06,
+      "loss": 0.6107,
+      "step": 4709
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4218494457276135e-06,
+      "loss": 0.5287,
+      "step": 4710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4198394592491596e-06,
+      "loss": 0.6454,
+      "step": 4711
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4178300408975395e-06,
+      "loss": 0.6192,
+      "step": 4712
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.415821191115212e-06,
+      "loss": 0.6116,
+      "step": 4713
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4138129103445045e-06,
+      "loss": 0.6563,
+      "step": 4714
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4118051990276216e-06,
+      "loss": 0.6144,
+      "step": 4715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4097980576066406e-06,
+      "loss": 0.6164,
+      "step": 4716
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.407791486523518e-06,
+      "loss": 0.6134,
+      "step": 4717
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4057854862200803e-06,
+      "loss": 0.5146,
+      "step": 4718
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4037800571380297e-06,
+      "loss": 0.6113,
+      "step": 4719
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4017751997189425e-06,
+      "loss": 0.6227,
+      "step": 4720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.39977091440427e-06,
+      "loss": 0.6123,
+      "step": 4721
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3977672016353332e-06,
+      "loss": 0.6045,
+      "step": 4722
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.395764061853336e-06,
+      "loss": 0.6124,
+      "step": 4723
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3937614954993467e-06,
+      "loss": 0.5268,
+      "step": 4724
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3917595030143143e-06,
+      "loss": 0.5443,
+      "step": 4725
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3897580848390577e-06,
+      "loss": 0.6303,
+      "step": 4726
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.387757241414269e-06,
+      "loss": 0.628,
+      "step": 4727
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3857569731805143e-06,
+      "loss": 0.6455,
+      "step": 4728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3837572805782368e-06,
+      "loss": 0.6023,
+      "step": 4729
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3817581640477477e-06,
+      "loss": 0.6183,
+      "step": 4730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.379759624029234e-06,
+      "loss": 0.5726,
+      "step": 4731
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.377761660962754e-06,
+      "loss": 0.6085,
+      "step": 4732
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3757642752882422e-06,
+      "loss": 0.605,
+      "step": 4733
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3737674674455002e-06,
+      "loss": 0.6227,
+      "step": 4734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3717712378742113e-06,
+      "loss": 0.6456,
+      "step": 4735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3697755870139233e-06,
+      "loss": 0.5699,
+      "step": 4736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3677805153040583e-06,
+      "loss": 0.6446,
+      "step": 4737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.365786023183916e-06,
+      "loss": 0.6109,
+      "step": 4738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.363792111092662e-06,
+      "loss": 0.6225,
+      "step": 4739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3617987794693358e-06,
+      "loss": 0.6025,
+      "step": 4740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3598060287528536e-06,
+      "loss": 0.6082,
+      "step": 4741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3578138593819975e-06,
+      "loss": 0.5503,
+      "step": 4742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.355822271795425e-06,
+      "loss": 0.6655,
+      "step": 4743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.353831266431665e-06,
+      "loss": 0.5413,
+      "step": 4744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3518408437291155e-06,
+      "loss": 0.6321,
+      "step": 4745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3498510041260524e-06,
+      "loss": 0.5994,
+      "step": 4746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3478617480606185e-06,
+      "loss": 0.563,
+      "step": 4747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.345873075970828e-06,
+      "loss": 0.548,
+      "step": 4748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.343884988294566e-06,
+      "loss": 0.545,
+      "step": 4749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.341897485469595e-06,
+      "loss": 0.6463,
+      "step": 4750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3399105679335397e-06,
+      "loss": 0.5465,
+      "step": 4751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3379242361239047e-06,
+      "loss": 0.6307,
+      "step": 4752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3359384904780586e-06,
+      "loss": 0.5513,
+      "step": 4753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3339533314332445e-06,
+      "loss": 0.5811,
+      "step": 4754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3319687594265757e-06,
+      "loss": 0.6157,
+      "step": 4755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.329984774895036e-06,
+      "loss": 0.6247,
+      "step": 4756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3280013782754773e-06,
+      "loss": 0.6028,
+      "step": 4757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3260185700046295e-06,
+      "loss": 0.6068,
+      "step": 4758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.324036350519086e-06,
+      "loss": 0.5466,
+      "step": 4759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3220547202553117e-06,
+      "loss": 0.5905,
+      "step": 4760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3200736796496425e-06,
+      "loss": 0.553,
+      "step": 4761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3180932291382872e-06,
+      "loss": 0.5551,
+      "step": 4762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.316113369157319e-06,
+      "loss": 0.5304,
+      "step": 4763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3141341001426882e-06,
+      "loss": 0.62,
+      "step": 4764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3121554225302083e-06,
+      "loss": 0.592,
+      "step": 4765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3101773367555664e-06,
+      "loss": 0.6085,
+      "step": 4766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.308199843254317e-06,
+      "loss": 0.5448,
+      "step": 4767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3062229424618867e-06,
+      "loss": 0.6195,
+      "step": 4768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3042466348135674e-06,
+      "loss": 0.5646,
+      "step": 4769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3022709207445275e-06,
+      "loss": 0.6487,
+      "step": 4770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3002958006897985e-06,
+      "loss": 0.6494,
+      "step": 4771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2983212750842826e-06,
+      "loss": 0.6108,
+      "step": 4772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.296347344362752e-06,
+      "loss": 0.5401,
+      "step": 4773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.294374008959846e-06,
+      "loss": 0.6212,
+      "step": 4774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2924012693100756e-06,
+      "loss": 0.6797,
+      "step": 4775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2904291258478218e-06,
+      "loss": 0.6357,
+      "step": 4776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2884575790073287e-06,
+      "loss": 0.6455,
+      "step": 4777
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.286486629222714e-06,
+      "loss": 0.5982,
+      "step": 4778
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2845162769279607e-06,
+      "loss": 0.6371,
+      "step": 4779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2825465225569214e-06,
+      "loss": 0.5728,
+      "step": 4780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.280577366543317e-06,
+      "loss": 0.6422,
+      "step": 4781
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2786088093207398e-06,
+      "loss": 0.6164,
+      "step": 4782
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.276640851322645e-06,
+      "loss": 0.6079,
+      "step": 4783
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2746734929823596e-06,
+      "loss": 0.6074,
+      "step": 4784
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.272706734733076e-06,
+      "loss": 0.6124,
+      "step": 4785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.270740577007854e-06,
+      "loss": 0.6274,
+      "step": 4786
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.268775020239627e-06,
+      "loss": 0.6532,
+      "step": 4787
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.266810064861188e-06,
+      "loss": 0.6392,
+      "step": 4788
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2648457113052053e-06,
+      "loss": 0.6377,
+      "step": 4789
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.262881960004209e-06,
+      "loss": 0.5429,
+      "step": 4790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.260918811390598e-06,
+      "loss": 0.617,
+      "step": 4791
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2589562658966402e-06,
+      "loss": 0.6515,
+      "step": 4792
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.256994323954466e-06,
+      "loss": 0.6592,
+      "step": 4793
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2550329859960816e-06,
+      "loss": 0.64,
+      "step": 4794
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.253072252453352e-06,
+      "loss": 0.6459,
+      "step": 4795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.251112123758013e-06,
+      "loss": 0.6215,
+      "step": 4796
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.249152600341665e-06,
+      "loss": 0.6423,
+      "step": 4797
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2471936826357766e-06,
+      "loss": 0.6333,
+      "step": 4798
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2452353710716846e-06,
+      "loss": 0.6497,
+      "step": 4799
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.243277666080588e-06,
+      "loss": 0.5888,
+      "step": 4800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.241320568093558e-06,
+      "loss": 0.6426,
+      "step": 4801
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.239364077541527e-06,
+      "loss": 0.563,
+      "step": 4802
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.237408194855297e-06,
+      "loss": 0.5348,
+      "step": 4803
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2354529204655313e-06,
+      "loss": 0.5749,
+      "step": 4804
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.233498254802767e-06,
+      "loss": 0.6151,
+      "step": 4805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.231544198297401e-06,
+      "loss": 0.5654,
+      "step": 4806
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.229590751379699e-06,
+      "loss": 0.5698,
+      "step": 4807
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.22763791447979e-06,
+      "loss": 0.5228,
+      "step": 4808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2256856880276708e-06,
+      "loss": 0.5537,
+      "step": 4809
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2237340724532007e-06,
+      "loss": 0.5349,
+      "step": 4810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2217830681861114e-06,
+      "loss": 0.5467,
+      "step": 4811
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.219832675655992e-06,
+      "loss": 0.6249,
+      "step": 4812
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2178828952923025e-06,
+      "loss": 0.6325,
+      "step": 4813
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.215933727524366e-06,
+      "loss": 0.6169,
+      "step": 4814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.21398517278137e-06,
+      "loss": 0.5518,
+      "step": 4815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2120372314923657e-06,
+      "loss": 0.5646,
+      "step": 4816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2100899040862746e-06,
+      "loss": 0.6364,
+      "step": 4817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.208143190991879e-06,
+      "loss": 0.5628,
+      "step": 4818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2061970926378257e-06,
+      "loss": 0.602,
+      "step": 4819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2042516094526272e-06,
+      "loss": 0.5464,
+      "step": 4820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2023067418646604e-06,
+      "loss": 0.5948,
+      "step": 4821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2003624903021653e-06,
+      "loss": 0.6424,
+      "step": 4822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1984188551932513e-06,
+      "loss": 0.6163,
+      "step": 4823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1964758369658857e-06,
+      "loss": 0.6526,
+      "step": 4824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1945334360479015e-06,
+      "loss": 0.5301,
+      "step": 4825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1925916528670007e-06,
+      "loss": 0.6166,
+      "step": 4826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1906504878507426e-06,
+      "loss": 0.6449,
+      "step": 4827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1887099414265534e-06,
+      "loss": 0.6508,
+      "step": 4828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1867700140217245e-06,
+      "loss": 0.6414,
+      "step": 4829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1848307060634093e-06,
+      "loss": 0.626,
+      "step": 4830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1828920179786246e-06,
+      "loss": 0.6605,
+      "step": 4831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.18095395019425e-06,
+      "loss": 0.5414,
+      "step": 4832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1790165031370314e-06,
+      "loss": 0.6135,
+      "step": 4833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.177079677233573e-06,
+      "loss": 0.6774,
+      "step": 4834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1751434729103494e-06,
+      "loss": 0.6051,
+      "step": 4835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.173207890593693e-06,
+      "loss": 0.6343,
+      "step": 4836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.171272930709798e-06,
+      "loss": 0.6331,
+      "step": 4837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.169338593684729e-06,
+      "loss": 0.6109,
+      "step": 4838
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.167404879944406e-06,
+      "loss": 0.625,
+      "step": 4839
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1654717899146125e-06,
+      "loss": 0.5387,
+      "step": 4840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1635393240210014e-06,
+      "loss": 0.6167,
+      "step": 4841
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.16160748268908e-06,
+      "loss": 0.5381,
+      "step": 4842
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.159676266344222e-06,
+      "loss": 0.6084,
+      "step": 4843
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1577456754116626e-06,
+      "loss": 0.5442,
+      "step": 4844
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1558157103165e-06,
+      "loss": 0.5463,
+      "step": 4845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1538863714836923e-06,
+      "loss": 0.6105,
+      "step": 4846
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1519576593380654e-06,
+      "loss": 0.6465,
+      "step": 4847
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.150029574304301e-06,
+      "loss": 0.5713,
+      "step": 4848
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.148102116806944e-06,
+      "loss": 0.5956,
+      "step": 4849
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1461752872704044e-06,
+      "loss": 0.6236,
+      "step": 4850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1442490861189502e-06,
+      "loss": 0.6191,
+      "step": 4851
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1423235137767147e-06,
+      "loss": 0.545,
+      "step": 4852
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1403985706676894e-06,
+      "loss": 0.6196,
+      "step": 4853
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1384742572157288e-06,
+      "loss": 0.5329,
+      "step": 4854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1365505738445475e-06,
+      "loss": 0.6433,
+      "step": 4855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1346275209777227e-06,
+      "loss": 0.5226,
+      "step": 4856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1327050990386915e-06,
+      "loss": 0.5965,
+      "step": 4857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.130783308450755e-06,
+      "loss": 0.6531,
+      "step": 4858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1288621496370724e-06,
+      "loss": 0.5899,
+      "step": 4859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1269416230206645e-06,
+      "loss": 0.6304,
+      "step": 4860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.125021729024413e-06,
+      "loss": 0.6606,
+      "step": 4861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.123102468071058e-06,
+      "loss": 0.624,
+      "step": 4862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.121183840583207e-06,
+      "loss": 0.6103,
+      "step": 4863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1192658469833195e-06,
+      "loss": 0.6128,
+      "step": 4864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.117348487693723e-06,
+      "loss": 0.5998,
+      "step": 4865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1154317631366e-06,
+      "loss": 0.6336,
+      "step": 4866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1135156737339953e-06,
+      "loss": 0.5569,
+      "step": 4867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.111600219907813e-06,
+      "loss": 0.5091,
+      "step": 4868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1096854020798164e-06,
+      "loss": 0.6301,
+      "step": 4869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1077712206716338e-06,
+      "loss": 0.6105,
+      "step": 4870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.105857676104748e-06,
+      "loss": 0.5482,
+      "step": 4871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1039447688005034e-06,
+      "loss": 0.6254,
+      "step": 4872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.102032499180104e-06,
+      "loss": 0.6232,
+      "step": 4873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1001208676646113e-06,
+      "loss": 0.6345,
+      "step": 4874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0982098746749524e-06,
+      "loss": 0.6118,
+      "step": 4875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0962995206319057e-06,
+      "loss": 0.5202,
+      "step": 4876
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0943898059561172e-06,
+      "loss": 0.6009,
+      "step": 4877
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0924807310680857e-06,
+      "loss": 0.63,
+      "step": 4878
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0905722963881713e-06,
+      "loss": 0.6123,
+      "step": 4879
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0886645023365937e-06,
+      "loss": 0.634,
+      "step": 4880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0867573493334293e-06,
+      "loss": 0.6318,
+      "step": 4881
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.084850837798618e-06,
+      "loss": 0.5999,
+      "step": 4882
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.082944968151953e-06,
+      "loss": 0.6078,
+      "step": 4883
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0810397408130904e-06,
+      "loss": 0.582,
+      "step": 4884
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0791351562015428e-06,
+      "loss": 0.6318,
+      "step": 4885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.077231214736679e-06,
+      "loss": 0.6301,
+      "step": 4886
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0753279168377333e-06,
+      "loss": 0.6405,
+      "step": 4887
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0734252629237892e-06,
+      "loss": 0.629,
+      "step": 4888
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0715232534137976e-06,
+      "loss": 0.645,
+      "step": 4889
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0696218887265606e-06,
+      "loss": 0.532,
+      "step": 4890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0677211692807408e-06,
+      "loss": 0.5643,
+      "step": 4891
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.065821095494859e-06,
+      "loss": 0.6141,
+      "step": 4892
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.06392166778729e-06,
+      "loss": 0.5418,
+      "step": 4893
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.062022886576275e-06,
+      "loss": 0.6138,
+      "step": 4894
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0601247522799063e-06,
+      "loss": 0.5963,
+      "step": 4895
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0582272653161328e-06,
+      "loss": 0.6198,
+      "step": 4896
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0563304261027643e-06,
+      "loss": 0.6239,
+      "step": 4897
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0544342350574673e-06,
+      "loss": 0.6013,
+      "step": 4898
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0525386925977626e-06,
+      "loss": 0.645,
+      "step": 4899
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0506437991410327e-06,
+      "loss": 0.6505,
+      "step": 4900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.048749555104516e-06,
+      "loss": 0.5967,
+      "step": 4901
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.046855960905307e-06,
+      "loss": 0.529,
+      "step": 4902
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.044963016960356e-06,
+      "loss": 0.6122,
+      "step": 4903
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0430707236864687e-06,
+      "loss": 0.6082,
+      "step": 4904
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0411790815003155e-06,
+      "loss": 0.6291,
+      "step": 4905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0392880908184154e-06,
+      "loss": 0.6548,
+      "step": 4906
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0373977520571464e-06,
+      "loss": 0.6022,
+      "step": 4907
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.035508065632743e-06,
+      "loss": 0.6176,
+      "step": 4908
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.033619031961297e-06,
+      "loss": 0.6257,
+      "step": 4909
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0317306514587525e-06,
+      "loss": 0.5487,
+      "step": 4910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0298429245409185e-06,
+      "loss": 0.6358,
+      "step": 4911
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.027955851623451e-06,
+      "loss": 0.5482,
+      "step": 4912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0260694331218643e-06,
+      "loss": 0.6069,
+      "step": 4913
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0241836694515338e-06,
+      "loss": 0.6268,
+      "step": 4914
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0222985610276847e-06,
+      "loss": 0.6298,
+      "step": 4915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0204141082653975e-06,
+      "loss": 0.5535,
+      "step": 4916
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0185303115796155e-06,
+      "loss": 0.6497,
+      "step": 4917
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0166471713851306e-06,
+      "loss": 0.6311,
+      "step": 4918
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0147646880965922e-06,
+      "loss": 0.6258,
+      "step": 4919
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.012882862128505e-06,
+      "loss": 0.6459,
+      "step": 4920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.01100169389523e-06,
+      "loss": 0.5428,
+      "step": 4921
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0091211838109797e-06,
+      "loss": 0.6387,
+      "step": 4922
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0072413322898293e-06,
+      "loss": 0.6476,
+      "step": 4923
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.005362139745702e-06,
+      "loss": 0.5859,
+      "step": 4924
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.003483606592376e-06,
+      "loss": 0.4878,
+      "step": 4925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.00160573324349e-06,
+      "loss": 0.62,
+      "step": 4926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9997285201125328e-06,
+      "loss": 0.5947,
+      "step": 4927
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.997851967612846e-06,
+      "loss": 0.6296,
+      "step": 4928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9959760761576337e-06,
+      "loss": 0.5377,
+      "step": 4929
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9941008461599465e-06,
+      "loss": 0.6375,
+      "step": 4930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9922262780326925e-06,
+      "loss": 0.6237,
+      "step": 4931
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.990352372188633e-06,
+      "loss": 0.6226,
+      "step": 4932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9884791290403858e-06,
+      "loss": 0.6212,
+      "step": 4933
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.986606549000419e-06,
+      "loss": 0.6404,
+      "step": 4934
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9847346324810597e-06,
+      "loss": 0.6428,
+      "step": 4935
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9828633798944854e-06,
+      "loss": 0.6357,
+      "step": 4936
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.980992791652725e-06,
+      "loss": 0.6434,
+      "step": 4937
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9791228681676687e-06,
+      "loss": 0.6353,
+      "step": 4938
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.977253609851055e-06,
+      "loss": 0.5277,
+      "step": 4939
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.975385017114473e-06,
+      "loss": 0.5316,
+      "step": 4940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.973517090369374e-06,
+      "loss": 0.6217,
+      "step": 4941
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.971649830027056e-06,
+      "loss": 0.6217,
+      "step": 4942
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9697832364986715e-06,
+      "loss": 0.6107,
+      "step": 4943
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9679173101952264e-06,
+      "loss": 0.5492,
+      "step": 4944
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9660520515275803e-06,
+      "loss": 0.6275,
+      "step": 4945
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9641874609064443e-06,
+      "loss": 0.6297,
+      "step": 4946
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.962323538742386e-06,
+      "loss": 0.5637,
+      "step": 4947
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.960460285445822e-06,
+      "loss": 0.6288,
+      "step": 4948
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9585977014270234e-06,
+      "loss": 0.5455,
+      "step": 4949
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9567357870961113e-06,
+      "loss": 0.63,
+      "step": 4950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.954874542863065e-06,
+      "loss": 0.6225,
+      "step": 4951
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9530139691377093e-06,
+      "loss": 0.5467,
+      "step": 4952
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9511540663297284e-06,
+      "loss": 0.6522,
+      "step": 4953
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.949294834848654e-06,
+      "loss": 0.6154,
+      "step": 4954
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9474362751038706e-06,
+      "loss": 0.645,
+      "step": 4955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9455783875046153e-06,
+      "loss": 0.5787,
+      "step": 4956
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.943721172459976e-06,
+      "loss": 0.671,
+      "step": 4957
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9418646303788967e-06,
+      "loss": 0.6393,
+      "step": 4958
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9400087616701695e-06,
+      "loss": 0.634,
+      "step": 4959
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.938153566742438e-06,
+      "loss": 0.5261,
+      "step": 4960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9362990460042e-06,
+      "loss": 0.6522,
+      "step": 4961
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9344451998638e-06,
+      "loss": 0.5641,
+      "step": 4962
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9325920287294413e-06,
+      "loss": 0.6176,
+      "step": 4963
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9307395330091714e-06,
+      "loss": 0.635,
+      "step": 4964
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9288877131108957e-06,
+      "loss": 0.6452,
+      "step": 4965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9270365694423654e-06,
+      "loss": 0.6517,
+      "step": 4966
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.925186102411185e-06,
+      "loss": 0.5783,
+      "step": 4967
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9233363124248093e-06,
+      "loss": 0.6323,
+      "step": 4968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9214871998905425e-06,
+      "loss": 0.6176,
+      "step": 4969
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9196387652155463e-06,
+      "loss": 0.6106,
+      "step": 4970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9177910088068265e-06,
+      "loss": 0.5499,
+      "step": 4971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9159439310712397e-06,
+      "loss": 0.6074,
+      "step": 4972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.914097532415497e-06,
+      "loss": 0.612,
+      "step": 4973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9122518132461553e-06,
+      "loss": 0.5998,
+      "step": 4974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9104067739696277e-06,
+      "loss": 0.6188,
+      "step": 4975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9085624149921713e-06,
+      "loss": 0.54,
+      "step": 4976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9067187367198991e-06,
+      "loss": 0.5346,
+      "step": 4977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9048757395587708e-06,
+      "loss": 0.5309,
+      "step": 4978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.903033423914596e-06,
+      "loss": 0.6127,
+      "step": 4979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9011917901930354e-06,
+      "loss": 0.6329,
+      "step": 4980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.899350838799598e-06,
+      "loss": 0.6134,
+      "step": 4981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8975105701396468e-06,
+      "loss": 0.5815,
+      "step": 4982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.89567098461839e-06,
+      "loss": 0.6062,
+      "step": 4983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8938320826408862e-06,
+      "loss": 0.5484,
+      "step": 4984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.891993864612045e-06,
+      "loss": 0.5401,
+      "step": 4985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.890156330936624e-06,
+      "loss": 0.6226,
+      "step": 4986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.888319482019229e-06,
+      "loss": 0.6112,
+      "step": 4987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8864833182643206e-06,
+      "loss": 0.6126,
+      "step": 4988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8846478400762002e-06,
+      "loss": 0.5545,
+      "step": 4989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8828130478590274e-06,
+      "loss": 0.5998,
+      "step": 4990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.880978942016803e-06,
+      "loss": 0.6299,
+      "step": 4991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8791455229533806e-06,
+      "loss": 0.6779,
+      "step": 4992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8773127910724591e-06,
+      "loss": 0.5406,
+      "step": 4993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.875480746777593e-06,
+      "loss": 0.5478,
+      "step": 4994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8736493904721791e-06,
+      "loss": 0.591,
+      "step": 4995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8718187225594642e-06,
+      "loss": 0.5452,
+      "step": 4996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.869988743442544e-06,
+      "loss": 0.6333,
+      "step": 4997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.868159453524363e-06,
+      "loss": 0.5911,
+      "step": 4998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8663308532077118e-06,
+      "loss": 0.6045,
+      "step": 4999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8645029428952338e-06,
+      "loss": 0.5425,
+      "step": 5000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8626757229894139e-06,
+      "loss": 0.6474,
+      "step": 5001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8608491938925922e-06,
+      "loss": 0.6111,
+      "step": 5002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.859023356006952e-06,
+      "loss": 0.5452,
+      "step": 5003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8571982097345243e-06,
+      "loss": 0.6195,
+      "step": 5004
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8553737554771883e-06,
+      "loss": 0.6537,
+      "step": 5005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8535499936366741e-06,
+      "loss": 0.5544,
+      "step": 5006
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.851726924614556e-06,
+      "loss": 0.6082,
+      "step": 5007
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8499045488122553e-06,
+      "loss": 0.6173,
+      "step": 5008
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.848082866631042e-06,
+      "loss": 0.6281,
+      "step": 5009
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8462618784720326e-06,
+      "loss": 0.6041,
+      "step": 5010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8444415847361935e-06,
+      "loss": 0.6016,
+      "step": 5011
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8426219858243354e-06,
+      "loss": 0.6149,
+      "step": 5012
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8408030821371147e-06,
+      "loss": 0.6062,
+      "step": 5013
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.83898487407504e-06,
+      "loss": 0.6027,
+      "step": 5014
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8371673620384612e-06,
+      "loss": 0.6263,
+      "step": 5015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.835350546427576e-06,
+      "loss": 0.6406,
+      "step": 5016
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8335344276424338e-06,
+      "loss": 0.6024,
+      "step": 5017
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8317190060829242e-06,
+      "loss": 0.6174,
+      "step": 5018
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8299042821487861e-06,
+      "loss": 0.6233,
+      "step": 5019
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8280902562396046e-06,
+      "loss": 0.6083,
+      "step": 5020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.826276928754811e-06,
+      "loss": 0.6361,
+      "step": 5021
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8244643000936807e-06,
+      "loss": 0.6378,
+      "step": 5022
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.822652370655341e-06,
+      "loss": 0.6455,
+      "step": 5023
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.82084114083876e-06,
+      "loss": 0.5605,
+      "step": 5024
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8190306110427509e-06,
+      "loss": 0.5371,
+      "step": 5025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8172207816659793e-06,
+      "loss": 0.624,
+      "step": 5026
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8154116531069504e-06,
+      "loss": 0.6224,
+      "step": 5027
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8136032257640157e-06,
+      "loss": 0.6323,
+      "step": 5028
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8117955000353765e-06,
+      "loss": 0.6455,
+      "step": 5029
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8099884763190762e-06,
+      "loss": 0.6452,
+      "step": 5030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.808182155013003e-06,
+      "loss": 0.5845,
+      "step": 5031
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8063765365148927e-06,
+      "loss": 0.5223,
+      "step": 5032
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8045716212223247e-06,
+      "loss": 0.6248,
+      "step": 5033
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8027674095327225e-06,
+      "loss": 0.6243,
+      "step": 5034
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.80096390184336e-06,
+      "loss": 0.6086,
+      "step": 5035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7991610985513509e-06,
+      "loss": 0.5179,
+      "step": 5036
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7973590000536556e-06,
+      "loss": 0.6201,
+      "step": 5037
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7955576067470765e-06,
+      "loss": 0.6282,
+      "step": 5038
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7937569190282672e-06,
+      "loss": 0.6246,
+      "step": 5039
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7919569372937185e-06,
+      "loss": 0.5194,
+      "step": 5040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7901576619397725e-06,
+      "loss": 0.644,
+      "step": 5041
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7883590933626117e-06,
+      "loss": 0.5782,
+      "step": 5042
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7865612319582632e-06,
+      "loss": 0.6035,
+      "step": 5043
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7847640781225982e-06,
+      "loss": 0.5558,
+      "step": 5044
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7829676322513346e-06,
+      "loss": 0.6395,
+      "step": 5045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7811718947400297e-06,
+      "loss": 0.5207,
+      "step": 5046
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.779376865984092e-06,
+      "loss": 0.5627,
+      "step": 5047
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7775825463787672e-06,
+      "loss": 0.5132,
+      "step": 5048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7757889363191484e-06,
+      "loss": 0.6338,
+      "step": 5049
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773996036200169e-06,
+      "loss": 0.6468,
+      "step": 5050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7722038464166124e-06,
+      "loss": 0.6339,
+      "step": 5051
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7704123673630986e-06,
+      "loss": 0.5473,
+      "step": 5052
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7686215994340972e-06,
+      "loss": 0.6116,
+      "step": 5053
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7668315430239163e-06,
+      "loss": 0.5991,
+      "step": 5054
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7650421985267103e-06,
+      "loss": 0.6042,
+      "step": 5055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7632535663364752e-06,
+      "loss": 0.6354,
+      "step": 5056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7614656468470508e-06,
+      "loss": 0.6201,
+      "step": 5057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7596784404521182e-06,
+      "loss": 0.6214,
+      "step": 5058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7578919475452072e-06,
+      "loss": 0.6465,
+      "step": 5059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7561061685196845e-06,
+      "loss": 0.6365,
+      "step": 5060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7543211037687613e-06,
+      "loss": 0.6204,
+      "step": 5061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7525367536854908e-06,
+      "loss": 0.6474,
+      "step": 5062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7507531186627734e-06,
+      "loss": 0.6344,
+      "step": 5063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.748970199093345e-06,
+      "loss": 0.6168,
+      "step": 5064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7471879953697906e-06,
+      "loss": 0.6362,
+      "step": 5065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7454065078845333e-06,
+      "loss": 0.5491,
+      "step": 5066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7436257370298393e-06,
+      "loss": 0.6017,
+      "step": 5067
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7418456831978176e-06,
+      "loss": 0.633,
+      "step": 5068
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7400663467804174e-06,
+      "loss": 0.6021,
+      "step": 5069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7382877281694355e-06,
+      "loss": 0.6075,
+      "step": 5070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7365098277565046e-06,
+      "loss": 0.6262,
+      "step": 5071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7347326459331015e-06,
+      "loss": 0.638,
+      "step": 5072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.732956183090545e-06,
+      "loss": 0.631,
+      "step": 5073
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7311804396199954e-06,
+      "loss": 0.6296,
+      "step": 5074
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7294054159124524e-06,
+      "loss": 0.5133,
+      "step": 5075
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7276311123587631e-06,
+      "loss": 0.6043,
+      "step": 5076
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7258575293496089e-06,
+      "loss": 0.6107,
+      "step": 5077
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7240846672755196e-06,
+      "loss": 0.6281,
+      "step": 5078
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7223125265268597e-06,
+      "loss": 0.5369,
+      "step": 5079
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7205411074938395e-06,
+      "loss": 0.5521,
+      "step": 5080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7187704105665054e-06,
+      "loss": 0.6198,
+      "step": 5081
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.717000436134752e-06,
+      "loss": 0.58,
+      "step": 5082
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7152311845883096e-06,
+      "loss": 0.6383,
+      "step": 5083
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.71346265631675e-06,
+      "loss": 0.5891,
+      "step": 5084
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7116948517094868e-06,
+      "loss": 0.5902,
+      "step": 5085
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7099277711557731e-06,
+      "loss": 0.6451,
+      "step": 5086
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7081614150447019e-06,
+      "loss": 0.608,
+      "step": 5087
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7063957837652118e-06,
+      "loss": 0.6485,
+      "step": 5088
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7046308777060744e-06,
+      "loss": 0.6193,
+      "step": 5089
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7028666972559082e-06,
+      "loss": 0.5358,
+      "step": 5090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7011032428031677e-06,
+      "loss": 0.621,
+      "step": 5091
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6993405147361485e-06,
+      "loss": 0.5498,
+      "step": 5092
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6975785134429857e-06,
+      "loss": 0.5329,
+      "step": 5093
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6958172393116574e-06,
+      "loss": 0.5532,
+      "step": 5094
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.694056692729979e-06,
+      "loss": 0.5367,
+      "step": 5095
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.692296874085605e-06,
+      "loss": 0.6007,
+      "step": 5096
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6905377837660313e-06,
+      "loss": 0.5448,
+      "step": 5097
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6887794221585935e-06,
+      "loss": 0.6248,
+      "step": 5098
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6870217896504631e-06,
+      "loss": 0.6325,
+      "step": 5099
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.685264886628658e-06,
+      "loss": 0.6039,
+      "step": 5100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6835087134800288e-06,
+      "loss": 0.6448,
+      "step": 5101
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.681753270591271e-06,
+      "loss": 0.5327,
+      "step": 5102
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6799985583489143e-06,
+      "loss": 0.6277,
+      "step": 5103
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6782445771393303e-06,
+      "loss": 0.5546,
+      "step": 5104
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.676491327348727e-06,
+      "loss": 0.5962,
+      "step": 5105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6747388093631572e-06,
+      "loss": 0.6229,
+      "step": 5106
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6729870235685069e-06,
+      "loss": 0.6364,
+      "step": 5107
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.671235970350502e-06,
+      "loss": 0.6195,
+      "step": 5108
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6694856500947081e-06,
+      "loss": 0.6039,
+      "step": 5109
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6677360631865298e-06,
+      "loss": 0.6725,
+      "step": 5110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6659872100112069e-06,
+      "loss": 0.6233,
+      "step": 5111
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6642390909538241e-06,
+      "loss": 0.6142,
+      "step": 5112
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6624917063992985e-06,
+      "loss": 0.5851,
+      "step": 5113
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6607450567323868e-06,
+      "loss": 0.6381,
+      "step": 5114
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6589991423376872e-06,
+      "loss": 0.6239,
+      "step": 5115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6572539635996298e-06,
+      "loss": 0.5675,
+      "step": 5116
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6555095209024908e-06,
+      "loss": 0.6244,
+      "step": 5117
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6537658146303776e-06,
+      "loss": 0.5385,
+      "step": 5118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6520228451672382e-06,
+      "loss": 0.5343,
+      "step": 5119
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6502806128968568e-06,
+      "loss": 0.6245,
+      "step": 5120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6485391182028576e-06,
+      "loss": 0.6507,
+      "step": 5121
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6467983614686995e-06,
+      "loss": 0.6357,
+      "step": 5122
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6450583430776828e-06,
+      "loss": 0.6044,
+      "step": 5123
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.643319063412942e-06,
+      "loss": 0.5042,
+      "step": 5124
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6415805228574506e-06,
+      "loss": 0.6209,
+      "step": 5125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6398427217940161e-06,
+      "loss": 0.6216,
+      "step": 5126
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6381056606052891e-06,
+      "loss": 0.5499,
+      "step": 5127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6363693396737518e-06,
+      "loss": 0.6318,
+      "step": 5128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6346337593817274e-06,
+      "loss": 0.5501,
+      "step": 5129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6328989201113738e-06,
+      "loss": 0.6149,
+      "step": 5130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6311648222446852e-06,
+      "loss": 0.5895,
+      "step": 5131
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6294314661634941e-06,
+      "loss": 0.6313,
+      "step": 5132
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6276988522494685e-06,
+      "loss": 0.5984,
+      "step": 5133
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6259669808841123e-06,
+      "loss": 0.5274,
+      "step": 5134
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.62423585244877e-06,
+      "loss": 0.619,
+      "step": 5135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6225054673246181e-06,
+      "loss": 0.5925,
+      "step": 5136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6207758258926714e-06,
+      "loss": 0.5917,
+      "step": 5137
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.619046928533779e-06,
+      "loss": 0.5992,
+      "step": 5138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6173187756286301e-06,
+      "loss": 0.6452,
+      "step": 5139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6155913675577445e-06,
+      "loss": 0.5427,
+      "step": 5140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.613864704701485e-06,
+      "loss": 0.5389,
+      "step": 5141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.612138787440044e-06,
+      "loss": 0.6308,
+      "step": 5142
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6104136161534522e-06,
+      "loss": 0.6098,
+      "step": 5143
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6086891912215758e-06,
+      "loss": 0.6061,
+      "step": 5144
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6069655130241173e-06,
+      "loss": 0.6217,
+      "step": 5145
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6052425819406126e-06,
+      "loss": 0.6243,
+      "step": 5146
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6035203983504377e-06,
+      "loss": 0.5508,
+      "step": 5147
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.601798962632799e-06,
+      "loss": 0.6062,
+      "step": 5148
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6000782751667416e-06,
+      "loss": 0.6408,
+      "step": 5149
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.598358336331141e-06,
+      "loss": 0.5916,
+      "step": 5150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5966391465047164e-06,
+      "loss": 0.6529,
+      "step": 5151
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5949207060660138e-06,
+      "loss": 0.6088,
+      "step": 5152
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5932030153934191e-06,
+      "loss": 0.6428,
+      "step": 5153
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5914860748651512e-06,
+      "loss": 0.6325,
+      "step": 5154
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5897698848592635e-06,
+      "loss": 0.6198,
+      "step": 5155
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5880544457536451e-06,
+      "loss": 0.5964,
+      "step": 5156
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5863397579260193e-06,
+      "loss": 0.5412,
+      "step": 5157
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5846258217539423e-06,
+      "loss": 0.57,
+      "step": 5158
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5829126376148097e-06,
+      "loss": 0.5505,
+      "step": 5159
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5812002058858467e-06,
+      "loss": 0.6436,
+      "step": 5160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5794885269441152e-06,
+      "loss": 0.5541,
+      "step": 5161
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5777776011665096e-06,
+      "loss": 0.6181,
+      "step": 5162
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5760674289297589e-06,
+      "loss": 0.605,
+      "step": 5163
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5743580106104294e-06,
+      "loss": 0.6398,
+      "step": 5164
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5726493465849157e-06,
+      "loss": 0.65,
+      "step": 5165
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.570941437229453e-06,
+      "loss": 0.6533,
+      "step": 5166
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5692342829201041e-06,
+      "loss": 0.6497,
+      "step": 5167
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.567527884032769e-06,
+      "loss": 0.5315,
+      "step": 5168
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.565822240943179e-06,
+      "loss": 0.6415,
+      "step": 5169
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5641173540269029e-06,
+      "loss": 0.5452,
+      "step": 5170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5624132236593392e-06,
+      "loss": 0.6049,
+      "step": 5171
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.560709850215722e-06,
+      "loss": 0.6288,
+      "step": 5172
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5590072340711177e-06,
+      "loss": 0.6111,
+      "step": 5173
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5573053756004253e-06,
+      "loss": 0.6125,
+      "step": 5174
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5556042751783772e-06,
+      "loss": 0.6336,
+      "step": 5175
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5539039331795424e-06,
+      "loss": 0.5347,
+      "step": 5176
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5522043499783168e-06,
+      "loss": 0.5938,
+      "step": 5177
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.550505525948936e-06,
+      "loss": 0.6254,
+      "step": 5178
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.548807461465463e-06,
+      "loss": 0.6011,
+      "step": 5179
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5471101569017955e-06,
+      "loss": 0.5588,
+      "step": 5180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5454136126316622e-06,
+      "loss": 0.5771,
+      "step": 5181
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5437178290286298e-06,
+      "loss": 0.6257,
+      "step": 5182
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5420228064660915e-06,
+      "loss": 0.6118,
+      "step": 5183
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.540328545317275e-06,
+      "loss": 0.5375,
+      "step": 5184
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5386350459552407e-06,
+      "loss": 0.5428,
+      "step": 5185
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5369423087528819e-06,
+      "loss": 0.6313,
+      "step": 5186
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.53525033408292e-06,
+      "loss": 0.656,
+      "step": 5187
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5335591223179164e-06,
+      "loss": 0.6284,
+      "step": 5188
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5318686738302563e-06,
+      "loss": 0.633,
+      "step": 5189
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5301789889921631e-06,
+      "loss": 0.6387,
+      "step": 5190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5284900681756886e-06,
+      "loss": 0.5561,
+      "step": 5191
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.526801911752716e-06,
+      "loss": 0.6299,
+      "step": 5192
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5251145200949608e-06,
+      "loss": 0.5187,
+      "step": 5193
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5234278935739732e-06,
+      "loss": 0.6067,
+      "step": 5194
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.521742032561131e-06,
+      "loss": 0.5881,
+      "step": 5195
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.520056937427644e-06,
+      "loss": 0.6115,
+      "step": 5196
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5183726085445555e-06,
+      "loss": 0.6,
+      "step": 5197
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5166890462827378e-06,
+      "loss": 0.6396,
+      "step": 5198
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5150062510128938e-06,
+      "loss": 0.631,
+      "step": 5199
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5133242231055622e-06,
+      "loss": 0.5852,
+      "step": 5200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.511642962931108e-06,
+      "loss": 0.6354,
+      "step": 5201
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5099624708597276e-06,
+      "loss": 0.6411,
+      "step": 5202
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.508282747261452e-06,
+      "loss": 0.6051,
+      "step": 5203
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.506603792506139e-06,
+      "loss": 0.6623,
+      "step": 5204
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.504925606963477e-06,
+      "loss": 0.6018,
+      "step": 5205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5032481910029895e-06,
+      "loss": 0.5302,
+      "step": 5206
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.501571544994026e-06,
+      "loss": 0.5188,
+      "step": 5207
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4998956693057682e-06,
+      "loss": 0.5441,
+      "step": 5208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4982205643072273e-06,
+      "loss": 0.6178,
+      "step": 5209
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4965462303672468e-06,
+      "loss": 0.6158,
+      "step": 5210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.494872667854496e-06,
+      "loss": 0.6193,
+      "step": 5211
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4931998771374822e-06,
+      "loss": 0.6291,
+      "step": 5212
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.491527858584535e-06,
+      "loss": 0.6146,
+      "step": 5213
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4898566125638163e-06,
+      "loss": 0.614,
+      "step": 5214
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4881861394433222e-06,
+      "loss": 0.5885,
+      "step": 5215
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.486516439590872e-06,
+      "loss": 0.5792,
+      "step": 5216
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4848475133741176e-06,
+      "loss": 0.6047,
+      "step": 5217
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4831793611605422e-06,
+      "loss": 0.6161,
+      "step": 5218
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4815119833174574e-06,
+      "loss": 0.6108,
+      "step": 5219
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4798453802120027e-06,
+      "loss": 0.6285,
+      "step": 5220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4781795522111486e-06,
+      "loss": 0.6095,
+      "step": 5221
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4765144996816927e-06,
+      "loss": 0.633,
+      "step": 5222
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4748502229902678e-06,
+      "loss": 0.5634,
+      "step": 5223
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4731867225033293e-06,
+      "loss": 0.651,
+      "step": 5224
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4715239985871643e-06,
+      "loss": 0.6445,
+      "step": 5225
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4698620516078882e-06,
+      "loss": 0.6392,
+      "step": 5226
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4682008819314481e-06,
+      "loss": 0.5097,
+      "step": 5227
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4665404899236145e-06,
+      "loss": 0.62,
+      "step": 5228
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4648808759499937e-06,
+      "loss": 0.6066,
+      "step": 5229
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4632220403760156e-06,
+      "loss": 0.632,
+      "step": 5230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4615639835669393e-06,
+      "loss": 0.6081,
+      "step": 5231
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4599067058878536e-06,
+      "loss": 0.6163,
+      "step": 5232
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4582502077036747e-06,
+      "loss": 0.54,
+      "step": 5233
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4565944893791472e-06,
+      "loss": 0.5177,
+      "step": 5234
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4549395512788467e-06,
+      "loss": 0.6188,
+      "step": 5235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4532853937671736e-06,
+      "loss": 0.5725,
+      "step": 5236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.451632017208358e-06,
+      "loss": 0.5443,
+      "step": 5237
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4499794219664575e-06,
+      "loss": 0.6331,
+      "step": 5238
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4483276084053567e-06,
+      "loss": 0.6272,
+      "step": 5239
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4466765768887697e-06,
+      "loss": 0.6282,
+      "step": 5240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4450263277802402e-06,
+      "loss": 0.6026,
+      "step": 5241
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.443376861443136e-06,
+      "loss": 0.6517,
+      "step": 5242
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.441728178240654e-06,
+      "loss": 0.5307,
+      "step": 5243
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.440080278535818e-06,
+      "loss": 0.589,
+      "step": 5244
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4384331626914799e-06,
+      "loss": 0.6269,
+      "step": 5245
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.436786831070318e-06,
+      "loss": 0.6544,
+      "step": 5246
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4351412840348421e-06,
+      "loss": 0.6356,
+      "step": 5247
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4334965219473834e-06,
+      "loss": 0.6125,
+      "step": 5248
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4318525451701038e-06,
+      "loss": 0.6216,
+      "step": 5249
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4302093540649909e-06,
+      "loss": 0.646,
+      "step": 5250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4285669489938586e-06,
+      "loss": 0.5477,
+      "step": 5251
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4269253303183516e-06,
+      "loss": 0.641,
+      "step": 5252
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.425284498399936e-06,
+      "loss": 0.6416,
+      "step": 5253
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.42364445359991e-06,
+      "loss": 0.6138,
+      "step": 5254
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4220051962793952e-06,
+      "loss": 0.5388,
+      "step": 5255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4203667267993393e-06,
+      "loss": 0.6564,
+      "step": 5256
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4187290455205188e-06,
+      "loss": 0.5819,
+      "step": 5257
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.417092152803533e-06,
+      "loss": 0.6182,
+      "step": 5258
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4154560490088142e-06,
+      "loss": 0.5867,
+      "step": 5259
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4138207344966154e-06,
+      "loss": 0.6251,
+      "step": 5260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4121862096270162e-06,
+      "loss": 0.5486,
+      "step": 5261
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4105524747599241e-06,
+      "loss": 0.5356,
+      "step": 5262
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4089195302550707e-06,
+      "loss": 0.6181,
+      "step": 5263
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.407287376472018e-06,
+      "loss": 0.6256,
+      "step": 5264
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4056560137701469e-06,
+      "loss": 0.6368,
+      "step": 5265
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4040254425086714e-06,
+      "loss": 0.6329,
+      "step": 5266
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4023956630466268e-06,
+      "loss": 0.5323,
+      "step": 5267
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4007666757428745e-06,
+      "loss": 0.5416,
+      "step": 5268
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3991384809561004e-06,
+      "loss": 0.6112,
+      "step": 5269
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.397511079044821e-06,
+      "loss": 0.5582,
+      "step": 5270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.395884470367373e-06,
+      "loss": 0.6043,
+      "step": 5271
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3942586552819203e-06,
+      "loss": 0.6278,
+      "step": 5272
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3926336341464518e-06,
+      "loss": 0.6137,
+      "step": 5273
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.391009407318782e-06,
+      "loss": 0.552,
+      "step": 5274
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.389385975156548e-06,
+      "loss": 0.5532,
+      "step": 5275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.387763338017218e-06,
+      "loss": 0.6406,
+      "step": 5276
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.386141496258077e-06,
+      "loss": 0.6086,
+      "step": 5277
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3845204502362442e-06,
+      "loss": 0.5946,
+      "step": 5278
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.382900200308655e-06,
+      "loss": 0.6521,
+      "step": 5279
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3812807468320744e-06,
+      "loss": 0.6252,
+      "step": 5280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3796620901630886e-06,
+      "loss": 0.624,
+      "step": 5281
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3780442306581132e-06,
+      "loss": 0.6256,
+      "step": 5282
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3764271686733844e-06,
+      "loss": 0.5323,
+      "step": 5283
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3748109045649639e-06,
+      "loss": 0.6405,
+      "step": 5284
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3731954386887375e-06,
+      "loss": 0.5326,
+      "step": 5285
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3715807714004153e-06,
+      "loss": 0.5497,
+      "step": 5286
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3699669030555308e-06,
+      "loss": 0.537,
+      "step": 5287
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.368353834009445e-06,
+      "loss": 0.6286,
+      "step": 5288
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3667415646173388e-06,
+      "loss": 0.6171,
+      "step": 5289
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3651300952342172e-06,
+      "loss": 0.5525,
+      "step": 5290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3635194262149131e-06,
+      "loss": 0.5367,
+      "step": 5291
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3619095579140801e-06,
+      "loss": 0.6173,
+      "step": 5292
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3603004906861933e-06,
+      "loss": 0.5539,
+      "step": 5293
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3586922248855577e-06,
+      "loss": 0.6399,
+      "step": 5294
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3570847608662963e-06,
+      "loss": 0.6084,
+      "step": 5295
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3554780989823574e-06,
+      "loss": 0.6564,
+      "step": 5296
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3538722395875125e-06,
+      "loss": 0.6059,
+      "step": 5297
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3522671830353573e-06,
+      "loss": 0.6569,
+      "step": 5298
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3506629296793079e-06,
+      "loss": 0.5384,
+      "step": 5299
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3490594798726087e-06,
+      "loss": 0.6303,
+      "step": 5300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3474568339683226e-06,
+      "loss": 0.6277,
+      "step": 5301
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3458549923193353e-06,
+      "loss": 0.6504,
+      "step": 5302
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3442539552783612e-06,
+      "loss": 0.6237,
+      "step": 5303
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3426537231979309e-06,
+      "loss": 0.5725,
+      "step": 5304
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3410542964303986e-06,
+      "loss": 0.5413,
+      "step": 5305
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.339455675327947e-06,
+      "loss": 0.6204,
+      "step": 5306
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3378578602425745e-06,
+      "loss": 0.6413,
+      "step": 5307
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3362608515261055e-06,
+      "loss": 0.5921,
+      "step": 5308
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3346646495301862e-06,
+      "loss": 0.5464,
+      "step": 5309
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.333069254606285e-06,
+      "loss": 0.6031,
+      "step": 5310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3314746671056917e-06,
+      "loss": 0.5476,
+      "step": 5311
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.329880887379521e-06,
+      "loss": 0.5916,
+      "step": 5312
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3282879157787076e-06,
+      "loss": 0.6361,
+      "step": 5313
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.326695752654007e-06,
+      "loss": 0.587,
+      "step": 5314
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3251043983560013e-06,
+      "loss": 0.6281,
+      "step": 5315
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.32351385323509e-06,
+      "loss": 0.6312,
+      "step": 5316
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3219241176414948e-06,
+      "loss": 0.5336,
+      "step": 5317
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3203351919252634e-06,
+      "loss": 0.5851,
+      "step": 5318
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3187470764362604e-06,
+      "loss": 0.6058,
+      "step": 5319
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3171597715241736e-06,
+      "loss": 0.549,
+      "step": 5320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3155732775385127e-06,
+      "loss": 0.6084,
+      "step": 5321
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3139875948286074e-06,
+      "loss": 0.6465,
+      "step": 5322
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3124027237436127e-06,
+      "loss": 0.6254,
+      "step": 5323
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3108186646325e-06,
+      "loss": 0.612,
+      "step": 5324
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3092354178440653e-06,
+      "loss": 0.6308,
+      "step": 5325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3076529837269237e-06,
+      "loss": 0.628,
+      "step": 5326
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3060713626295108e-06,
+      "loss": 0.6348,
+      "step": 5327
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3044905549000858e-06,
+      "loss": 0.6431,
+      "step": 5328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3029105608867294e-06,
+      "loss": 0.528,
+      "step": 5329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3013313809373396e-06,
+      "loss": 0.624,
+      "step": 5330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2997530153996373e-06,
+      "loss": 0.6074,
+      "step": 5331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.298175464621163e-06,
+      "loss": 0.638,
+      "step": 5332
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2965987289492788e-06,
+      "loss": 0.6165,
+      "step": 5333
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2950228087311644e-06,
+      "loss": 0.5811,
+      "step": 5334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2934477043138272e-06,
+      "loss": 0.6017,
+      "step": 5335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2918734160440872e-06,
+      "loss": 0.6216,
+      "step": 5336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2902999442685888e-06,
+      "loss": 0.5271,
+      "step": 5337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2887272893337943e-06,
+      "loss": 0.6079,
+      "step": 5338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2871554515859874e-06,
+      "loss": 0.5563,
+      "step": 5339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2855844313712734e-06,
+      "loss": 0.6541,
+      "step": 5340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2840142290355739e-06,
+      "loss": 0.6313,
+      "step": 5341
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2824448449246358e-06,
+      "loss": 0.6407,
+      "step": 5342
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.28087627938402e-06,
+      "loss": 0.6176,
+      "step": 5343
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2793085327591103e-06,
+      "loss": 0.6259,
+      "step": 5344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2777416053951096e-06,
+      "loss": 0.6172,
+      "step": 5345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2761754976370394e-06,
+      "loss": 0.6452,
+      "step": 5346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.274610209829744e-06,
+      "loss": 0.6202,
+      "step": 5347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2730457423178839e-06,
+      "loss": 0.5649,
+      "step": 5348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2714820954459395e-06,
+      "loss": 0.6022,
+      "step": 5349
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2699192695582113e-06,
+      "loss": 0.6145,
+      "step": 5350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2683572649988168e-06,
+      "loss": 0.592,
+      "step": 5351
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.266796082111698e-06,
+      "loss": 0.5182,
+      "step": 5352
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2652357212406101e-06,
+      "loss": 0.62,
+      "step": 5353
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2636761827291315e-06,
+      "loss": 0.6405,
+      "step": 5354
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2621174669206577e-06,
+      "loss": 0.6051,
+      "step": 5355
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2605595741584015e-06,
+      "loss": 0.6284,
+      "step": 5356
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.259002504785397e-06,
+      "loss": 0.5437,
+      "step": 5357
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.257446259144494e-06,
+      "loss": 0.6466,
+      "step": 5358
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2558908375783675e-06,
+      "loss": 0.6161,
+      "step": 5359
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2543362404295035e-06,
+      "loss": 0.5981,
+      "step": 5360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2527824680402095e-06,
+      "loss": 0.6065,
+      "step": 5361
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2512295207526126e-06,
+      "loss": 0.6255,
+      "step": 5362
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.249677398908656e-06,
+      "loss": 0.6027,
+      "step": 5363
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2481261028501013e-06,
+      "loss": 0.6023,
+      "step": 5364
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2465756329185303e-06,
+      "loss": 0.6271,
+      "step": 5365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.245025989455343e-06,
+      "loss": 0.6205,
+      "step": 5366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2434771728017553e-06,
+      "loss": 0.5373,
+      "step": 5367
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2419291832988017e-06,
+      "loss": 0.6324,
+      "step": 5368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2403820212873347e-06,
+      "loss": 0.5176,
+      "step": 5369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2388356871080232e-06,
+      "loss": 0.6045,
+      "step": 5370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2372901811013581e-06,
+      "loss": 0.6238,
+      "step": 5371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2357455036076444e-06,
+      "loss": 0.6152,
+      "step": 5372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2342016549670038e-06,
+      "loss": 0.5241,
+      "step": 5373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2326586355193786e-06,
+      "loss": 0.6056,
+      "step": 5374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2311164456045255e-06,
+      "loss": 0.6359,
+      "step": 5375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.229575085562022e-06,
+      "loss": 0.6104,
+      "step": 5376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2280345557312607e-06,
+      "loss": 0.5945,
+      "step": 5377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2264948564514489e-06,
+      "loss": 0.5479,
+      "step": 5378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2249559880616173e-06,
+      "loss": 0.5233,
+      "step": 5379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.223417950900609e-06,
+      "loss": 0.6175,
+      "step": 5380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.221880745307083e-06,
+      "loss": 0.6143,
+      "step": 5381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2203443716195213e-06,
+      "loss": 0.6202,
+      "step": 5382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2188088301762158e-06,
+      "loss": 0.6493,
+      "step": 5383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2172741213152794e-06,
+      "loss": 0.5953,
+      "step": 5384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2157402453746398e-06,
+      "loss": 0.6087,
+      "step": 5385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2142072026920427e-06,
+      "loss": 0.5819,
+      "step": 5386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.212674993605047e-06,
+      "loss": 0.6341,
+      "step": 5387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2111436184510338e-06,
+      "loss": 0.6242,
+      "step": 5388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.209613077567196e-06,
+      "loss": 0.6358,
+      "step": 5389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2080833712905432e-06,
+      "loss": 0.5414,
+      "step": 5390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2065544999579043e-06,
+      "loss": 0.6193,
+      "step": 5391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2050264639059217e-06,
+      "loss": 0.5284,
+      "step": 5392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2034992634710524e-06,
+      "loss": 0.6297,
+      "step": 5393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.201972898989574e-06,
+      "loss": 0.5994,
+      "step": 5394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.200447370797576e-06,
+      "loss": 0.5492,
+      "step": 5395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1989226792309667e-06,
+      "loss": 0.5995,
+      "step": 5396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1973988246254664e-06,
+      "loss": 0.6064,
+      "step": 5397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1958758073166155e-06,
+      "loss": 0.6313,
+      "step": 5398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.194353627639765e-06,
+      "loss": 0.616,
+      "step": 5399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1928322859300883e-06,
+      "loss": 0.5418,
+      "step": 5400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1913117825225678e-06,
+      "loss": 0.5325,
+      "step": 5401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1897921177520038e-06,
+      "loss": 0.5399,
+      "step": 5402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1882732919530143e-06,
+      "loss": 0.5951,
+      "step": 5403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1867553054600289e-06,
+      "loss": 0.6602,
+      "step": 5404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.185238158607292e-06,
+      "loss": 0.6423,
+      "step": 5405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1837218517288684e-06,
+      "loss": 0.6392,
+      "step": 5406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1822063851586324e-06,
+      "loss": 0.618,
+      "step": 5407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1806917592302763e-06,
+      "loss": 0.5371,
+      "step": 5408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1791779742773053e-06,
+      "loss": 0.628,
+      "step": 5409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1776650306330413e-06,
+      "loss": 0.636,
+      "step": 5410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1761529286306177e-06,
+      "loss": 0.6152,
+      "step": 5411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1746416686029887e-06,
+      "loss": 0.6076,
+      "step": 5412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1731312508829178e-06,
+      "loss": 0.6424,
+      "step": 5413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1716216758029848e-06,
+      "loss": 0.6044,
+      "step": 5414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1701129436955816e-06,
+      "loss": 0.6181,
+      "step": 5415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1686050548929201e-06,
+      "loss": 0.6157,
+      "step": 5416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1670980097270208e-06,
+      "loss": 0.6309,
+      "step": 5417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1655918085297224e-06,
+      "loss": 0.6152,
+      "step": 5418
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.164086451632675e-06,
+      "loss": 0.6223,
+      "step": 5419
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1625819393673443e-06,
+      "loss": 0.5589,
+      "step": 5420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.16107827206501e-06,
+      "loss": 0.6224,
+      "step": 5421
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.159575450056764e-06,
+      "loss": 0.5583,
+      "step": 5422
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1580734736735134e-06,
+      "loss": 0.5536,
+      "step": 5423
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1565723432459807e-06,
+      "loss": 0.6365,
+      "step": 5424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1550720591047e-06,
+      "loss": 0.628,
+      "step": 5425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.153572621580019e-06,
+      "loss": 0.5423,
+      "step": 5426
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1520740310020995e-06,
+      "loss": 0.6471,
+      "step": 5427
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.150576287700918e-06,
+      "loss": 0.6522,
+      "step": 5428
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1490793920062614e-06,
+      "loss": 0.6105,
+      "step": 5429
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1475833442477353e-06,
+      "loss": 0.5567,
+      "step": 5430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1460881447547522e-06,
+      "loss": 0.622,
+      "step": 5431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1445937938565426e-06,
+      "loss": 0.5992,
+      "step": 5432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1431002918821471e-06,
+      "loss": 0.6255,
+      "step": 5433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1416076391604197e-06,
+      "loss": 0.5035,
+      "step": 5434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1401158360200314e-06,
+      "loss": 0.6438,
+      "step": 5435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1386248827894614e-06,
+      "loss": 0.6217,
+      "step": 5436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1371347797970029e-06,
+      "loss": 0.6162,
+      "step": 5437
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1356455273707633e-06,
+      "loss": 0.6382,
+      "step": 5438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1341571258386601e-06,
+      "loss": 0.6213,
+      "step": 5439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1326695755284279e-06,
+      "loss": 0.5449,
+      "step": 5440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1311828767676086e-06,
+      "loss": 0.5449,
+      "step": 5441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1296970298835613e-06,
+      "loss": 0.5363,
+      "step": 5442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1282120352034548e-06,
+      "loss": 0.6398,
+      "step": 5443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1267278930542707e-06,
+      "loss": 0.6506,
+      "step": 5444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1252446037628022e-06,
+      "loss": 0.5348,
+      "step": 5445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1237621676556548e-06,
+      "loss": 0.5863,
+      "step": 5446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.12228058505925e-06,
+      "loss": 0.6591,
+      "step": 5447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1207998562998156e-06,
+      "loss": 0.5376,
+      "step": 5448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.119319981703395e-06,
+      "loss": 0.5614,
+      "step": 5449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1178409615958419e-06,
+      "loss": 0.5499,
+      "step": 5450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1163627963028233e-06,
+      "loss": 0.5814,
+      "step": 5451
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.114885486149815e-06,
+      "loss": 0.6298,
+      "step": 5452
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1134090314621082e-06,
+      "loss": 0.631,
+      "step": 5453
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1119334325648063e-06,
+      "loss": 0.6434,
+      "step": 5454
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1104586897828196e-06,
+      "loss": 0.6328,
+      "step": 5455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.108984803440873e-06,
+      "loss": 0.5457,
+      "step": 5456
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1075117738635027e-06,
+      "loss": 0.5348,
+      "step": 5457
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1060396013750535e-06,
+      "loss": 0.5865,
+      "step": 5458
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1045682862996876e-06,
+      "loss": 0.6318,
+      "step": 5459
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1030978289613725e-06,
+      "loss": 0.6138,
+      "step": 5460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1016282296838887e-06,
+      "loss": 0.6251,
+      "step": 5461
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.100159488790829e-06,
+      "loss": 0.5594,
+      "step": 5462
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0986916066055953e-06,
+      "loss": 0.5684,
+      "step": 5463
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0972245834514006e-06,
+      "loss": 0.6323,
+      "step": 5464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0957584196512716e-06,
+      "loss": 0.6053,
+      "step": 5465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0942931155280411e-06,
+      "loss": 0.5306,
+      "step": 5466
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.092828671404358e-06,
+      "loss": 0.5739,
+      "step": 5467
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0913650876026776e-06,
+      "loss": 0.6189,
+      "step": 5468
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0899023644452673e-06,
+      "loss": 0.6275,
+      "step": 5469
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0884405022542026e-06,
+      "loss": 0.6108,
+      "step": 5470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0869795013513757e-06,
+      "loss": 0.6449,
+      "step": 5471
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0855193620584831e-06,
+      "loss": 0.5528,
+      "step": 5472
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0840600846970333e-06,
+      "loss": 0.6414,
+      "step": 5473
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0826016695883457e-06,
+      "loss": 0.5995,
+      "step": 5474
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0811441170535498e-06,
+      "loss": 0.6251,
+      "step": 5475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0796874274135827e-06,
+      "loss": 0.5566,
+      "step": 5476
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0782316009891964e-06,
+      "loss": 0.6167,
+      "step": 5477
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.076776638100948e-06,
+      "loss": 0.6415,
+      "step": 5478
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0753225390692085e-06,
+      "loss": 0.6403,
+      "step": 5479
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0738693042141562e-06,
+      "loss": 0.6411,
+      "step": 5480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.072416933855777e-06,
+      "loss": 0.6259,
+      "step": 5481
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.070965428313873e-06,
+      "loss": 0.6168,
+      "step": 5482
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0695147879080497e-06,
+      "loss": 0.5452,
+      "step": 5483
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0680650129577252e-06,
+      "loss": 0.5877,
+      "step": 5484
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0666161037821254e-06,
+      "loss": 0.5561,
+      "step": 5485
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0651680607002861e-06,
+      "loss": 0.5224,
+      "step": 5486
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0637208840310525e-06,
+      "loss": 0.6429,
+      "step": 5487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0622745740930807e-06,
+      "loss": 0.579,
+      "step": 5488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0608291312048335e-06,
+      "loss": 0.5348,
+      "step": 5489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0593845556845828e-06,
+      "loss": 0.5537,
+      "step": 5490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.057940847850412e-06,
+      "loss": 0.6132,
+      "step": 5491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0564980080202114e-06,
+      "loss": 0.5247,
+      "step": 5492
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0550560365116785e-06,
+      "loss": 0.6197,
+      "step": 5493
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0536149336423256e-06,
+      "loss": 0.627,
+      "step": 5494
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0521746997294674e-06,
+      "loss": 0.5264,
+      "step": 5495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0507353350902305e-06,
+      "loss": 0.6028,
+      "step": 5496
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0492968400415487e-06,
+      "loss": 0.6407,
+      "step": 5497
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0478592149001648e-06,
+      "loss": 0.596,
+      "step": 5498
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0464224599826301e-06,
+      "loss": 0.61,
+      "step": 5499
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0449865756053057e-06,
+      "loss": 0.5694,
+      "step": 5500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0435515620843589e-06,
+      "loss": 0.5605,
+      "step": 5501
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0421174197357664e-06,
+      "loss": 0.5345,
+      "step": 5502
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0406841488753107e-06,
+      "loss": 0.515,
+      "step": 5503
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0392517498185872e-06,
+      "loss": 0.5991,
+      "step": 5504
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0378202228809942e-06,
+      "loss": 0.6187,
+      "step": 5505
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.036389568377743e-06,
+      "loss": 0.6268,
+      "step": 5506
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0349597866238487e-06,
+      "loss": 0.6027,
+      "step": 5507
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0335308779341357e-06,
+      "loss": 0.6643,
+      "step": 5508
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.032102842623236e-06,
+      "loss": 0.5292,
+      "step": 5509
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0306756810055895e-06,
+      "loss": 0.6125,
+      "step": 5510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0292493933954417e-06,
+      "loss": 0.6026,
+      "step": 5511
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0278239801068518e-06,
+      "loss": 0.6434,
+      "step": 5512
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0263994414536793e-06,
+      "loss": 0.6082,
+      "step": 5513
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.024975777749595e-06,
+      "loss": 0.6493,
+      "step": 5514
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0235529893080742e-06,
+      "loss": 0.5626,
+      "step": 5515
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0221310764424042e-06,
+      "loss": 0.632,
+      "step": 5516
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.020710039465675e-06,
+      "loss": 0.617,
+      "step": 5517
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0192898786907861e-06,
+      "loss": 0.5406,
+      "step": 5518
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0178705944304447e-06,
+      "loss": 0.6347,
+      "step": 5519
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0164521869971612e-06,
+      "loss": 0.6111,
+      "step": 5520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.015034656703257e-06,
+      "loss": 0.5281,
+      "step": 5521
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0136180038608583e-06,
+      "loss": 0.5315,
+      "step": 5522
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0122022287818972e-06,
+      "loss": 0.6391,
+      "step": 5523
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0107873317781165e-06,
+      "loss": 0.6425,
+      "step": 5524
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0093733131610621e-06,
+      "loss": 0.6009,
+      "step": 5525
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0079601732420875e-06,
+      "loss": 0.6137,
+      "step": 5526
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0065479123323507e-06,
+      "loss": 0.6248,
+      "step": 5527
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0051365307428217e-06,
+      "loss": 0.5805,
+      "step": 5528
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.00372602878427e-06,
+      "loss": 0.6372,
+      "step": 5529
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0023164067672775e-06,
+      "loss": 0.6089,
+      "step": 5530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0009076650022287e-06,
+      "loss": 0.6206,
+      "step": 5531
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.994998037993142e-07,
+      "loss": 0.628,
+      "step": 5532
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.980928234685328e-07,
+      "loss": 0.6206,
+      "step": 5533
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.96686724319686e-07,
+      "loss": 0.5876,
+      "step": 5534
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.952815066623862e-07,
+      "loss": 0.6243,
+      "step": 5535
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.938771708060479e-07,
+      "loss": 0.5967,
+      "step": 5536
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.92473717059893e-07,
+      "loss": 0.6141,
+      "step": 5537
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.91071145732948e-07,
+      "loss": 0.5592,
+      "step": 5538
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.89669457134046e-07,
+      "loss": 0.5941,
+      "step": 5539
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.882686515718237e-07,
+      "loss": 0.6514,
+      "step": 5540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.868687293547274e-07,
+      "loss": 0.5868,
+      "step": 5541
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.85469690791007e-07,
+      "loss": 0.6327,
+      "step": 5542
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.840715361887176e-07,
+      "loss": 0.6395,
+      "step": 5543
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.82674265855718e-07,
+      "loss": 0.5536,
+      "step": 5544
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.812778800996747e-07,
+      "loss": 0.5331,
+      "step": 5545
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.798823792280571e-07,
+      "loss": 0.5857,
+      "step": 5546
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.784877635481438e-07,
+      "loss": 0.6393,
+      "step": 5547
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.770940333670147e-07,
+      "loss": 0.6097,
+      "step": 5548
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.757011889915557e-07,
+      "loss": 0.629,
+      "step": 5549
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.743092307284585e-07,
+      "loss": 0.584,
+      "step": 5550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.729181588842184e-07,
+      "loss": 0.6167,
+      "step": 5551
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.71527973765135e-07,
+      "loss": 0.6206,
+      "step": 5552
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.701386756773174e-07,
+      "loss": 0.5344,
+      "step": 5553
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.687502649266718e-07,
+      "loss": 0.6019,
+      "step": 5554
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.673627418189164e-07,
+      "loss": 0.6212,
+      "step": 5555
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.659761066595692e-07,
+      "loss": 0.6385,
+      "step": 5556
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.645903597539547e-07,
+      "loss": 0.5425,
+      "step": 5557
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.632055014071983e-07,
+      "loss": 0.5537,
+      "step": 5558
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.61821531924237e-07,
+      "loss": 0.5996,
+      "step": 5559
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.60438451609806e-07,
+      "loss": 0.558,
+      "step": 5560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.590562607684457e-07,
+      "loss": 0.5678,
+      "step": 5561
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.576749597045015e-07,
+      "loss": 0.617,
+      "step": 5562
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.562945487221237e-07,
+      "loss": 0.6187,
+      "step": 5563
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.549150281252633e-07,
+      "loss": 0.5902,
+      "step": 5564
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.53536398217681e-07,
+      "loss": 0.6227,
+      "step": 5565
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.521586593029347e-07,
+      "loss": 0.6317,
+      "step": 5566
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.507818116843931e-07,
+      "loss": 0.5391,
+      "step": 5567
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.494058556652224e-07,
+      "loss": 0.5502,
+      "step": 5568
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.480307915483961e-07,
+      "loss": 0.598,
+      "step": 5569
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.466566196366884e-07,
+      "loss": 0.6012,
+      "step": 5570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.452833402326816e-07,
+      "loss": 0.5355,
+      "step": 5571
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.439109536387575e-07,
+      "loss": 0.6221,
+      "step": 5572
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.425394601571026e-07,
+      "loss": 0.6187,
+      "step": 5573
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.411688600897067e-07,
+      "loss": 0.5122,
+      "step": 5574
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.397991537383627e-07,
+      "loss": 0.538,
+      "step": 5575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.384303414046658e-07,
+      "loss": 0.618,
+      "step": 5576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.370624233900183e-07,
+      "loss": 0.601,
+      "step": 5577
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.356953999956203e-07,
+      "loss": 0.62,
+      "step": 5578
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.343292715224789e-07,
+      "loss": 0.6182,
+      "step": 5579
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.329640382714028e-07,
+      "loss": 0.5355,
+      "step": 5580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.315997005430022e-07,
+      "loss": 0.6044,
+      "step": 5581
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.302362586376901e-07,
+      "loss": 0.6625,
+      "step": 5582
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.288737128556868e-07,
+      "loss": 0.6316,
+      "step": 5583
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.275120634970103e-07,
+      "loss": 0.529,
+      "step": 5584
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.261513108614823e-07,
+      "loss": 0.5186,
+      "step": 5585
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.247914552487286e-07,
+      "loss": 0.6518,
+      "step": 5586
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.234324969581743e-07,
+      "loss": 0.6257,
+      "step": 5587
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.220744362890522e-07,
+      "loss": 0.5365,
+      "step": 5588
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.207172735403924e-07,
+      "loss": 0.5294,
+      "step": 5589
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.193610090110305e-07,
+      "loss": 0.5365,
+      "step": 5590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.180056429996004e-07,
+      "loss": 0.6359,
+      "step": 5591
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.166511758045444e-07,
+      "loss": 0.592,
+      "step": 5592
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.152976077241005e-07,
+      "loss": 0.6216,
+      "step": 5593
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.139449390563137e-07,
+      "loss": 0.5574,
+      "step": 5594
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.125931700990276e-07,
+      "loss": 0.652,
+      "step": 5595
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.112423011498894e-07,
+      "loss": 0.6237,
+      "step": 5596
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.098923325063469e-07,
+      "loss": 0.6141,
+      "step": 5597
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.085432644656511e-07,
+      "loss": 0.6153,
+      "step": 5598
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.071950973248517e-07,
+      "loss": 0.6185,
+      "step": 5599
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.058478313808062e-07,
+      "loss": 0.535,
+      "step": 5600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.045014669301671e-07,
+      "loss": 0.5412,
+      "step": 5601
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.031560042693921e-07,
+      "loss": 0.6047,
+      "step": 5602
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.018114436947373e-07,
+      "loss": 0.596,
+      "step": 5603
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.004677855022654e-07,
+      "loss": 0.5715,
+      "step": 5604
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.99125029987834e-07,
+      "loss": 0.5075,
+      "step": 5605
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.97783177447108e-07,
+      "loss": 0.5458,
+      "step": 5606
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.964422281755486e-07,
+      "loss": 0.6074,
+      "step": 5607
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.951021824684214e-07,
+      "loss": 0.6177,
+      "step": 5608
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.937630406207909e-07,
+      "loss": 0.647,
+      "step": 5609
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.924248029275234e-07,
+      "loss": 0.605,
+      "step": 5610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.91087469683285e-07,
+      "loss": 0.6341,
+      "step": 5611
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.897510411825466e-07,
+      "loss": 0.6131,
+      "step": 5612
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.884155177195752e-07,
+      "loss": 0.603,
+      "step": 5613
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.870808995884406e-07,
+      "loss": 0.6182,
+      "step": 5614
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.85747187083012e-07,
+      "loss": 0.597,
+      "step": 5615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.844143804969624e-07,
+      "loss": 0.6308,
+      "step": 5616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.830824801237609e-07,
+      "loss": 0.6122,
+      "step": 5617
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.817514862566817e-07,
+      "loss": 0.5756,
+      "step": 5618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.804213991887956e-07,
+      "loss": 0.6261,
+      "step": 5619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.790922192129759e-07,
+      "loss": 0.6304,
+      "step": 5620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.77763946621894e-07,
+      "loss": 0.6438,
+      "step": 5621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.764365817080245e-07,
+      "loss": 0.602,
+      "step": 5622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.751101247636379e-07,
+      "loss": 0.6151,
+      "step": 5623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.737845760808111e-07,
+      "loss": 0.6084,
+      "step": 5624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.724599359514152e-07,
+      "loss": 0.6563,
+      "step": 5625
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.711362046671234e-07,
+      "loss": 0.5136,
+      "step": 5626
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.698133825194094e-07,
+      "loss": 0.634,
+      "step": 5627
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.684914697995445e-07,
+      "loss": 0.5985,
+      "step": 5628
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.671704667986036e-07,
+      "loss": 0.5972,
+      "step": 5629
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.658503738074564e-07,
+      "loss": 0.6236,
+      "step": 5630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.645311911167781e-07,
+      "loss": 0.6046,
+      "step": 5631
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.63212919017038e-07,
+      "loss": 0.6285,
+      "step": 5632
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.618955577985077e-07,
+      "loss": 0.6297,
+      "step": 5633
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.605791077512576e-07,
+      "loss": 0.6003,
+      "step": 5634
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.592635691651563e-07,
+      "loss": 0.5498,
+      "step": 5635
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.579489423298748e-07,
+      "loss": 0.583,
+      "step": 5636
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.566352275348805e-07,
+      "loss": 0.6403,
+      "step": 5637
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.553224250694414e-07,
+      "loss": 0.6518,
+      "step": 5638
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.540105352226235e-07,
+      "loss": 0.6004,
+      "step": 5639
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.526995582832914e-07,
+      "loss": 0.616,
+      "step": 5640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.513894945401125e-07,
+      "loss": 0.5891,
+      "step": 5641
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.500803442815475e-07,
+      "loss": 0.604,
+      "step": 5642
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.487721077958616e-07,
+      "loss": 0.6073,
+      "step": 5643
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.474647853711149e-07,
+      "loss": 0.6155,
+      "step": 5644
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.461583772951676e-07,
+      "loss": 0.6211,
+      "step": 5645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.448528838556763e-07,
+      "loss": 0.6195,
+      "step": 5646
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.435483053401011e-07,
+      "loss": 0.5254,
+      "step": 5647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.422446420356966e-07,
+      "loss": 0.6141,
+      "step": 5648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.409418942295172e-07,
+      "loss": 0.6007,
+      "step": 5649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.396400622084161e-07,
+      "loss": 0.6056,
+      "step": 5650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.383391462590429e-07,
+      "loss": 0.6431,
+      "step": 5651
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.370391466678468e-07,
+      "loss": 0.6132,
+      "step": 5652
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.357400637210777e-07,
+      "loss": 0.619,
+      "step": 5653
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.344418977047786e-07,
+      "loss": 0.5466,
+      "step": 5654
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.331446489047956e-07,
+      "loss": 0.6305,
+      "step": 5655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.318483176067699e-07,
+      "loss": 0.5808,
+      "step": 5656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.305529040961413e-07,
+      "loss": 0.6179,
+      "step": 5657
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.292584086581457e-07,
+      "loss": 0.6145,
+      "step": 5658
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.279648315778216e-07,
+      "loss": 0.5898,
+      "step": 5659
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.266721731400012e-07,
+      "loss": 0.6174,
+      "step": 5660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.253804336293153e-07,
+      "loss": 0.6433,
+      "step": 5661
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.240896133301929e-07,
+      "loss": 0.6013,
+      "step": 5662
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.227997125268605e-07,
+      "loss": 0.6219,
+      "step": 5663
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.215107315033405e-07,
+      "loss": 0.5738,
+      "step": 5664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.202226705434568e-07,
+      "loss": 0.6142,
+      "step": 5665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.18935529930826e-07,
+      "loss": 0.5312,
+      "step": 5666
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.176493099488664e-07,
+      "loss": 0.6111,
+      "step": 5667
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.163640108807897e-07,
+      "loss": 0.5853,
+      "step": 5668
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.150796330096077e-07,
+      "loss": 0.5902,
+      "step": 5669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.137961766181263e-07,
+      "loss": 0.6157,
+      "step": 5670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.125136419889524e-07,
+      "loss": 0.6289,
+      "step": 5671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.112320294044873e-07,
+      "loss": 0.6127,
+      "step": 5672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.099513391469304e-07,
+      "loss": 0.587,
+      "step": 5673
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.086715714982767e-07,
+      "loss": 0.5227,
+      "step": 5674
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.073927267403186e-07,
+      "loss": 0.6068,
+      "step": 5675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.061148051546458e-07,
+      "loss": 0.6421,
+      "step": 5676
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.048378070226454e-07,
+      "loss": 0.6407,
+      "step": 5677
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.035617326254996e-07,
+      "loss": 0.612,
+      "step": 5678
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.022865822441872e-07,
+      "loss": 0.545,
+      "step": 5679
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.010123561594857e-07,
+      "loss": 0.6023,
+      "step": 5680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.997390546519668e-07,
+      "loss": 0.6256,
+      "step": 5681
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.984666780019984e-07,
+      "loss": 0.5608,
+      "step": 5682
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.971952264897476e-07,
+      "loss": 0.619,
+      "step": 5683
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.959247003951753e-07,
+      "loss": 0.6466,
+      "step": 5684
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.946550999980391e-07,
+      "loss": 0.6009,
+      "step": 5685
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.933864255778928e-07,
+      "loss": 0.6073,
+      "step": 5686
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.921186774140866e-07,
+      "loss": 0.6271,
+      "step": 5687
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.908518557857653e-07,
+      "loss": 0.5184,
+      "step": 5688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.895859609718737e-07,
+      "loss": 0.6304,
+      "step": 5689
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.883209932511477e-07,
+      "loss": 0.6179,
+      "step": 5690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.870569529021211e-07,
+      "loss": 0.547,
+      "step": 5691
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.857938402031251e-07,
+      "loss": 0.539,
+      "step": 5692
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.845316554322835e-07,
+      "loss": 0.5523,
+      "step": 5693
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.832703988675195e-07,
+      "loss": 0.6051,
+      "step": 5694
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.82010070786548e-07,
+      "loss": 0.6491,
+      "step": 5695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.807506714668817e-07,
+      "loss": 0.5421,
+      "step": 5696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.794922011858292e-07,
+      "loss": 0.5954,
+      "step": 5697
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.782346602204921e-07,
+      "loss": 0.5488,
+      "step": 5698
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.76978048847769e-07,
+      "loss": 0.5455,
+      "step": 5699
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.757223673443559e-07,
+      "loss": 0.5148,
+      "step": 5700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.744676159867403e-07,
+      "loss": 0.6209,
+      "step": 5701
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.73213795051207e-07,
+      "loss": 0.6014,
+      "step": 5702
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.719609048138344e-07,
+      "loss": 0.5306,
+      "step": 5703
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.70708945550499e-07,
+      "loss": 0.6164,
+      "step": 5704
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.694579175368677e-07,
+      "loss": 0.6107,
+      "step": 5705
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.682078210484078e-07,
+      "loss": 0.6366,
+      "step": 5706
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.669586563603782e-07,
+      "loss": 0.6121,
+      "step": 5707
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.657104237478314e-07,
+      "loss": 0.5893,
+      "step": 5708
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.644631234856175e-07,
+      "loss": 0.634,
+      "step": 5709
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.632167558483805e-07,
+      "loss": 0.5356,
+      "step": 5710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.619713211105561e-07,
+      "loss": 0.5831,
+      "step": 5711
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.607268195463813e-07,
+      "loss": 0.6021,
+      "step": 5712
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.594832514298811e-07,
+      "loss": 0.6363,
+      "step": 5713
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.582406170348777e-07,
+      "loss": 0.5265,
+      "step": 5714
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.569989166349872e-07,
+      "loss": 0.6012,
+      "step": 5715
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.557581505036199e-07,
+      "loss": 0.6098,
+      "step": 5716
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.545183189139821e-07,
+      "loss": 0.519,
+      "step": 5717
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.532794221390705e-07,
+      "loss": 0.6601,
+      "step": 5718
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.520414604516818e-07,
+      "loss": 0.6079,
+      "step": 5719
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.508044341244014e-07,
+      "loss": 0.6009,
+      "step": 5720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.495683434296113e-07,
+      "loss": 0.6159,
+      "step": 5721
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.483331886394863e-07,
+      "loss": 0.5524,
+      "step": 5722
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.470989700259951e-07,
+      "loss": 0.65,
+      "step": 5723
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.458656878609028e-07,
+      "loss": 0.5879,
+      "step": 5724
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.446333424157653e-07,
+      "loss": 0.6205,
+      "step": 5725
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.434019339619336e-07,
+      "loss": 0.6266,
+      "step": 5726
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.421714627705523e-07,
+      "loss": 0.6205,
+      "step": 5727
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.40941929112558e-07,
+      "loss": 0.6021,
+      "step": 5728
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.397133332586848e-07,
+      "loss": 0.6339,
+      "step": 5729
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.384856754794551e-07,
+      "loss": 0.6028,
+      "step": 5730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.372589560451904e-07,
+      "loss": 0.5999,
+      "step": 5731
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.360331752260014e-07,
+      "loss": 0.6242,
+      "step": 5732
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.348083332917927e-07,
+      "loss": 0.6089,
+      "step": 5733
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.335844305122636e-07,
+      "loss": 0.6194,
+      "step": 5734
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.323614671569045e-07,
+      "loss": 0.5978,
+      "step": 5735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.311394434950031e-07,
+      "loss": 0.5414,
+      "step": 5736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.29918359795635e-07,
+      "loss": 0.605,
+      "step": 5737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.286982163276724e-07,
+      "loss": 0.5993,
+      "step": 5738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.274790133597792e-07,
+      "loss": 0.6098,
+      "step": 5739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.262607511604103e-07,
+      "loss": 0.6398,
+      "step": 5740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.250434299978182e-07,
+      "loss": 0.5098,
+      "step": 5741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.238270501400435e-07,
+      "loss": 0.5247,
+      "step": 5742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.226116118549236e-07,
+      "loss": 0.6383,
+      "step": 5743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.213971154100852e-07,
+      "loss": 0.5192,
+      "step": 5744
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.201835610729491e-07,
+      "loss": 0.5406,
+      "step": 5745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.189709491107272e-07,
+      "loss": 0.612,
+      "step": 5746
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.17759279790427e-07,
+      "loss": 0.5661,
+      "step": 5747
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.165485533788457e-07,
+      "loss": 0.545,
+      "step": 5748
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.15338770142574e-07,
+      "loss": 0.6113,
+      "step": 5749
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.141299303479948e-07,
+      "loss": 0.6054,
+      "step": 5750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.129220342612825e-07,
+      "loss": 0.6148,
+      "step": 5751
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.117150821484031e-07,
+      "loss": 0.6046,
+      "step": 5752
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.105090742751197e-07,
+      "loss": 0.5308,
+      "step": 5753
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.093040109069809e-07,
+      "loss": 0.5354,
+      "step": 5754
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.080998923093302e-07,
+      "loss": 0.5252,
+      "step": 5755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.068967187473047e-07,
+      "loss": 0.6039,
+      "step": 5756
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.05694490485831e-07,
+      "loss": 0.6342,
+      "step": 5757
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.044932077896271e-07,
+      "loss": 0.6343,
+      "step": 5758
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.032928709232062e-07,
+      "loss": 0.5995,
+      "step": 5759
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.0209348015087e-07,
+      "loss": 0.6032,
+      "step": 5760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.008950357367134e-07,
+      "loss": 0.6332,
+      "step": 5761
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.99697537944622e-07,
+      "loss": 0.6411,
+      "step": 5762
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.98500987038273e-07,
+      "loss": 0.6317,
+      "step": 5763
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.973053832811356e-07,
+      "loss": 0.5959,
+      "step": 5764
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.961107269364714e-07,
+      "loss": 0.5676,
+      "step": 5765
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.949170182673321e-07,
+      "loss": 0.6232,
+      "step": 5766
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.937242575365594e-07,
+      "loss": 0.5439,
+      "step": 5767
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.925324450067906e-07,
+      "loss": 0.6091,
+      "step": 5768
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9134158094045e-07,
+      "loss": 0.5183,
+      "step": 5769
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.901516655997536e-07,
+      "loss": 0.5946,
+      "step": 5770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.889626992467114e-07,
+      "loss": 0.6112,
+      "step": 5771
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.87774682143122e-07,
+      "loss": 0.6142,
+      "step": 5772
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.865876145505746e-07,
+      "loss": 0.6089,
+      "step": 5773
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.854014967304507e-07,
+      "loss": 0.6097,
+      "step": 5774
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.842163289439224e-07,
+      "loss": 0.6391,
+      "step": 5775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.830321114519511e-07,
+      "loss": 0.6313,
+      "step": 5776
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.818488445152921e-07,
+      "loss": 0.6336,
+      "step": 5777
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.806665283944891e-07,
+      "loss": 0.6448,
+      "step": 5778
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.79485163349875e-07,
+      "loss": 0.5478,
+      "step": 5779
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.783047496415779e-07,
+      "loss": 0.6111,
+      "step": 5780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.771252875295126e-07,
+      "loss": 0.6117,
+      "step": 5781
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.759467772733841e-07,
+      "loss": 0.5865,
+      "step": 5782
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.747692191326915e-07,
+      "loss": 0.628,
+      "step": 5783
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.735926133667204e-07,
+      "loss": 0.6251,
+      "step": 5784
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.724169602345487e-07,
+      "loss": 0.6062,
+      "step": 5785
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.712422599950441e-07,
+      "loss": 0.6126,
+      "step": 5786
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.70068512906864e-07,
+      "loss": 0.606,
+      "step": 5787
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.688957192284562e-07,
+      "loss": 0.6309,
+      "step": 5788
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.677238792180596e-07,
+      "loss": 0.6302,
+      "step": 5789
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.665529931337017e-07,
+      "loss": 0.6199,
+      "step": 5790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.653830612331996e-07,
+      "loss": 0.5855,
+      "step": 5791
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.642140837741634e-07,
+      "loss": 0.6109,
+      "step": 5792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.630460610139893e-07,
+      "loss": 0.6531,
+      "step": 5793
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.618789932098646e-07,
+      "loss": 0.637,
+      "step": 5794
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.607128806187674e-07,
+      "loss": 0.6237,
+      "step": 5795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.595477234974646e-07,
+      "loss": 0.5579,
+      "step": 5796
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.583835221025125e-07,
+      "loss": 0.5821,
+      "step": 5797
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.572202766902569e-07,
+      "loss": 0.6487,
+      "step": 5798
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.560579875168321e-07,
+      "loss": 0.6077,
+      "step": 5799
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.548966548381663e-07,
+      "loss": 0.6466,
+      "step": 5800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.537362789099716e-07,
+      "loss": 0.5329,
+      "step": 5801
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.52576859987753e-07,
+      "loss": 0.6147,
+      "step": 5802
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.514183983268024e-07,
+      "loss": 0.6277,
+      "step": 5803
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.502608941822014e-07,
+      "loss": 0.6021,
+      "step": 5804
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.491043478088238e-07,
+      "loss": 0.6313,
+      "step": 5805
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.479487594613276e-07,
+      "loss": 0.5429,
+      "step": 5806
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.467941293941654e-07,
+      "loss": 0.6046,
+      "step": 5807
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.456404578615732e-07,
+      "loss": 0.6274,
+      "step": 5808
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.444877451175797e-07,
+      "loss": 0.596,
+      "step": 5809
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.433359914160009e-07,
+      "loss": 0.6134,
+      "step": 5810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.421851970104409e-07,
+      "loss": 0.5384,
+      "step": 5811
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.410353621542959e-07,
+      "loss": 0.6083,
+      "step": 5812
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.398864871007476e-07,
+      "loss": 0.5845,
+      "step": 5813
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.387385721027672e-07,
+      "loss": 0.559,
+      "step": 5814
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.375916174131152e-07,
+      "loss": 0.6331,
+      "step": 5815
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.36445623284338e-07,
+      "loss": 0.5917,
+      "step": 5816
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.353005899687758e-07,
+      "loss": 0.6185,
+      "step": 5817
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.341565177185516e-07,
+      "loss": 0.5974,
+      "step": 5818
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.330134067855814e-07,
+      "loss": 0.6184,
+      "step": 5819
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.318712574215657e-07,
+      "loss": 0.6208,
+      "step": 5820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.307300698779956e-07,
+      "loss": 0.6167,
+      "step": 5821
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.2958984440615e-07,
+      "loss": 0.6012,
+      "step": 5822
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.28450581257094e-07,
+      "loss": 0.5459,
+      "step": 5823
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.273122806816845e-07,
+      "loss": 0.6024,
+      "step": 5824
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.261749429305641e-07,
+      "loss": 0.6027,
+      "step": 5825
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.250385682541638e-07,
+      "loss": 0.5842,
+      "step": 5826
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.239031569027015e-07,
+      "loss": 0.5805,
+      "step": 5827
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.227687091261842e-07,
+      "loss": 0.5928,
+      "step": 5828
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.216352251744079e-07,
+      "loss": 0.5417,
+      "step": 5829
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.205027052969526e-07,
+      "loss": 0.6063,
+      "step": 5830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.193711497431915e-07,
+      "loss": 0.5512,
+      "step": 5831
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.182405587622803e-07,
+      "loss": 0.5976,
+      "step": 5832
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.171109326031654e-07,
+      "loss": 0.5356,
+      "step": 5833
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.159822715145785e-07,
+      "loss": 0.5452,
+      "step": 5834
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.148545757450402e-07,
+      "loss": 0.593,
+      "step": 5835
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.137278455428603e-07,
+      "loss": 0.6392,
+      "step": 5836
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.126020811561323e-07,
+      "loss": 0.5409,
+      "step": 5837
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.114772828327392e-07,
+      "loss": 0.6037,
+      "step": 5838
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.103534508203512e-07,
+      "loss": 0.5975,
+      "step": 5839
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.092305853664249e-07,
+      "loss": 0.543,
+      "step": 5840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.08108686718204e-07,
+      "loss": 0.5494,
+      "step": 5841
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.069877551227221e-07,
+      "loss": 0.6042,
+      "step": 5842
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.058677908267951e-07,
+      "loss": 0.5683,
+      "step": 5843
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.047487940770308e-07,
+      "loss": 0.5326,
+      "step": 5844
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.036307651198209e-07,
+      "loss": 0.6279,
+      "step": 5845
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.025137042013446e-07,
+      "loss": 0.6518,
+      "step": 5846
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.013976115675668e-07,
+      "loss": 0.5935,
+      "step": 5847
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.002824874642427e-07,
+      "loss": 0.6199,
+      "step": 5848
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.991683321369107e-07,
+      "loss": 0.6067,
+      "step": 5849
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.980551458308981e-07,
+      "loss": 0.6143,
+      "step": 5850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.969429287913176e-07,
+      "loss": 0.5187,
+      "step": 5851
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.958316812630672e-07,
+      "loss": 0.6177,
+      "step": 5852
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.94721403490835e-07,
+      "loss": 0.5344,
+      "step": 5853
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.936120957190939e-07,
+      "loss": 0.638,
+      "step": 5854
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.925037581921006e-07,
+      "loss": 0.6123,
+      "step": 5855
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.913963911539033e-07,
+      "loss": 0.633,
+      "step": 5856
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.902899948483326e-07,
+      "loss": 0.6251,
+      "step": 5857
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.891845695190046e-07,
+      "loss": 0.6233,
+      "step": 5858
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.880801154093269e-07,
+      "loss": 0.5938,
+      "step": 5859
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.869766327624876e-07,
+      "loss": 0.6212,
+      "step": 5860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.858741218214637e-07,
+      "loss": 0.6449,
+      "step": 5861
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.847725828290174e-07,
+      "loss": 0.5561,
+      "step": 5862
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.836720160276971e-07,
+      "loss": 0.5988,
+      "step": 5863
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.825724216598361e-07,
+      "loss": 0.5795,
+      "step": 5864
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.814737999675568e-07,
+      "loss": 0.5381,
+      "step": 5865
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.803761511927647e-07,
+      "loss": 0.6129,
+      "step": 5866
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.792794755771497e-07,
+      "loss": 0.5373,
+      "step": 5867
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.781837733621914e-07,
+      "loss": 0.5951,
+      "step": 5868
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.770890447891531e-07,
+      "loss": 0.6171,
+      "step": 5869
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.759952900990817e-07,
+      "loss": 0.6091,
+      "step": 5870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.749025095328137e-07,
+      "loss": 0.6499,
+      "step": 5871
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.738107033309687e-07,
+      "loss": 0.5357,
+      "step": 5872
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.727198717339511e-07,
+      "loss": 0.6207,
+      "step": 5873
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.71630014981952e-07,
+      "loss": 0.6058,
+      "step": 5874
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.705411333149474e-07,
+      "loss": 0.5258,
+      "step": 5875
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.694532269726977e-07,
+      "loss": 0.606,
+      "step": 5876
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.683662961947522e-07,
+      "loss": 0.622,
+      "step": 5877
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.672803412204409e-07,
+      "loss": 0.5634,
+      "step": 5878
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.661953622888811e-07,
+      "loss": 0.6163,
+      "step": 5879
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.65111359638974e-07,
+      "loss": 0.6089,
+      "step": 5880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.640283335094083e-07,
+      "loss": 0.5549,
+      "step": 5881
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.629462841386546e-07,
+      "loss": 0.6013,
+      "step": 5882
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.618652117649709e-07,
+      "loss": 0.6095,
+      "step": 5883
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.607851166263995e-07,
+      "loss": 0.6109,
+      "step": 5884
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.597059989607656e-07,
+      "loss": 0.6207,
+      "step": 5885
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.586278590056821e-07,
+      "loss": 0.6,
+      "step": 5886
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.575506969985439e-07,
+      "loss": 0.6232,
+      "step": 5887
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.564745131765315e-07,
+      "loss": 0.6078,
+      "step": 5888
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.553993077766124e-07,
+      "loss": 0.5652,
+      "step": 5889
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.54325081035535e-07,
+      "loss": 0.5061,
+      "step": 5890
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.532518331898345e-07,
+      "loss": 0.6266,
+      "step": 5891
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.521795644758288e-07,
+      "loss": 0.5986,
+      "step": 5892
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.511082751296232e-07,
+      "loss": 0.6043,
+      "step": 5893
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.500379653871035e-07,
+      "loss": 0.5444,
+      "step": 5894
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.48968635483943e-07,
+      "loss": 0.6524,
+      "step": 5895
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.479002856555982e-07,
+      "loss": 0.6613,
+      "step": 5896
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.468329161373087e-07,
+      "loss": 0.5596,
+      "step": 5897
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.457665271640993e-07,
+      "loss": 0.5284,
+      "step": 5898
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.44701118970779e-07,
+      "loss": 0.6199,
+      "step": 5899
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.436366917919395e-07,
+      "loss": 0.5658,
+      "step": 5900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.425732458619587e-07,
+      "loss": 0.5935,
+      "step": 5901
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.415107814149978e-07,
+      "loss": 0.5391,
+      "step": 5902
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.404492986850001e-07,
+      "loss": 0.6366,
+      "step": 5903
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.393887979056939e-07,
+      "loss": 0.5249,
+      "step": 5904
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.383292793105927e-07,
+      "loss": 0.5268,
+      "step": 5905
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.37270743132991e-07,
+      "loss": 0.6427,
+      "step": 5906
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.362131896059703e-07,
+      "loss": 0.6157,
+      "step": 5907
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.351566189623919e-07,
+      "loss": 0.5265,
+      "step": 5908
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.341010314349043e-07,
+      "loss": 0.645,
+      "step": 5909
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.330464272559361e-07,
+      "loss": 0.515,
+      "step": 5910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.319928066577013e-07,
+      "loss": 0.6176,
+      "step": 5911
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.309401698721989e-07,
+      "loss": 0.6316,
+      "step": 5912
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.298885171312084e-07,
+      "loss": 0.6171,
+      "step": 5913
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.288378486662937e-07,
+      "loss": 0.6286,
+      "step": 5914
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.277881647088024e-07,
+      "loss": 0.6478,
+      "step": 5915
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.267394654898633e-07,
+      "loss": 0.6091,
+      "step": 5916
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.256917512403919e-07,
+      "loss": 0.5946,
+      "step": 5917
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.246450221910843e-07,
+      "loss": 0.611,
+      "step": 5918
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.235992785724203e-07,
+      "loss": 0.6163,
+      "step": 5919
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.225545206146632e-07,
+      "loss": 0.6153,
+      "step": 5920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.215107485478588e-07,
+      "loss": 0.5371,
+      "step": 5921
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.204679626018345e-07,
+      "loss": 0.6197,
+      "step": 5922
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.19426163006202e-07,
+      "loss": 0.6007,
+      "step": 5923
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.183853499903574e-07,
+      "loss": 0.5663,
+      "step": 5924
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.173455237834768e-07,
+      "loss": 0.6129,
+      "step": 5925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.163066846145198e-07,
+      "loss": 0.6096,
+      "step": 5926
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.15268832712229e-07,
+      "loss": 0.6048,
+      "step": 5927
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1423196830513e-07,
+      "loss": 0.6199,
+      "step": 5928
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.13196091621529e-07,
+      "loss": 0.6324,
+      "step": 5929
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.121612028895184e-07,
+      "loss": 0.529,
+      "step": 5930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.11127302336969e-07,
+      "loss": 0.6225,
+      "step": 5931
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.100943901915373e-07,
+      "loss": 0.6477,
+      "step": 5932
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.090624666806604e-07,
+      "loss": 0.6248,
+      "step": 5933
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.080315320315577e-07,
+      "loss": 0.5961,
+      "step": 5934
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.070015864712302e-07,
+      "loss": 0.6097,
+      "step": 5935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.059726302264645e-07,
+      "loss": 0.6149,
+      "step": 5936
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.049446635238253e-07,
+      "loss": 0.5373,
+      "step": 5937
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.039176865896622e-07,
+      "loss": 0.6421,
+      "step": 5938
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.028916996501043e-07,
+      "loss": 0.6087,
+      "step": 5939
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.018667029310658e-07,
+      "loss": 0.5775,
+      "step": 5940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.008426966582386e-07,
+      "loss": 0.5238,
+      "step": 5941
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.998196810571026e-07,
+      "loss": 0.5987,
+      "step": 5942
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.987976563529128e-07,
+      "loss": 0.5908,
+      "step": 5943
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.977766227707126e-07,
+      "loss": 0.5914,
+      "step": 5944
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.96756580535322e-07,
+      "loss": 0.6081,
+      "step": 5945
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.957375298713452e-07,
+      "loss": 0.6122,
+      "step": 5946
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.947194710031655e-07,
+      "loss": 0.6147,
+      "step": 5947
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.937024041549532e-07,
+      "loss": 0.6118,
+      "step": 5948
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.926863295506546e-07,
+      "loss": 0.5923,
+      "step": 5949
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.91671247414e-07,
+      "loss": 0.5184,
+      "step": 5950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.90657157968501e-07,
+      "loss": 0.5668,
+      "step": 5951
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.896440614374504e-07,
+      "loss": 0.6055,
+      "step": 5952
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.886319580439219e-07,
+      "loss": 0.5341,
+      "step": 5953
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.87620848010772e-07,
+      "loss": 0.5336,
+      "step": 5954
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.866107315606361e-07,
+      "loss": 0.6219,
+      "step": 5955
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.856016089159349e-07,
+      "loss": 0.602,
+      "step": 5956
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.84593480298865e-07,
+      "loss": 0.6295,
+      "step": 5957
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.835863459314072e-07,
+      "loss": 0.5261,
+      "step": 5958
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.825802060353246e-07,
+      "loss": 0.6159,
+      "step": 5959
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.815750608321584e-07,
+      "loss": 0.606,
+      "step": 5960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.805709105432316e-07,
+      "loss": 0.553,
+      "step": 5961
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.795677553896494e-07,
+      "loss": 0.522,
+      "step": 5962
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.785655955922969e-07,
+      "loss": 0.628,
+      "step": 5963
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.775644313718386e-07,
+      "loss": 0.5353,
+      "step": 5964
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7656426294872383e-07,
+      "loss": 0.5899,
+      "step": 5965
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7556509054317914e-07,
+      "loss": 0.6135,
+      "step": 5966
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7456691437521243e-07,
+      "loss": 0.6118,
+      "step": 5967
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7356973466461176e-07,
+      "loss": 0.5505,
+      "step": 5968
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.725735516309493e-07,
+      "loss": 0.6369,
+      "step": 5969
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7157836549357186e-07,
+      "loss": 0.6459,
+      "step": 5970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7058417647161294e-07,
+      "loss": 0.6069,
+      "step": 5971
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6959098478398244e-07,
+      "loss": 0.5407,
+      "step": 5972
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.685987906493711e-07,
+      "loss": 0.6081,
+      "step": 5973
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6760759428625123e-07,
+      "loss": 0.6056,
+      "step": 5974
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6661739591287505e-07,
+      "loss": 0.6112,
+      "step": 5975
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6562819574727304e-07,
+      "loss": 0.6105,
+      "step": 5976
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.646399940072599e-07,
+      "loss": 0.631,
+      "step": 5977
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.636527909104277e-07,
+      "loss": 0.5477,
+      "step": 5978
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6266658667414866e-07,
+      "loss": 0.6116,
+      "step": 5979
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6168138151557516e-07,
+      "loss": 0.5868,
+      "step": 5980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6069717565164205e-07,
+      "loss": 0.6258,
+      "step": 5981
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.597139692990593e-07,
+      "loss": 0.5307,
+      "step": 5982
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5873176267432205e-07,
+      "loss": 0.6357,
+      "step": 5983
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5775055599370235e-07,
+      "loss": 0.6098,
+      "step": 5984
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5677034947325174e-07,
+      "loss": 0.6514,
+      "step": 5985
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5579114332880257e-07,
+      "loss": 0.6114,
+      "step": 5986
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5481293777596746e-07,
+      "loss": 0.6084,
+      "step": 5987
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.538357330301363e-07,
+      "loss": 0.6169,
+      "step": 5988
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.528595293064819e-07,
+      "loss": 0.6103,
+      "step": 5989
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5188432681995517e-07,
+      "loss": 0.5366,
+      "step": 5990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.509101257852855e-07,
+      "loss": 0.5322,
+      "step": 5991
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.499369264169828e-07,
+      "loss": 0.5451,
+      "step": 5992
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4896472892933693e-07,
+      "loss": 0.6239,
+      "step": 5993
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.479935335364155e-07,
+      "loss": 0.637,
+      "step": 5994
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.47023340452068e-07,
+      "loss": 0.5246,
+      "step": 5995
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.460541498899207e-07,
+      "loss": 0.5655,
+      "step": 5996
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.450859620633807e-07,
+      "loss": 0.6343,
+      "step": 5997
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4411877718563314e-07,
+      "loss": 0.6372,
+      "step": 5998
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.431525954696436e-07,
+      "loss": 0.6402,
+      "step": 5999
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4218741712815425e-07,
+      "loss": 0.5288,
+      "step": 6000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.412232423736912e-07,
+      "loss": 0.5281,
+      "step": 6001
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.402600714185551e-07,
+      "loss": 0.6152,
+      "step": 6002
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3929790447482633e-07,
+      "loss": 0.5789,
+      "step": 6003
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3833674175436604e-07,
+      "loss": 0.5307,
+      "step": 6004
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3737658346881205e-07,
+      "loss": 0.6062,
+      "step": 6005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3641742982958203e-07,
+      "loss": 0.6082,
+      "step": 6006
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.354592810478747e-07,
+      "loss": 0.6476,
+      "step": 6007
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.345021373346636e-07,
+      "loss": 0.6178,
+      "step": 6008
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.335459989007029e-07,
+      "loss": 0.6476,
+      "step": 6009
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3259086595652477e-07,
+      "loss": 0.6449,
+      "step": 6010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.316367387124404e-07,
+      "loss": 0.6282,
+      "step": 6011
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.306836173785406e-07,
+      "loss": 0.5995,
+      "step": 6012
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.297315021646931e-07,
+      "loss": 0.6277,
+      "step": 6013
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2878039328054457e-07,
+      "loss": 0.6117,
+      "step": 6014
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.278302909355203e-07,
+      "loss": 0.5907,
+      "step": 6015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.26881195338823e-07,
+      "loss": 0.6232,
+      "step": 6016
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.259331066994349e-07,
+      "loss": 0.6314,
+      "step": 6017
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2498602522611674e-07,
+      "loss": 0.6596,
+      "step": 6018
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.240399511274057e-07,
+      "loss": 0.624,
+      "step": 6019
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.230948846116195e-07,
+      "loss": 0.6199,
+      "step": 6020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2215082588685285e-07,
+      "loss": 0.5662,
+      "step": 6021
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.212077751609778e-07,
+      "loss": 0.5846,
+      "step": 6022
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2026573264164426e-07,
+      "loss": 0.5856,
+      "step": 6023
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.193246985362837e-07,
+      "loss": 0.5881,
+      "step": 6024
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.183846730521007e-07,
+      "loss": 0.6389,
+      "step": 6025
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.17445656396081e-07,
+      "loss": 0.5532,
+      "step": 6026
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.165076487749864e-07,
+      "loss": 0.6263,
+      "step": 6027
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.155706503953577e-07,
+      "loss": 0.5291,
+      "step": 6028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1463466146351217e-07,
+      "loss": 0.6354,
+      "step": 6029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1369968218554714e-07,
+      "loss": 0.5532,
+      "step": 6030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.127657127673346e-07,
+      "loss": 0.6149,
+      "step": 6031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.118327534145278e-07,
+      "loss": 0.6267,
+      "step": 6032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.109008043325541e-07,
+      "loss": 0.6145,
+      "step": 6033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.099698657266199e-07,
+      "loss": 0.5545,
+      "step": 6034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.090399378017085e-07,
+      "loss": 0.6401,
+      "step": 6035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0811102076258266e-07,
+      "loss": 0.5179,
+      "step": 6036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0718311481378e-07,
+      "loss": 0.6432,
+      "step": 6037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0625622015961763e-07,
+      "loss": 0.601,
+      "step": 6038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.053303370041878e-07,
+      "loss": 0.5436,
+      "step": 6039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.044054655513613e-07,
+      "loss": 0.6078,
+      "step": 6040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.034816060047858e-07,
+      "loss": 0.6125,
+      "step": 6041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0255875856788807e-07,
+      "loss": 0.5394,
+      "step": 6042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.016369234438683e-07,
+      "loss": 0.6712,
+      "step": 6043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.007161008357069e-07,
+      "loss": 0.5312,
+      "step": 6044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.997962909461611e-07,
+      "loss": 0.6076,
+      "step": 6045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.988774939777623e-07,
+      "loss": 0.6566,
+      "step": 6046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9795971013282197e-07,
+      "loss": 0.5315,
+      "step": 6047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9704293961342733e-07,
+      "loss": 0.5248,
+      "step": 6048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.96127182621443e-07,
+      "loss": 0.6104,
+      "step": 6049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.952124393585094e-07,
+      "loss": 0.5598,
+      "step": 6050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9429871002604426e-07,
+      "loss": 0.6047,
+      "step": 6051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.933859948252422e-07,
+      "loss": 0.5469,
+      "step": 6052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9247429395707357e-07,
+      "loss": 0.5629,
+      "step": 6053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9156360762228784e-07,
+      "loss": 0.6338,
+      "step": 6054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.90653936021409e-07,
+      "loss": 0.6086,
+      "step": 6055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8974527935473685e-07,
+      "loss": 0.5216,
+      "step": 6056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.888376378223507e-07,
+      "loss": 0.5531,
+      "step": 6057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8793101162410417e-07,
+      "loss": 0.5079,
+      "step": 6058
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8702540095962637e-07,
+      "loss": 0.5593,
+      "step": 6059
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8612080602832555e-07,
+      "loss": 0.6289,
+      "step": 6060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.852172270293852e-07,
+      "loss": 0.6405,
+      "step": 6061
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.84314664161764e-07,
+      "loss": 0.6059,
+      "step": 6062
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8341311762419784e-07,
+      "loss": 0.6342,
+      "step": 6063
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8251258761519803e-07,
+      "loss": 0.5886,
+      "step": 6064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8161307433305486e-07,
+      "loss": 0.6071,
+      "step": 6065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8071457797583035e-07,
+      "loss": 0.5356,
+      "step": 6066
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.798170987413663e-07,
+      "loss": 0.6161,
+      "step": 6067
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.789206368272774e-07,
+      "loss": 0.5481,
+      "step": 6068
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.78025192430958e-07,
+      "loss": 0.6008,
+      "step": 6069
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7713076574957483e-07,
+      "loss": 0.5605,
+      "step": 6070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.762373569800737e-07,
+      "loss": 0.5165,
+      "step": 6071
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.753449663191744e-07,
+      "loss": 0.6032,
+      "step": 6072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.74453593963372e-07,
+      "loss": 0.5359,
+      "step": 6073
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7356324010893894e-07,
+      "loss": 0.5943,
+      "step": 6074
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7267390495192235e-07,
+      "loss": 0.5987,
+      "step": 6075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.717855886881444e-07,
+      "loss": 0.5474,
+      "step": 6076
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.708982915132064e-07,
+      "loss": 0.5958,
+      "step": 6077
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.700120136224805e-07,
+      "loss": 0.6218,
+      "step": 6078
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.691267552111183e-07,
+      "loss": 0.6143,
+      "step": 6079
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.682425164740427e-07,
+      "loss": 0.5932,
+      "step": 6080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.673592976059581e-07,
+      "loss": 0.4992,
+      "step": 6081
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6647709880133773e-07,
+      "loss": 0.6105,
+      "step": 6082
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.655959202544362e-07,
+      "loss": 0.5375,
+      "step": 6083
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6471576215927897e-07,
+      "loss": 0.6126,
+      "step": 6084
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.638366247096692e-07,
+      "loss": 0.6251,
+      "step": 6085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6295850809918433e-07,
+      "loss": 0.6478,
+      "step": 6086
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.620814125211769e-07,
+      "loss": 0.6373,
+      "step": 6087
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6120533816877467e-07,
+      "loss": 0.5249,
+      "step": 6088
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.603302852348822e-07,
+      "loss": 0.59,
+      "step": 6089
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5945625391217765e-07,
+      "loss": 0.5906,
+      "step": 6090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5858324439311364e-07,
+      "loss": 0.616,
+      "step": 6091
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.577112568699187e-07,
+      "loss": 0.5578,
+      "step": 6092
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5684029153459585e-07,
+      "loss": 0.56,
+      "step": 6093
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5597034857892396e-07,
+      "loss": 0.5652,
+      "step": 6094
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5510142819445647e-07,
+      "loss": 0.5297,
+      "step": 6095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5423353057252197e-07,
+      "loss": 0.5993,
+      "step": 6096
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5336665590422147e-07,
+      "loss": 0.5391,
+      "step": 6097
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.52500804380434e-07,
+      "loss": 0.5316,
+      "step": 6098
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5163597619181146e-07,
+      "loss": 0.6272,
+      "step": 6099
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5077217152877927e-07,
+      "loss": 0.6374,
+      "step": 6100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.499093905815415e-07,
+      "loss": 0.5952,
+      "step": 6101
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.490476335400733e-07,
+      "loss": 0.6037,
+      "step": 6102
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.481869005941252e-07,
+      "loss": 0.6036,
+      "step": 6103
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4732719193322215e-07,
+      "loss": 0.5483,
+      "step": 6104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.46468507746664e-07,
+      "loss": 0.5681,
+      "step": 6105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.456108482235254e-07,
+      "loss": 0.6235,
+      "step": 6106
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4475421355265326e-07,
+      "loss": 0.5241,
+      "step": 6107
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.438986039226727e-07,
+      "loss": 0.629,
+      "step": 6108
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.430440195219803e-07,
+      "loss": 0.6272,
+      "step": 6109
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4219046053874604e-07,
+      "loss": 0.5318,
+      "step": 6110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4133792716091684e-07,
+      "loss": 0.5881,
+      "step": 6111
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4048641957621087e-07,
+      "loss": 0.5328,
+      "step": 6112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.396359379721237e-07,
+      "loss": 0.5895,
+      "step": 6113
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.387864825359227e-07,
+      "loss": 0.5933,
+      "step": 6114
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3793805345465e-07,
+      "loss": 0.5961,
+      "step": 6115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3709065091512115e-07,
+      "loss": 0.5421,
+      "step": 6116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3624427510392575e-07,
+      "loss": 0.6363,
+      "step": 6117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.353989262074292e-07,
+      "loss": 0.6344,
+      "step": 6118
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3455460441176814e-07,
+      "loss": 0.5584,
+      "step": 6119
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3371130990285493e-07,
+      "loss": 0.6148,
+      "step": 6120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.328690428663744e-07,
+      "loss": 0.6389,
+      "step": 6121
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3202780348778665e-07,
+      "loss": 0.6195,
+      "step": 6122
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3118759195232273e-07,
+      "loss": 0.6212,
+      "step": 6123
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3034840844499097e-07,
+      "loss": 0.614,
+      "step": 6124
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.295102531505717e-07,
+      "loss": 0.5445,
+      "step": 6125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.286731262536175e-07,
+      "loss": 0.6137,
+      "step": 6126
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2783702793845683e-07,
+      "loss": 0.589,
+      "step": 6127
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.270019583891898e-07,
+      "loss": 0.6064,
+      "step": 6128
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.261679177896909e-07,
+      "loss": 0.5849,
+      "step": 6129
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.253349063236083e-07,
+      "loss": 0.5811,
+      "step": 6130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.24502924174363e-07,
+      "loss": 0.532,
+      "step": 6131
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.236719715251496e-07,
+      "loss": 0.598,
+      "step": 6132
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2284204855893673e-07,
+      "loss": 0.6093,
+      "step": 6133
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.220131554584649e-07,
+      "loss": 0.5989,
+      "step": 6134
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2118529240624754e-07,
+      "loss": 0.5476,
+      "step": 6135
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2035845958457324e-07,
+      "loss": 0.5352,
+      "step": 6136
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1953265717550364e-07,
+      "loss": 0.6249,
+      "step": 6137
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1870788536087106e-07,
+      "loss": 0.6217,
+      "step": 6138
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.17884144322283e-07,
+      "loss": 0.5699,
+      "step": 6139
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1706143424111933e-07,
+      "loss": 0.5547,
+      "step": 6140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.162397552985319e-07,
+      "loss": 0.5445,
+      "step": 6141
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.154191076754487e-07,
+      "loss": 0.6236,
+      "step": 6142
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1459949155256796e-07,
+      "loss": 0.6267,
+      "step": 6143
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.137809071103598e-07,
+      "loss": 0.5486,
+      "step": 6144
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1296335452907055e-07,
+      "loss": 0.6191,
+      "step": 6145
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1214683398871736e-07,
+      "loss": 0.6562,
+      "step": 6146
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.113313456690886e-07,
+      "loss": 0.538,
+      "step": 6147
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1051688974974904e-07,
+      "loss": 0.645,
+      "step": 6148
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0970346641003346e-07,
+      "loss": 0.5872,
+      "step": 6149
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0889107582904974e-07,
+      "loss": 0.5341,
+      "step": 6150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0807971818567817e-07,
+      "loss": 0.6037,
+      "step": 6151
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.07269393658573e-07,
+      "loss": 0.5948,
+      "step": 6152
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0646010242615875e-07,
+      "loss": 0.5918,
+      "step": 6153
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.056518446666351e-07,
+      "loss": 0.6229,
+      "step": 6154
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0484462055797194e-07,
+      "loss": 0.6039,
+      "step": 6155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0403843027791144e-07,
+      "loss": 0.6243,
+      "step": 6156
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0323327400397117e-07,
+      "loss": 0.6119,
+      "step": 6157
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0242915191343756e-07,
+      "loss": 0.5225,
+      "step": 6158
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0162606418337014e-07,
+      "loss": 0.5969,
+      "step": 6159
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0082401099060244e-07,
+      "loss": 0.6114,
+      "step": 6160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.000229925117387e-07,
+      "loss": 0.5418,
+      "step": 6161
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9922300892315517e-07,
+      "loss": 0.5499,
+      "step": 6162
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.984240604010008e-07,
+      "loss": 0.5942,
+      "step": 6163
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9762614712119607e-07,
+      "loss": 0.6024,
+      "step": 6164
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.968292692594338e-07,
+      "loss": 0.6059,
+      "step": 6165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.960334269911802e-07,
+      "loss": 0.5447,
+      "step": 6166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.952386204916713e-07,
+      "loss": 0.5906,
+      "step": 6167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.944448499359148e-07,
+      "loss": 0.5963,
+      "step": 6168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9365211549869375e-07,
+      "loss": 0.6392,
+      "step": 6169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.928604173545585e-07,
+      "loss": 0.587,
+      "step": 6170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.920697556778357e-07,
+      "loss": 0.603,
+      "step": 6171
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.912801306426205e-07,
+      "loss": 0.5532,
+      "step": 6172
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9049154242278055e-07,
+      "loss": 0.6457,
+      "step": 6173
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.897039911919558e-07,
+      "loss": 0.5562,
+      "step": 6174
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.88917477123557e-07,
+      "loss": 0.5936,
+      "step": 6175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.881320003907667e-07,
+      "loss": 0.6318,
+      "step": 6176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.873475611665416e-07,
+      "loss": 0.6085,
+      "step": 6177
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.865641596236057e-07,
+      "loss": 0.5135,
+      "step": 6178
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8578179593445777e-07,
+      "loss": 0.621,
+      "step": 6179
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.85000470271366e-07,
+      "loss": 0.6298,
+      "step": 6180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8422018280637067e-07,
+      "loss": 0.5906,
+      "step": 6181
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.834409337112842e-07,
+      "loss": 0.612,
+      "step": 6182
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.826627231576906e-07,
+      "loss": 0.5354,
+      "step": 6183
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8188555131694326e-07,
+      "loss": 0.6267,
+      "step": 6184
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.811094183601687e-07,
+      "loss": 0.621,
+      "step": 6185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.80334324458264e-07,
+      "loss": 0.591,
+      "step": 6186
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.795602697818972e-07,
+      "loss": 0.6001,
+      "step": 6187
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.787872545015069e-07,
+      "loss": 0.5576,
+      "step": 6188
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.780152787873053e-07,
+      "loss": 0.6326,
+      "step": 6189
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.772443428092736e-07,
+      "loss": 0.5729,
+      "step": 6190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7647444673716496e-07,
+      "loss": 0.6124,
+      "step": 6191
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7570559074050205e-07,
+      "loss": 0.5621,
+      "step": 6192
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7493777498857954e-07,
+      "loss": 0.6325,
+      "step": 6193
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.741709996504649e-07,
+      "loss": 0.6225,
+      "step": 6194
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.734052648949931e-07,
+      "loss": 0.5554,
+      "step": 6195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7264057089077255e-07,
+      "loss": 0.618,
+      "step": 6196
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.71876917806182e-07,
+      "loss": 0.5438,
+      "step": 6197
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7111430580936957e-07,
+      "loss": 0.5982,
+      "step": 6198
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.703527350682561e-07,
+      "loss": 0.6244,
+      "step": 6199
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6959220575053067e-07,
+      "loss": 0.625,
+      "step": 6200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6883271802365606e-07,
+      "loss": 0.5734,
+      "step": 6201
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.680742720548646e-07,
+      "loss": 0.6023,
+      "step": 6202
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6731686801115766e-07,
+      "loss": 0.63,
+      "step": 6203
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6656050605930904e-07,
+      "loss": 0.5952,
+      "step": 6204
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6580518636586114e-07,
+      "loss": 0.63,
+      "step": 6205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.650509090971304e-07,
+      "loss": 0.6006,
+      "step": 6206
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.642976744192e-07,
+      "loss": 0.6204,
+      "step": 6207
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.635454824979261e-07,
+      "loss": 0.6432,
+      "step": 6208
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6279433349893357e-07,
+      "loss": 0.6421,
+      "step": 6209
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.620442275876184e-07,
+      "loss": 0.6356,
+      "step": 6210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6129516492914673e-07,
+      "loss": 0.6115,
+      "step": 6211
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.60547145688454e-07,
+      "loss": 0.5219,
+      "step": 6212
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5980017003024884e-07,
+      "loss": 0.6475,
+      "step": 6213
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5905423811900755e-07,
+      "loss": 0.6207,
+      "step": 6214
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5830935011897707e-07,
+      "loss": 0.6061,
+      "step": 6215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.575655061941745e-07,
+      "loss": 0.5899,
+      "step": 6216
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5682270650838723e-07,
+      "loss": 0.5961,
+      "step": 6217
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.560809512251722e-07,
+      "loss": 0.6269,
+      "step": 6218
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5534024050785754e-07,
+      "loss": 0.6512,
+      "step": 6219
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5460057451954124e-07,
+      "loss": 0.6251,
+      "step": 6220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5386195342308964e-07,
+      "loss": 0.6299,
+      "step": 6221
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.531243773811415e-07,
+      "loss": 0.6117,
+      "step": 6222
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.523878465561019e-07,
+      "loss": 0.6337,
+      "step": 6223
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.516523611101496e-07,
+      "loss": 0.5866,
+      "step": 6224
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5091792120523153e-07,
+      "loss": 0.6445,
+      "step": 6225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.50184527003064e-07,
+      "loss": 0.5263,
+      "step": 6226
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4945217866513274e-07,
+      "loss": 0.6435,
+      "step": 6227
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.487208763526949e-07,
+      "loss": 0.6251,
+      "step": 6228
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4799062022677514e-07,
+      "loss": 0.6424,
+      "step": 6229
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.472614104481708e-07,
+      "loss": 0.5289,
+      "step": 6230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.465332471774451e-07,
+      "loss": 0.6432,
+      "step": 6231
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.458061305749332e-07,
+      "loss": 0.6477,
+      "step": 6232
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.450800608007403e-07,
+      "loss": 0.6317,
+      "step": 6233
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.443550380147386e-07,
+      "loss": 0.6453,
+      "step": 6234
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4363106237657196e-07,
+      "loss": 0.6346,
+      "step": 6235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4290813404565296e-07,
+      "loss": 0.6182,
+      "step": 6236
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.421862531811636e-07,
+      "loss": 0.6106,
+      "step": 6237
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.414654199420552e-07,
+      "loss": 0.5233,
+      "step": 6238
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.407456344870479e-07,
+      "loss": 0.5312,
+      "step": 6239
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4002689697463224e-07,
+      "loss": 0.6017,
+      "step": 6240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3930920756306597e-07,
+      "loss": 0.5178,
+      "step": 6241
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3859256641037996e-07,
+      "loss": 0.5678,
+      "step": 6242
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.378769736743697e-07,
+      "loss": 0.65,
+      "step": 6243
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3716242951260238e-07,
+      "loss": 0.5943,
+      "step": 6244
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3644893408241443e-07,
+      "loss": 0.6351,
+      "step": 6245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3573648754091072e-07,
+      "loss": 0.5517,
+      "step": 6246
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.350250900449641e-07,
+      "loss": 0.6199,
+      "step": 6247
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3431474175121926e-07,
+      "loss": 0.6096,
+      "step": 6248
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.336054428160872e-07,
+      "loss": 0.6082,
+      "step": 6249
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3289719339574857e-07,
+      "loss": 0.6138,
+      "step": 6250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3218999364615357e-07,
+      "loss": 0.6369,
+      "step": 6251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3148384372302046e-07,
+      "loss": 0.5368,
+      "step": 6252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.307787437818365e-07,
+      "loss": 0.6164,
+      "step": 6253
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.300746939778592e-07,
+      "loss": 0.6165,
+      "step": 6254
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2937169446611342e-07,
+      "loss": 0.644,
+      "step": 6255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2866974540139141e-07,
+      "loss": 0.6293,
+      "step": 6256
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2796884693825738e-07,
+      "loss": 0.6166,
+      "step": 6257
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2726899923104173e-07,
+      "loss": 0.6413,
+      "step": 6258
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.26570202433844e-07,
+      "loss": 0.6238,
+      "step": 6259
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2587245670053382e-07,
+      "loss": 0.5977,
+      "step": 6260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2517576218474724e-07,
+      "loss": 0.6175,
+      "step": 6261
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2448011903988987e-07,
+      "loss": 0.5373,
+      "step": 6262
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2378552741913585e-07,
+      "loss": 0.603,
+      "step": 6263
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2309198747542782e-07,
+      "loss": 0.5278,
+      "step": 6264
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2239949936147588e-07,
+      "loss": 0.6051,
+      "step": 6265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2170806322976023e-07,
+      "loss": 0.5908,
+      "step": 6266
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2101767923252804e-07,
+      "loss": 0.5432,
+      "step": 6267
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2032834752179599e-07,
+      "loss": 0.5502,
+      "step": 6268
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1964006824934713e-07,
+      "loss": 0.6213,
+      "step": 6269
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1895284156673523e-07,
+      "loss": 0.5415,
+      "step": 6270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1826666762527982e-07,
+      "loss": 0.6306,
+      "step": 6271
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1758154657607168e-07,
+      "loss": 0.6311,
+      "step": 6272
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1689747856996625e-07,
+      "loss": 0.6072,
+      "step": 6273
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1621446375759026e-07,
+      "loss": 0.6505,
+      "step": 6274
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.155325022893362e-07,
+      "loss": 0.601,
+      "step": 6275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.148515943153645e-07,
+      "loss": 0.6027,
+      "step": 6276
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1417173998560693e-07,
+      "loss": 0.6277,
+      "step": 6277
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1349293944975923e-07,
+      "loss": 0.5926,
+      "step": 6278
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1281519285728803e-07,
+      "loss": 0.6128,
+      "step": 6279
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1213850035742557e-07,
+      "loss": 0.6131,
+      "step": 6280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1146286209917265e-07,
+      "loss": 0.5855,
+      "step": 6281
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1078827823130032e-07,
+      "loss": 0.6221,
+      "step": 6282
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1011474890234418e-07,
+      "loss": 0.6166,
+      "step": 6283
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0944227426060948e-07,
+      "loss": 0.5466,
+      "step": 6284
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0877085445416889e-07,
+      "loss": 0.6532,
+      "step": 6285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0810048963086248e-07,
+      "loss": 0.625,
+      "step": 6286
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0743117993829774e-07,
+      "loss": 0.5987,
+      "step": 6287
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0676292552385068e-07,
+      "loss": 0.5799,
+      "step": 6288
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.060957265346647e-07,
+      "loss": 0.6084,
+      "step": 6289
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.054295831176506e-07,
+      "loss": 0.5485,
+      "step": 6290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0476449541948662e-07,
+      "loss": 0.611,
+      "step": 6291
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.041004635866195e-07,
+      "loss": 0.6022,
+      "step": 6292
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0343748776526117e-07,
+      "loss": 0.581,
+      "step": 6293
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0277556810139433e-07,
+      "loss": 0.5945,
+      "step": 6294
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0211470474076567e-07,
+      "loss": 0.6178,
+      "step": 6295
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0145489782889326e-07,
+      "loss": 0.5203,
+      "step": 6296
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.007961475110587e-07,
+      "loss": 0.6428,
+      "step": 6297
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0013845393231312e-07,
+      "loss": 0.6198,
+      "step": 6298
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9948181723747406e-07,
+      "loss": 0.6037,
+      "step": 6299
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9882623757112586e-07,
+      "loss": 0.6028,
+      "step": 6300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.98171715077623e-07,
+      "loss": 0.5958,
+      "step": 6301
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9751824990108414e-07,
+      "loss": 0.5337,
+      "step": 6302
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9686584218539528e-07,
+      "loss": 0.5821,
+      "step": 6303
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.962144920742115e-07,
+      "loss": 0.6059,
+      "step": 6304
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9556419971095363e-07,
+      "loss": 0.5621,
+      "step": 6305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9491496523880937e-07,
+      "loss": 0.6086,
+      "step": 6306
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.942667888007338e-07,
+      "loss": 0.5241,
+      "step": 6307
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.936196705394505e-07,
+      "loss": 0.5808,
+      "step": 6308
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9297361059744834e-07,
+      "loss": 0.5242,
+      "step": 6309
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.923286091169829e-07,
+      "loss": 0.6422,
+      "step": 6310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9168466624007788e-07,
+      "loss": 0.6532,
+      "step": 6311
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9104178210852252e-07,
+      "loss": 0.6171,
+      "step": 6312
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9039995686387536e-07,
+      "loss": 0.6086,
+      "step": 6313
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.897591906474594e-07,
+      "loss": 0.6552,
+      "step": 6314
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8911948360036514e-07,
+      "loss": 0.5273,
+      "step": 6315
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.884808358634499e-07,
+      "loss": 0.6075,
+      "step": 6316
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8784324757733842e-07,
+      "loss": 0.6054,
+      "step": 6317
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8720671888242058e-07,
+      "loss": 0.5158,
+      "step": 6318
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.865712499188549e-07,
+      "loss": 0.6376,
+      "step": 6319
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8593684082656493e-07,
+      "loss": 0.5482,
+      "step": 6320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.853034917452423e-07,
+      "loss": 0.6044,
+      "step": 6321
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8467120281434438e-07,
+      "loss": 0.6022,
+      "step": 6322
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8403997417309471e-07,
+      "loss": 0.5269,
+      "step": 6323
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.834098059604833e-07,
+      "loss": 0.6132,
+      "step": 6324
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8278069831526857e-07,
+      "loss": 0.6085,
+      "step": 6325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8215265137597305e-07,
+      "loss": 0.6057,
+      "step": 6326
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8152566528088778e-07,
+      "loss": 0.6013,
+      "step": 6327
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8089974016806788e-07,
+      "loss": 0.6108,
+      "step": 6328
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.802748761753359e-07,
+      "loss": 0.5964,
+      "step": 6329
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7965107344028232e-07,
+      "loss": 0.5496,
+      "step": 6330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7902833210026228e-07,
+      "loss": 0.641,
+      "step": 6331
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7840665229239606e-07,
+      "loss": 0.604,
+      "step": 6332
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7778603415357366e-07,
+      "loss": 0.5389,
+      "step": 6333
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.771664778204485e-07,
+      "loss": 0.5939,
+      "step": 6334
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.765479834294398e-07,
+      "loss": 0.5378,
+      "step": 6335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7593055111673641e-07,
+      "loss": 0.6097,
+      "step": 6336
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.75314181018289e-07,
+      "loss": 0.6263,
+      "step": 6337
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7469887326981794e-07,
+      "loss": 0.6434,
+      "step": 6338
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.740846280068076e-07,
+      "loss": 0.6082,
+      "step": 6339
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7347144536450867e-07,
+      "loss": 0.5839,
+      "step": 6340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.728593254779376e-07,
+      "loss": 0.6226,
+      "step": 6341
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.722482684818788e-07,
+      "loss": 0.613,
+      "step": 6342
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7163827451088023e-07,
+      "loss": 0.611,
+      "step": 6343
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.710293436992566e-07,
+      "loss": 0.6144,
+      "step": 6344
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7042147618109019e-07,
+      "loss": 0.6176,
+      "step": 6345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6981467209022607e-07,
+      "loss": 0.6235,
+      "step": 6346
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6920893156027686e-07,
+      "loss": 0.6461,
+      "step": 6347
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6860425472462193e-07,
+      "loss": 0.497,
+      "step": 6348
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6800064171640429e-07,
+      "loss": 0.6172,
+      "step": 6349
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6739809266853424e-07,
+      "loss": 0.5968,
+      "step": 6350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.667966077136879e-07,
+      "loss": 0.6068,
+      "step": 6351
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6619618698430538e-07,
+      "loss": 0.6216,
+      "step": 6352
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.655968306125938e-07,
+      "loss": 0.6372,
+      "step": 6353
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6499853873052695e-07,
+      "loss": 0.6356,
+      "step": 6354
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6440131146984227e-07,
+      "loss": 0.6041,
+      "step": 6355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.638051489620429e-07,
+      "loss": 0.6173,
+      "step": 6356
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6321005133839884e-07,
+      "loss": 0.6188,
+      "step": 6357
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6261601872994472e-07,
+      "loss": 0.5279,
+      "step": 6358
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6202305126748085e-07,
+      "loss": 0.5475,
+      "step": 6359
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6143114908157342e-07,
+      "loss": 0.6077,
+      "step": 6360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.608403123025537e-07,
+      "loss": 0.5848,
+      "step": 6361
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6025054106051817e-07,
+      "loss": 0.5924,
+      "step": 6362
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.59661835485328e-07,
+      "loss": 0.5939,
+      "step": 6363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.590741957066122e-07,
+      "loss": 0.5728,
+      "step": 6364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.584876218537612e-07,
+      "loss": 0.6164,
+      "step": 6365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.579021140559356e-07,
+      "loss": 0.6065,
+      "step": 6366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5731767244205666e-07,
+      "loss": 0.5428,
+      "step": 6367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5673429714081422e-07,
+      "loss": 0.6245,
+      "step": 6368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5615198828066057e-07,
+      "loss": 0.5427,
+      "step": 6369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5557074598981593e-07,
+      "loss": 0.5796,
+      "step": 6370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5499057039626342e-07,
+      "loss": 0.6154,
+      "step": 6371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5441146162775255e-07,
+      "loss": 0.6252,
+      "step": 6372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.53833419811798e-07,
+      "loss": 0.5386,
+      "step": 6373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5325644507567848e-07,
+      "loss": 0.6305,
+      "step": 6374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.526805375464391e-07,
+      "loss": 0.6187,
+      "step": 6375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5210569735088775e-07,
+      "loss": 0.6083,
+      "step": 6376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5153192461559996e-07,
+      "loss": 0.6523,
+      "step": 6377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5095921946691516e-07,
+      "loss": 0.6133,
+      "step": 6378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5038758203093695e-07,
+      "loss": 0.6368,
+      "step": 6379
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4981701243353464e-07,
+      "loss": 0.5406,
+      "step": 6380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4924751080034162e-07,
+      "loss": 0.6056,
+      "step": 6381
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4867907725675812e-07,
+      "loss": 0.6349,
+      "step": 6382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4811171192794628e-07,
+      "loss": 0.5512,
+      "step": 6383
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.475454149388361e-07,
+      "loss": 0.6221,
+      "step": 6384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4698018641412015e-07,
+      "loss": 0.6308,
+      "step": 6385
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4641602647825658e-07,
+      "loss": 0.6204,
+      "step": 6386
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4585293525546718e-07,
+      "loss": 0.647,
+      "step": 6387
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4529091286973994e-07,
+      "loss": 0.6453,
+      "step": 6388
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4472995944482704e-07,
+      "loss": 0.6078,
+      "step": 6389
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4417007510424464e-07,
+      "loss": 0.6198,
+      "step": 6390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.436112599712741e-07,
+      "loss": 0.5989,
+      "step": 6391
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4305351416896152e-07,
+      "loss": 0.6073,
+      "step": 6392
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4249683782011636e-07,
+      "loss": 0.6034,
+      "step": 6393
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4194123104731394e-07,
+      "loss": 0.547,
+      "step": 6394
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4138669397289473e-07,
+      "loss": 0.5428,
+      "step": 6395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4083322671896048e-07,
+      "loss": 0.5795,
+      "step": 6396
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4028082940738097e-07,
+      "loss": 0.5986,
+      "step": 6397
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.397295021597883e-07,
+      "loss": 0.6077,
+      "step": 6398
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.391792450975793e-07,
+      "loss": 0.5355,
+      "step": 6399
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3863005834191535e-07,
+      "loss": 0.5885,
+      "step": 6400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.380819420137225e-07,
+      "loss": 0.6079,
+      "step": 6401
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3753489623369087e-07,
+      "loss": 0.6213,
+      "step": 6402
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.369889211222747e-07,
+      "loss": 0.5854,
+      "step": 6403
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.364440167996922e-07,
+      "loss": 0.5284,
+      "step": 6404
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.359001833859258e-07,
+      "loss": 0.6269,
+      "step": 6405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3535742100072303e-07,
+      "loss": 0.6197,
+      "step": 6406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.348157297635949e-07,
+      "loss": 0.6077,
+      "step": 6407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.342751097938161e-07,
+      "loss": 0.6474,
+      "step": 6408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.337355612104274e-07,
+      "loss": 0.6317,
+      "step": 6409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.331970841322311e-07,
+      "loss": 0.5516,
+      "step": 6410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3265967867779504e-07,
+      "loss": 0.5193,
+      "step": 6411
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3212334496545022e-07,
+      "loss": 0.5272,
+      "step": 6412
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3158808311329374e-07,
+      "loss": 0.6079,
+      "step": 6413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3105389323918417e-07,
+      "loss": 0.6057,
+      "step": 6414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.305207754607446e-07,
+      "loss": 0.6,
+      "step": 6415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2998872989536339e-07,
+      "loss": 0.616,
+      "step": 6416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2945775666019178e-07,
+      "loss": 0.5576,
+      "step": 6417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2892785587214406e-07,
+      "loss": 0.5192,
+      "step": 6418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.283990276479008e-07,
+      "loss": 0.6386,
+      "step": 6419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2787127210390326e-07,
+      "loss": 0.5847,
+      "step": 6420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2734458935636018e-07,
+      "loss": 0.6204,
+      "step": 6421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2681897952124046e-07,
+      "loss": 0.5468,
+      "step": 6422
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2629444271427982e-07,
+      "loss": 0.5388,
+      "step": 6423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2577097905097423e-07,
+      "loss": 0.5443,
+      "step": 6424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2524858864658763e-07,
+      "loss": 0.5914,
+      "step": 6425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.247272716161446e-07,
+      "loss": 0.6126,
+      "step": 6426
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.242070280744334e-07,
+      "loss": 0.5993,
+      "step": 6427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2368785813600792e-07,
+      "loss": 0.6133,
+      "step": 6428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2316976191518338e-07,
+      "loss": 0.5923,
+      "step": 6429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2265273952604072e-07,
+      "loss": 0.5952,
+      "step": 6430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2213679108242283e-07,
+      "loss": 0.5936,
+      "step": 6431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.21621916697936e-07,
+      "loss": 0.5257,
+      "step": 6432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2110811648595177e-07,
+      "loss": 0.5968,
+      "step": 6433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2059539055960355e-07,
+      "loss": 0.6325,
+      "step": 6434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2008373903178828e-07,
+      "loss": 0.517,
+      "step": 6435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1957316201516744e-07,
+      "loss": 0.6211,
+      "step": 6436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.190636596221656e-07,
+      "loss": 0.6409,
+      "step": 6437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1855523196496966e-07,
+      "loss": 0.6431,
+      "step": 6438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1804787915553063e-07,
+      "loss": 0.586,
+      "step": 6439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1754160130556302e-07,
+      "loss": 0.6023,
+      "step": 6440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1703639852654325e-07,
+      "loss": 0.5915,
+      "step": 6441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1653227092971398e-07,
+      "loss": 0.6218,
+      "step": 6442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1602921862607863e-07,
+      "loss": 0.5177,
+      "step": 6443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1552724172640417e-07,
+      "loss": 0.6083,
+      "step": 6444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1502634034122107e-07,
+      "loss": 0.54,
+      "step": 6445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1452651458082386e-07,
+      "loss": 0.6039,
+      "step": 6446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1402776455526842e-07,
+      "loss": 0.5825,
+      "step": 6447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1353009037437523e-07,
+      "loss": 0.6085,
+      "step": 6448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1303349214772829e-07,
+      "loss": 0.6182,
+      "step": 6449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1253796998467237e-07,
+      "loss": 0.6344,
+      "step": 6450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1204352399431739e-07,
+      "loss": 0.602,
+      "step": 6451
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1155015428553573e-07,
+      "loss": 0.6476,
+      "step": 6452
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1105786096696214e-07,
+      "loss": 0.5493,
+      "step": 6453
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1056664414699602e-07,
+      "loss": 0.6322,
+      "step": 6454
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1007650393379755e-07,
+      "loss": 0.6222,
+      "step": 6455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0958744043529201e-07,
+      "loss": 0.5419,
+      "step": 6456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0909945375916497e-07,
+      "loss": 0.6332,
+      "step": 6457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0861254401286825e-07,
+      "loss": 0.6185,
+      "step": 6458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0812671130361329e-07,
+      "loss": 0.6273,
+      "step": 6459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0764195573837677e-07,
+      "loss": 0.5361,
+      "step": 6460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0715827742389717e-07,
+      "loss": 0.525,
+      "step": 6461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0667567646667542e-07,
+      "loss": 0.6062,
+      "step": 6462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0619415297297597e-07,
+      "loss": 0.6268,
+      "step": 6463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0571370704882622e-07,
+      "loss": 0.615,
+      "step": 6464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0523433880001433e-07,
+      "loss": 0.6007,
+      "step": 6465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0475604833209363e-07,
+      "loss": 0.6163,
+      "step": 6466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0427883575037934e-07,
+      "loss": 0.6083,
+      "step": 6467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0380270115994906e-07,
+      "loss": 0.5246,
+      "step": 6468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0332764466564227e-07,
+      "loss": 0.6104,
+      "step": 6469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0285366637206251e-07,
+      "loss": 0.5696,
+      "step": 6470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.023807663835752e-07,
+      "loss": 0.5404,
+      "step": 6471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0190894480430924e-07,
+      "loss": 0.5877,
+      "step": 6472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0143820173815433e-07,
+      "loss": 0.6363,
+      "step": 6473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0096853728876365e-07,
+      "loss": 0.6235,
+      "step": 6474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0049995155955284e-07,
+      "loss": 0.6301,
+      "step": 6475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0003244465370043e-07,
+      "loss": 0.6355,
+      "step": 6476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.956601667414633e-08,
+      "loss": 0.542,
+      "step": 6477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.91006677235945e-08,
+      "loss": 0.6103,
+      "step": 6478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.863639790451019e-08,
+      "loss": 0.6409,
+      "step": 6479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.81732073191205e-08,
+      "loss": 0.5227,
+      "step": 6480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.771109606941554e-08,
+      "loss": 0.6192,
+      "step": 6481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.725006425714833e-08,
+      "loss": 0.6194,
+      "step": 6482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.679011198383381e-08,
+      "loss": 0.5895,
+      "step": 6483
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.633123935074873e-08,
+      "loss": 0.6108,
+      "step": 6484
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.587344645893282e-08,
+      "loss": 0.5631,
+      "step": 6485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.541673340918766e-08,
+      "loss": 0.6427,
+      "step": 6486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.496110030207673e-08,
+      "loss": 0.6031,
+      "step": 6487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.45065472379264e-08,
+      "loss": 0.5978,
+      "step": 6488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.405307431682498e-08,
+      "loss": 0.5342,
+      "step": 6489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.360068163862312e-08,
+      "loss": 0.6133,
+      "step": 6490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.314936930293283e-08,
+      "loss": 0.633,
+      "step": 6491
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.269913740912961e-08,
+      "loss": 0.5807,
+      "step": 6492
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.224998605634916e-08,
+      "loss": 0.654,
+      "step": 6493
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.180191534349069e-08,
+      "loss": 0.5957,
+      "step": 6494
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.135492536921586e-08,
+      "loss": 0.6172,
+      "step": 6495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.090901623194648e-08,
+      "loss": 0.6166,
+      "step": 6496
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.046418802986845e-08,
+      "loss": 0.5502,
+      "step": 6497
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.002044086092843e-08,
+      "loss": 0.6322,
+      "step": 6498
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.957777482283548e-08,
+      "loss": 0.6218,
+      "step": 6499
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.913619001305995e-08,
+      "loss": 0.6452,
+      "step": 6500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.869568652883465e-08,
+      "loss": 0.614,
+      "step": 6501
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.825626446715529e-08,
+      "loss": 0.601,
+      "step": 6502
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.781792392477728e-08,
+      "loss": 0.6069,
+      "step": 6503
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.73806649982195e-08,
+      "loss": 0.5677,
+      "step": 6504
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.694448778376163e-08,
+      "loss": 0.5377,
+      "step": 6505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.650939237744682e-08,
+      "loss": 0.6298,
+      "step": 6506
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.607537887507788e-08,
+      "loss": 0.6232,
+      "step": 6507
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.564244737222116e-08,
+      "loss": 0.6089,
+      "step": 6508
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.521059796420427e-08,
+      "loss": 0.6193,
+      "step": 6509
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.477983074611562e-08,
+      "loss": 0.5319,
+      "step": 6510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.435014581280654e-08,
+      "loss": 0.5541,
+      "step": 6511
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.392154325888858e-08,
+      "loss": 0.5301,
+      "step": 6512
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.34940231787379e-08,
+      "loss": 0.6082,
+      "step": 6513
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.306758566648865e-08,
+      "loss": 0.5612,
+      "step": 6514
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.264223081603961e-08,
+      "loss": 0.5154,
+      "step": 6515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.221795872104865e-08,
+      "loss": 0.5505,
+      "step": 6516
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.179476947493714e-08,
+      "loss": 0.6035,
+      "step": 6517
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.137266317088665e-08,
+      "loss": 0.6048,
+      "step": 6518
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.095163990184173e-08,
+      "loss": 0.6324,
+      "step": 6519
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.053169976050767e-08,
+      "loss": 0.6464,
+      "step": 6520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.011284283935105e-08,
+      "loss": 0.6099,
+      "step": 6521
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.969506923060032e-08,
+      "loss": 0.602,
+      "step": 6522
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.92783790262447e-08,
+      "loss": 0.623,
+      "step": 6523
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.88627723180363e-08,
+      "loss": 0.616,
+      "step": 6524
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.84482491974875e-08,
+      "loss": 0.5249,
+      "step": 6525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.803480975587197e-08,
+      "loss": 0.5976,
+      "step": 6526
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.76224540842252e-08,
+      "loss": 0.6178,
+      "step": 6527
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.721118227334456e-08,
+      "loss": 0.6355,
+      "step": 6528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.680099441378819e-08,
+      "loss": 0.6515,
+      "step": 6529
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.639189059587382e-08,
+      "loss": 0.5417,
+      "step": 6530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.598387090968495e-08,
+      "loss": 0.5958,
+      "step": 6531
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.55769354450614e-08,
+      "loss": 0.5423,
+      "step": 6532
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.517108429160758e-08,
+      "loss": 0.5416,
+      "step": 6533
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.476631753868812e-08,
+      "loss": 0.5425,
+      "step": 6534
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.436263527542787e-08,
+      "loss": 0.5245,
+      "step": 6535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.396003759071512e-08,
+      "loss": 0.6068,
+      "step": 6536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.355852457319734e-08,
+      "loss": 0.5958,
+      "step": 6537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.315809631128435e-08,
+      "loss": 0.6092,
+      "step": 6538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.275875289314616e-08,
+      "loss": 0.6299,
+      "step": 6539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.23604944067141e-08,
+      "loss": 0.5269,
+      "step": 6540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.196332093968083e-08,
+      "loss": 0.6293,
+      "step": 6541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.156723257950138e-08,
+      "loss": 0.623,
+      "step": 6542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.117222941338986e-08,
+      "loss": 0.5505,
+      "step": 6543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.077831152832226e-08,
+      "loss": 0.5602,
+      "step": 6544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.038547901103531e-08,
+      "loss": 0.5994,
+      "step": 6545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.999373194802761e-08,
+      "loss": 0.6133,
+      "step": 6546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.960307042555737e-08,
+      "loss": 0.6145,
+      "step": 6547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.92134945296452e-08,
+      "loss": 0.6329,
+      "step": 6548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.882500434607198e-08,
+      "loss": 0.5924,
+      "step": 6549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.843759996037924e-08,
+      "loss": 0.6369,
+      "step": 6550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.805128145786987e-08,
+      "loss": 0.616,
+      "step": 6551
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.766604892360751e-08,
+      "loss": 0.5304,
+      "step": 6552
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.72819024424165e-08,
+      "loss": 0.6078,
+      "step": 6553
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.689884209888254e-08,
+      "loss": 0.5822,
+      "step": 6554
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.651686797735202e-08,
+      "loss": 0.511,
+      "step": 6555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.613598016193157e-08,
+      "loss": 0.597,
+      "step": 6556
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.575617873648909e-08,
+      "loss": 0.5315,
+      "step": 6557
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.537746378465382e-08,
+      "loss": 0.5466,
+      "step": 6558
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.499983538981402e-08,
+      "loss": 0.6233,
+      "step": 6559
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.462329363512154e-08,
+      "loss": 0.6197,
+      "step": 6560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.424783860348616e-08,
+      "loss": 0.6346,
+      "step": 6561
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.387347037758007e-08,
+      "loss": 0.6009,
+      "step": 6562
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.350018903983513e-08,
+      "loss": 0.5375,
+      "step": 6563
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.312799467244446e-08,
+      "loss": 0.6103,
+      "step": 6564
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.275688735736141e-08,
+      "loss": 0.6354,
+      "step": 6565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.238686717630116e-08,
+      "loss": 0.6156,
+      "step": 6566
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.201793421073799e-08,
+      "loss": 0.6303,
+      "step": 6567
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.165008854190802e-08,
+      "loss": 0.6019,
+      "step": 6568
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.128333025080646e-08,
+      "loss": 0.6174,
+      "step": 6569
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.091765941818984e-08,
+      "loss": 0.6008,
+      "step": 6570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.055307612457706e-08,
+      "loss": 0.6197,
+      "step": 6571
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.018958045024392e-08,
+      "loss": 0.6268,
+      "step": 6572
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.982717247523028e-08,
+      "loss": 0.6362,
+      "step": 6573
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.946585227933455e-08,
+      "loss": 0.5567,
+      "step": 6574
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9105619942115275e-08,
+      "loss": 0.6088,
+      "step": 6575
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.874647554289292e-08,
+      "loss": 0.528,
+      "step": 6576
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8388419160747555e-08,
+      "loss": 0.5722,
+      "step": 6577
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.803145087451945e-08,
+      "loss": 0.6414,
+      "step": 6578
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.767557076281016e-08,
+      "loss": 0.6057,
+      "step": 6579
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.732077890398036e-08,
+      "loss": 0.5727,
+      "step": 6580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.696707537615254e-08,
+      "loss": 0.6013,
+      "step": 6581
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.661446025720829e-08,
+      "loss": 0.555,
+      "step": 6582
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.626293362479107e-08,
+      "loss": 0.5902,
+      "step": 6583
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.5912495556302295e-08,
+      "loss": 0.6056,
+      "step": 6584
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.5563146128906344e-08,
+      "loss": 0.5948,
+      "step": 6585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.521488541952613e-08,
+      "loss": 0.6163,
+      "step": 6586
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4867713504845325e-08,
+      "loss": 0.6279,
+      "step": 6587
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4521630461307204e-08,
+      "loss": 0.5799,
+      "step": 6588
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4176636365117495e-08,
+      "loss": 0.6172,
+      "step": 6589
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.383273129223987e-08,
+      "loss": 0.6012,
+      "step": 6590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.348991531839875e-08,
+      "loss": 0.6269,
+      "step": 6591
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.314818851907933e-08,
+      "loss": 0.5265,
+      "step": 6592
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2807550969526386e-08,
+      "loss": 0.6212,
+      "step": 6593
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.246800274474439e-08,
+      "loss": 0.6392,
+      "step": 6594
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2129543919500203e-08,
+      "loss": 0.6514,
+      "step": 6595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.179217456831753e-08,
+      "loss": 0.5212,
+      "step": 6596
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.145589476548307e-08,
+      "loss": 0.6098,
+      "step": 6597
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.112070458504259e-08,
+      "loss": 0.5451,
+      "step": 6598
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.078660410080094e-08,
+      "loss": 0.6384,
+      "step": 6599
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0453593386323716e-08,
+      "loss": 0.5977,
+      "step": 6600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0121672514937823e-08,
+      "loss": 0.5984,
+      "step": 6601
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.979084155972813e-08,
+      "loss": 0.6096,
+      "step": 6602
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9461100593541365e-08,
+      "loss": 0.6032,
+      "step": 6603
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.913244968898279e-08,
+      "loss": 0.6012,
+      "step": 6604
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8804888918417837e-08,
+      "loss": 0.505,
+      "step": 6605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.847841835397216e-08,
+      "loss": 0.6187,
+      "step": 6606
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8153038067532686e-08,
+      "loss": 0.5987,
+      "step": 6607
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.782874813074434e-08,
+      "loss": 0.5573,
+      "step": 6608
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7505548615012223e-08,
+      "loss": 0.6038,
+      "step": 6609
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.718343959150273e-08,
+      "loss": 0.5853,
+      "step": 6610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.686242113114081e-08,
+      "loss": 0.5356,
+      "step": 6611
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.654249330461158e-08,
+      "loss": 0.5518,
+      "step": 6612
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.62236561823598e-08,
+      "loss": 0.6245,
+      "step": 6613
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.590590983459153e-08,
+      "loss": 0.5993,
+      "step": 6614
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.558925433127026e-08,
+      "loss": 0.6097,
+      "step": 6615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5273689742120766e-08,
+      "loss": 0.6302,
+      "step": 6616
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.495921613662746e-08,
+      "loss": 0.5881,
+      "step": 6617
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.464583358403496e-08,
+      "loss": 0.5998,
+      "step": 6618
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.43335421533464e-08,
+      "loss": 0.5843,
+      "step": 6619
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.402234191332566e-08,
+      "loss": 0.6209,
+      "step": 6620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.371223293249627e-08,
+      "loss": 0.6412,
+      "step": 6621
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3403215279141376e-08,
+      "loss": 0.5951,
+      "step": 6622
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.309528902130322e-08,
+      "loss": 0.6348,
+      "step": 6623
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2788454226784236e-08,
+      "loss": 0.613,
+      "step": 6624
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.248271096314705e-08,
+      "loss": 0.6162,
+      "step": 6625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2178059297713345e-08,
+      "loss": 0.5751,
+      "step": 6626
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.187449929756393e-08,
+      "loss": 0.5879,
+      "step": 6627
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.157203102954088e-08,
+      "loss": 0.6036,
+      "step": 6628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1270654560243706e-08,
+      "loss": 0.5885,
+      "step": 6629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.097036995603321e-08,
+      "loss": 0.5524,
+      "step": 6630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.067117728302927e-08,
+      "loss": 0.6426,
+      "step": 6631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.037307660711087e-08,
+      "loss": 0.6344,
+      "step": 6632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.007606799391717e-08,
+      "loss": 0.6357,
+      "step": 6633
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.978015150884695e-08,
+      "loss": 0.6229,
+      "step": 6634
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.948532721705811e-08,
+      "loss": 0.6325,
+      "step": 6635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.91915951834676e-08,
+      "loss": 0.6187,
+      "step": 6636
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.889895547275313e-08,
+      "loss": 0.5989,
+      "step": 6637
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.860740814935038e-08,
+      "loss": 0.6007,
+      "step": 6638
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.831695327745633e-08,
+      "loss": 0.6079,
+      "step": 6639
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8027590921025924e-08,
+      "loss": 0.5164,
+      "step": 6640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.773932114377321e-08,
+      "loss": 0.5084,
+      "step": 6641
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.745214400917407e-08,
+      "loss": 0.6276,
+      "step": 6642
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.716605958046071e-08,
+      "loss": 0.5883,
+      "step": 6643
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.688106792062662e-08,
+      "loss": 0.6273,
+      "step": 6644
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6597169092424944e-08,
+      "loss": 0.6143,
+      "step": 6645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6314363158366226e-08,
+      "loss": 0.6188,
+      "step": 6646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.603265018072288e-08,
+      "loss": 0.6395,
+      "step": 6647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5752030221524736e-08,
+      "loss": 0.6296,
+      "step": 6648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.547250334256236e-08,
+      "loss": 0.5405,
+      "step": 6649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.51940696053843e-08,
+      "loss": 0.583,
+      "step": 6650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4916729071298724e-08,
+      "loss": 0.5609,
+      "step": 6651
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.464048180137403e-08,
+      "loss": 0.6006,
+      "step": 6652
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4365327856436534e-08,
+      "loss": 0.5831,
+      "step": 6653
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.409126729707335e-08,
+      "loss": 0.5905,
+      "step": 6654
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3818300183629525e-08,
+      "loss": 0.6238,
+      "step": 6655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.354642657621032e-08,
+      "loss": 0.6356,
+      "step": 6656
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.327564653467896e-08,
+      "loss": 0.6106,
+      "step": 6657
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3005960118658865e-08,
+      "loss": 0.6285,
+      "step": 6658
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.273736738753308e-08,
+      "loss": 0.5384,
+      "step": 6659
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.246986840044208e-08,
+      "loss": 0.6078,
+      "step": 6660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.220346321628709e-08,
+      "loss": 0.6244,
+      "step": 6661
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.193815189372895e-08,
+      "loss": 0.5523,
+      "step": 6662
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.167393449118539e-08,
+      "loss": 0.534,
+      "step": 6663
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1410811066834855e-08,
+      "loss": 0.52,
+      "step": 6664
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1148781678614345e-08,
+      "loss": 0.5455,
+      "step": 6665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.08878463842216e-08,
+      "loss": 0.6291,
+      "step": 6666
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.062800524111065e-08,
+      "loss": 0.5433,
+      "step": 6667
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.036925830649684e-08,
+      "loss": 0.5615,
+      "step": 6668
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.011160563735349e-08,
+      "loss": 0.6346,
+      "step": 6669
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.985504729041244e-08,
+      "loss": 0.6298,
+      "step": 6670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.959958332216739e-08,
+      "loss": 0.6254,
+      "step": 6671
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9345213788867232e-08,
+      "loss": 0.5465,
+      "step": 6672
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9091938746522718e-08,
+      "loss": 0.6042,
+      "step": 6673
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8839758250902572e-08,
+      "loss": 0.511,
+      "step": 6674
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8588672357534043e-08,
+      "loss": 0.6456,
+      "step": 6675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8338681121704014e-08,
+      "loss": 0.5279,
+      "step": 6676
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8089784598457903e-08,
+      "loss": 0.6388,
+      "step": 6677
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.784198284260131e-08,
+      "loss": 0.5252,
+      "step": 6678
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7595275908697262e-08,
+      "loss": 0.6164,
+      "step": 6679
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7349663851067853e-08,
+      "loss": 0.5521,
+      "step": 6680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7105146723795385e-08,
+      "loss": 0.5685,
+      "step": 6681
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6861724580719562e-08,
+      "loss": 0.5346,
+      "step": 6682
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6619397475439735e-08,
+      "loss": 0.5881,
+      "step": 6683
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6378165461313776e-08,
+      "loss": 0.5193,
+      "step": 6684
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6138028591459752e-08,
+      "loss": 0.587,
+      "step": 6685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5898986918752588e-08,
+      "loss": 0.6254,
+      "step": 6686
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.566104049582685e-08,
+      "loss": 0.5257,
+      "step": 6687
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5424189375076735e-08,
+      "loss": 0.5874,
+      "step": 6688
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.518843360865386e-08,
+      "loss": 0.5608,
+      "step": 6689
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4953773248470038e-08,
+      "loss": 0.6399,
+      "step": 6690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4720208346195596e-08,
+      "loss": 0.6163,
+      "step": 6691
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.448773895325829e-08,
+      "loss": 0.6196,
+      "step": 6692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4256365120846058e-08,
+      "loss": 0.511,
+      "step": 6693
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4026086899905376e-08,
+      "loss": 0.6236,
+      "step": 6694
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3796904341141236e-08,
+      "loss": 0.6247,
+      "step": 6695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3568817495017716e-08,
+      "loss": 0.5172,
+      "step": 6696
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3341826411756863e-08,
+      "loss": 0.6103,
+      "step": 6697
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3115931141340366e-08,
+      "loss": 0.5522,
+      "step": 6698
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2891131733507876e-08,
+      "loss": 0.593,
+      "step": 6699
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2667428237758692e-08,
+      "loss": 0.5538,
+      "step": 6700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2444820703349523e-08,
+      "loss": 0.5115,
+      "step": 6701
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2223309179296714e-08,
+      "loss": 0.5747,
+      "step": 6702
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.200289371437514e-08,
+      "loss": 0.6388,
+      "step": 6703
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1783574357117642e-08,
+      "loss": 0.5785,
+      "step": 6704
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.156535115581726e-08,
+      "loss": 0.604,
+      "step": 6705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1348224158523333e-08,
+      "loss": 0.6225,
+      "step": 6706
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.113219341304651e-08,
+      "loss": 0.618,
+      "step": 6707
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0917258966953735e-08,
+      "loss": 0.6183,
+      "step": 6708
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.07034208675716e-08,
+      "loss": 0.6026,
+      "step": 6709
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0490679161986328e-08,
+      "loss": 0.649,
+      "step": 6710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0279033897040445e-08,
+      "loss": 0.6001,
+      "step": 6711
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0068485119336123e-08,
+      "loss": 0.5974,
+      "step": 6712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9859032875235164e-08,
+      "loss": 0.6018,
+      "step": 6713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9650677210856227e-08,
+      "loss": 0.6351,
+      "step": 6714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.944341817207762e-08,
+      "loss": 0.6006,
+      "step": 6715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.923725580453617e-08,
+      "loss": 0.6194,
+      "step": 6716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9032190153625564e-08,
+      "loss": 0.523,
+      "step": 6717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8828221264500235e-08,
+      "loss": 0.6024,
+      "step": 6718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8625349182072592e-08,
+      "loss": 0.6456,
+      "step": 6719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8423573951012465e-08,
+      "loss": 0.6289,
+      "step": 6720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8222895615748748e-08,
+      "loss": 0.5754,
+      "step": 6721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8023314220469434e-08,
+      "loss": 0.5366,
+      "step": 6722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7824829809119925e-08,
+      "loss": 0.6087,
+      "step": 6723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7627442425404707e-08,
+      "loss": 0.6452,
+      "step": 6724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.743115211278734e-08,
+      "loss": 0.6144,
+      "step": 6725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.723595891448826e-08,
+      "loss": 0.5296,
+      "step": 6726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7041862873487524e-08,
+      "loss": 0.6075,
+      "step": 6727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6848864032522615e-08,
+      "loss": 0.6069,
+      "step": 6728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6656962434091206e-08,
+      "loss": 0.6245,
+      "step": 6729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6466158120447272e-08,
+      "loss": 0.5361,
+      "step": 6730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6276451133603878e-08,
+      "loss": 0.5273,
+      "step": 6731
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6087841515333714e-08,
+      "loss": 0.6315,
+      "step": 6732
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.590032930716634e-08,
+      "loss": 0.6361,
+      "step": 6733
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5713914550389843e-08,
+      "loss": 0.6331,
+      "step": 6734
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5528597286051937e-08,
+      "loss": 0.6102,
+      "step": 6735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.53443775549561e-08,
+      "loss": 0.5804,
+      "step": 6736
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5161255397667663e-08,
+      "loss": 0.6067,
+      "step": 6737
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.497923085450714e-08,
+      "loss": 0.6105,
+      "step": 6738
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4798303965554706e-08,
+      "loss": 0.6548,
+      "step": 6739
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4618474770649038e-08,
+      "loss": 0.6081,
+      "step": 6740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.44397433093868e-08,
+      "loss": 0.5687,
+      "step": 6741
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4262109621122622e-08,
+      "loss": 0.5942,
+      "step": 6742
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4085573744970216e-08,
+      "loss": 0.5908,
+      "step": 6743
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.391013571980071e-08,
+      "loss": 0.5492,
+      "step": 6744
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.373579558424487e-08,
+      "loss": 0.5864,
+      "step": 6745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3562553376689214e-08,
+      "loss": 0.6216,
+      "step": 6746
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3390409135281001e-08,
+      "loss": 0.6276,
+      "step": 6747
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.321936289792436e-08,
+      "loss": 0.6171,
+      "step": 6748
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3049414702281938e-08,
+      "loss": 0.5435,
+      "step": 6749
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2880564585775468e-08,
+      "loss": 0.5486,
+      "step": 6750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.27128125855841e-08,
+      "loss": 0.5245,
+      "step": 6751
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2546158738643843e-08,
+      "loss": 0.5414,
+      "step": 6752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.238060308165201e-08,
+      "loss": 0.5548,
+      "step": 6753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2216145651061107e-08,
+      "loss": 0.5755,
+      "step": 6754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2052786483083834e-08,
+      "loss": 0.5504,
+      "step": 6755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1890525613690307e-08,
+      "loss": 0.6013,
+      "step": 6756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1729363078608057e-08,
+      "loss": 0.6072,
+      "step": 6757
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1569298913324811e-08,
+      "loss": 0.5267,
+      "step": 6758
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.141033315308404e-08,
+      "loss": 0.6131,
+      "step": 6759
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.125246583288886e-08,
+      "loss": 0.6189,
+      "step": 6760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1095696987500348e-08,
+      "loss": 0.6384,
+      "step": 6761
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0940026651437562e-08,
+      "loss": 0.61,
+      "step": 6762
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0785454858977529e-08,
+      "loss": 0.621,
+      "step": 6763
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0631981644155243e-08,
+      "loss": 0.6328,
+      "step": 6764
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0479607040763673e-08,
+      "loss": 0.6573,
+      "step": 6765
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0328331082355425e-08,
+      "loss": 0.5923,
+      "step": 6766
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0178153802239409e-08,
+      "loss": 0.6066,
+      "step": 6767
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0029075233483066e-08,
+      "loss": 0.5724,
+      "step": 6768
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.881095408912355e-09,
+      "loss": 0.5893,
+      "step": 6769
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.734214361111215e-09,
+      "loss": 0.5501,
+      "step": 6770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.588432122420999e-09,
+      "loss": 0.5804,
+      "step": 6771
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.443748724941027e-09,
+      "loss": 0.5631,
+      "step": 6772
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.300164200530815e-09,
+      "loss": 0.5897,
+      "step": 6773
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.157678580805074e-09,
+      "loss": 0.5917,
+      "step": 6774
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.016291897138152e-09,
+      "loss": 0.5715,
+      "step": 6775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.876004180662368e-09,
+      "loss": 0.5954,
+      "step": 6776
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.7368154622669e-09,
+      "loss": 0.6166,
+      "step": 6777
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.598725772600569e-09,
+      "loss": 0.6078,
+      "step": 6778
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.46173514206905e-09,
+      "loss": 0.6013,
+      "step": 6779
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.325843600836547e-09,
+      "loss": 0.5353,
+      "step": 6780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.191051178824683e-09,
+      "loss": 0.5368,
+      "step": 6781
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.057357905714158e-09,
+      "loss": 0.6159,
+      "step": 6782
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.924763810941982e-09,
+      "loss": 0.615,
+      "step": 6783
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.793268923705356e-09,
+      "loss": 0.6138,
+      "step": 6784
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.662873272957227e-09,
+      "loss": 0.6354,
+      "step": 6785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.533576887410188e-09,
+      "loss": 0.5773,
+      "step": 6786
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.405379795533685e-09,
+      "loss": 0.6111,
+      "step": 6787
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.278282025555139e-09,
+      "loss": 0.5692,
+      "step": 6788
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.1522836054604975e-09,
+      "loss": 0.6369,
+      "step": 6789
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.027384562994233e-09,
+      "loss": 0.5436,
+      "step": 6790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.903584925657125e-09,
+      "loss": 0.6037,
+      "step": 6791
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.780884720708481e-09,
+      "loss": 0.6069,
+      "step": 6792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.659283975166131e-09,
+      "loss": 0.6502,
+      "step": 6793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.5387827158053255e-09,
+      "loss": 0.6279,
+      "step": 6794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.419380969159283e-09,
+      "loss": 0.5953,
+      "step": 6795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.301078761519197e-09,
+      "loss": 0.6195,
+      "step": 6796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.183876118934229e-09,
+      "loss": 0.6176,
+      "step": 6797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.067773067210958e-09,
+      "loss": 0.6147,
+      "step": 6798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.95276963191449e-09,
+      "loss": 0.5808,
+      "step": 6799
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.838865838366792e-09,
+      "loss": 0.6298,
+      "step": 6800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.726061711649467e-09,
+      "loss": 0.6156,
+      "step": 6801
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.614357276600424e-09,
+      "loss": 0.5224,
+      "step": 6802
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.503752557815545e-09,
+      "loss": 0.5991,
+      "step": 6803
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3942475796492366e-09,
+      "loss": 0.5359,
+      "step": 6804
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.285842366213878e-09,
+      "loss": 0.6188,
+      "step": 6805
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.178536941378154e-09,
+      "loss": 0.571,
+      "step": 6806
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.072331328770941e-09,
+      "loss": 0.5943,
+      "step": 6807
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9672255517768665e-09,
+      "loss": 0.5576,
+      "step": 6808
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.86321963353964e-09,
+      "loss": 0.632,
+      "step": 6809
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.760313596960387e-09,
+      "loss": 0.6522,
+      "step": 6810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.658507464698203e-09,
+      "loss": 0.6154,
+      "step": 6811
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.557801259169048e-09,
+      "loss": 0.6184,
+      "step": 6812
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4581950025479605e-09,
+      "loss": 0.605,
+      "step": 6813
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.359688716767396e-09,
+      "loss": 0.6132,
+      "step": 6814
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.262282423517783e-09,
+      "loss": 0.5891,
+      "step": 6815
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1659761442464085e-09,
+      "loss": 0.6191,
+      "step": 6816
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.070769900159088e-09,
+      "loss": 0.6402,
+      "step": 6817
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9766637122196085e-09,
+      "loss": 0.6209,
+      "step": 6818
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.883657601149171e-09,
+      "loss": 0.6048,
+      "step": 6819
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.791751587426951e-09,
+      "loss": 0.6045,
+      "step": 6820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.700945691289537e-09,
+      "loss": 0.6043,
+      "step": 6821
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.611239932732047e-09,
+      "loss": 0.6235,
+      "step": 6822
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.522634331505903e-09,
+      "loss": 0.6223,
+      "step": 6823
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.435128907122165e-09,
+      "loss": 0.6096,
+      "step": 6824
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.348723678847643e-09,
+      "loss": 0.613,
+      "step": 6825
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2634186657093393e-09,
+      "loss": 0.5387,
+      "step": 6826
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1792138864894517e-09,
+      "loss": 0.62,
+      "step": 6827
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.096109359729815e-09,
+      "loss": 0.5551,
+      "step": 6828
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0141051037285705e-09,
+      "loss": 0.6197,
+      "step": 6829
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9332011365429404e-09,
+      "loss": 0.5565,
+      "step": 6830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8533974759864523e-09,
+      "loss": 0.6214,
+      "step": 6831
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7746941396322726e-09,
+      "loss": 0.6628,
+      "step": 6832
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.697091144808761e-09,
+      "loss": 0.6278,
+      "step": 6833
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.620588508603916e-09,
+      "loss": 0.5929,
+      "step": 6834
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.545186247863707e-09,
+      "loss": 0.6096,
+      "step": 6835
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.470884379189853e-09,
+      "loss": 0.6207,
+      "step": 6836
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.397682918943156e-09,
+      "loss": 0.6179,
+      "step": 6837
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3255818832423894e-09,
+      "loss": 0.6155,
+      "step": 6838
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.254581287963187e-09,
+      "loss": 0.5645,
+      "step": 6839
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.184681148739154e-09,
+      "loss": 0.6333,
+      "step": 6840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.115881480961313e-09,
+      "loss": 0.5418,
+      "step": 6841
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.048182299779211e-09,
+      "loss": 0.6334,
+      "step": 6842
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.981583620099814e-09,
+      "loss": 0.518,
+      "step": 6843
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9160854565869467e-09,
+      "loss": 0.5497,
+      "step": 6844
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.851687823662962e-09,
+      "loss": 0.5976,
+      "step": 6845
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.788390735507628e-09,
+      "loss": 0.5126,
+      "step": 6846
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7261942060581295e-09,
+      "loss": 0.6044,
+      "step": 6847
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6650982490096223e-09,
+      "loss": 0.6237,
+      "step": 6848
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6051028778146794e-09,
+      "loss": 0.5371,
+      "step": 6849
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5462081056843992e-09,
+      "loss": 0.5461,
+      "step": 6850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4884139455861868e-09,
+      "loss": 0.6135,
+      "step": 6851
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4317204102459736e-09,
+      "loss": 0.5952,
+      "step": 6852
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3761275121465523e-09,
+      "loss": 0.5816,
+      "step": 6853
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.321635263530352e-09,
+      "loss": 0.6057,
+      "step": 6854
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2682436763949978e-09,
+      "loss": 0.5521,
+      "step": 6855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2159527624971967e-09,
+      "loss": 0.599,
+      "step": 6856
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1647625333505163e-09,
+      "loss": 0.6031,
+      "step": 6857
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1146730002270511e-09,
+      "loss": 0.5911,
+      "step": 6858
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0656841741552015e-09,
+      "loss": 0.6036,
+      "step": 6859
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0177960659230046e-09,
+      "loss": 0.5937,
+      "step": 6860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.710086860742485e-10,
+      "loss": 0.5945,
+      "step": 6861
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.253220449112476e-10,
+      "loss": 0.6291,
+      "step": 6862
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.80736152493733e-10,
+      "loss": 0.6506,
+      "step": 6863
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.372510186388516e-10,
+      "loss": 0.601,
+      "step": 6864
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.948666529217219e-10,
+      "loss": 0.5633,
+      "step": 6865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.535830646754339e-10,
+      "loss": 0.6223,
+      "step": 6866
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.134002629899384e-10,
+      "loss": 0.5498,
+      "step": 6867
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.743182567131578e-10,
+      "loss": 0.6287,
+      "step": 6868
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.36337054450431e-10,
+      "loss": 0.6369,
+      "step": 6869
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.994566645650679e-10,
+      "loss": 0.6173,
+      "step": 6870
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.636770951772396e-10,
+      "loss": 0.604,
+      "step": 6871
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.289983541661992e-10,
+      "loss": 0.5921,
+      "step": 6872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.954204491669501e-10,
+      "loss": 0.5158,
+      "step": 6873
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.629433875735778e-10,
+      "loss": 0.6162,
+      "step": 6874
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3156717653702883e-10,
+      "loss": 0.6356,
+      "step": 6875
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0129182296622104e-10,
+      "loss": 0.6084,
+      "step": 6876
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7211733352748856e-10,
+      "loss": 0.596,
+      "step": 6877
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.440437146445819e-10,
+      "loss": 0.5916,
+      "step": 6878
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.170709724986676e-10,
+      "loss": 0.6067,
+      "step": 6879
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.91199113029994e-10,
+      "loss": 0.598,
+      "step": 6880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6642814193456045e-10,
+      "loss": 0.5873,
+      "step": 6881
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4275806466633743e-10,
+      "loss": 0.6287,
+      "step": 6882
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.201888864378221e-10,
+      "loss": 0.5181,
+      "step": 6883
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.987206122189278e-10,
+      "loss": 0.6279,
+      "step": 6884
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7835324673587394e-10,
+      "loss": 0.5433,
+      "step": 6885
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5908679447340646e-10,
+      "loss": 0.5979,
+      "step": 6886
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4092125967479776e-10,
+      "loss": 0.601,
+      "step": 6887
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2385664633851603e-10,
+      "loss": 0.6191,
+      "step": 6888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0789295822322132e-10,
+      "loss": 0.6757,
+      "step": 6889
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.303019884387976e-11,
+      "loss": 0.6438,
+      "step": 6890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.926837147231859e-11,
+      "loss": 0.6145,
+      "step": 6891
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.660747913944665e-11,
+      "loss": 0.615,
+      "step": 6892
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.5047524632478865e-11,
+      "loss": 0.5469,
+      "step": 6893
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4588510497711735e-11,
+      "loss": 0.6302,
+      "step": 6894
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.52304390371927e-11,
+      "loss": 0.6056,
+      "step": 6895
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6973312312050804e-11,
+      "loss": 0.5999,
+      "step": 6896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.981713213972114e-11,
+      "loss": 0.5949,
+      "step": 6897
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3761900097275515e-11,
+      "loss": 0.658,
+      "step": 6898
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.807617516426448e-12,
+      "loss": 0.5925,
+      "step": 6899
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.954285489633392e-12,
+      "loss": 0.6155,
+      "step": 6900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.201904863996518e-12,
+      "loss": 0.6086,
+      "step": 6901
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.504762462527069e-13,
+      "loss": 0.599,
+      "step": 6902
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.5722,
+      "step": 6903
+    },
+    {
+      "epoch": 1.0,
+      "step": 6903,
+      "total_flos": 6.0981125513216e+16,
+      "train_loss": 0.009743342240029757,
+      "train_runtime": 4572.0401,
+      "train_samples_per_second": 1545.933,
+      "train_steps_per_second": 1.51
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 6903,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 6.0981125513216e+16,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}