diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,60025 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.666666666666668e-08,
+      "loss": 4.2933,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3333333333333336e-07,
+      "loss": 5.5119,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 4.5678,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.666666666666667e-07,
+      "loss": 4.7196,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3333333333333335e-07,
+      "loss": 4.5042,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 4.0815,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.666666666666667e-07,
+      "loss": 4.7187,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.333333333333335e-07,
+      "loss": 4.8599,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 4.442,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 4.0855,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.333333333333334e-07,
+      "loss": 4.2954,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 4.3323,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.666666666666668e-07,
+      "loss": 4.5357,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.333333333333334e-07,
+      "loss": 4.9926,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 4.6079,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.066666666666667e-06,
+      "loss": 4.5357,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1333333333333334e-06,
+      "loss": 4.5712,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 4.4225,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2666666666666669e-06,
+      "loss": 3.5053,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 3.8387,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 4.912,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4666666666666669e-06,
+      "loss": 4.251,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5333333333333334e-06,
+      "loss": 4.5226,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 3.8137,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 3.479,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7333333333333336e-06,
+      "loss": 4.1615,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 3.2372,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8666666666666669e-06,
+      "loss": 3.2985,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9333333333333336e-06,
+      "loss": 3.6659,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 4.5036,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0666666666666666e-06,
+      "loss": 2.8007,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.133333333333334e-06,
+      "loss": 2.795,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2e-06,
+      "loss": 3.1444,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.266666666666667e-06,
+      "loss": 3.4938,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3333333333333336e-06,
+      "loss": 3.7747,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 3.8282,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.466666666666667e-06,
+      "loss": 4.7914,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5333333333333338e-06,
+      "loss": 2.8539,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6e-06,
+      "loss": 2.4869,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 2.0564,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7333333333333336e-06,
+      "loss": 3.2159,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 3.3486,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.866666666666667e-06,
+      "loss": 3.0701,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9333333333333338e-06,
+      "loss": 3.8482,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 2.2338,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.066666666666667e-06,
+      "loss": 3.08,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.133333333333334e-06,
+      "loss": 3.2021,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 2.709,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.266666666666667e-06,
+      "loss": 2.9274,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 2.7837,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 3.8124,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4666666666666672e-06,
+      "loss": 3.1106,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5333333333333335e-06,
+      "loss": 3.7357,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 3.3175,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6666666666666666e-06,
+      "loss": 2.8638,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7333333333333337e-06,
+      "loss": 3.5825,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 2.7092,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.866666666666667e-06,
+      "loss": 2.7498,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9333333333333335e-06,
+      "loss": 3.2523,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.7717,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.066666666666667e-06,
+      "loss": 4.5202,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.133333333333333e-06,
+      "loss": 3.8026,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 1.8961,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.266666666666668e-06,
+      "loss": 3.6237,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.333333333333334e-06,
+      "loss": 2.5701,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4e-06,
+      "loss": 3.5209,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4666666666666665e-06,
+      "loss": 3.7411,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.533333333333334e-06,
+      "loss": 2.6935,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 2.3858,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 3.1032,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7333333333333335e-06,
+      "loss": 3.615,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 3.4406,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.866666666666667e-06,
+      "loss": 2.8169,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.933333333333334e-06,
+      "loss": 1.9669,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 3.8514,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.0666666666666676e-06,
+      "loss": 3.298,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.133333333333334e-06,
+      "loss": 3.2946,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2e-06,
+      "loss": 2.0763,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2666666666666665e-06,
+      "loss": 3.9673,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 3.5898,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 2.766,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.466666666666667e-06,
+      "loss": 3.0955,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.533333333333334e-06,
+      "loss": 2.4074,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 3.3946,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.666666666666667e-06,
+      "loss": 2.1353,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.733333333333334e-06,
+      "loss": 2.8928,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8e-06,
+      "loss": 2.6594,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8666666666666675e-06,
+      "loss": 2.8122,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.933333333333335e-06,
+      "loss": 2.8166,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-06,
+      "loss": 3.7096,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.066666666666667e-06,
+      "loss": 3.5962,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.133333333333334e-06,
+      "loss": 2.8742,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 3.6359,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.266666666666668e-06,
+      "loss": 3.2908,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.333333333333333e-06,
+      "loss": 4.3522,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 3.0058,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.466666666666667e-06,
+      "loss": 2.9743,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.533333333333334e-06,
+      "loss": 3.138,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 2.9636,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 2.5023,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.733333333333334e-06,
+      "loss": 3.1038,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 2.3478,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.866666666666667e-06,
+      "loss": 2.8765,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.9333333333333344e-06,
+      "loss": 2.7639,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7e-06,
+      "loss": 3.8603,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.066666666666667e-06,
+      "loss": 2.9888,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.133333333333334e-06,
+      "loss": 2.4059,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 3.2029,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.266666666666668e-06,
+      "loss": 2.6066,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 2.0874,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.4e-06,
+      "loss": 3.6698,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 2.3297,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.533333333333334e-06,
+      "loss": 2.4459,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 3.1341,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.666666666666667e-06,
+      "loss": 3.6682,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.733333333333334e-06,
+      "loss": 2.7486,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 2.2661,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.866666666666667e-06,
+      "loss": 2.3888,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.933333333333334e-06,
+      "loss": 2.8409,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 3.3906,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.066666666666667e-06,
+      "loss": 2.4848,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.133333333333334e-06,
+      "loss": 3.1851,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.2e-06,
+      "loss": 4.0076,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.266666666666667e-06,
+      "loss": 2.6639,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 3.9118,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 2.3095,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.466666666666668e-06,
+      "loss": 3.2617,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 2.5881,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.6e-06,
+      "loss": 3.063,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 3.4773,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.733333333333333e-06,
+      "loss": 3.26,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.8e-06,
+      "loss": 2.2576,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.866666666666668e-06,
+      "loss": 3.0813,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.933333333333333e-06,
+      "loss": 2.4005,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-06,
+      "loss": 1.928,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 1.5051,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.133333333333335e-06,
+      "loss": 1.8026,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 3.1605,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.266666666666667e-06,
+      "loss": 3.677,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 2.573,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.4e-06,
+      "loss": 4.0978,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.466666666666667e-06,
+      "loss": 2.2895,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.533333333333334e-06,
+      "loss": 2.4096,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 2.4762,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.666666666666667e-06,
+      "loss": 2.849,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.733333333333334e-06,
+      "loss": 3.2764,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 3.1781,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.866666666666668e-06,
+      "loss": 4.1019,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.933333333333334e-06,
+      "loss": 3.6493,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 3.3117,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0066666666666666e-05,
+      "loss": 2.6779,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0133333333333335e-05,
+      "loss": 2.9406,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.02e-05,
+      "loss": 3.9508,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0266666666666668e-05,
+      "loss": 2.8167,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0333333333333335e-05,
+      "loss": 3.5587,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.04e-05,
+      "loss": 3.5881,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0466666666666668e-05,
+      "loss": 2.9942,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0533333333333333e-05,
+      "loss": 3.5651,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0600000000000002e-05,
+      "loss": 3.2932,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 3.412,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0733333333333333e-05,
+      "loss": 3.5347,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 1.9562,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0866666666666667e-05,
+      "loss": 2.5026,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0933333333333334e-05,
+      "loss": 2.6285,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 2.5184,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1066666666666669e-05,
+      "loss": 3.1378,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1133333333333334e-05,
+      "loss": 2.6477,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 2.7095,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1266666666666668e-05,
+      "loss": 2.4116,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1333333333333334e-05,
+      "loss": 2.1314,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.14e-05,
+      "loss": 3.4205,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1466666666666668e-05,
+      "loss": 2.9413,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1533333333333334e-05,
+      "loss": 3.7892,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.16e-05,
+      "loss": 3.5739,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 3.2331,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1733333333333335e-05,
+      "loss": 2.0237,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.18e-05,
+      "loss": 3.2918,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.186666666666667e-05,
+      "loss": 3.2144,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1933333333333335e-05,
+      "loss": 3.0643,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-05,
+      "loss": 3.2304,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.206666666666667e-05,
+      "loss": 4.0435,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2133333333333335e-05,
+      "loss": 1.9489,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.22e-05,
+      "loss": 2.7057,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2266666666666667e-05,
+      "loss": 2.6728,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2333333333333334e-05,
+      "loss": 2.7149,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2400000000000002e-05,
+      "loss": 3.0875,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2466666666666667e-05,
+      "loss": 2.514,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2533333333333336e-05,
+      "loss": 3.3154,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 2.9255,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2666666666666667e-05,
+      "loss": 2.9082,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2733333333333336e-05,
+      "loss": 2.8262,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 2.7956,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2866666666666667e-05,
+      "loss": 3.0438,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2933333333333334e-05,
+      "loss": 3.1871,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 2.1825,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3066666666666668e-05,
+      "loss": 3.5393,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3133333333333334e-05,
+      "loss": 2.7879,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3200000000000002e-05,
+      "loss": 4.6184,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3266666666666668e-05,
+      "loss": 3.3483,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 2.8329,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 3.3093,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3466666666666668e-05,
+      "loss": 2.1591,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3533333333333333e-05,
+      "loss": 2.8405,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 2.7651,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3666666666666667e-05,
+      "loss": 3.4977,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3733333333333335e-05,
+      "loss": 2.5857,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.38e-05,
+      "loss": 2.7792,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3866666666666669e-05,
+      "loss": 3.2431,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3933333333333334e-05,
+      "loss": 3.195,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4e-05,
+      "loss": 2.1388,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4066666666666669e-05,
+      "loss": 2.5721,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4133333333333334e-05,
+      "loss": 3.5551,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4200000000000001e-05,
+      "loss": 3.5043,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4266666666666668e-05,
+      "loss": 2.4144,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4333333333333334e-05,
+      "loss": 2.3008,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 1.9461,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4466666666666668e-05,
+      "loss": 2.9322,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4533333333333335e-05,
+      "loss": 2.3639,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.46e-05,
+      "loss": 2.4887,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 2.262,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4733333333333335e-05,
+      "loss": 2.8818,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.48e-05,
+      "loss": 2.8382,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4866666666666668e-05,
+      "loss": 3.5164,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4933333333333335e-05,
+      "loss": 2.6706,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 2.3828,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5066666666666668e-05,
+      "loss": 3.1831,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5133333333333335e-05,
+      "loss": 4.0644,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 3.4847,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5266666666666667e-05,
+      "loss": 2.653,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5333333333333334e-05,
+      "loss": 3.2539,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.54e-05,
+      "loss": 3.2927,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.546666666666667e-05,
+      "loss": 2.9569,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5533333333333333e-05,
+      "loss": 2.6044,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5600000000000003e-05,
+      "loss": 2.9668,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5666666666666667e-05,
+      "loss": 2.652,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5733333333333334e-05,
+      "loss": 3.3727,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.58e-05,
+      "loss": 2.755,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.586666666666667e-05,
+      "loss": 2.2876,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5933333333333336e-05,
+      "loss": 2.9842,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 2.9475,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.606666666666667e-05,
+      "loss": 3.2021,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6133333333333334e-05,
+      "loss": 2.7995,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.62e-05,
+      "loss": 1.9904,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6266666666666668e-05,
+      "loss": 2.4949,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6333333333333335e-05,
+      "loss": 3.3996,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.64e-05,
+      "loss": 2.4739,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.646666666666667e-05,
+      "loss": 2.8194,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6533333333333333e-05,
+      "loss": 2.8638,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.66e-05,
+      "loss": 3.632,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 2.8138,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6733333333333335e-05,
+      "loss": 3.4731,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 3.5483,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.686666666666667e-05,
+      "loss": 2.0005,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6933333333333336e-05,
+      "loss": 2.3123,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7e-05,
+      "loss": 3.1361,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.706666666666667e-05,
+      "loss": 3.6121,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7133333333333334e-05,
+      "loss": 3.2895,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.72e-05,
+      "loss": 2.8106,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.726666666666667e-05,
+      "loss": 2.8489,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 3.5915,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7400000000000003e-05,
+      "loss": 3.0289,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7466666666666667e-05,
+      "loss": 3.8622,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7533333333333337e-05,
+      "loss": 3.2679,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.76e-05,
+      "loss": 2.8722,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7666666666666668e-05,
+      "loss": 3.1121,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7733333333333335e-05,
+      "loss": 2.9872,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7800000000000002e-05,
+      "loss": 1.4382,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7866666666666666e-05,
+      "loss": 3.1455,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7933333333333333e-05,
+      "loss": 2.4033,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8e-05,
+      "loss": 2.5713,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8066666666666668e-05,
+      "loss": 3.8718,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8133333333333335e-05,
+      "loss": 2.996,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 2.2063,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.826666666666667e-05,
+      "loss": 2.7209,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 2.366,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 3.3631,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8466666666666667e-05,
+      "loss": 1.8302,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8533333333333334e-05,
+      "loss": 4.3794,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.86e-05,
+      "loss": 2.1688,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 2.9545,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8733333333333336e-05,
+      "loss": 2.7229,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.88e-05,
+      "loss": 1.9973,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.886666666666667e-05,
+      "loss": 2.2278,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8933333333333334e-05,
+      "loss": 2.2168,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9e-05,
+      "loss": 3.5272,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9066666666666668e-05,
+      "loss": 3.9567,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9133333333333335e-05,
+      "loss": 1.7651,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 2.3812,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.926666666666667e-05,
+      "loss": 2.4078,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9333333333333333e-05,
+      "loss": 2.8164,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.94e-05,
+      "loss": 3.5021,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9466666666666668e-05,
+      "loss": 4.1566,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9533333333333335e-05,
+      "loss": 2.5986,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 2.6839,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9666666666666666e-05,
+      "loss": 2.848,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9733333333333336e-05,
+      "loss": 4.486,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.98e-05,
+      "loss": 3.2076,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9866666666666667e-05,
+      "loss": 2.4385,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9933333333333334e-05,
+      "loss": 2.508,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 2.8764,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999999475523208e-05,
+      "loss": 2.2813,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999997902092877e-05,
+      "loss": 3.0674,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999995279709183e-05,
+      "loss": 2.6392,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999991608372392e-05,
+      "loss": 3.6036,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999986888082895e-05,
+      "loss": 3.3022,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999981118841182e-05,
+      "loss": 2.2984,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999974300647867e-05,
+      "loss": 3.0575,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999996643350365e-05,
+      "loss": 2.538,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999957517409375e-05,
+      "loss": 3.6064,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999994755236596e-05,
+      "loss": 1.9906,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999993653837446e-05,
+      "loss": 3.2155,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999992447543603e-05,
+      "loss": 3.0904,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999911363551933e-05,
+      "loss": 3.8838,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999897202723546e-05,
+      "loss": 2.9924,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999881992952353e-05,
+      "loss": 2.9884,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999986573423995e-05,
+      "loss": 2.5421,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999848426588038e-05,
+      "loss": 3.2431,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999983006999844e-05,
+      "loss": 2.2875,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999981066447308e-05,
+      "loss": 3.0848,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999979021001399e-05,
+      "loss": 2.9645,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999768706623318e-05,
+      "loss": 3.3551,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999746154303318e-05,
+      "loss": 3.0388,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999722553056354e-05,
+      "loss": 3.0695,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999969790288491e-05,
+      "loss": 2.3498,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999672203791564e-05,
+      "loss": 2.593,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999645455779014e-05,
+      "loss": 2.2464,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999617658850067e-05,
+      "loss": 3.0089,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999958881300763e-05,
+      "loss": 3.6637,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999558918254746e-05,
+      "loss": 2.6916,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999952797459453e-05,
+      "loss": 2.6628,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999949598203025e-05,
+      "loss": 2.3816,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999462940565242e-05,
+      "loss": 3.2282,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999428850202985e-05,
+      "loss": 3.6746,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999939371094705e-05,
+      "loss": 3.9716,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999357522801125e-05,
+      "loss": 3.2169,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999320285769e-05,
+      "loss": 2.3433,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999281999854586e-05,
+      "loss": 2.4421,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99992426650619e-05,
+      "loss": 3.1333,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999202281395064e-05,
+      "loss": 3.1044,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999916084885832e-05,
+      "loss": 2.0566,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999911836745601e-05,
+      "loss": 3.7373,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999074837192588e-05,
+      "loss": 3.0114,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999030258072624e-05,
+      "loss": 2.2761,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999898463010079e-05,
+      "loss": 2.4042,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999893795328188e-05,
+      "loss": 2.4976,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998890227620784e-05,
+      "loss": 2.082,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999884145312251e-05,
+      "loss": 3.2536,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998791629792172e-05,
+      "loss": 2.1896,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998740757634998e-05,
+      "loss": 2.726,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998688836656322e-05,
+      "loss": 3.3639,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998635866861597e-05,
+      "loss": 1.924,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999858184825637e-05,
+      "loss": 3.0413,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998526780846314e-05,
+      "loss": 3.0301,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998470664637205e-05,
+      "loss": 3.1059,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998413499634927e-05,
+      "loss": 3.1296,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998355285845473e-05,
+      "loss": 4.0811,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998296023274958e-05,
+      "loss": 2.5442,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998235711929593e-05,
+      "loss": 2.6286,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998174351815704e-05,
+      "loss": 2.5547,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998111942939727e-05,
+      "loss": 2.4427,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998048485308213e-05,
+      "loss": 2.6273,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997983978927813e-05,
+      "loss": 2.0925,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997918423805295e-05,
+      "loss": 2.8035,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997851819947537e-05,
+      "loss": 1.8798,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997784167361526e-05,
+      "loss": 2.5603,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997715466054357e-05,
+      "loss": 3.594,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997645716033233e-05,
+      "loss": 2.1482,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999757491730548e-05,
+      "loss": 3.0647,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997503069878515e-05,
+      "loss": 2.4434,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997430173759876e-05,
+      "loss": 2.9617,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997356228957213e-05,
+      "loss": 2.5608,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999728123547828e-05,
+      "loss": 2.5009,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997205193330943e-05,
+      "loss": 3.1555,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997128102523186e-05,
+      "loss": 2.3955,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999704996306308e-05,
+      "loss": 1.849,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996970774958836e-05,
+      "loss": 3.2385,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996890538218756e-05,
+      "loss": 3.0807,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996809252851254e-05,
+      "loss": 2.2904,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996726918864857e-05,
+      "loss": 3.4125,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996643536268202e-05,
+      "loss": 3.3575,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996559105070038e-05,
+      "loss": 3.5882,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999647362527922e-05,
+      "loss": 3.9226,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999638709690471e-05,
+      "loss": 3.9351,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999629951995559e-05,
+      "loss": 3.6388,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996210894441047e-05,
+      "loss": 3.3352,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996121220370374e-05,
+      "loss": 3.0401,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996030497752977e-05,
+      "loss": 3.616,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995938726598374e-05,
+      "loss": 2.3772,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999584590691619e-05,
+      "loss": 3.2202,
+      "step": 389
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995752038716166e-05,
+      "loss": 3.0972,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999565712200814e-05,
+      "loss": 2.9456,
+      "step": 391
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999556115680208e-05,
+      "loss": 2.7784,
+      "step": 392
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999546414310804e-05,
+      "loss": 2.832,
+      "step": 393
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995366080936206e-05,
+      "loss": 1.1589,
+      "step": 394
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995266970296856e-05,
+      "loss": 2.6524,
+      "step": 395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999516681120039e-05,
+      "loss": 2.6977,
+      "step": 396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995065603657317e-05,
+      "loss": 3.3339,
+      "step": 397
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999496334767825e-05,
+      "loss": 3.1973,
+      "step": 398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994860043273915e-05,
+      "loss": 2.5056,
+      "step": 399
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994755690455154e-05,
+      "loss": 3.3793,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994650289232898e-05,
+      "loss": 2.6249,
+      "step": 401
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999454383961822e-05,
+      "loss": 1.773,
+      "step": 402
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999443634162228e-05,
+      "loss": 3.9671,
+      "step": 403
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999432779525635e-05,
+      "loss": 3.7492,
+      "step": 404
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994218200531823e-05,
+      "loss": 1.6801,
+      "step": 405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999410755746019e-05,
+      "loss": 2.6153,
+      "step": 406
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993995866053057e-05,
+      "loss": 2.8102,
+      "step": 407
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993883126322142e-05,
+      "loss": 2.5999,
+      "step": 408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999376933827927e-05,
+      "loss": 3.7101,
+      "step": 409
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999365450193638e-05,
+      "loss": 1.7806,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999353861730551e-05,
+      "loss": 3.5163,
+      "step": 411
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993421684398825e-05,
+      "loss": 3.3887,
+      "step": 412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993303703228585e-05,
+      "loss": 2.8079,
+      "step": 413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999318467380716e-05,
+      "loss": 2.4403,
+      "step": 414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999306459614705e-05,
+      "loss": 3.3692,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992943470260845e-05,
+      "loss": 2.666,
+      "step": 416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999282129616124e-05,
+      "loss": 3.098,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992698073861067e-05,
+      "loss": 3.4611,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992573803373242e-05,
+      "loss": 2.1981,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99924484847108e-05,
+      "loss": 2.6451,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992322117886886e-05,
+      "loss": 3.0457,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992194702914762e-05,
+      "loss": 2.9373,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999206623980779e-05,
+      "loss": 2.8628,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991936728579438e-05,
+      "loss": 1.9978,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991806169243302e-05,
+      "loss": 3.7635,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999167456181307e-05,
+      "loss": 3.4176,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991541906302552e-05,
+      "loss": 2.8507,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999140820272566e-05,
+      "loss": 1.5132,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991273451096414e-05,
+      "loss": 3.2811,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991137651428957e-05,
+      "loss": 3.2372,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999100080373753e-05,
+      "loss": 2.5815,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990862908036492e-05,
+      "loss": 2.7844,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990723964340303e-05,
+      "loss": 3.3884,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990583972663534e-05,
+      "loss": 3.3219,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999044293302088e-05,
+      "loss": 3.1955,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990300845427123e-05,
+      "loss": 3.2649,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999015770989718e-05,
+      "loss": 3.1334,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990013526446056e-05,
+      "loss": 2.7044,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989868295088876e-05,
+      "loss": 2.4971,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998972201584088e-05,
+      "loss": 1.8091,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989574688717405e-05,
+      "loss": 2.833,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989426313733912e-05,
+      "loss": 2.453,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989276890905957e-05,
+      "loss": 2.8455,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998912642024922e-05,
+      "loss": 3.3875,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988974901779482e-05,
+      "loss": 2.0311,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988822335512638e-05,
+      "loss": 2.6549,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998866872146469e-05,
+      "loss": 2.3045,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998851405965175e-05,
+      "loss": 2.8222,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988358350090045e-05,
+      "loss": 3.0799,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998820159279591e-05,
+      "loss": 3.3891,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998804378778578e-05,
+      "loss": 3.1212,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987884935076213e-05,
+      "loss": 2.8597,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987725034683875e-05,
+      "loss": 2.4997,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998756408662553e-05,
+      "loss": 2.1669,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998740209091807e-05,
+      "loss": 3.2485,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987239047578482e-05,
+      "loss": 2.3873,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987074956623872e-05,
+      "loss": 1.6535,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998690981807145e-05,
+      "loss": 3.2595,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986743631938536e-05,
+      "loss": 3.1836,
+      "step": 459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986576398242566e-05,
+      "loss": 4.4084,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986408117001082e-05,
+      "loss": 3.0843,
+      "step": 461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986238788231733e-05,
+      "loss": 3.2802,
+      "step": 462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986068411952283e-05,
+      "loss": 2.4752,
+      "step": 463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985896988180607e-05,
+      "loss": 2.6303,
+      "step": 464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998572451693468e-05,
+      "loss": 2.8853,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985550998232598e-05,
+      "loss": 3.3157,
+      "step": 466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985376432092558e-05,
+      "loss": 2.6368,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985200818532873e-05,
+      "loss": 2.7704,
+      "step": 468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985024157571972e-05,
+      "loss": 2.3403,
+      "step": 469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998484644922837e-05,
+      "loss": 2.2194,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998466769352072e-05,
+      "loss": 3.5631,
+      "step": 471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984487890467773e-05,
+      "loss": 2.3152,
+      "step": 472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998430704008838e-05,
+      "loss": 2.9691,
+      "step": 473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998412514240152e-05,
+      "loss": 3.0195,
+      "step": 474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983942197426272e-05,
+      "loss": 4.3899,
+      "step": 475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983758205181824e-05,
+      "loss": 2.2119,
+      "step": 476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983573165687473e-05,
+      "loss": 3.2836,
+      "step": 477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983387078962634e-05,
+      "loss": 2.0206,
+      "step": 478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983199945026822e-05,
+      "loss": 3.4608,
+      "step": 479
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983011763899674e-05,
+      "loss": 2.8475,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998282253560092e-05,
+      "loss": 3.3883,
+      "step": 481
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982632260150417e-05,
+      "loss": 3.3658,
+      "step": 482
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982440937568118e-05,
+      "loss": 2.8299,
+      "step": 483
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982248567874098e-05,
+      "loss": 2.1797,
+      "step": 484
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998205515108853e-05,
+      "loss": 2.6288,
+      "step": 485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981860687231708e-05,
+      "loss": 2.191,
+      "step": 486
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981665176324023e-05,
+      "loss": 3.0569,
+      "step": 487
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998146861838599e-05,
+      "loss": 2.1342,
+      "step": 488
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998127101343822e-05,
+      "loss": 2.453,
+      "step": 489
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998107236150145e-05,
+      "loss": 2.0634,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980872662596514e-05,
+      "loss": 2.8386,
+      "step": 491
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980671916744356e-05,
+      "loss": 3.4988,
+      "step": 492
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998047012396603e-05,
+      "loss": 2.7527,
+      "step": 493
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980267284282718e-05,
+      "loss": 3.1989,
+      "step": 494
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980063397715685e-05,
+      "loss": 3.1244,
+      "step": 495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979858464286317e-05,
+      "loss": 4.2765,
+      "step": 496
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979652484016117e-05,
+      "loss": 2.8788,
+      "step": 497
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997944545692669e-05,
+      "loss": 2.3947,
+      "step": 498
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979237383039745e-05,
+      "loss": 3.1048,
+      "step": 499
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997902826237712e-05,
+      "loss": 2.4374,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997881809496074e-05,
+      "loss": 2.243,
+      "step": 501
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978606880812657e-05,
+      "loss": 2.8722,
+      "step": 502
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978394619955024e-05,
+      "loss": 3.2385,
+      "step": 503
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978181312410104e-05,
+      "loss": 2.6369,
+      "step": 504
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977966958200276e-05,
+      "loss": 2.3733,
+      "step": 505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977751557348028e-05,
+      "loss": 2.4693,
+      "step": 506
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977535109875944e-05,
+      "loss": 2.4365,
+      "step": 507
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977317615806738e-05,
+      "loss": 2.4102,
+      "step": 508
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977099075163216e-05,
+      "loss": 3.0946,
+      "step": 509
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997687948796831e-05,
+      "loss": 3.8172,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976658854245046e-05,
+      "loss": 3.0114,
+      "step": 511
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976437174016575e-05,
+      "loss": 1.7131,
+      "step": 512
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976214447306145e-05,
+      "loss": 2.8564,
+      "step": 513
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997599067413712e-05,
+      "loss": 3.0361,
+      "step": 514
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975765854532974e-05,
+      "loss": 2.6796,
+      "step": 515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997553998851729e-05,
+      "loss": 2.9344,
+      "step": 516
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975313076113754e-05,
+      "loss": 3.4702,
+      "step": 517
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997508511734618e-05,
+      "loss": 3.4101,
+      "step": 518
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997485611223847e-05,
+      "loss": 2.2819,
+      "step": 519
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997462606081465e-05,
+      "loss": 2.4279,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974394963098846e-05,
+      "loss": 3.0414,
+      "step": 521
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974162819115306e-05,
+      "loss": 3.1298,
+      "step": 522
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973929628888378e-05,
+      "loss": 2.1534,
+      "step": 523
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997369539244252e-05,
+      "loss": 1.9004,
+      "step": 524
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973460109802306e-05,
+      "loss": 2.6227,
+      "step": 525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973223780992414e-05,
+      "loss": 3.5883,
+      "step": 526
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997298640603764e-05,
+      "loss": 1.6453,
+      "step": 527
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997274798496287e-05,
+      "loss": 2.7514,
+      "step": 528
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972508517793125e-05,
+      "loss": 2.9985,
+      "step": 529
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997226800455352e-05,
+      "loss": 3.1809,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972026445269285e-05,
+      "loss": 2.4321,
+      "step": 531
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971783839965756e-05,
+      "loss": 2.7298,
+      "step": 532
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971540188668386e-05,
+      "loss": 3.2781,
+      "step": 533
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971295491402725e-05,
+      "loss": 3.1229,
+      "step": 534
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971049748194448e-05,
+      "loss": 2.3091,
+      "step": 535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997080295906933e-05,
+      "loss": 2.8672,
+      "step": 536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970555124053256e-05,
+      "loss": 2.6915,
+      "step": 537
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970306243172223e-05,
+      "loss": 1.5469,
+      "step": 538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997005631645234e-05,
+      "loss": 2.6656,
+      "step": 539
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969805343919822e-05,
+      "loss": 2.0797,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969553325600994e-05,
+      "loss": 3.4097,
+      "step": 541
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969300261522292e-05,
+      "loss": 3.4053,
+      "step": 542
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969046151710266e-05,
+      "loss": 2.5786,
+      "step": 543
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996879099619156e-05,
+      "loss": 3.1153,
+      "step": 544
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968534794992947e-05,
+      "loss": 3.7956,
+      "step": 545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968277548141302e-05,
+      "loss": 3.2557,
+      "step": 546
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968019255663608e-05,
+      "loss": 2.9464,
+      "step": 547
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967759917586953e-05,
+      "loss": 3.7973,
+      "step": 548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967499533938544e-05,
+      "loss": 4.1609,
+      "step": 549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967238104745695e-05,
+      "loss": 3.0693,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996697563003583e-05,
+      "loss": 2.8666,
+      "step": 551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966712109836476e-05,
+      "loss": 3.4788,
+      "step": 552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966447544175285e-05,
+      "loss": 2.9226,
+      "step": 553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996618193308e-05,
+      "loss": 2.7589,
+      "step": 554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996591527657848e-05,
+      "loss": 2.9108,
+      "step": 555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965647574698705e-05,
+      "loss": 2.4675,
+      "step": 556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996537882746875e-05,
+      "loss": 2.3051,
+      "step": 557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965109034916806e-05,
+      "loss": 2.3718,
+      "step": 558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964838197071173e-05,
+      "loss": 2.9184,
+      "step": 559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964566313960265e-05,
+      "loss": 2.332,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964293385612595e-05,
+      "loss": 2.3615,
+      "step": 561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964019412056798e-05,
+      "loss": 3.7697,
+      "step": 562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963744393321606e-05,
+      "loss": 2.3907,
+      "step": 563
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963468329435872e-05,
+      "loss": 2.6988,
+      "step": 564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963191220428552e-05,
+      "loss": 3.1312,
+      "step": 565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996291306632871e-05,
+      "loss": 2.1049,
+      "step": 566
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996263386716553e-05,
+      "loss": 4.2314,
+      "step": 567
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962353622968296e-05,
+      "loss": 3.5554,
+      "step": 568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99620723337664e-05,
+      "loss": 2.5264,
+      "step": 569
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961789999589357e-05,
+      "loss": 2.5037,
+      "step": 570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961506620466774e-05,
+      "loss": 1.8222,
+      "step": 571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996122219642838e-05,
+      "loss": 2.4549,
+      "step": 572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960936727504007e-05,
+      "loss": 2.7507,
+      "step": 573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960650213723604e-05,
+      "loss": 2.4945,
+      "step": 574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996036265511722e-05,
+      "loss": 3.4476,
+      "step": 575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960074051715022e-05,
+      "loss": 3.0118,
+      "step": 576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959784403547283e-05,
+      "loss": 2.6252,
+      "step": 577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959493710644385e-05,
+      "loss": 2.699,
+      "step": 578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959201973036816e-05,
+      "loss": 2.7513,
+      "step": 579
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995890919075519e-05,
+      "loss": 2.4159,
+      "step": 580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958615363830202e-05,
+      "loss": 3.3491,
+      "step": 581
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958320492292686e-05,
+      "loss": 2.6507,
+      "step": 582
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995802457617357e-05,
+      "loss": 2.5653,
+      "step": 583
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995772761550389e-05,
+      "loss": 1.2759,
+      "step": 584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957429610314797e-05,
+      "loss": 1.3155,
+      "step": 585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957130560637553e-05,
+      "loss": 2.3748,
+      "step": 586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995683046650353e-05,
+      "loss": 2.5546,
+      "step": 587
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956529327944198e-05,
+      "loss": 2.764,
+      "step": 588
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995622714499115e-05,
+      "loss": 4.0103,
+      "step": 589
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995592391767608e-05,
+      "loss": 2.0315,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99556196460308e-05,
+      "loss": 3.2007,
+      "step": 591
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955314330087225e-05,
+      "loss": 2.9096,
+      "step": 592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955007969877384e-05,
+      "loss": 2.9724,
+      "step": 593
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954700565433406e-05,
+      "loss": 3.4275,
+      "step": 594
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995439211678754e-05,
+      "loss": 3.1237,
+      "step": 595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954082623972143e-05,
+      "loss": 3.0477,
+      "step": 596
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953772087019674e-05,
+      "loss": 2.6921,
+      "step": 597
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995346050596271e-05,
+      "loss": 2.6666,
+      "step": 598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953147880833935e-05,
+      "loss": 2.8506,
+      "step": 599
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995283421166614e-05,
+      "loss": 1.5483,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995251949849223e-05,
+      "loss": 2.6479,
+      "step": 601
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995220374134522e-05,
+      "loss": 2.2203,
+      "step": 602
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995188694025822e-05,
+      "loss": 2.2069,
+      "step": 603
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951569095264473e-05,
+      "loss": 3.4474,
+      "step": 604
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995125020639731e-05,
+      "loss": 3.044,
+      "step": 605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995093027369019e-05,
+      "loss": 2.7634,
+      "step": 606
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950609297176668e-05,
+      "loss": 3.5653,
+      "step": 607
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995028727689041e-05,
+      "loss": 2.3554,
+      "step": 608
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99499642128652e-05,
+      "loss": 2.2194,
+      "step": 609
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994964010513492e-05,
+      "loss": 1.9859,
+      "step": 610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949314953733573e-05,
+      "loss": 3.0724,
+      "step": 611
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948988758695263e-05,
+      "loss": 3.449,
+      "step": 612
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994866152005421e-05,
+      "loss": 2.0916,
+      "step": 613
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994833323784473e-05,
+      "loss": 3.1038,
+      "step": 614
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948003912101274e-05,
+      "loss": 2.3568,
+      "step": 615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994767354285837e-05,
+      "loss": 2.9764,
+      "step": 616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994734213015068e-05,
+      "loss": 1.7688,
+      "step": 617
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947009674012975e-05,
+      "loss": 2.7583,
+      "step": 618
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946676174480115e-05,
+      "loss": 3.0937,
+      "step": 619
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946341631587086e-05,
+      "loss": 2.0932,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946006045368987e-05,
+      "loss": 3.0623,
+      "step": 621
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994566941586101e-05,
+      "loss": 2.1918,
+      "step": 622
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945331743098476e-05,
+      "loss": 1.9083,
+      "step": 623
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944993027116798e-05,
+      "loss": 2.9481,
+      "step": 624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944653267951507e-05,
+      "loss": 2.9722,
+      "step": 625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944312465638244e-05,
+      "loss": 3.5768,
+      "step": 626
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943970620212752e-05,
+      "loss": 2.3906,
+      "step": 627
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943627731710896e-05,
+      "loss": 3.7709,
+      "step": 628
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943283800168643e-05,
+      "loss": 4.7137,
+      "step": 629
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942938825622064e-05,
+      "loss": 2.9163,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942592808107354e-05,
+      "loss": 2.2021,
+      "step": 631
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942245747660797e-05,
+      "loss": 3.4779,
+      "step": 632
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941897644318805e-05,
+      "loss": 3.2088,
+      "step": 633
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941548498117894e-05,
+      "loss": 2.8764,
+      "step": 634
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994119830909469e-05,
+      "loss": 3.1355,
+      "step": 635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940847077285918e-05,
+      "loss": 2.9318,
+      "step": 636
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994049480272842e-05,
+      "loss": 4.0351,
+      "step": 637
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994014148545916e-05,
+      "loss": 3.2479,
+      "step": 638
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939787125515188e-05,
+      "loss": 2.2145,
+      "step": 639
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939431722933678e-05,
+      "loss": 1.816,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939075277751914e-05,
+      "loss": 3.5039,
+      "step": 641
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993871779000728e-05,
+      "loss": 3.2401,
+      "step": 642
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938359259737273e-05,
+      "loss": 1.6436,
+      "step": 643
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993799968697951e-05,
+      "loss": 3.8362,
+      "step": 644
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937639071771704e-05,
+      "loss": 4.1314,
+      "step": 645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937277414151677e-05,
+      "loss": 2.234,
+      "step": 646
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936914714157375e-05,
+      "loss": 2.7954,
+      "step": 647
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936550971826835e-05,
+      "loss": 2.3384,
+      "step": 648
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936186187198214e-05,
+      "loss": 2.2882,
+      "step": 649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993582036030978e-05,
+      "loss": 1.7459,
+      "step": 650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935453491199902e-05,
+      "loss": 3.1631,
+      "step": 651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935085579907064e-05,
+      "loss": 1.8801,
+      "step": 652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993471662646986e-05,
+      "loss": 3.7503,
+      "step": 653
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934346630926988e-05,
+      "loss": 2.6151,
+      "step": 654
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933975593317263e-05,
+      "loss": 4.2847,
+      "step": 655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933603513679604e-05,
+      "loss": 2.5657,
+      "step": 656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993323039205304e-05,
+      "loss": 2.9585,
+      "step": 657
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932856228476705e-05,
+      "loss": 2.7569,
+      "step": 658
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932481022989857e-05,
+      "loss": 2.5367,
+      "step": 659
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932104775631847e-05,
+      "loss": 2.3448,
+      "step": 660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993172748644214e-05,
+      "loss": 2.3982,
+      "step": 661
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931349155460315e-05,
+      "loss": 2.7121,
+      "step": 662
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930969782726058e-05,
+      "loss": 3.3197,
+      "step": 663
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993058936827916e-05,
+      "loss": 2.9171,
+      "step": 664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993020791215953e-05,
+      "loss": 3.0479,
+      "step": 665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929825414407175e-05,
+      "loss": 2.0655,
+      "step": 666
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929441875062218e-05,
+      "loss": 3.2928,
+      "step": 667
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929057294164894e-05,
+      "loss": 2.9501,
+      "step": 668
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992867167175554e-05,
+      "loss": 1.9734,
+      "step": 669
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992828500787461e-05,
+      "loss": 1.546,
+      "step": 670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992789730256266e-05,
+      "loss": 3.4348,
+      "step": 671
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992750855586036e-05,
+      "loss": 2.2814,
+      "step": 672
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992711876780849e-05,
+      "loss": 1.9695,
+      "step": 673
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926727938447935e-05,
+      "loss": 2.9891,
+      "step": 674
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926336067819686e-05,
+      "loss": 2.8082,
+      "step": 675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925943155964857e-05,
+      "loss": 2.6423,
+      "step": 676
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925549202924657e-05,
+      "loss": 3.5556,
+      "step": 677
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925154208740412e-05,
+      "loss": 1.7126,
+      "step": 678
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924758173453555e-05,
+      "loss": 2.2051,
+      "step": 679
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924361097105624e-05,
+      "loss": 2.6423,
+      "step": 680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923962979738277e-05,
+      "loss": 3.441,
+      "step": 681
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923563821393272e-05,
+      "loss": 2.6574,
+      "step": 682
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992316362211248e-05,
+      "loss": 3.1888,
+      "step": 683
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992276238193788e-05,
+      "loss": 1.9128,
+      "step": 684
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922360100911553e-05,
+      "loss": 2.7822,
+      "step": 685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921956779075707e-05,
+      "loss": 2.8444,
+      "step": 686
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921552416472644e-05,
+      "loss": 1.4174,
+      "step": 687
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 2.9448,
+      "step": 688
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992074056913464e-05,
+      "loss": 3.0182,
+      "step": 689
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992033308448486e-05,
+      "loss": 2.9418,
+      "step": 690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919924559238173e-05,
+      "loss": 2.7078,
+      "step": 691
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919514993437445e-05,
+      "loss": 3.3076,
+      "step": 692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919104387125632e-05,
+      "loss": 2.8846,
+      "step": 693
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918692740345804e-05,
+      "loss": 3.4149,
+      "step": 694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918280053141144e-05,
+      "loss": 2.2122,
+      "step": 695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917866325554936e-05,
+      "loss": 3.0999,
+      "step": 696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991745155763058e-05,
+      "loss": 2.8522,
+      "step": 697
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917035749411585e-05,
+      "loss": 4.1362,
+      "step": 698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916618900941567e-05,
+      "loss": 2.9066,
+      "step": 699
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916201012264255e-05,
+      "loss": 2.4822,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915782083423473e-05,
+      "loss": 2.106,
+      "step": 701
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915362114463172e-05,
+      "loss": 2.5937,
+      "step": 702
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914941105427403e-05,
+      "loss": 1.8859,
+      "step": 703
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991451905636033e-05,
+      "loss": 4.0667,
+      "step": 704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914095967306224e-05,
+      "loss": 2.7613,
+      "step": 705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913671838309466e-05,
+      "loss": 2.341,
+      "step": 706
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991324666941454e-05,
+      "loss": 2.2684,
+      "step": 707
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912820460666046e-05,
+      "loss": 3.1457,
+      "step": 708
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912393212108692e-05,
+      "loss": 3.7368,
+      "step": 709
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911964923787295e-05,
+      "loss": 2.2107,
+      "step": 710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911535595746783e-05,
+      "loss": 3.0765,
+      "step": 711
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911105228032186e-05,
+      "loss": 1.9517,
+      "step": 712
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910673820688652e-05,
+      "loss": 1.8646,
+      "step": 713
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910241373761426e-05,
+      "loss": 2.7998,
+      "step": 714
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990980788729588e-05,
+      "loss": 3.3131,
+      "step": 715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909373361337475e-05,
+      "loss": 2.3373,
+      "step": 716
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908937795931798e-05,
+      "loss": 3.6792,
+      "step": 717
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908501191124535e-05,
+      "loss": 2.9602,
+      "step": 718
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908063546961482e-05,
+      "loss": 2.4594,
+      "step": 719
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990762486348855e-05,
+      "loss": 2.37,
+      "step": 720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907185140751753e-05,
+      "loss": 2.3284,
+      "step": 721
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906744378797214e-05,
+      "loss": 3.2902,
+      "step": 722
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906302577671166e-05,
+      "loss": 3.7178,
+      "step": 723
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990585973741996e-05,
+      "loss": 3.1146,
+      "step": 724
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905415858090036e-05,
+      "loss": 2.5103,
+      "step": 725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904970939727962e-05,
+      "loss": 2.3222,
+      "step": 726
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904524982380408e-05,
+      "loss": 3.1396,
+      "step": 727
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904077986094153e-05,
+      "loss": 3.3551,
+      "step": 728
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903629950916083e-05,
+      "loss": 2.892,
+      "step": 729
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903180876893195e-05,
+      "loss": 2.5833,
+      "step": 730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902730764072593e-05,
+      "loss": 3.8777,
+      "step": 731
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902279612501494e-05,
+      "loss": 2.0763,
+      "step": 732
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901827422227225e-05,
+      "loss": 2.7588,
+      "step": 733
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901374193297212e-05,
+      "loss": 3.2113,
+      "step": 734
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900919925759e-05,
+      "loss": 3.0979,
+      "step": 735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900464619660243e-05,
+      "loss": 3.3412,
+      "step": 736
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900008275048696e-05,
+      "loss": 3.5514,
+      "step": 737
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899550891972224e-05,
+      "loss": 2.7229,
+      "step": 738
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989909247047881e-05,
+      "loss": 3.3643,
+      "step": 739
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989863301061654e-05,
+      "loss": 2.4911,
+      "step": 740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989817251243361e-05,
+      "loss": 3.3636,
+      "step": 741
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897710975978323e-05,
+      "loss": 2.2355,
+      "step": 742
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989724840129909e-05,
+      "loss": 2.3284,
+      "step": 743
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989678478844443e-05,
+      "loss": 2.2167,
+      "step": 744
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896320137462984e-05,
+      "loss": 2.8071,
+      "step": 745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895854448403482e-05,
+      "loss": 2.8768,
+      "step": 746
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895387721314776e-05,
+      "loss": 2.7613,
+      "step": 747
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894919956245825e-05,
+      "loss": 1.9599,
+      "step": 748
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894451153245695e-05,
+      "loss": 3.6077,
+      "step": 749
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9893981312363563e-05,
+      "loss": 3.6963,
+      "step": 750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893510433648705e-05,
+      "loss": 2.991,
+      "step": 751
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893038517150526e-05,
+      "loss": 2.2963,
+      "step": 752
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892565562918518e-05,
+      "loss": 3.586,
+      "step": 753
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892091571002295e-05,
+      "loss": 3.6683,
+      "step": 754
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989161654145158e-05,
+      "loss": 3.1654,
+      "step": 755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891140474316197e-05,
+      "loss": 2.6669,
+      "step": 756
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989066336964608e-05,
+      "loss": 3.0648,
+      "step": 757
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890185227491285e-05,
+      "loss": 2.1763,
+      "step": 758
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889706047901956e-05,
+      "loss": 3.3128,
+      "step": 759
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889225830928365e-05,
+      "loss": 2.5753,
+      "step": 760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888744576620883e-05,
+      "loss": 1.9525,
+      "step": 761
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988826228502999e-05,
+      "loss": 2.4246,
+      "step": 762
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887778956206274e-05,
+      "loss": 2.5591,
+      "step": 763
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887294590200437e-05,
+      "loss": 3.0955,
+      "step": 764
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886809187063285e-05,
+      "loss": 3.1434,
+      "step": 765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886322746845738e-05,
+      "loss": 1.6039,
+      "step": 766
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885835269598818e-05,
+      "loss": 1.5803,
+      "step": 767
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988534675537366e-05,
+      "loss": 2.3875,
+      "step": 768
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884857204221503e-05,
+      "loss": 2.3524,
+      "step": 769
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884366616193707e-05,
+      "loss": 2.6065,
+      "step": 770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883874991341724e-05,
+      "loss": 3.6837,
+      "step": 771
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988338232971713e-05,
+      "loss": 2.3641,
+      "step": 772
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882888631371596e-05,
+      "loss": 3.0345,
+      "step": 773
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882393896356915e-05,
+      "loss": 1.9784,
+      "step": 774
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988189812472498e-05,
+      "loss": 2.5945,
+      "step": 775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881401316527795e-05,
+      "loss": 2.7687,
+      "step": 776
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880903471817473e-05,
+      "loss": 3.0439,
+      "step": 777
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880404590646233e-05,
+      "loss": 2.9799,
+      "step": 778
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987990467306641e-05,
+      "loss": 2.8154,
+      "step": 779
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987940371913044e-05,
+      "loss": 3.2012,
+      "step": 780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878901728890875e-05,
+      "loss": 2.732,
+      "step": 781
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878398702400364e-05,
+      "loss": 2.4528,
+      "step": 782
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877894639711678e-05,
+      "loss": 3.5686,
+      "step": 783
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877389540877686e-05,
+      "loss": 3.0596,
+      "step": 784
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876883405951378e-05,
+      "loss": 2.4482,
+      "step": 785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876376234985838e-05,
+      "loss": 2.8124,
+      "step": 786
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987586802803427e-05,
+      "loss": 3.0339,
+      "step": 787
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875358785149982e-05,
+      "loss": 3.1534,
+      "step": 788
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874848506386392e-05,
+      "loss": 3.5857,
+      "step": 789
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987433719179702e-05,
+      "loss": 3.4699,
+      "step": 790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873824841435505e-05,
+      "loss": 2.3601,
+      "step": 791
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987331145535559e-05,
+      "loss": 2.7605,
+      "step": 792
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987279703361113e-05,
+      "loss": 2.6022,
+      "step": 793
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872281576256078e-05,
+      "loss": 3.5395,
+      "step": 794
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987176508334451e-05,
+      "loss": 3.8354,
+      "step": 795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98712475549306e-05,
+      "loss": 3.0984,
+      "step": 796
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870728991068636e-05,
+      "loss": 3.2392,
+      "step": 797
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870209391813013e-05,
+      "loss": 2.1407,
+      "step": 798
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869688757218233e-05,
+      "loss": 1.8836,
+      "step": 799
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869167087338908e-05,
+      "loss": 2.9355,
+      "step": 800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868644382229762e-05,
+      "loss": 3.3825,
+      "step": 801
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868120641945622e-05,
+      "loss": 2.3738,
+      "step": 802
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867595866541423e-05,
+      "loss": 3.8296,
+      "step": 803
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867070056072215e-05,
+      "loss": 2.3617,
+      "step": 804
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866543210593154e-05,
+      "loss": 3.8,
+      "step": 805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866015330159504e-05,
+      "loss": 3.1213,
+      "step": 806
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865486414826634e-05,
+      "loss": 2.093,
+      "step": 807
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864956464650027e-05,
+      "loss": 2.0172,
+      "step": 808
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986442547968527e-05,
+      "loss": 2.0151,
+      "step": 809
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986389345998806e-05,
+      "loss": 2.3621,
+      "step": 810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863360405614213e-05,
+      "loss": 3.4312,
+      "step": 811
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986282631661963e-05,
+      "loss": 3.0139,
+      "step": 812
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862291193060344e-05,
+      "loss": 2.9669,
+      "step": 813
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861755034992483e-05,
+      "loss": 3.3474,
+      "step": 814
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986121784247229e-05,
+      "loss": 3.2668,
+      "step": 815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860679615556112e-05,
+      "loss": 2.7858,
+      "step": 816
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860140354300406e-05,
+      "loss": 2.4815,
+      "step": 817
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985960005876174e-05,
+      "loss": 3.9251,
+      "step": 818
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859058728996788e-05,
+      "loss": 3.553,
+      "step": 819
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858516365062334e-05,
+      "loss": 3.6177,
+      "step": 820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857972967015268e-05,
+      "loss": 2.8363,
+      "step": 821
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985742853491259e-05,
+      "loss": 2.649,
+      "step": 822
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856883068811405e-05,
+      "loss": 2.6592,
+      "step": 823
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856336568768936e-05,
+      "loss": 3.1833,
+      "step": 824
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855789034842504e-05,
+      "loss": 1.8583,
+      "step": 825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855240467089545e-05,
+      "loss": 3.1154,
+      "step": 826
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854690865567604e-05,
+      "loss": 2.0456,
+      "step": 827
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854140230334323e-05,
+      "loss": 3.4769,
+      "step": 828
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985358856144747e-05,
+      "loss": 3.9953,
+      "step": 829
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853035858964907e-05,
+      "loss": 1.326,
+      "step": 830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985248212294461e-05,
+      "loss": 2.8795,
+      "step": 831
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985192735344467e-05,
+      "loss": 2.8329,
+      "step": 832
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851371550523272e-05,
+      "loss": 2.4213,
+      "step": 833
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850814714238718e-05,
+      "loss": 3.4153,
+      "step": 834
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850256844649422e-05,
+      "loss": 0.9851,
+      "step": 835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98496979418139e-05,
+      "loss": 3.2508,
+      "step": 836
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849138005790775e-05,
+      "loss": 2.7923,
+      "step": 837
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984857703663879e-05,
+      "loss": 3.416,
+      "step": 838
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848015034416776e-05,
+      "loss": 3.1152,
+      "step": 839
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847451999183692e-05,
+      "loss": 2.5577,
+      "step": 840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98468879309986e-05,
+      "loss": 2.3699,
+      "step": 841
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846322829920663e-05,
+      "loss": 3.0857,
+      "step": 842
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984575669600916e-05,
+      "loss": 3.2395,
+      "step": 843
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845189529323473e-05,
+      "loss": 3.1092,
+      "step": 844
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98446213299231e-05,
+      "loss": 3.3601,
+      "step": 845
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844052097867638e-05,
+      "loss": 2.1255,
+      "step": 846
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98434818332168e-05,
+      "loss": 4.1439,
+      "step": 847
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98429105360304e-05,
+      "loss": 3.2642,
+      "step": 848
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842338206368375e-05,
+      "loss": 1.8523,
+      "step": 849
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841764844290744e-05,
+      "loss": 2.3738,
+      "step": 850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841190449857662e-05,
+      "loss": 3.9256,
+      "step": 851
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840615023129372e-05,
+      "loss": 3.1947,
+      "step": 852
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984003856416624e-05,
+      "loss": 2.5036,
+      "step": 853
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839461073028733e-05,
+      "loss": 2.1231,
+      "step": 854
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838882549777426e-05,
+      "loss": 3.8735,
+      "step": 855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838302994473e-05,
+      "loss": 1.5643,
+      "step": 856
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983772240717625e-05,
+      "loss": 2.8567,
+      "step": 857
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837140787948082e-05,
+      "loss": 1.9932,
+      "step": 858
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98365581368495e-05,
+      "loss": 2.7731,
+      "step": 859
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835974453941623e-05,
+      "loss": 3.0073,
+      "step": 860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983538973928567e-05,
+      "loss": 2.3678,
+      "step": 861
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983480399294299e-05,
+      "loss": 2.1541,
+      "step": 862
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834217214975008e-05,
+      "loss": 3.4918,
+      "step": 863
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833629405443283e-05,
+      "loss": 2.3786,
+      "step": 864
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983304056440948e-05,
+      "loss": 3.0617,
+      "step": 865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832450691935353e-05,
+      "loss": 2.512,
+      "step": 866
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831859788082778e-05,
+      "loss": 2.4902,
+      "step": 867
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983126785291375e-05,
+      "loss": 2.5997,
+      "step": 868
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983067488649035e-05,
+      "loss": 2.8511,
+      "step": 869
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983008088887478e-05,
+      "loss": 2.8375,
+      "step": 870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829485860129347e-05,
+      "loss": 1.9814,
+      "step": 871
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828889800316467e-05,
+      "loss": 2.2311,
+      "step": 872
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828292709498665e-05,
+      "loss": 3.3454,
+      "step": 873
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982769458773857e-05,
+      "loss": 2.8069,
+      "step": 874
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827095435098926e-05,
+      "loss": 2.2216,
+      "step": 875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982649525164258e-05,
+      "loss": 3.8249,
+      "step": 876
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825894037432486e-05,
+      "loss": 3.7763,
+      "step": 877
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825291792531717e-05,
+      "loss": 3.9217,
+      "step": 878
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824688517003433e-05,
+      "loss": 3.044,
+      "step": 879
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824084210910924e-05,
+      "loss": 2.9989,
+      "step": 880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982347887431758e-05,
+      "loss": 4.8197,
+      "step": 881
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982287250728689e-05,
+      "loss": 3.3275,
+      "step": 882
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822265109882463e-05,
+      "loss": 1.8326,
+      "step": 883
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821656682168013e-05,
+      "loss": 3.1634,
+      "step": 884
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821047224207362e-05,
+      "loss": 1.7484,
+      "step": 885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820436736064437e-05,
+      "loss": 2.3024,
+      "step": 886
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819825217803275e-05,
+      "loss": 4.2258,
+      "step": 887
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819212669488026e-05,
+      "loss": 4.1054,
+      "step": 888
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981859909118294e-05,
+      "loss": 3.7739,
+      "step": 889
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817984482952378e-05,
+      "loss": 3.8589,
+      "step": 890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981736884486081e-05,
+      "loss": 4.4928,
+      "step": 891
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816752176972815e-05,
+      "loss": 3.1866,
+      "step": 892
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816134479353076e-05,
+      "loss": 3.0173,
+      "step": 893
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815515752066386e-05,
+      "loss": 2.6746,
+      "step": 894
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814895995177653e-05,
+      "loss": 2.5787,
+      "step": 895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814275208751882e-05,
+      "loss": 2.8286,
+      "step": 896
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981365339285419e-05,
+      "loss": 3.1161,
+      "step": 897
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813030547549806e-05,
+      "loss": 3.3945,
+      "step": 898
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812406672904058e-05,
+      "loss": 2.499,
+      "step": 899
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811781768982392e-05,
+      "loss": 2.5743,
+      "step": 900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811155835850355e-05,
+      "loss": 2.9781,
+      "step": 901
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981052887357361e-05,
+      "loss": 2.2962,
+      "step": 902
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809900882217915e-05,
+      "loss": 2.8133,
+      "step": 903
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809271861849147e-05,
+      "loss": 1.9355,
+      "step": 904
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808641812533286e-05,
+      "loss": 2.5126,
+      "step": 905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808010734336425e-05,
+      "loss": 1.7734,
+      "step": 906
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807378627324753e-05,
+      "loss": 3.211,
+      "step": 907
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806745491564588e-05,
+      "loss": 3.1789,
+      "step": 908
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806111327122332e-05,
+      "loss": 2.2472,
+      "step": 909
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980547613406451e-05,
+      "loss": 2.123,
+      "step": 910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804839912457746e-05,
+      "loss": 2.8453,
+      "step": 911
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804202662368782e-05,
+      "loss": 2.3842,
+      "step": 912
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803564383864465e-05,
+      "loss": 3.663,
+      "step": 913
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802925077011742e-05,
+      "loss": 2.3202,
+      "step": 914
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802284741877674e-05,
+      "loss": 2.3461,
+      "step": 915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980164337852943e-05,
+      "loss": 2.5091,
+      "step": 916
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980100098703429e-05,
+      "loss": 2.4754,
+      "step": 917
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800357567459633e-05,
+      "loss": 2.9057,
+      "step": 918
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979971311987295e-05,
+      "loss": 2.4505,
+      "step": 919
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799067644341844e-05,
+      "loss": 1.7782,
+      "step": 920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979842114093402e-05,
+      "loss": 2.6352,
+      "step": 921
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797773609717295e-05,
+      "loss": 2.4391,
+      "step": 922
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797125050759593e-05,
+      "loss": 1.9266,
+      "step": 923
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796475464128943e-05,
+      "loss": 1.7579,
+      "step": 924
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795824849893483e-05,
+      "loss": 2.2545,
+      "step": 925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979517320812146e-05,
+      "loss": 2.429,
+      "step": 926
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794520538881222e-05,
+      "loss": 3.645,
+      "step": 927
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793866842241245e-05,
+      "loss": 3.2109,
+      "step": 928
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979321211827009e-05,
+      "loss": 2.4282,
+      "step": 929
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792556367036432e-05,
+      "loss": 2.8099,
+      "step": 930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791899588609063e-05,
+      "loss": 3.4913,
+      "step": 931
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791241783056874e-05,
+      "loss": 4.5703,
+      "step": 932
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979058295044886e-05,
+      "loss": 3.1634,
+      "step": 933
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789923090854138e-05,
+      "loss": 3.1361,
+      "step": 934
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789262204341918e-05,
+      "loss": 3.4564,
+      "step": 935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788600290981525e-05,
+      "loss": 3.4557,
+      "step": 936
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787937350842394e-05,
+      "loss": 3.4332,
+      "step": 937
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978727338399406e-05,
+      "loss": 3.8361,
+      "step": 938
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786608390506176e-05,
+      "loss": 1.0592,
+      "step": 939
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978594237044849e-05,
+      "loss": 2.5076,
+      "step": 940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785275323890867e-05,
+      "loss": 1.4526,
+      "step": 941
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784607250903278e-05,
+      "loss": 2.0525,
+      "step": 942
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783938151555803e-05,
+      "loss": 1.9729,
+      "step": 943
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783268025918622e-05,
+      "loss": 2.6845,
+      "step": 944
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782596874062028e-05,
+      "loss": 4.1295,
+      "step": 945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781924696056426e-05,
+      "loss": 2.3462,
+      "step": 946
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781251491972324e-05,
+      "loss": 4.114,
+      "step": 947
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780577261880336e-05,
+      "loss": 2.7218,
+      "step": 948
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779902005851187e-05,
+      "loss": 3.0482,
+      "step": 949
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977922572395571e-05,
+      "loss": 2.926,
+      "step": 950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778548416264837e-05,
+      "loss": 2.8731,
+      "step": 951
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977787008284962e-05,
+      "loss": 3.143,
+      "step": 952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977719072378121e-05,
+      "loss": 2.1973,
+      "step": 953
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776510339130874e-05,
+      "loss": 3.1868,
+      "step": 954
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775828928969976e-05,
+      "loss": 3.8777,
+      "step": 955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775146493369996e-05,
+      "loss": 1.8597,
+      "step": 956
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774463032402514e-05,
+      "loss": 3.8031,
+      "step": 957
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773778546139228e-05,
+      "loss": 3.5441,
+      "step": 958
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773093034651928e-05,
+      "loss": 2.3109,
+      "step": 959
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977240649801253e-05,
+      "loss": 3.7239,
+      "step": 960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771718936293047e-05,
+      "loss": 2.841,
+      "step": 961
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97710303495656e-05,
+      "loss": 2.2197,
+      "step": 962
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770340737902417e-05,
+      "loss": 3.035,
+      "step": 963
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769650101375835e-05,
+      "loss": 2.0158,
+      "step": 964
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97689584400583e-05,
+      "loss": 2.8878,
+      "step": 965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768265754022366e-05,
+      "loss": 2.8392,
+      "step": 966
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767572043340687e-05,
+      "loss": 3.6075,
+      "step": 967
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766877308086038e-05,
+      "loss": 1.8854,
+      "step": 968
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766181548331283e-05,
+      "loss": 2.7456,
+      "step": 969
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765484764149413e-05,
+      "loss": 3.4188,
+      "step": 970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764786955613514e-05,
+      "loss": 2.7482,
+      "step": 971
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764088122796785e-05,
+      "loss": 2.6491,
+      "step": 972
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763388265772527e-05,
+      "loss": 1.8656,
+      "step": 973
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762687384614152e-05,
+      "loss": 2.6335,
+      "step": 974
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976198547939518e-05,
+      "loss": 3.9693,
+      "step": 975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976128255018924e-05,
+      "loss": 2.6706,
+      "step": 976
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760578597070064e-05,
+      "loss": 2.2208,
+      "step": 977
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759873620111492e-05,
+      "loss": 2.3877,
+      "step": 978
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759167619387474e-05,
+      "loss": 3.5069,
+      "step": 979
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758460594972068e-05,
+      "loss": 2.3238,
+      "step": 980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757752546939434e-05,
+      "loss": 2.904,
+      "step": 981
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975704347536385e-05,
+      "loss": 3.5287,
+      "step": 982
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756333380319687e-05,
+      "loss": 2.4599,
+      "step": 983
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975562226188143e-05,
+      "loss": 2.3268,
+      "step": 984
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754910120123675e-05,
+      "loss": 3.0736,
+      "step": 985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754196955121125e-05,
+      "loss": 2.2762,
+      "step": 986
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975348276694858e-05,
+      "loss": 3.1459,
+      "step": 987
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752767555680967e-05,
+      "loss": 2.2893,
+      "step": 988
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97520513213933e-05,
+      "loss": 2.8897,
+      "step": 989
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751334064160708e-05,
+      "loss": 2.257,
+      "step": 990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750615784058427e-05,
+      "loss": 1.6723,
+      "step": 991
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749896481161807e-05,
+      "loss": 3.4862,
+      "step": 992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97491761555463e-05,
+      "loss": 2.5168,
+      "step": 993
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748454807287458e-05,
+      "loss": 1.2839,
+      "step": 994
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747732436460955e-05,
+      "loss": 4.0738,
+      "step": 995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747009043142556e-05,
+      "loss": 2.1561,
+      "step": 996
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746284627408143e-05,
+      "loss": 3.7232,
+      "step": 997
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974555918933371e-05,
+      "loss": 3.2896,
+      "step": 998
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974483272899535e-05,
+      "loss": 2.9798,
+      "step": 999
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744105246469264e-05,
+      "loss": 2.0935,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974337674183176e-05,
+      "loss": 2.8534,
+      "step": 1001
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742647215159255e-05,
+      "loss": 4.3149,
+      "step": 1002
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741916666528273e-05,
+      "loss": 4.0613,
+      "step": 1003
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974118509601545e-05,
+      "loss": 2.3783,
+      "step": 1004
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740452503697518e-05,
+      "loss": 2.2033,
+      "step": 1005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739718889651328e-05,
+      "loss": 2.5424,
+      "step": 1006
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973898425395383e-05,
+      "loss": 3.6531,
+      "step": 1007
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738248596682078e-05,
+      "loss": 2.8453,
+      "step": 1008
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973751191791325e-05,
+      "loss": 3.0196,
+      "step": 1009
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736774217724614e-05,
+      "loss": 0.9197,
+      "step": 1010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736035496193552e-05,
+      "loss": 2.2568,
+      "step": 1011
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973529575339755e-05,
+      "loss": 2.3589,
+      "step": 1012
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734554989414212e-05,
+      "loss": 3.3122,
+      "step": 1013
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733813204321233e-05,
+      "loss": 2.6589,
+      "step": 1014
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733070398196423e-05,
+      "loss": 3.86,
+      "step": 1015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732326571117703e-05,
+      "loss": 2.4102,
+      "step": 1016
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731581723163094e-05,
+      "loss": 2.945,
+      "step": 1017
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730835854410726e-05,
+      "loss": 2.6282,
+      "step": 1018
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730088964938842e-05,
+      "loss": 3.5551,
+      "step": 1019
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729341054825783e-05,
+      "loss": 2.3609,
+      "step": 1020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728592124150003e-05,
+      "loss": 3.082,
+      "step": 1021
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972784217299006e-05,
+      "loss": 3.1901,
+      "step": 1022
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727091201424623e-05,
+      "loss": 3.2051,
+      "step": 1023
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726339209532462e-05,
+      "loss": 3.9922,
+      "step": 1024
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972558619739246e-05,
+      "loss": 2.9471,
+      "step": 1025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724832165083607e-05,
+      "loss": 3.3637,
+      "step": 1026
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972407711268499e-05,
+      "loss": 1.9433,
+      "step": 1027
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723321040275816e-05,
+      "loss": 2.1976,
+      "step": 1028
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972256394793539e-05,
+      "loss": 2.7193,
+      "step": 1029
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972180583574313e-05,
+      "loss": 2.3749,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721046703778563e-05,
+      "loss": 1.5336,
+      "step": 1031
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972028655212131e-05,
+      "loss": 2.4131,
+      "step": 1032
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971952538085111e-05,
+      "loss": 2.4065,
+      "step": 1033
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971876319004781e-05,
+      "loss": 2.6428,
+      "step": 1034
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717999979791356e-05,
+      "loss": 2.8441,
+      "step": 1035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717235750161808e-05,
+      "loss": 2.4207,
+      "step": 1036
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716470501239327e-05,
+      "loss": 2.8745,
+      "step": 1037
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715704233104188e-05,
+      "loss": 2.5637,
+      "step": 1038
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714936945836764e-05,
+      "loss": 2.8427,
+      "step": 1039
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714168639517543e-05,
+      "loss": 2.2047,
+      "step": 1040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971339931422712e-05,
+      "loss": 2.485,
+      "step": 1041
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971262897004619e-05,
+      "loss": 2.2089,
+      "step": 1042
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711857607055553e-05,
+      "loss": 2.866,
+      "step": 1043
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971108522533613e-05,
+      "loss": 2.1044,
+      "step": 1044
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710311824968942e-05,
+      "loss": 3.2523,
+      "step": 1045
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709537406035106e-05,
+      "loss": 2.8677,
+      "step": 1046
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970876196861586e-05,
+      "loss": 3.7139,
+      "step": 1047
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707985512792544e-05,
+      "loss": 3.5907,
+      "step": 1048
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707208038646605e-05,
+      "loss": 3.1036,
+      "step": 1049
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9706429546259592e-05,
+      "loss": 2.6016,
+      "step": 1050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705650035713172e-05,
+      "loss": 3.7879,
+      "step": 1051
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704869507089105e-05,
+      "loss": 2.7432,
+      "step": 1052
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970408796046927e-05,
+      "loss": 1.9733,
+      "step": 1053
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970330539593565e-05,
+      "loss": 1.6589,
+      "step": 1054
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702521813570322e-05,
+      "loss": 2.0651,
+      "step": 1055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970173721345549e-05,
+      "loss": 2.9539,
+      "step": 1056
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970095159567345e-05,
+      "loss": 2.0049,
+      "step": 1057
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700164960306612e-05,
+      "loss": 2.9624,
+      "step": 1058
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969937730743749e-05,
+      "loss": 2.4031,
+      "step": 1059
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698588637148705e-05,
+      "loss": 3.0742,
+      "step": 1060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697798949522983e-05,
+      "loss": 2.477,
+      "step": 1061
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697008244643163e-05,
+      "loss": 3.1747,
+      "step": 1062
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969621652259218e-05,
+      "loss": 2.5953,
+      "step": 1063
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695423783453086e-05,
+      "loss": 2.3631,
+      "step": 1064
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694630027309035e-05,
+      "loss": 2.114,
+      "step": 1065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969383525424329e-05,
+      "loss": 1.9189,
+      "step": 1066
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693039464339216e-05,
+      "loss": 3.331,
+      "step": 1067
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692242657680286e-05,
+      "loss": 3.8493,
+      "step": 1068
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969144483435009e-05,
+      "loss": 4.2373,
+      "step": 1069
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690645994432307e-05,
+      "loss": 2.496,
+      "step": 1070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689846138010736e-05,
+      "loss": 2.8863,
+      "step": 1071
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689045265169272e-05,
+      "loss": 2.5791,
+      "step": 1072
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688243375991935e-05,
+      "loss": 3.286,
+      "step": 1073
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968744047056283e-05,
+      "loss": 2.7699,
+      "step": 1074
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686636548966177e-05,
+      "loss": 3.9351,
+      "step": 1075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685831611286312e-05,
+      "loss": 2.7144,
+      "step": 1076
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685025657607663e-05,
+      "loss": 2.3764,
+      "step": 1077
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684218688014773e-05,
+      "loss": 3.357,
+      "step": 1078
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683410702592284e-05,
+      "loss": 1.8856,
+      "step": 1079
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682601701424958e-05,
+      "loss": 2.7734,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681791684597653e-05,
+      "loss": 2.6181,
+      "step": 1081
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680980652195334e-05,
+      "loss": 2.6534,
+      "step": 1082
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680168604303074e-05,
+      "loss": 1.8387,
+      "step": 1083
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679355541006056e-05,
+      "loss": 3.1424,
+      "step": 1084
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678541462389564e-05,
+      "loss": 2.2163,
+      "step": 1085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967772636853899e-05,
+      "loss": 2.2341,
+      "step": 1086
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967691025953984e-05,
+      "loss": 2.9522,
+      "step": 1087
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676093135477713e-05,
+      "loss": 1.2785,
+      "step": 1088
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675274996438324e-05,
+      "loss": 2.9574,
+      "step": 1089
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674455842507494e-05,
+      "loss": 3.2596,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673635673771145e-05,
+      "loss": 1.2542,
+      "step": 1091
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672814490315312e-05,
+      "loss": 2.6168,
+      "step": 1092
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967199229222613e-05,
+      "loss": 2.9484,
+      "step": 1093
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967116907958985e-05,
+      "loss": 2.2062,
+      "step": 1094
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670344852492814e-05,
+      "loss": 2.6624,
+      "step": 1095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669519611021485e-05,
+      "loss": 2.0751,
+      "step": 1096
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668693355262428e-05,
+      "loss": 2.0121,
+      "step": 1097
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667866085302312e-05,
+      "loss": 3.3372,
+      "step": 1098
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667037801227914e-05,
+      "loss": 2.8934,
+      "step": 1099
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666208503126115e-05,
+      "loss": 3.0462,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665378191083907e-05,
+      "loss": 2.418,
+      "step": 1101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664546865188384e-05,
+      "loss": 2.8958,
+      "step": 1102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966371452552675e-05,
+      "loss": 2.2621,
+      "step": 1103
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662881172186313e-05,
+      "loss": 3.2713,
+      "step": 1104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966204680525449e-05,
+      "loss": 3.4431,
+      "step": 1105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661211424818798e-05,
+      "loss": 3.3638,
+      "step": 1106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660375030966868e-05,
+      "loss": 2.254,
+      "step": 1107
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659537623786428e-05,
+      "loss": 2.5935,
+      "step": 1108
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965869920336533e-05,
+      "loss": 2.5149,
+      "step": 1109
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657859769791506e-05,
+      "loss": 3.7677,
+      "step": 1110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965701932315302e-05,
+      "loss": 2.1623,
+      "step": 1111
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656177863538025e-05,
+      "loss": 3.6034,
+      "step": 1112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655335391034787e-05,
+      "loss": 3.6113,
+      "step": 1113
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965449190573168e-05,
+      "loss": 3.6682,
+      "step": 1114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965364740771718e-05,
+      "loss": 2.6009,
+      "step": 1115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965280189707987e-05,
+      "loss": 3.4957,
+      "step": 1116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965195537390844e-05,
+      "loss": 2.4701,
+      "step": 1117
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965110783829169e-05,
+      "loss": 3.5485,
+      "step": 1118
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965025929031852e-05,
+      "loss": 2.6511,
+      "step": 1119
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649409730077934e-05,
+      "loss": 2.7395,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648559157659054e-05,
+      "loss": 3.8641,
+      "step": 1121
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96477075731511e-05,
+      "loss": 3.4058,
+      "step": 1122
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96468549766434e-05,
+      "loss": 1.8407,
+      "step": 1123
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646001368225382e-05,
+      "loss": 3.2496,
+      "step": 1124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964514674798659e-05,
+      "loss": 1.4114,
+      "step": 1125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644291116016666e-05,
+      "loss": 3.2121,
+      "step": 1126
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964343447240537e-05,
+      "loss": 2.7698,
+      "step": 1127
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642576817242553e-05,
+      "loss": 2.6822,
+      "step": 1128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641718150618177e-05,
+      "loss": 3.0543,
+      "step": 1129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640858472622316e-05,
+      "loss": 2.1284,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963999778334515e-05,
+      "loss": 3.2924,
+      "step": 1131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639136082876954e-05,
+      "loss": 2.6594,
+      "step": 1132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638273371308118e-05,
+      "loss": 3.2417,
+      "step": 1133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637409648729142e-05,
+      "loss": 1.8997,
+      "step": 1134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963654491523062e-05,
+      "loss": 2.4587,
+      "step": 1135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635679170903258e-05,
+      "loss": 2.9246,
+      "step": 1136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634812415837876e-05,
+      "loss": 3.1911,
+      "step": 1137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963394465012539e-05,
+      "loss": 3.2688,
+      "step": 1138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963307587385682e-05,
+      "loss": 3.7672,
+      "step": 1139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632206087123296e-05,
+      "loss": 3.2342,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631335290016063e-05,
+      "loss": 3.6208,
+      "step": 1141
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630463482626454e-05,
+      "loss": 2.4573,
+      "step": 1142
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629590665045925e-05,
+      "loss": 2.7313,
+      "step": 1143
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962871683736603e-05,
+      "loss": 2.4497,
+      "step": 1144
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627841999678422e-05,
+      "loss": 2.999,
+      "step": 1145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626966152074876e-05,
+      "loss": 3.6978,
+      "step": 1146
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626089294647256e-05,
+      "loss": 2.1761,
+      "step": 1147
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962521142748755e-05,
+      "loss": 2.2419,
+      "step": 1148
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624332550687834e-05,
+      "loss": 2.9257,
+      "step": 1149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623452664340305e-05,
+      "loss": 3.6914,
+      "step": 1150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622571768537254e-05,
+      "loss": 2.8717,
+      "step": 1151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621689863371083e-05,
+      "loss": 2.4524,
+      "step": 1152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620806948934297e-05,
+      "loss": 2.7469,
+      "step": 1153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961992302531952e-05,
+      "loss": 3.8772,
+      "step": 1154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619038092619465e-05,
+      "loss": 3.4666,
+      "step": 1155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618152150926953e-05,
+      "loss": 2.837,
+      "step": 1156
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617265200334924e-05,
+      "loss": 2.4407,
+      "step": 1157
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961637724093641e-05,
+      "loss": 3.884,
+      "step": 1158
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961548827282455e-05,
+      "loss": 2.8858,
+      "step": 1159
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614598296092603e-05,
+      "loss": 2.8456,
+      "step": 1160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613707310833913e-05,
+      "loss": 1.9449,
+      "step": 1161
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612815317141947e-05,
+      "loss": 3.3443,
+      "step": 1162
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611922315110268e-05,
+      "loss": 3.2466,
+      "step": 1163
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611028304832547e-05,
+      "loss": 2.6147,
+      "step": 1164
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610133286402565e-05,
+      "loss": 2.8495,
+      "step": 1165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96092372599142e-05,
+      "loss": 4.0532,
+      "step": 1166
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608340225461447e-05,
+      "loss": 2.8752,
+      "step": 1167
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607442183138403e-05,
+      "loss": 2.8825,
+      "step": 1168
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606543133039254e-05,
+      "loss": 3.6893,
+      "step": 1169
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605643075258323e-05,
+      "loss": 1.5953,
+      "step": 1170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960474200989001e-05,
+      "loss": 3.299,
+      "step": 1171
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960383993702884e-05,
+      "loss": 2.1121,
+      "step": 1172
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602936856769432e-05,
+      "loss": 3.4373,
+      "step": 1173
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602032769206517e-05,
+      "loss": 2.7025,
+      "step": 1174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960112767443493e-05,
+      "loss": 2.663,
+      "step": 1175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600221572549607e-05,
+      "loss": 2.7229,
+      "step": 1176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599314463645597e-05,
+      "loss": 2.1539,
+      "step": 1177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598406347818056e-05,
+      "loss": 2.4061,
+      "step": 1178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597497225162233e-05,
+      "loss": 2.6713,
+      "step": 1179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596587095773496e-05,
+      "loss": 2.6841,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959567595974731e-05,
+      "loss": 2.9503,
+      "step": 1181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594763817179254e-05,
+      "loss": 2.7094,
+      "step": 1182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593850668165002e-05,
+      "loss": 2.3871,
+      "step": 1183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959293651280034e-05,
+      "loss": 3.6402,
+      "step": 1184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592021351181163e-05,
+      "loss": 3.7053,
+      "step": 1185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591105183403463e-05,
+      "loss": 3.3821,
+      "step": 1186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590188009563342e-05,
+      "loss": 2.176,
+      "step": 1187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958926982975701e-05,
+      "loss": 3.3238,
+      "step": 1188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588350644080777e-05,
+      "loss": 2.7572,
+      "step": 1189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958743045263106e-05,
+      "loss": 2.7957,
+      "step": 1190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958650925550439e-05,
+      "loss": 3.5404,
+      "step": 1191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958558705279739e-05,
+      "loss": 3.1395,
+      "step": 1192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584663844606792e-05,
+      "loss": 1.9391,
+      "step": 1193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583739631029446e-05,
+      "loss": 3.9691,
+      "step": 1194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582814412162288e-05,
+      "loss": 2.5547,
+      "step": 1195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581888188102375e-05,
+      "loss": 3.1083,
+      "step": 1196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580960958946865e-05,
+      "loss": 3.0866,
+      "step": 1197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958003272479301e-05,
+      "loss": 2.4436,
+      "step": 1198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957910348573819e-05,
+      "loss": 4.0389,
+      "step": 1199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957817324187987e-05,
+      "loss": 2.9695,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577241993315634e-05,
+      "loss": 3.873,
+      "step": 1201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957630974014316e-05,
+      "loss": 2.1232,
+      "step": 1202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957537648246024e-05,
+      "loss": 3.1511,
+      "step": 1203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574442220364768e-05,
+      "loss": 3.2416,
+      "step": 1204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957350695395474e-05,
+      "loss": 2.0445,
+      "step": 1205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957257068332827e-05,
+      "loss": 3.2386,
+      "step": 1206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571633408583558e-05,
+      "loss": 1.8596,
+      "step": 1207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570695129818928e-05,
+      "loss": 3.3555,
+      "step": 1208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569755847132796e-05,
+      "loss": 3.3899,
+      "step": 1209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956881556062369e-05,
+      "loss": 3.2039,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567874270390242e-05,
+      "loss": 2.9763,
+      "step": 1211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956693197653119e-05,
+      "loss": 3.5904,
+      "step": 1212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565988679145375e-05,
+      "loss": 3.2588,
+      "step": 1213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565044378331745e-05,
+      "loss": 3.0678,
+      "step": 1214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956409907418935e-05,
+      "loss": 2.2442,
+      "step": 1215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563152766817356e-05,
+      "loss": 2.3278,
+      "step": 1216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562205456315017e-05,
+      "loss": 2.5902,
+      "step": 1217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561257142781706e-05,
+      "loss": 2.1461,
+      "step": 1218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560307826316892e-05,
+      "loss": 3.6075,
+      "step": 1219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559357507020163e-05,
+      "loss": 1.6113,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558406184991195e-05,
+      "loss": 3.3414,
+      "step": 1221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955745386032978e-05,
+      "loss": 3.0593,
+      "step": 1222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556500533135817e-05,
+      "loss": 1.8551,
+      "step": 1223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555546203509297e-05,
+      "loss": 2.7098,
+      "step": 1224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955459087155033e-05,
+      "loss": 3.4937,
+      "step": 1225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553634537359123e-05,
+      "loss": 3.0414,
+      "step": 1226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552677201035993e-05,
+      "loss": 3.5143,
+      "step": 1227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551718862681363e-05,
+      "loss": 2.7086,
+      "step": 1228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550759522395753e-05,
+      "loss": 2.5238,
+      "step": 1229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549799180279793e-05,
+      "loss": 3.3282,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548837836434227e-05,
+      "loss": 2.199,
+      "step": 1231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547875490959884e-05,
+      "loss": 3.2734,
+      "step": 1232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954691214395772e-05,
+      "loss": 2.1913,
+      "step": 1233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545947795528777e-05,
+      "loss": 3.1175,
+      "step": 1234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544982445774217e-05,
+      "loss": 1.7422,
+      "step": 1235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544016094795294e-05,
+      "loss": 2.2256,
+      "step": 1236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543048742693383e-05,
+      "loss": 2.1147,
+      "step": 1237
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542080389569947e-05,
+      "loss": 3.24,
+      "step": 1238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541111035526563e-05,
+      "loss": 4.3408,
+      "step": 1239
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540140680664915e-05,
+      "loss": 1.8674,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539169325086786e-05,
+      "loss": 2.7479,
+      "step": 1241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538196968894067e-05,
+      "loss": 2.3637,
+      "step": 1242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537223612188752e-05,
+      "loss": 3.0458,
+      "step": 1243
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953624925507295e-05,
+      "loss": 3.6755,
+      "step": 1244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535273897648857e-05,
+      "loss": 2.8725,
+      "step": 1245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534297540018787e-05,
+      "loss": 3.4656,
+      "step": 1246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533320182285152e-05,
+      "loss": 3.4219,
+      "step": 1247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953234182455048e-05,
+      "loss": 3.3352,
+      "step": 1248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531362466917388e-05,
+      "loss": 2.0987,
+      "step": 1249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953038210948861e-05,
+      "loss": 2.817,
+      "step": 1250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529400752366983e-05,
+      "loss": 2.663,
+      "step": 1251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528418395655443e-05,
+      "loss": 2.5967,
+      "step": 1252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527435039457035e-05,
+      "loss": 2.1047,
+      "step": 1253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952645068387491e-05,
+      "loss": 3.0115,
+      "step": 1254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525465329012322e-05,
+      "loss": 4.421,
+      "step": 1255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952447897497263e-05,
+      "loss": 3.1776,
+      "step": 1256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95234916218593e-05,
+      "loss": 2.6592,
+      "step": 1257
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522503269775897e-05,
+      "loss": 3.3508,
+      "step": 1258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95215139188261e-05,
+      "loss": 1.8699,
+      "step": 1259
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952052356911368e-05,
+      "loss": 2.4382,
+      "step": 1260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951953222074252e-05,
+      "loss": 1.6127,
+      "step": 1261
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518539873816617e-05,
+      "loss": 3.8472,
+      "step": 1262
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517546528440054e-05,
+      "loss": 2.81,
+      "step": 1263
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516552184717036e-05,
+      "loss": 2.8424,
+      "step": 1264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515556842751863e-05,
+      "loss": 2.6819,
+      "step": 1265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514560502648936e-05,
+      "loss": 2.8288,
+      "step": 1266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513563164512773e-05,
+      "loss": 2.4013,
+      "step": 1267
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951256482844799e-05,
+      "loss": 1.5787,
+      "step": 1268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511565494559298e-05,
+      "loss": 3.331,
+      "step": 1269
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 2.9698,
+      "step": 1270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950956383372963e-05,
+      "loss": 2.3039,
+      "step": 1271
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508561506998613e-05,
+      "loss": 3.3427,
+      "step": 1272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950755818286362e-05,
+      "loss": 1.9696,
+      "step": 1273
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95065538614299e-05,
+      "loss": 3.7201,
+      "step": 1274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505548542802805e-05,
+      "loss": 1.7835,
+      "step": 1275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950454222708778e-05,
+      "loss": 3.1803,
+      "step": 1276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503534914390393e-05,
+      "loss": 1.9598,
+      "step": 1277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502526604816293e-05,
+      "loss": 2.5744,
+      "step": 1278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950151729847126e-05,
+      "loss": 3.3385,
+      "step": 1279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950050699546116e-05,
+      "loss": 3.1327,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949949569589197e-05,
+      "loss": 2.263,
+      "step": 1281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498483399869767e-05,
+      "loss": 2.7243,
+      "step": 1282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949747010750074e-05,
+      "loss": 4.0374,
+      "step": 1283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949645581889118e-05,
+      "loss": 3.9411,
+      "step": 1284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949544053414748e-05,
+      "loss": 3.3078,
+      "step": 1285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494424253376133e-05,
+      "loss": 1.6954,
+      "step": 1286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493406976683752e-05,
+      "loss": 0.8448,
+      "step": 1287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492388704177036e-05,
+      "loss": 2.9006,
+      "step": 1288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491369435962802e-05,
+      "loss": 2.6007,
+      "step": 1289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490349172147964e-05,
+      "loss": 2.3221,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948932791283954e-05,
+      "loss": 2.3399,
+      "step": 1291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488305658144666e-05,
+      "loss": 1.9226,
+      "step": 1292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487282408170564e-05,
+      "loss": 2.4074,
+      "step": 1293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486258163024567e-05,
+      "loss": 3.4556,
+      "step": 1294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485232922814117e-05,
+      "loss": 3.0075,
+      "step": 1295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484206687646753e-05,
+      "loss": 2.1895,
+      "step": 1296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948317945763013e-05,
+      "loss": 3.3498,
+      "step": 1297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948215123287199e-05,
+      "loss": 1.0429,
+      "step": 1298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94811220134802e-05,
+      "loss": 3.0613,
+      "step": 1299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480091799562706e-05,
+      "loss": 1.743,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479060591227586e-05,
+      "loss": 2.6053,
+      "step": 1301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478028388583e-05,
+      "loss": 2.6321,
+      "step": 1302
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947699519173723e-05,
+      "loss": 3.3507,
+      "step": 1303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475961000798645e-05,
+      "loss": 2.8452,
+      "step": 1304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947492581587573e-05,
+      "loss": 3.154,
+      "step": 1305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947388963707707e-05,
+      "loss": 1.9152,
+      "step": 1306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947285246451136e-05,
+      "loss": 2.3711,
+      "step": 1307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947181429828739e-05,
+      "loss": 2.3375,
+      "step": 1308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9470775138514063e-05,
+      "loss": 1.4465,
+      "step": 1309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469734985300373e-05,
+      "loss": 3.2832,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946869383875544e-05,
+      "loss": 2.6509,
+      "step": 1311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467651698988464e-05,
+      "loss": 3.3199,
+      "step": 1312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466608566108768e-05,
+      "loss": 2.7798,
+      "step": 1313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465564440225768e-05,
+      "loss": 2.3491,
+      "step": 1314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464519321448988e-05,
+      "loss": 2.6452,
+      "step": 1315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463473209888063e-05,
+      "loss": 3.554,
+      "step": 1316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462426105652713e-05,
+      "loss": 1.8777,
+      "step": 1317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461378008852785e-05,
+      "loss": 1.1169,
+      "step": 1318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460328919598216e-05,
+      "loss": 3.0887,
+      "step": 1319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459278837999048e-05,
+      "loss": 2.4271,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945822776416543e-05,
+      "loss": 2.1419,
+      "step": 1321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945717569820762e-05,
+      "loss": 1.9341,
+      "step": 1322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945612264023597e-05,
+      "loss": 2.8708,
+      "step": 1323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455068590360943e-05,
+      "loss": 1.563,
+      "step": 1324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454013548693103e-05,
+      "loss": 3.7754,
+      "step": 1325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945295751534312e-05,
+      "loss": 3.0629,
+      "step": 1326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451900490421766e-05,
+      "loss": 3.3756,
+      "step": 1327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450842474039914e-05,
+      "loss": 2.8833,
+      "step": 1328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449783466308553e-05,
+      "loss": 2.9509,
+      "step": 1329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448723467338765e-05,
+      "loss": 1.994,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447662477241737e-05,
+      "loss": 2.558,
+      "step": 1331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944660049612876e-05,
+      "loss": 2.1696,
+      "step": 1332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445537524111238e-05,
+      "loss": 3.0887,
+      "step": 1333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444473561300666e-05,
+      "loss": 2.8445,
+      "step": 1334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944340860780865e-05,
+      "loss": 3.1515,
+      "step": 1335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442342663746903e-05,
+      "loss": 2.343,
+      "step": 1336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944127572922723e-05,
+      "loss": 2.647,
+      "step": 1337
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944020780436155e-05,
+      "loss": 2.7309,
+      "step": 1338
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943913888926189e-05,
+      "loss": 2.9226,
+      "step": 1339
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438068984040366e-05,
+      "loss": 2.83,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943699808880921e-05,
+      "loss": 3.5139,
+      "step": 1341
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943592620368075e-05,
+      "loss": 3.0993,
+      "step": 1342
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434853328767426e-05,
+      "loss": 2.4562,
+      "step": 1343
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943377946418178e-05,
+      "loss": 2.8747,
+      "step": 1344
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432704610036448e-05,
+      "loss": 3.321,
+      "step": 1345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943162876644418e-05,
+      "loss": 2.492,
+      "step": 1346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430551933517833e-05,
+      "loss": 1.7562,
+      "step": 1347
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942947411137035e-05,
+      "loss": 2.7406,
+      "step": 1348
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428395300114803e-05,
+      "loss": 3.3155,
+      "step": 1349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9427315499864345e-05,
+      "loss": 3.342,
+      "step": 1350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426234710732244e-05,
+      "loss": 3.2145,
+      "step": 1351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942515293283187e-05,
+      "loss": 3.9369,
+      "step": 1352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424070166276695e-05,
+      "loss": 2.0684,
+      "step": 1353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94229864111803e-05,
+      "loss": 2.7521,
+      "step": 1354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421901667656364e-05,
+      "loss": 2.6003,
+      "step": 1355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420815935818673e-05,
+      "loss": 2.4918,
+      "step": 1356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941972921578111e-05,
+      "loss": 3.747,
+      "step": 1357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418641507657673e-05,
+      "loss": 2.3854,
+      "step": 1358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417552811562457e-05,
+      "loss": 3.0734,
+      "step": 1359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416463127609655e-05,
+      "loss": 2.1385,
+      "step": 1360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415372455913577e-05,
+      "loss": 2.316,
+      "step": 1361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414280796588625e-05,
+      "loss": 3.2612,
+      "step": 1362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413188149749312e-05,
+      "loss": 2.7403,
+      "step": 1363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941209451551025e-05,
+      "loss": 2.1354,
+      "step": 1364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410999893986157e-05,
+      "loss": 3.3563,
+      "step": 1365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940990428529185e-05,
+      "loss": 3.0104,
+      "step": 1366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9408807689542257e-05,
+      "loss": 3.333,
+      "step": 1367
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407710106852405e-05,
+      "loss": 2.4119,
+      "step": 1368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406611537337425e-05,
+      "loss": 2.8258,
+      "step": 1369
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405511981112553e-05,
+      "loss": 2.2422,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404411438293126e-05,
+      "loss": 2.1916,
+      "step": 1371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940330990899459e-05,
+      "loss": 3.6327,
+      "step": 1372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402207393332483e-05,
+      "loss": 3.0866,
+      "step": 1373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401103891422455e-05,
+      "loss": 1.3166,
+      "step": 1374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399999403380266e-05,
+      "loss": 1.9838,
+      "step": 1375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398893929321763e-05,
+      "loss": 1.8537,
+      "step": 1376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397787469362908e-05,
+      "loss": 2.6427,
+      "step": 1377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396680023619767e-05,
+      "loss": 3.0937,
+      "step": 1378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93955715922085e-05,
+      "loss": 2.8522,
+      "step": 1379
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394462175245382e-05,
+      "loss": 3.4511,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939335177284678e-05,
+      "loss": 3.5323,
+      "step": 1381
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392240385129174e-05,
+      "loss": 2.8752,
+      "step": 1382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391128012209144e-05,
+      "loss": 2.1216,
+      "step": 1383
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939001465420337e-05,
+      "loss": 2.006,
+      "step": 1384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388900311228636e-05,
+      "loss": 1.9633,
+      "step": 1385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387784983401837e-05,
+      "loss": 3.699,
+      "step": 1386
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386668670839965e-05,
+      "loss": 3.6948,
+      "step": 1387
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385551373660113e-05,
+      "loss": 2.5877,
+      "step": 1388
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938443309197948e-05,
+      "loss": 3.2475,
+      "step": 1389
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383313825915372e-05,
+      "loss": 2.7405,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382193575585193e-05,
+      "loss": 3.2195,
+      "step": 1391
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381072341106453e-05,
+      "loss": 1.9355,
+      "step": 1392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937995012259676e-05,
+      "loss": 2.616,
+      "step": 1393
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378826920173835e-05,
+      "loss": 2.6908,
+      "step": 1394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377702733955493e-05,
+      "loss": 3.6732,
+      "step": 1395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937657756405966e-05,
+      "loss": 2.1133,
+      "step": 1396
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375451410604358e-05,
+      "loss": 1.8884,
+      "step": 1397
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374324273707717e-05,
+      "loss": 2.5154,
+      "step": 1398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373196153487962e-05,
+      "loss": 2.8224,
+      "step": 1399
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937206705006344e-05,
+      "loss": 2.5166,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370936963552574e-05,
+      "loss": 3.282,
+      "step": 1401
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9369805894073917e-05,
+      "loss": 2.0425,
+      "step": 1402
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936867384174611e-05,
+      "loss": 2.8744,
+      "step": 1403
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367540806687894e-05,
+      "loss": 2.4593,
+      "step": 1404
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366406789018127e-05,
+      "loss": 3.8015,
+      "step": 1405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365271788855756e-05,
+      "loss": 2.8512,
+      "step": 1406
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936413580631984e-05,
+      "loss": 2.7485,
+      "step": 1407
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362998841529542e-05,
+      "loss": 2.8955,
+      "step": 1408
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361860894604116e-05,
+      "loss": 2.8621,
+      "step": 1409
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360721965662934e-05,
+      "loss": 3.187,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359582054825462e-05,
+      "loss": 2.6571,
+      "step": 1411
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935844116221127e-05,
+      "loss": 2.7402,
+      "step": 1412
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357299287940037e-05,
+      "loss": 2.8248,
+      "step": 1413
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356156432131533e-05,
+      "loss": 2.3989,
+      "step": 1414
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355012594905645e-05,
+      "loss": 2.8015,
+      "step": 1415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353867776382357e-05,
+      "loss": 2.8319,
+      "step": 1416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352721976681747e-05,
+      "loss": 3.3669,
+      "step": 1417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351575195924014e-05,
+      "loss": 1.6652,
+      "step": 1418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935042743422944e-05,
+      "loss": 3.6618,
+      "step": 1419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349278691718426e-05,
+      "loss": 1.9709,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934812896851147e-05,
+      "loss": 2.2035,
+      "step": 1421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346978264729173e-05,
+      "loss": 1.3242,
+      "step": 1422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934582658049223e-05,
+      "loss": 3.3364,
+      "step": 1423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934467391592146e-05,
+      "loss": 3.7768,
+      "step": 1424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343520271137764e-05,
+      "loss": 2.4741,
+      "step": 1425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342365646262158e-05,
+      "loss": 2.6153,
+      "step": 1426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341210041415755e-05,
+      "loss": 2.8541,
+      "step": 1427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340053456719768e-05,
+      "loss": 3.5882,
+      "step": 1428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338895892295527e-05,
+      "loss": 2.9758,
+      "step": 1429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337737348264448e-05,
+      "loss": 2.0189,
+      "step": 1430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336577824748057e-05,
+      "loss": 2.9079,
+      "step": 1431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335417321867988e-05,
+      "loss": 2.1185,
+      "step": 1432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334255839745966e-05,
+      "loss": 4.439,
+      "step": 1433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333093378503832e-05,
+      "loss": 2.6778,
+      "step": 1434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331929938263515e-05,
+      "loss": 2.6597,
+      "step": 1435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330765519147058e-05,
+      "loss": 2.9318,
+      "step": 1436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329600121276607e-05,
+      "loss": 3.1551,
+      "step": 1437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9328433744774403e-05,
+      "loss": 1.9802,
+      "step": 1438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327266389762787e-05,
+      "loss": 2.0633,
+      "step": 1439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326098056364224e-05,
+      "loss": 2.6533,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324928744701256e-05,
+      "loss": 1.7769,
+      "step": 1441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932375845489654e-05,
+      "loss": 3.2159,
+      "step": 1442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322587187072835e-05,
+      "loss": 3.534,
+      "step": 1443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321414941353006e-05,
+      "loss": 2.6138,
+      "step": 1444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320241717860007e-05,
+      "loss": 3.218,
+      "step": 1445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931906751671691e-05,
+      "loss": 2.6048,
+      "step": 1446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931789233804688e-05,
+      "loss": 2.9304,
+      "step": 1447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931671618197319e-05,
+      "loss": 3.3579,
+      "step": 1448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315539048619212e-05,
+      "loss": 4.0668,
+      "step": 1449
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9314360938108427e-05,
+      "loss": 2.0344,
+      "step": 1450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313181850564402e-05,
+      "loss": 1.8135,
+      "step": 1451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931200178611083e-05,
+      "loss": 2.5207,
+      "step": 1452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310820744871486e-05,
+      "loss": 1.6594,
+      "step": 1453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930963872697026e-05,
+      "loss": 2.998,
+      "step": 1454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930845573253114e-05,
+      "loss": 2.7194,
+      "step": 1455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307271761678214e-05,
+      "loss": 2.8914,
+      "step": 1456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306086814535678e-05,
+      "loss": 2.6411,
+      "step": 1457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304900891227825e-05,
+      "loss": 2.7349,
+      "step": 1458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303713991879052e-05,
+      "loss": 2.2438,
+      "step": 1459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302526116613863e-05,
+      "loss": 1.8662,
+      "step": 1460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930133726555686e-05,
+      "loss": 2.9676,
+      "step": 1461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300147438832747e-05,
+      "loss": 2.533,
+      "step": 1462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298956636566326e-05,
+      "loss": 2.7232,
+      "step": 1463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297764858882516e-05,
+      "loss": 3.2355,
+      "step": 1464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296572105906323e-05,
+      "loss": 2.6354,
+      "step": 1465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295378377762862e-05,
+      "loss": 2.4022,
+      "step": 1466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294183674577353e-05,
+      "loss": 2.4341,
+      "step": 1467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292987996475113e-05,
+      "loss": 2.924,
+      "step": 1468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291791343581557e-05,
+      "loss": 1.5304,
+      "step": 1469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290593716022218e-05,
+      "loss": 3.763,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289395113922714e-05,
+      "loss": 3.7981,
+      "step": 1471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928819553740878e-05,
+      "loss": 3.4349,
+      "step": 1472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928699498660624e-05,
+      "loss": 3.0044,
+      "step": 1473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928579346164103e-05,
+      "loss": 1.7501,
+      "step": 1474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928459096263918e-05,
+      "loss": 3.8038,
+      "step": 1475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283387489726827e-05,
+      "loss": 4.6918,
+      "step": 1476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282183043030216e-05,
+      "loss": 2.3636,
+      "step": 1477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928097762267568e-05,
+      "loss": 1.5182,
+      "step": 1478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927977122878967e-05,
+      "loss": 3.1023,
+      "step": 1479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278563861498726e-05,
+      "loss": 3.8535,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277355520929494e-05,
+      "loss": 3.213,
+      "step": 1481
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276146207208727e-05,
+      "loss": 3.2875,
+      "step": 1482
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274935920463273e-05,
+      "loss": 3.7036,
+      "step": 1483
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273724660820086e-05,
+      "loss": 1.7225,
+      "step": 1484
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927251242840623e-05,
+      "loss": 2.1004,
+      "step": 1485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927129922334885e-05,
+      "loss": 1.8918,
+      "step": 1486
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927008504577521e-05,
+      "loss": 2.2546,
+      "step": 1487
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268869895812673e-05,
+      "loss": 2.6128,
+      "step": 1488
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267653773588702e-05,
+      "loss": 2.5265,
+      "step": 1489
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266436679230866e-05,
+      "loss": 2.604,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926521861286683e-05,
+      "loss": 2.9903,
+      "step": 1491
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263999574624357e-05,
+      "loss": 3.4401,
+      "step": 1492
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262779564631326e-05,
+      "loss": 2.8849,
+      "step": 1493
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926155858301571e-05,
+      "loss": 2.5351,
+      "step": 1494
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926033662990558e-05,
+      "loss": 3.2881,
+      "step": 1495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925911370542912e-05,
+      "loss": 3.5306,
+      "step": 1496
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257889809714604e-05,
+      "loss": 3.0358,
+      "step": 1497
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256664942890412e-05,
+      "loss": 2.0631,
+      "step": 1498
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925543910508503e-05,
+      "loss": 4.19,
+      "step": 1499
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254212296427043e-05,
+      "loss": 2.1581,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252984517045136e-05,
+      "loss": 2.8864,
+      "step": 1501
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251755767068095e-05,
+      "loss": 3.3801,
+      "step": 1502
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250526046624816e-05,
+      "loss": 1.6149,
+      "step": 1503
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249295355844286e-05,
+      "loss": 3.0476,
+      "step": 1504
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248063694855603e-05,
+      "loss": 3.0495,
+      "step": 1505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246831063787956e-05,
+      "loss": 3.2365,
+      "step": 1506
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924559746277065e-05,
+      "loss": 2.8157,
+      "step": 1507
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244362891933077e-05,
+      "loss": 2.1489,
+      "step": 1508
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9243127351404743e-05,
+      "loss": 3.2433,
+      "step": 1509
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924189084131525e-05,
+      "loss": 3.726,
+      "step": 1510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92406533617943e-05,
+      "loss": 2.1306,
+      "step": 1511
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239414912971697e-05,
+      "loss": 2.8156,
+      "step": 1512
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923817549497735e-05,
+      "loss": 3.1028,
+      "step": 1513
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236935107941272e-05,
+      "loss": 3.5099,
+      "step": 1514
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923569375199357e-05,
+      "loss": 3.7079,
+      "step": 1515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923445142726446e-05,
+      "loss": 1.6136,
+      "step": 1516
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233208133884253e-05,
+      "loss": 1.7859,
+      "step": 1517
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231963871983367e-05,
+      "loss": 4.1348,
+      "step": 1518
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923071864169231e-05,
+      "loss": 3.5028,
+      "step": 1519
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922947244314172e-05,
+      "loss": 2.7175,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92282252764623e-05,
+      "loss": 2.4301,
+      "step": 1521
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226977141784873e-05,
+      "loss": 3.1383,
+      "step": 1522
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225728039240375e-05,
+      "loss": 1.4532,
+      "step": 1523
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922447796895982e-05,
+      "loss": 4.0255,
+      "step": 1524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922322693107434e-05,
+      "loss": 3.6112,
+      "step": 1525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922197492571516e-05,
+      "loss": 2.8705,
+      "step": 1526
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922072195301361e-05,
+      "loss": 1.9947,
+      "step": 1527
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219468013101123e-05,
+      "loss": 2.6994,
+      "step": 1528
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218213106109234e-05,
+      "loss": 2.3827,
+      "step": 1529
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9216957232169567e-05,
+      "loss": 2.2938,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215700391413867e-05,
+      "loss": 3.0603,
+      "step": 1531
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214442583973965e-05,
+      "loss": 2.2116,
+      "step": 1532
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9213183809981806e-05,
+      "loss": 3.5773,
+      "step": 1533
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211924069569422e-05,
+      "loss": 3.8905,
+      "step": 1534
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210663362868956e-05,
+      "loss": 3.287,
+      "step": 1535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920940169001265e-05,
+      "loss": 2.5956,
+      "step": 1536
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920813905113285e-05,
+      "loss": 3.9766,
+      "step": 1537
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9206875446362005e-05,
+      "loss": 3.2364,
+      "step": 1538
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920561087583265e-05,
+      "loss": 2.6652,
+      "step": 1539
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204345339677442e-05,
+      "loss": 3.6079,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203078838029125e-05,
+      "loss": 2.4055,
+      "step": 1541
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201811371020553e-05,
+      "loss": 3.3629,
+      "step": 1542
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920054293878467e-05,
+      "loss": 3.2815,
+      "step": 1543
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919927354145454e-05,
+      "loss": 1.669,
+      "step": 1544
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9198003179163308e-05,
+      "loss": 3.7119,
+      "step": 1545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919673185204423e-05,
+      "loss": 2.3191,
+      "step": 1546
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9195459560230665e-05,
+      "loss": 3.4411,
+      "step": 1547
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919418630385607e-05,
+      "loss": 2.2787,
+      "step": 1548
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9192912083054003e-05,
+      "loss": 3.1964,
+      "step": 1549
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9191636897958123e-05,
+      "loss": 3.7047,
+      "step": 1550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190360748702192e-05,
+      "loss": 2.3148,
+      "step": 1551
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189083635420077e-05,
+      "loss": 2.2762,
+      "step": 1552
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918780555824573e-05,
+      "loss": 3.1861,
+      "step": 1553
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9186526517313227e-05,
+      "loss": 1.5788,
+      "step": 1554
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9185246512756727e-05,
+      "loss": 2.5753,
+      "step": 1555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183965544710495e-05,
+      "loss": 1.9341,
+      "step": 1556
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182683613308905e-05,
+      "loss": 3.5029,
+      "step": 1557
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918140071868642e-05,
+      "loss": 3.256,
+      "step": 1558
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180116860977613e-05,
+      "loss": 3.5985,
+      "step": 1559
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178832040317153e-05,
+      "loss": 2.4787,
+      "step": 1560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177546256839814e-05,
+      "loss": 2.4462,
+      "step": 1561
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9176259510680463e-05,
+      "loss": 4.0957,
+      "step": 1562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174971801974084e-05,
+      "loss": 2.9104,
+      "step": 1563
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173683130855737e-05,
+      "loss": 3.3947,
+      "step": 1564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917239349746061e-05,
+      "loss": 2.7584,
+      "step": 1565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171102901923978e-05,
+      "loss": 1.5315,
+      "step": 1566
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169811344381215e-05,
+      "loss": 3.7364,
+      "step": 1567
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168518824967797e-05,
+      "loss": 3.3508,
+      "step": 1568
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916722534381931e-05,
+      "loss": 2.5687,
+      "step": 1569
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916593090107143e-05,
+      "loss": 3.5681,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164635496859937e-05,
+      "loss": 2.7093,
+      "step": 1571
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916333913132072e-05,
+      "loss": 3.831,
+      "step": 1572
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9162041804589752e-05,
+      "loss": 3.0645,
+      "step": 1573
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916074351680312e-05,
+      "loss": 1.8889,
+      "step": 1574
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159444268097012e-05,
+      "loss": 2.7129,
+      "step": 1575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915814405860771e-05,
+      "loss": 2.9618,
+      "step": 1576
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91568428884716e-05,
+      "loss": 3.6628,
+      "step": 1577
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155540757825168e-05,
+      "loss": 3.0971,
+      "step": 1578
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154237666805005e-05,
+      "loss": 2.9188,
+      "step": 1579
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91529336155478e-05,
+      "loss": 2.3277,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915162860419033e-05,
+      "loss": 3.4236,
+      "step": 1581
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9150322632869498e-05,
+      "loss": 2.7531,
+      "step": 1582
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914901570172229e-05,
+      "loss": 3.3102,
+      "step": 1583
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147707810885798e-05,
+      "loss": 3.0898,
+      "step": 1584
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146398960497213e-05,
+      "loss": 3.4707,
+      "step": 1585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145089150693823e-05,
+      "loss": 2.8496,
+      "step": 1586
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143778381613026e-05,
+      "loss": 1.8994,
+      "step": 1587
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142466653392317e-05,
+      "loss": 2.9074,
+      "step": 1588
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141153966169287e-05,
+      "loss": 2.8138,
+      "step": 1589
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913984032008163e-05,
+      "loss": 2.7389,
+      "step": 1590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913852571526714e-05,
+      "loss": 2.6462,
+      "step": 1591
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913721015186372e-05,
+      "loss": 3.5136,
+      "step": 1592
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9135893630009362e-05,
+      "loss": 2.4912,
+      "step": 1593
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9134576149842164e-05,
+      "loss": 3.3714,
+      "step": 1594
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133257711500318e-05,
+      "loss": 3.8549,
+      "step": 1595
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913193831512213e-05,
+      "loss": 3.6759,
+      "step": 1596
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130617960845996e-05,
+      "loss": 3.1557,
+      "step": 1597
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912929664881041e-05,
+      "loss": 2.6425,
+      "step": 1598
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127974379153983e-05,
+      "loss": 2.9589,
+      "step": 1599
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126651152015404e-05,
+      "loss": 3.4849,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9125326967533476e-05,
+      "loss": 2.4223,
+      "step": 1601
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124001825847103e-05,
+      "loss": 3.4542,
+      "step": 1602
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122675727095285e-05,
+      "loss": 2.231,
+      "step": 1603
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912134867141712e-05,
+      "loss": 3.1357,
+      "step": 1604
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120020658951814e-05,
+      "loss": 3.7772,
+      "step": 1605
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118691689838668e-05,
+      "loss": 3.4823,
+      "step": 1606
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117361764217087e-05,
+      "loss": 2.5404,
+      "step": 1607
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911603088222657e-05,
+      "loss": 1.8382,
+      "step": 1608
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114699044006725e-05,
+      "loss": 3.247,
+      "step": 1609
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911336624969725e-05,
+      "loss": 3.8336,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911203249943795e-05,
+      "loss": 3.9986,
+      "step": 1611
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9110697793368733e-05,
+      "loss": 2.4225,
+      "step": 1612
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109362131629604e-05,
+      "loss": 3.4594,
+      "step": 1613
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910802551436066e-05,
+      "loss": 2.1827,
+      "step": 1614
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910668794170212e-05,
+      "loss": 2.0438,
+      "step": 1615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105349413794272e-05,
+      "loss": 3.3954,
+      "step": 1616
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104009930777533e-05,
+      "loss": 3.3576,
+      "step": 1617
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102669492792406e-05,
+      "loss": 3.6934,
+      "step": 1618
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101328099979496e-05,
+      "loss": 3.1396,
+      "step": 1619
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099985752479505e-05,
+      "loss": 1.6888,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098642450433248e-05,
+      "loss": 2.7677,
+      "step": 1621
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9097298193981623e-05,
+      "loss": 4.1135,
+      "step": 1622
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909595298326564e-05,
+      "loss": 2.9965,
+      "step": 1623
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094606818426403e-05,
+      "loss": 2.4318,
+      "step": 1624
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093259699605125e-05,
+      "loss": 2.8123,
+      "step": 1625
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9091911626943104e-05,
+      "loss": 3.3254,
+      "step": 1626
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9090562600581748e-05,
+      "loss": 2.5094,
+      "step": 1627
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908921262066257e-05,
+      "loss": 3.7,
+      "step": 1628
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908786168732717e-05,
+      "loss": 3.5378,
+      "step": 1629
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908650980071726e-05,
+      "loss": 1.6163,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908515696097464e-05,
+      "loss": 2.4568,
+      "step": 1631
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9083803168241225e-05,
+      "loss": 3.397,
+      "step": 1632
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908244842265901e-05,
+      "loss": 3.7423,
+      "step": 1633
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081092724370112e-05,
+      "loss": 2.4791,
+      "step": 1634
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9079736073516735e-05,
+      "loss": 2.8611,
+      "step": 1635
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078378470241183e-05,
+      "loss": 2.8555,
+      "step": 1636
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077019914685864e-05,
+      "loss": 3.2678,
+      "step": 1637
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075660406993285e-05,
+      "loss": 2.9599,
+      "step": 1638
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907429994730605e-05,
+      "loss": 2.5563,
+      "step": 1639
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072938535766864e-05,
+      "loss": 2.0784,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071576172518534e-05,
+      "loss": 2.9013,
+      "step": 1641
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907021285770397e-05,
+      "loss": 3.2673,
+      "step": 1642
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906884859146617e-05,
+      "loss": 3.2874,
+      "step": 1643
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067483373948245e-05,
+      "loss": 3.7238,
+      "step": 1644
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066117205293393e-05,
+      "loss": 2.9956,
+      "step": 1645
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906475008564493e-05,
+      "loss": 2.6699,
+      "step": 1646
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906338201514625e-05,
+      "loss": 2.9217,
+      "step": 1647
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906201299394086e-05,
+      "loss": 1.9319,
+      "step": 1648
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9060643022172364e-05,
+      "loss": 2.6944,
+      "step": 1649
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905927209998447e-05,
+      "loss": 3.9137,
+      "step": 1650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057900227520977e-05,
+      "loss": 2.2517,
+      "step": 1651
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9056527404925788e-05,
+      "loss": 3.8194,
+      "step": 1652
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055153632342907e-05,
+      "loss": 2.8435,
+      "step": 1653
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905377890991644e-05,
+      "loss": 2.6472,
+      "step": 1654
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905240323779058e-05,
+      "loss": 3.3741,
+      "step": 1655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9051026616109637e-05,
+      "loss": 3.3457,
+      "step": 1656
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049649045018008e-05,
+      "loss": 3.6064,
+      "step": 1657
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9048270524660197e-05,
+      "loss": 2.914,
+      "step": 1658
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90468910551808e-05,
+      "loss": 2.3293,
+      "step": 1659
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904551063672452e-05,
+      "loss": 1.7979,
+      "step": 1660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044129269436155e-05,
+      "loss": 2.5333,
+      "step": 1661
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042746953460607e-05,
+      "loss": 3.7278,
+      "step": 1662
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904136368894287e-05,
+      "loss": 2.4601,
+      "step": 1663
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039979476028044e-05,
+      "loss": 1.9426,
+      "step": 1664
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038594314861328e-05,
+      "loss": 2.4921,
+      "step": 1665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903720820558802e-05,
+      "loss": 1.959,
+      "step": 1666
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035821148353507e-05,
+      "loss": 3.0538,
+      "step": 1667
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90344331433033e-05,
+      "loss": 3.1739,
+      "step": 1668
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903304419058298e-05,
+      "loss": 2.8314,
+      "step": 1669
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031654290338256e-05,
+      "loss": 3.0542,
+      "step": 1670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9030263442714908e-05,
+      "loss": 3.2548,
+      "step": 1671
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9028871647858836e-05,
+      "loss": 1.9078,
+      "step": 1672
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027478905916035e-05,
+      "loss": 4.0244,
+      "step": 1673
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9026085217032592e-05,
+      "loss": 3.2161,
+      "step": 1674
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90246905813547e-05,
+      "loss": 3.1822,
+      "step": 1675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023294999028654e-05,
+      "loss": 3.7484,
+      "step": 1676
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902189847020084e-05,
+      "loss": 2.9264,
+      "step": 1677
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020500995017747e-05,
+      "loss": 2.5169,
+      "step": 1678
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019102573625966e-05,
+      "loss": 3.0524,
+      "step": 1679
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017703206172187e-05,
+      "loss": 1.7298,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901630289280319e-05,
+      "loss": 2.6579,
+      "step": 1681
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901490163366587e-05,
+      "loss": 2.1324,
+      "step": 1682
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013499428907202e-05,
+      "loss": 2.6113,
+      "step": 1683
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012096278674283e-05,
+      "loss": 3.1188,
+      "step": 1684
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010692183114285e-05,
+      "loss": 2.7638,
+      "step": 1685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90092871423745e-05,
+      "loss": 2.9885,
+      "step": 1686
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007881156602307e-05,
+      "loss": 2.2596,
+      "step": 1687
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900647422594519e-05,
+      "loss": 2.8266,
+      "step": 1688
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9005066350550724e-05,
+      "loss": 2.1911,
+      "step": 1689
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900365753056659e-05,
+      "loss": 3.7387,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900224776614057e-05,
+      "loss": 1.8979,
+      "step": 1691
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900083705742054e-05,
+      "loss": 2.7454,
+      "step": 1692
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899942540455448e-05,
+      "loss": 3.4865,
+      "step": 1693
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899801280769046e-05,
+      "loss": 2.9827,
+      "step": 1694
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996599266976658e-05,
+      "loss": 1.648,
+      "step": 1695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8995184782561343e-05,
+      "loss": 3.5892,
+      "step": 1696
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993769354592897e-05,
+      "loss": 3.2651,
+      "step": 1697
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992352983219785e-05,
+      "loss": 3.8522,
+      "step": 1698
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8990935668590583e-05,
+      "loss": 3.4263,
+      "step": 1699
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989517410853956e-05,
+      "loss": 3.5332,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988098210158673e-05,
+      "loss": 2.5355,
+      "step": 1701
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8986678066653604e-05,
+      "loss": 3.7723,
+      "step": 1702
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985256980487713e-05,
+      "loss": 2.7523,
+      "step": 1703
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983834951810068e-05,
+      "loss": 2.6537,
+      "step": 1704
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898241198076983e-05,
+      "loss": 3.2671,
+      "step": 1705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980988067516265e-05,
+      "loss": 2.7306,
+      "step": 1706
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8979563212198737e-05,
+      "loss": 2.2349,
+      "step": 1707
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89781374149667e-05,
+      "loss": 2.1271,
+      "step": 1708
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8976710675969715e-05,
+      "loss": 3.5301,
+      "step": 1709
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975282995357448e-05,
+      "loss": 3.0749,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973854373279646e-05,
+      "loss": 3.468,
+      "step": 1711
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897242480988617e-05,
+      "loss": 3.4987,
+      "step": 1712
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970994305326973e-05,
+      "loss": 2.6123,
+      "step": 1713
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896956285975211e-05,
+      "loss": 2.2308,
+      "step": 1714
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8968130473311732e-05,
+      "loss": 3.3003,
+      "step": 1715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8966697146156092e-05,
+      "loss": 3.5497,
+      "step": 1716
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965262878435534e-05,
+      "loss": 1.9606,
+      "step": 1717
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8963827670300512e-05,
+      "loss": 4.114,
+      "step": 1718
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962391521901565e-05,
+      "loss": 2.4815,
+      "step": 1719
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896095443338935e-05,
+      "loss": 2.5034,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959516404914597e-05,
+      "loss": 2.2799,
+      "step": 1721
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895807743662816e-05,
+      "loss": 4.333,
+      "step": 1722
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956637528680973e-05,
+      "loss": 2.112,
+      "step": 1723
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895519668122408e-05,
+      "loss": 2.237,
+      "step": 1724
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8953754894408617e-05,
+      "loss": 2.9375,
+      "step": 1725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895231216838582e-05,
+      "loss": 3.6295,
+      "step": 1726
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950868503307026e-05,
+      "loss": 3.5757,
+      "step": 1727
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894942389932367e-05,
+      "loss": 2.6412,
+      "step": 1728
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8947978356587283e-05,
+      "loss": 3.1481,
+      "step": 1729
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946531875249496e-05,
+      "loss": 1.4969,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945084455462032e-05,
+      "loss": 1.7614,
+      "step": 1731
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943636097376728e-05,
+      "loss": 3.6874,
+      "step": 1732
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8942186801145505e-05,
+      "loss": 2.7118,
+      "step": 1733
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940736566920386e-05,
+      "loss": 1.8759,
+      "step": 1734
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89392853948535e-05,
+      "loss": 2.3361,
+      "step": 1735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937833285097067e-05,
+      "loss": 2.5408,
+      "step": 1736
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8936380237803398e-05,
+      "loss": 3.1286,
+      "step": 1737
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8934926253124922e-05,
+      "loss": 3.2417,
+      "step": 1738
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893347133121415e-05,
+      "loss": 2.3961,
+      "step": 1739
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8932015472223692e-05,
+      "loss": 2.9164,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893055867630627e-05,
+      "loss": 2.3545,
+      "step": 1741
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892910094361469e-05,
+      "loss": 3.1221,
+      "step": 1742
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927642274301862e-05,
+      "loss": 3.2812,
+      "step": 1743
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926182668520794e-05,
+      "loss": 3.6227,
+      "step": 1744
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892472212642459e-05,
+      "loss": 3.2721,
+      "step": 1745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923260648166457e-05,
+      "loss": 3.6103,
+      "step": 1746
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89217982338997e-05,
+      "loss": 2.9313,
+      "step": 1747
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892033488377771e-05,
+      "loss": 2.8008,
+      "step": 1748
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8918870597953996e-05,
+      "loss": 3.6821,
+      "step": 1749
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8917405376582144e-05,
+      "loss": 2.383,
+      "step": 1750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891593921981586e-05,
+      "loss": 2.872,
+      "step": 1751
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891447212780893e-05,
+      "loss": 2.6987,
+      "step": 1752
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913004100715247e-05,
+      "loss": 2.9015,
+      "step": 1753
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89115351386888e-05,
+      "loss": 2.7344,
+      "step": 1754
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891006524188368e-05,
+      "loss": 3.159,
+      "step": 1755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908594410454068e-05,
+      "loss": 2.6954,
+      "step": 1756
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8907122644554245e-05,
+      "loss": 2.3207,
+      "step": 1757
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905649944338596e-05,
+      "loss": 2.524,
+      "step": 1758
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8904176309961606e-05,
+      "loss": 2.3576,
+      "step": 1759
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902701741577844e-05,
+      "loss": 1.9231,
+      "step": 1760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8901226239341986e-05,
+      "loss": 3.3714,
+      "step": 1761
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8899749803408808e-05,
+      "loss": 3.5592,
+      "step": 1762
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889827243393318e-05,
+      "loss": 2.6573,
+      "step": 1763
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896794131070073e-05,
+      "loss": 2.8328,
+      "step": 1764
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889531489497455e-05,
+      "loss": 2.6337,
+      "step": 1765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8893834725801782e-05,
+      "loss": 3.5298,
+      "step": 1766
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892353623707033e-05,
+      "loss": 2.6356,
+      "step": 1767
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890871588845653e-05,
+      "loss": 2.9572,
+      "step": 1768
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889388621373107e-05,
+      "loss": 2.3944,
+      "step": 1769
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8887904721444955e-05,
+      "loss": 2.344,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886419889216846e-05,
+      "loss": 1.5985,
+      "step": 1771
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884934124844534e-05,
+      "loss": 1.1906,
+      "step": 1772
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888344742848387e-05,
+      "loss": 2.4629,
+      "step": 1773
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88819598002908e-05,
+      "loss": 2.9576,
+      "step": 1774
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880471240421365e-05,
+      "loss": 3.0377,
+      "step": 1775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878981749031718e-05,
+      "loss": 2.4923,
+      "step": 1776
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877491326278093e-05,
+      "loss": 3.1126,
+      "step": 1777
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875999972316826e-05,
+      "loss": 2.9987,
+      "step": 1778
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887450768730436e-05,
+      "loss": 2.8644,
+      "step": 1779
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873014471397225e-05,
+      "loss": 2.456,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8871520324752054e-05,
+      "loss": 2.1099,
+      "step": 1781
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887002524752557e-05,
+      "loss": 2.093,
+      "step": 1782
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868529239874608e-05,
+      "loss": 2.8003,
+      "step": 1783
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886703230195609e-05,
+      "loss": 3.9029,
+      "step": 1784
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8865534433927034e-05,
+      "loss": 2.39,
+      "step": 1785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864035635944564e-05,
+      "loss": 3.1016,
+      "step": 1786
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862535908165897e-05,
+      "loss": 3.2267,
+      "step": 1787
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8861035250748343e-05,
+      "loss": 2.8283,
+      "step": 1788
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859533663849318e-05,
+      "loss": 1.681,
+      "step": 1789
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858031147626326e-05,
+      "loss": 2.275,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8856527702236983e-05,
+      "loss": 4.0711,
+      "step": 1791
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855023327838984e-05,
+      "loss": 3.4388,
+      "step": 1792
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853518024590136e-05,
+      "loss": 1.9073,
+      "step": 1793
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885201179264834e-05,
+      "loss": 1.8399,
+      "step": 1794
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885050463217159e-05,
+      "loss": 2.849,
+      "step": 1795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8848996543317982e-05,
+      "loss": 1.4415,
+      "step": 1796
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847487526245703e-05,
+      "loss": 2.4538,
+      "step": 1797
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845977581113048e-05,
+      "loss": 1.7769,
+      "step": 1798
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8844466708078398e-05,
+      "loss": 3.0858,
+      "step": 1799
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8842954907300236e-05,
+      "loss": 3.7206,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884144217893715e-05,
+      "loss": 3.1099,
+      "step": 1801
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883992852314781e-05,
+      "loss": 3.2163,
+      "step": 1802
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838413940091e-05,
+      "loss": 2.3265,
+      "step": 1803
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836898429925586e-05,
+      "loss": 3.0455,
+      "step": 1804
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883538199281054e-05,
+      "loss": 2.7937,
+      "step": 1805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883386462890493e-05,
+      "loss": 2.9215,
+      "step": 1806
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883234633836792e-05,
+      "loss": 3.8912,
+      "step": 1807
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883082712135877e-05,
+      "loss": 2.5025,
+      "step": 1808
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8829306978036837e-05,
+      "loss": 2.8465,
+      "step": 1809
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8827785908561585e-05,
+      "loss": 1.5213,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882626391309256e-05,
+      "loss": 1.9027,
+      "step": 1811
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824740991789417e-05,
+      "loss": 1.296,
+      "step": 1812
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88232171448119e-05,
+      "loss": 2.1325,
+      "step": 1813
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8821692372319852e-05,
+      "loss": 1.3526,
+      "step": 1814
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820166674473217e-05,
+      "loss": 1.3229,
+      "step": 1815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8818640051432036e-05,
+      "loss": 3.0855,
+      "step": 1816
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881711250335644e-05,
+      "loss": 4.6315,
+      "step": 1817
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815584030406663e-05,
+      "loss": 3.176,
+      "step": 1818
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814054632743038e-05,
+      "loss": 2.5201,
+      "step": 1819
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881252431052599e-05,
+      "loss": 2.6783,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881099306391604e-05,
+      "loss": 3.8805,
+      "step": 1821
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880946089307381e-05,
+      "loss": 2.4956,
+      "step": 1822
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880792779816002e-05,
+      "loss": 3.4165,
+      "step": 1823
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806393779335483e-05,
+      "loss": 3.4475,
+      "step": 1824
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880485883676111e-05,
+      "loss": 2.5671,
+      "step": 1825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8803322970597907e-05,
+      "loss": 2.7302,
+      "step": 1826
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8801786181006985e-05,
+      "loss": 2.7153,
+      "step": 1827
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800248468149545e-05,
+      "loss": 4.1207,
+      "step": 1828
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879870983218688e-05,
+      "loss": 2.6612,
+      "step": 1829
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879717027328039e-05,
+      "loss": 1.8234,
+      "step": 1830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879562979159157e-05,
+      "loss": 3.0106,
+      "step": 1831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8794088387282e-05,
+      "loss": 2.6041,
+      "step": 1832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879254606051338e-05,
+      "loss": 2.0753,
+      "step": 1833
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791002811447483e-05,
+      "loss": 2.6382,
+      "step": 1834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8789458640246193e-05,
+      "loss": 2.5075,
+      "step": 1835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8787913547071485e-05,
+      "loss": 1.9273,
+      "step": 1836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786367532085426e-05,
+      "loss": 1.5005,
+      "step": 1837
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8784820595450198e-05,
+      "loss": 0.8559,
+      "step": 1838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878327273732806e-05,
+      "loss": 3.23,
+      "step": 1839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8781723957881374e-05,
+      "loss": 1.6892,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8780174257272603e-05,
+      "loss": 3.4822,
+      "step": 1841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778623635664305e-05,
+      "loss": 2.4373,
+      "step": 1842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8777072093219127e-05,
+      "loss": 3.078,
+      "step": 1843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775519630099822e-05,
+      "loss": 2.5629,
+      "step": 1844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8773966246469238e-05,
+      "loss": 3.553,
+      "step": 1845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772411942490315e-05,
+      "loss": 2.6662,
+      "step": 1846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770856718326094e-05,
+      "loss": 1.8203,
+      "step": 1847
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876930057413971e-05,
+      "loss": 2.9681,
+      "step": 1848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8767743510094395e-05,
+      "loss": 3.611,
+      "step": 1849
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876618552635348e-05,
+      "loss": 1.8227,
+      "step": 1850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8764626623080385e-05,
+      "loss": 3.6043,
+      "step": 1851
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763066800438638e-05,
+      "loss": 2.2575,
+      "step": 1852
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876150605859185e-05,
+      "loss": 2.8149,
+      "step": 1853
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8759944397703748e-05,
+      "loss": 2.3303,
+      "step": 1854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758381817938126e-05,
+      "loss": 2.6743,
+      "step": 1855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875681831945891e-05,
+      "loss": 1.7432,
+      "step": 1856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755253902430082e-05,
+      "loss": 1.0671,
+      "step": 1857
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875368856701576e-05,
+      "loss": 1.7466,
+      "step": 1858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752122313380137e-05,
+      "loss": 3.7392,
+      "step": 1859
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87505551416875e-05,
+      "loss": 1.5142,
+      "step": 1860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874898705210224e-05,
+      "loss": 1.3185,
+      "step": 1861
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747418044788848e-05,
+      "loss": 3.1069,
+      "step": 1862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8745848119911897e-05,
+      "loss": 1.9044,
+      "step": 1863
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874427727763607e-05,
+      "loss": 2.9553,
+      "step": 1864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874270551812614e-05,
+      "loss": 1.6221,
+      "step": 1865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874113284154698e-05,
+      "loss": 2.3153,
+      "step": 1866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739559248063553e-05,
+      "loss": 1.9743,
+      "step": 1867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873798473784092e-05,
+      "loss": 2.8787,
+      "step": 1868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736409311044244e-05,
+      "loss": 3.2924,
+      "step": 1869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734832967838775e-05,
+      "loss": 2.2703,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8733255708389872e-05,
+      "loss": 2.3221,
+      "step": 1871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731677532862975e-05,
+      "loss": 3.7376,
+      "step": 1872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873009844142363e-05,
+      "loss": 2.578,
+      "step": 1873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728518434237476e-05,
+      "loss": 3.3061,
+      "step": 1874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8726937511470247e-05,
+      "loss": 1.684,
+      "step": 1875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872535567328778e-05,
+      "loss": 2.7527,
+      "step": 1876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723772919855997e-05,
+      "loss": 1.4237,
+      "step": 1877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872218925134092e-05,
+      "loss": 2.3349,
+      "step": 1878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720604667908673e-05,
+      "loss": 2.7697,
+      "step": 1879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871901916972547e-05,
+      "loss": 3.0854,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717432756957626e-05,
+      "loss": 3.0948,
+      "step": 1881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871584542977154e-05,
+      "loss": 1.6648,
+      "step": 1882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871425718833372e-05,
+      "loss": 3.3696,
+      "step": 1883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8712668032810767e-05,
+      "loss": 2.7585,
+      "step": 1884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711077963369377e-05,
+      "loss": 3.6921,
+      "step": 1885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870948698017633e-05,
+      "loss": 3.5588,
+      "step": 1886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870789508339853e-05,
+      "loss": 3.7236,
+      "step": 1887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870630227320294e-05,
+      "loss": 3.6035,
+      "step": 1888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704708549756657e-05,
+      "loss": 1.276,
+      "step": 1889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8703113913226847e-05,
+      "loss": 3.429,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701518363780776e-05,
+      "loss": 3.2947,
+      "step": 1891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8699921901585814e-05,
+      "loss": 3.07,
+      "step": 1892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698324526809423e-05,
+      "loss": 2.3589,
+      "step": 1893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869672623961916e-05,
+      "loss": 3.5586,
+      "step": 1894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695127040182678e-05,
+      "loss": 2.9745,
+      "step": 1895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8693526928667724e-05,
+      "loss": 3.8458,
+      "step": 1896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869192590524214e-05,
+      "loss": 3.6516,
+      "step": 1897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690323970073874e-05,
+      "loss": 3.8726,
+      "step": 1898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8688721123330952e-05,
+      "loss": 3.0112,
+      "step": 1899
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687117365181514e-05,
+      "loss": 3.7521,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8685512695793777e-05,
+      "loss": 1.9349,
+      "step": 1901
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8683907115336073e-05,
+      "loss": 2.856,
+      "step": 1902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682300623976815e-05,
+      "loss": 1.8186,
+      "step": 1903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868069322188452e-05,
+      "loss": 3.7768,
+      "step": 1904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867908490922779e-05,
+      "loss": 2.6325,
+      "step": 1905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8677475686175336e-05,
+      "loss": 2.4993,
+      "step": 1906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675865552895955e-05,
+      "loss": 2.2425,
+      "step": 1907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8674254509558544e-05,
+      "loss": 1.3933,
+      "step": 1908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8672642556332093e-05,
+      "loss": 2.5653,
+      "step": 1909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867102969338569e-05,
+      "loss": 3.464,
+      "step": 1910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669415920888517e-05,
+      "loss": 2.8981,
+      "step": 1911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8667801239009845e-05,
+      "loss": 2.279,
+      "step": 1912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866618564791906e-05,
+      "loss": 3.9647,
+      "step": 1913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8664569147785615e-05,
+      "loss": 2.3101,
+      "step": 1914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662951738779077e-05,
+      "loss": 3.0595,
+      "step": 1915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866133342106911e-05,
+      "loss": 3.035,
+      "step": 1916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865971419482547e-05,
+      "loss": 2.2881,
+      "step": 1917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8658094060218e-05,
+      "loss": 4.031,
+      "step": 1918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656473017416644e-05,
+      "loss": 2.1139,
+      "step": 1919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865485106659145e-05,
+      "loss": 2.9826,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865322820791254e-05,
+      "loss": 2.2913,
+      "step": 1921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8651604441550154e-05,
+      "loss": 4.2653,
+      "step": 1922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864997976767462e-05,
+      "loss": 3.0337,
+      "step": 1923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864835418645635e-05,
+      "loss": 2.4432,
+      "step": 1924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8646727698065865e-05,
+      "loss": 3.4196,
+      "step": 1925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8645100302673775e-05,
+      "loss": 2.886,
+      "step": 1926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643472000450786e-05,
+      "loss": 2.6876,
+      "step": 1927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86418427915677e-05,
+      "loss": 2.3358,
+      "step": 1928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640212676195415e-05,
+      "loss": 1.9351,
+      "step": 1929
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863858165450492e-05,
+      "loss": 3.5705,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86369497266673e-05,
+      "loss": 3.2536,
+      "step": 1931
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863531689285374e-05,
+      "loss": 3.0048,
+      "step": 1932
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863368315323552e-05,
+      "loss": 2.5203,
+      "step": 1933
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8632048507984e-05,
+      "loss": 1.86,
+      "step": 1934
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863041295727066e-05,
+      "loss": 3.1827,
+      "step": 1935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8628776501267052e-05,
+      "loss": 2.6251,
+      "step": 1936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862713914014484e-05,
+      "loss": 3.3543,
+      "step": 1937
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862550087407577e-05,
+      "loss": 3.035,
+      "step": 1938
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862386170323169e-05,
+      "loss": 2.035,
+      "step": 1939
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862222162778454e-05,
+      "loss": 3.1713,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8620580647906365e-05,
+      "loss": 2.8238,
+      "step": 1941
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8618938763769283e-05,
+      "loss": 2.7095,
+      "step": 1942
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861729597554553e-05,
+      "loss": 2.0638,
+      "step": 1943
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861565228340742e-05,
+      "loss": 3.9203,
+      "step": 1944
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614007687527374e-05,
+      "loss": 2.9551,
+      "step": 1945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86123621880779e-05,
+      "loss": 2.4667,
+      "step": 1946
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8610715785231603e-05,
+      "loss": 1.8472,
+      "step": 1947
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8609068479161182e-05,
+      "loss": 2.3004,
+      "step": 1948
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860742027003944e-05,
+      "loss": 3.831,
+      "step": 1949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605771158039253e-05,
+      "loss": 1.7136,
+      "step": 1950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604121143333615e-05,
+      "loss": 2.7969,
+      "step": 1951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602470226095602e-05,
+      "loss": 3.3804,
+      "step": 1952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8600818406498388e-05,
+      "loss": 3.1501,
+      "step": 1953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599165684715238e-05,
+      "loss": 3.3228,
+      "step": 1954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8597512060919523e-05,
+      "loss": 2.6053,
+      "step": 1955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8595857535284692e-05,
+      "loss": 1.6986,
+      "step": 1956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85942021079843e-05,
+      "loss": 3.6995,
+      "step": 1957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592545779191993e-05,
+      "loss": 3.0144,
+      "step": 1958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590888549081514e-05,
+      "loss": 2.1068,
+      "step": 1959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85892304178267e-05,
+      "loss": 1.1415,
+      "step": 1960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858757138560148e-05,
+      "loss": 2.8391,
+      "step": 1961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858591145257987e-05,
+      "loss": 3.0905,
+      "step": 1962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584250618936e-05,
+      "loss": 3.838,
+      "step": 1963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8582588884844086e-05,
+      "loss": 3.6343,
+      "step": 1964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580926250478425e-05,
+      "loss": 2.2294,
+      "step": 1965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857926271601343e-05,
+      "loss": 3.2932,
+      "step": 1966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857759828162359e-05,
+      "loss": 3.3163,
+      "step": 1967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575932947483503e-05,
+      "loss": 1.749,
+      "step": 1968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857426671376785e-05,
+      "loss": 2.891,
+      "step": 1969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572599580651415e-05,
+      "loss": 3.4546,
+      "step": 1970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857093154830907e-05,
+      "loss": 1.4464,
+      "step": 1971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569262616915784e-05,
+      "loss": 2.484,
+      "step": 1972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856759278664662e-05,
+      "loss": 2.3109,
+      "step": 1973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8565922057676738e-05,
+      "loss": 2.1041,
+      "step": 1974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564250430181387e-05,
+      "loss": 3.5076,
+      "step": 1975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8562577904335913e-05,
+      "loss": 2.1174,
+      "step": 1976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8560904480315757e-05,
+      "loss": 2.8325,
+      "step": 1977
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8559230158296454e-05,
+      "loss": 2.5092,
+      "step": 1978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855755493845363e-05,
+      "loss": 3.2625,
+      "step": 1979
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8555878820963014e-05,
+      "loss": 3.9796,
+      "step": 1980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8554201806000417e-05,
+      "loss": 3.1883,
+      "step": 1981
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855252389374175e-05,
+      "loss": 2.663,
+      "step": 1982
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8550845084363023e-05,
+      "loss": 3.0096,
+      "step": 1983
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8549165378040328e-05,
+      "loss": 3.1638,
+      "step": 1984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547484774949865e-05,
+      "loss": 3.3577,
+      "step": 1985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545803275267923e-05,
+      "loss": 2.7147,
+      "step": 1986
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544120879170874e-05,
+      "loss": 3.2912,
+      "step": 1987
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542437586835202e-05,
+      "loss": 2.8619,
+      "step": 1988
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540753398437473e-05,
+      "loss": 1.9991,
+      "step": 1989
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539068314154355e-05,
+      "loss": 1.6821,
+      "step": 1990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8537382334162596e-05,
+      "loss": 3.5979,
+      "step": 1991
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535695458639056e-05,
+      "loss": 3.5531,
+      "step": 1992
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534007687760682e-05,
+      "loss": 1.978,
+      "step": 1993
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8532319021704502e-05,
+      "loss": 3.4583,
+      "step": 1994
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8530629460647658e-05,
+      "loss": 2.8939,
+      "step": 1995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8528939004767377e-05,
+      "loss": 1.8932,
+      "step": 1996
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8527247654240977e-05,
+      "loss": 3.7513,
+      "step": 1997
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8525555409245877e-05,
+      "loss": 1.7713,
+      "step": 1998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852386226995958e-05,
+      "loss": 2.4142,
+      "step": 1999
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8522168236559693e-05,
+      "loss": 2.9392,
+      "step": 2000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520473309223912e-05,
+      "loss": 2.8712,
+      "step": 2001
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8518777488130024e-05,
+      "loss": 2.0286,
+      "step": 2002
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8517080773455913e-05,
+      "loss": 2.6217,
+      "step": 2003
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851538316537956e-05,
+      "loss": 1.329,
+      "step": 2004
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8513684664079033e-05,
+      "loss": 3.0273,
+      "step": 2005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8511985269732496e-05,
+      "loss": 3.057,
+      "step": 2006
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8510284982518213e-05,
+      "loss": 3.0305,
+      "step": 2007
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8508583802614534e-05,
+      "loss": 2.4097,
+      "step": 2008
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85068817301999e-05,
+      "loss": 3.1741,
+      "step": 2009
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505178765452853e-05,
+      "loss": 1.9085,
+      "step": 2010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850347490855203e-05,
+      "loss": 1.9621,
+      "step": 2011
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501770159676157e-05,
+      "loss": 3.7082,
+      "step": 2012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500064519004047e-05,
+      "loss": 2.2103,
+      "step": 2013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8498357986714624e-05,
+      "loss": 2.5855,
+      "step": 2014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8496650562986888e-05,
+      "loss": 2.8391,
+      "step": 2015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849494224799994e-05,
+      "loss": 3.1887,
+      "step": 2016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8493233041932977e-05,
+      "loss": 3.1405,
+      "step": 2017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849152294496529e-05,
+      "loss": 3.3259,
+      "step": 2018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8489811957276254e-05,
+      "loss": 2.1641,
+      "step": 2019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8488100079045345e-05,
+      "loss": 3.0571,
+      "step": 2020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486387310452135e-05,
+      "loss": 3.0793,
+      "step": 2021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8484673651676283e-05,
+      "loss": 4.4729,
+      "step": 2022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848295910289754e-05,
+      "loss": 3.1301,
+      "step": 2023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848124366429576e-05,
+      "loss": 2.2275,
+      "step": 2024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847952733605088e-05,
+      "loss": 3.1808,
+      "step": 2025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847781011834294e-05,
+      "loss": 1.9494,
+      "step": 2026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8476092011352063e-05,
+      "loss": 3.0274,
+      "step": 2027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8474373015258472e-05,
+      "loss": 2.0411,
+      "step": 2028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8472653130242485e-05,
+      "loss": 2.7652,
+      "step": 2029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847093235648451e-05,
+      "loss": 2.2793,
+      "step": 2030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8469210694165042e-05,
+      "loss": 1.5559,
+      "step": 2031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846748814346468e-05,
+      "loss": 2.2997,
+      "step": 2032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846576470456411e-05,
+      "loss": 2.2532,
+      "step": 2033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846404037764411e-05,
+      "loss": 4.5855,
+      "step": 2034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8462315162885563e-05,
+      "loss": 3.2321,
+      "step": 2035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846058906046943e-05,
+      "loss": 2.1805,
+      "step": 2036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458862070576767e-05,
+      "loss": 1.4328,
+      "step": 2037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845713419338873e-05,
+      "loss": 3.1779,
+      "step": 2038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8455405429086576e-05,
+      "loss": 2.2966,
+      "step": 2039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8453675777851627e-05,
+      "loss": 2.1721,
+      "step": 2040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8451945239865326e-05,
+      "loss": 3.3164,
+      "step": 2041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84502138153092e-05,
+      "loss": 1.7634,
+      "step": 2042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448481504364856e-05,
+      "loss": 2.0837,
+      "step": 2043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844674830721402e-05,
+      "loss": 2.5744,
+      "step": 2044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8445014224038485e-05,
+      "loss": 2.604,
+      "step": 2045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8443279255020153e-05,
+      "loss": 2.3623,
+      "step": 2046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8441543400341012e-05,
+      "loss": 2.9434,
+      "step": 2047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843980666018315e-05,
+      "loss": 2.8102,
+      "step": 2048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8438069034728738e-05,
+      "loss": 2.7308,
+      "step": 2049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8436330524160048e-05,
+      "loss": 2.8057,
+      "step": 2050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843459112865944e-05,
+      "loss": 2.0539,
+      "step": 2051
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432850848409367e-05,
+      "loss": 2.9524,
+      "step": 2052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843110968359238e-05,
+      "loss": 2.6717,
+      "step": 2053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429367634391116e-05,
+      "loss": 2.4724,
+      "step": 2054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8427624700988308e-05,
+      "loss": 3.8499,
+      "step": 2055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425880883566784e-05,
+      "loss": 3.1435,
+      "step": 2056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842413618230946e-05,
+      "loss": 3.7977,
+      "step": 2057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842239059739935e-05,
+      "loss": 3.5129,
+      "step": 2058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8420644129019555e-05,
+      "loss": 2.5844,
+      "step": 2059
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8418896777353272e-05,
+      "loss": 2.6971,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8417148542583788e-05,
+      "loss": 2.891,
+      "step": 2061
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8415399424894492e-05,
+      "loss": 2.1604,
+      "step": 2062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841364942446885e-05,
+      "loss": 1.9484,
+      "step": 2063
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411898541490433e-05,
+      "loss": 3.2131,
+      "step": 2064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84101467761429e-05,
+      "loss": 3.1506,
+      "step": 2065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8408394128610005e-05,
+      "loss": 3.266,
+      "step": 2066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8406640599075585e-05,
+      "loss": 3.2392,
+      "step": 2067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840488618772359e-05,
+      "loss": 2.526,
+      "step": 2068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403130894738038e-05,
+      "loss": 2.9149,
+      "step": 2069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401374720303054e-05,
+      "loss": 4.2412,
+      "step": 2070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399617664602856e-05,
+      "loss": 3.6199,
+      "step": 2071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8397859727821747e-05,
+      "loss": 3.3933,
+      "step": 2072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839610091014413e-05,
+      "loss": 2.8497,
+      "step": 2073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8394341211754496e-05,
+      "loss": 3.8683,
+      "step": 2074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8392580632837423e-05,
+      "loss": 3.1267,
+      "step": 2075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83908191735776e-05,
+      "loss": 2.7493,
+      "step": 2076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838905683415978e-05,
+      "loss": 3.2585,
+      "step": 2077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387293614768843e-05,
+      "loss": 2.5551,
+      "step": 2078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8385529515589726e-05,
+      "loss": 3.5972,
+      "step": 2079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8383764536807486e-05,
+      "loss": 2.7715,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838199867860725e-05,
+      "loss": 2.8805,
+      "step": 2081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8380231941174258e-05,
+      "loss": 2.6985,
+      "step": 2082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837846432469383e-05,
+      "loss": 2.0088,
+      "step": 2083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376695829351378e-05,
+      "loss": 2.4869,
+      "step": 2084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837492645533241e-05,
+      "loss": 1.5283,
+      "step": 2085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8373156202822532e-05,
+      "loss": 3.0413,
+      "step": 2086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8371385072007423e-05,
+      "loss": 2.0018,
+      "step": 2087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369613063072875e-05,
+      "loss": 2.8626,
+      "step": 2088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836784017620476e-05,
+      "loss": 2.5693,
+      "step": 2089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836606641158905e-05,
+      "loss": 2.4969,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83642917694118e-05,
+      "loss": 3.243,
+      "step": 2091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362516249859164e-05,
+      "loss": 3.9576,
+      "step": 2092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8360739853117383e-05,
+      "loss": 3.0542,
+      "step": 2093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358962579372797e-05,
+      "loss": 2.9345,
+      "step": 2094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835718442881183e-05,
+      "loss": 1.8075,
+      "step": 2095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8355405401621e-05,
+      "loss": 2.5042,
+      "step": 2096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353625497986928e-05,
+      "loss": 2.9544,
+      "step": 2097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835184471809631e-05,
+      "loss": 3.136,
+      "step": 2098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835006306213594e-05,
+      "loss": 1.7784,
+      "step": 2099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8348280530292712e-05,
+      "loss": 2.4933,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8346497122753603e-05,
+      "loss": 1.333,
+      "step": 2101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8344712839705682e-05,
+      "loss": 2.5748,
+      "step": 2102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8342927681336112e-05,
+      "loss": 3.0211,
+      "step": 2103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834114164783215e-05,
+      "loss": 2.2183,
+      "step": 2104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8339354739381138e-05,
+      "loss": 2.3019,
+      "step": 2105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337566956170523e-05,
+      "loss": 2.8687,
+      "step": 2106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8335778298387828e-05,
+      "loss": 2.0584,
+      "step": 2107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8333988766220676e-05,
+      "loss": 2.8589,
+      "step": 2108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8332198359856782e-05,
+      "loss": 3.1642,
+      "step": 2109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833040707948395e-05,
+      "loss": 3.0769,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832861492529008e-05,
+      "loss": 1.9517,
+      "step": 2111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832682189746316e-05,
+      "loss": 3.291,
+      "step": 2112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8325027996191267e-05,
+      "loss": 2.9499,
+      "step": 2113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8323233221662574e-05,
+      "loss": 1.9607,
+      "step": 2114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8321437574065347e-05,
+      "loss": 2.7667,
+      "step": 2115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831964105358794e-05,
+      "loss": 2.741,
+      "step": 2116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83178436604188e-05,
+      "loss": 3.5001,
+      "step": 2117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831604539474646e-05,
+      "loss": 2.7478,
+      "step": 2118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8314246256759556e-05,
+      "loss": 3.4252,
+      "step": 2119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831244624664681e-05,
+      "loss": 2.3162,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310645364597028e-05,
+      "loss": 3.1171,
+      "step": 2121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8308843610799122e-05,
+      "loss": 1.7505,
+      "step": 2122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830704098544208e-05,
+      "loss": 2.2529,
+      "step": 2123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305237488714995e-05,
+      "loss": 2.809,
+      "step": 2124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8303433120807043e-05,
+      "loss": 1.9665,
+      "step": 2125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301627881907494e-05,
+      "loss": 2.5376,
+      "step": 2126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8299821772205706e-05,
+      "loss": 3.1105,
+      "step": 2127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8298014791891138e-05,
+      "loss": 3.2688,
+      "step": 2128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8296206941153333e-05,
+      "loss": 3.4968,
+      "step": 2129
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829439822018192e-05,
+      "loss": 2.2243,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829258862916663e-05,
+      "loss": 2.5942,
+      "step": 2131
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829077816829728e-05,
+      "loss": 2.2864,
+      "step": 2132
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828896683776378e-05,
+      "loss": 3.8472,
+      "step": 2133
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287154637756125e-05,
+      "loss": 2.1201,
+      "step": 2134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8285341568464416e-05,
+      "loss": 3.7174,
+      "step": 2135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283527630078827e-05,
+      "loss": 2.6589,
+      "step": 2136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8281712822789632e-05,
+      "loss": 1.3827,
+      "step": 2137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279897146787204e-05,
+      "loss": 2.4819,
+      "step": 2138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827808060226199e-05,
+      "loss": 3.2442,
+      "step": 2139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827626318940454e-05,
+      "loss": 3.0322,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8274444908405495e-05,
+      "loss": 2.3627,
+      "step": 2141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8272625759455582e-05,
+      "loss": 2.1947,
+      "step": 2142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827080574274562e-05,
+      "loss": 3.5203,
+      "step": 2143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8268984858466524e-05,
+      "loss": 2.5158,
+      "step": 2144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8267163106809288e-05,
+      "loss": 1.9335,
+      "step": 2145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265340487965017e-05,
+      "loss": 2.7127,
+      "step": 2146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826351700212489e-05,
+      "loss": 3.0267,
+      "step": 2147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8261692649480174e-05,
+      "loss": 2.7694,
+      "step": 2148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.825986743022225e-05,
+      "loss": 3.2806,
+      "step": 2149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8258041344542567e-05,
+      "loss": 4.1162,
+      "step": 2150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825621439263267e-05,
+      "loss": 2.6251,
+      "step": 2151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254386574684205e-05,
+      "loss": 2.1011,
+      "step": 2152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82525578908889e-05,
+      "loss": 2.953,
+      "step": 2153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825072834143857e-05,
+      "loss": 2.5724,
+      "step": 2154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824889792652513e-05,
+      "loss": 1.7107,
+      "step": 2155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824706664634058e-05,
+      "loss": 2.3506,
+      "step": 2156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824523450107702e-05,
+      "loss": 1.9678,
+      "step": 2157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8243401490926623e-05,
+      "loss": 1.983,
+      "step": 2158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241567616081674e-05,
+      "loss": 2.4387,
+      "step": 2159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8239732876734525e-05,
+      "loss": 3.1743,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8237897273077644e-05,
+      "loss": 2.364,
+      "step": 2161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823606080530357e-05,
+      "loss": 3.384,
+      "step": 2162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8234223473604946e-05,
+      "loss": 3.3937,
+      "step": 2163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823238527817449e-05,
+      "loss": 4.2953,
+      "step": 2164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8230546219205032e-05,
+      "loss": 3.4478,
+      "step": 2165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822870629688947e-05,
+      "loss": 1.8759,
+      "step": 2166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8226865511420812e-05,
+      "loss": 3.5339,
+      "step": 2167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822502386299214e-05,
+      "loss": 2.7102,
+      "step": 2168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223181351796642e-05,
+      "loss": 3.3677,
+      "step": 2169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822133797802758e-05,
+      "loss": 3.2748,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8219493741878325e-05,
+      "loss": 1.8711,
+      "step": 2171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217648643542326e-05,
+      "loss": 3.1412,
+      "step": 2172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821580268321312e-05,
+      "loss": 1.2631,
+      "step": 2173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8213955861084342e-05,
+      "loss": 2.6135,
+      "step": 2174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8212108177349722e-05,
+      "loss": 3.338,
+      "step": 2175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8210259632203063e-05,
+      "loss": 1.7011,
+      "step": 2176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8208410225838276e-05,
+      "loss": 2.6439,
+      "step": 2177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820655995844935e-05,
+      "loss": 3.4102,
+      "step": 2178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204708830230372e-05,
+      "loss": 2.7503,
+      "step": 2179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8202856841375517e-05,
+      "loss": 2.3219,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8201003992079054e-05,
+      "loss": 2.9458,
+      "step": 2181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8199150282535334e-05,
+      "loss": 2.5313,
+      "step": 2182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197295712938798e-05,
+      "loss": 6.0492,
+      "step": 2183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819544028348399e-05,
+      "loss": 1.8313,
+      "step": 2184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819358399436553e-05,
+      "loss": 2.367,
+      "step": 2185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819172684577814e-05,
+      "loss": 2.3319,
+      "step": 2186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818986883791662e-05,
+      "loss": 3.5261,
+      "step": 2187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818800997097587e-05,
+      "loss": 3.1968,
+      "step": 2188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186150245150876e-05,
+      "loss": 2.8329,
+      "step": 2189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8184289660636715e-05,
+      "loss": 2.5205,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8182428217628555e-05,
+      "loss": 3.4808,
+      "step": 2191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8180565916321646e-05,
+      "loss": 2.9584,
+      "step": 2192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817870275691134e-05,
+      "loss": 2.7375,
+      "step": 2193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8176838739593078e-05,
+      "loss": 2.6905,
+      "step": 2194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817497386456238e-05,
+      "loss": 1.9753,
+      "step": 2195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817310813201486e-05,
+      "loss": 3.1847,
+      "step": 2196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171241542146235e-05,
+      "loss": 3.6514,
+      "step": 2197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8169374095152298e-05,
+      "loss": 2.4281,
+      "step": 2198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816750579122893e-05,
+      "loss": 3.7548,
+      "step": 2199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816563663057211e-05,
+      "loss": 2.7062,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816376661337791e-05,
+      "loss": 1.3326,
+      "step": 2201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8161895739842476e-05,
+      "loss": 0.9343,
+      "step": 2202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8160024010162062e-05,
+      "loss": 1.7658,
+      "step": 2203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158151424533002e-05,
+      "loss": 2.9097,
+      "step": 2204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815627798315172e-05,
+      "loss": 4.1627,
+      "step": 2205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815440368621473e-05,
+      "loss": 3.9313,
+      "step": 2206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8152528533918642e-05,
+      "loss": 2.4672,
+      "step": 2207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8150652526460146e-05,
+      "loss": 2.4472,
+      "step": 2208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814877566403603e-05,
+      "loss": 2.9885,
+      "step": 2209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8146897946843162e-05,
+      "loss": 2.0689,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8145019375078517e-05,
+      "loss": 2.6374,
+      "step": 2211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8143139948939138e-05,
+      "loss": 3.354,
+      "step": 2212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8141259668622174e-05,
+      "loss": 2.5542,
+      "step": 2213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813937853432485e-05,
+      "loss": 2.8913,
+      "step": 2214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81374965462445e-05,
+      "loss": 2.8034,
+      "step": 2215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8135613704578525e-05,
+      "loss": 3.0009,
+      "step": 2216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8133730009524436e-05,
+      "loss": 3.7413,
+      "step": 2217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8131845461279813e-05,
+      "loss": 2.4246,
+      "step": 2218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8129960060042345e-05,
+      "loss": 3.4691,
+      "step": 2219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81280738060098e-05,
+      "loss": 3.6276,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8126186699380035e-05,
+      "loss": 2.6119,
+      "step": 2221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124298740351004e-05,
+      "loss": 2.6957,
+      "step": 2222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8122409929120735e-05,
+      "loss": 2.8149,
+      "step": 2223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8120520265887364e-05,
+      "loss": 2.3496,
+      "step": 2224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8118629750849106e-05,
+      "loss": 2.3291,
+      "step": 2225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116738384204268e-05,
+      "loss": 3.8768,
+      "step": 2226
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114846166151243e-05,
+      "loss": 2.175,
+      "step": 2227
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8112953096888517e-05,
+      "loss": 2.9011,
+      "step": 2228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111059176614665e-05,
+      "loss": 1.9886,
+      "step": 2229
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810916440552835e-05,
+      "loss": 2.9595,
+      "step": 2230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8107268783828326e-05,
+      "loss": 2.5282,
+      "step": 2231
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810537231171343e-05,
+      "loss": 2.9192,
+      "step": 2232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81034749893826e-05,
+      "loss": 2.0979,
+      "step": 2233
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8101576817034853e-05,
+      "loss": 2.9543,
+      "step": 2234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099677794869297e-05,
+      "loss": 2.6951,
+      "step": 2235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809777792308513e-05,
+      "loss": 2.518,
+      "step": 2236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8095877201881647e-05,
+      "loss": 2.3421,
+      "step": 2237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093975631458215e-05,
+      "loss": 3.4244,
+      "step": 2238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8092073212014307e-05,
+      "loss": 3.2032,
+      "step": 2239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 2.774,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8088265826863363e-05,
+      "loss": 3.2236,
+      "step": 2241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8086360861555707e-05,
+      "loss": 3.4848,
+      "step": 2242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8084455048026327e-05,
+      "loss": 1.4918,
+      "step": 2243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808254838647513e-05,
+      "loss": 2.8943,
+      "step": 2244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808064087710212e-05,
+      "loss": 2.3999,
+      "step": 2245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8078732520107386e-05,
+      "loss": 3.1358,
+      "step": 2246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8076823315691106e-05,
+      "loss": 2.1726,
+      "step": 2247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074913264053547e-05,
+      "loss": 2.6837,
+      "step": 2248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807300236539506e-05,
+      "loss": 2.4431,
+      "step": 2249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8071090619916095e-05,
+      "loss": 1.623,
+      "step": 2250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806917802781718e-05,
+      "loss": 1.512,
+      "step": 2251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8067264589298945e-05,
+      "loss": 2.6651,
+      "step": 2252
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8065350304562095e-05,
+      "loss": 1.3569,
+      "step": 2253
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806343517380743e-05,
+      "loss": 3.5078,
+      "step": 2254
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061519197235835e-05,
+      "loss": 2.4922,
+      "step": 2255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8059602375048294e-05,
+      "loss": 2.1457,
+      "step": 2256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805768470744587e-05,
+      "loss": 2.6631,
+      "step": 2257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8055766194629717e-05,
+      "loss": 2.1876,
+      "step": 2258
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8053846836801076e-05,
+      "loss": 3.8026,
+      "step": 2259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8051926634161282e-05,
+      "loss": 3.3052,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8050005586911757e-05,
+      "loss": 2.6368,
+      "step": 2261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8048083695254006e-05,
+      "loss": 2.3809,
+      "step": 2262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8046160959389625e-05,
+      "loss": 2.3907,
+      "step": 2263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8044237379520305e-05,
+      "loss": 1.6256,
+      "step": 2264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804231295584782e-05,
+      "loss": 3.3641,
+      "step": 2265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040387688574027e-05,
+      "loss": 2.9987,
+      "step": 2266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038461577900883e-05,
+      "loss": 2.9319,
+      "step": 2267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8036534624030428e-05,
+      "loss": 3.1594,
+      "step": 2268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8034606827164795e-05,
+      "loss": 3.5721,
+      "step": 2269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8032678187506187e-05,
+      "loss": 3.0457,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8030748705256923e-05,
+      "loss": 2.2667,
+      "step": 2271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802881838061939e-05,
+      "loss": 3.2492,
+      "step": 2272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8026887213796072e-05,
+      "loss": 2.2953,
+      "step": 2273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8024955204989537e-05,
+      "loss": 2.337,
+      "step": 2274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802302235440245e-05,
+      "loss": 3.0215,
+      "step": 2275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802108866223755e-05,
+      "loss": 1.4162,
+      "step": 2276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801915412869768e-05,
+      "loss": 2.6862,
+      "step": 2277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8017218753985758e-05,
+      "loss": 1.9813,
+      "step": 2278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80152825383048e-05,
+      "loss": 0.7469,
+      "step": 2279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8013345481857903e-05,
+      "loss": 3.1869,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011407584848255e-05,
+      "loss": 2.4434,
+      "step": 2281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8009468847479134e-05,
+      "loss": 1.2749,
+      "step": 2282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8007529269953902e-05,
+      "loss": 3.7738,
+      "step": 2283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8005588852476018e-05,
+      "loss": 3.532,
+      "step": 2284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8003647595249016e-05,
+      "loss": 2.5137,
+      "step": 2285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8001705498476526e-05,
+      "loss": 2.3688,
+      "step": 2286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7999762562362266e-05,
+      "loss": 3.4265,
+      "step": 2287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7997818787110043e-05,
+      "loss": 2.9921,
+      "step": 2288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799587417292375e-05,
+      "loss": 2.4103,
+      "step": 2289
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799392872000736e-05,
+      "loss": 1.8488,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7991982428564953e-05,
+      "loss": 3.3272,
+      "step": 2291
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7990035298800682e-05,
+      "loss": 1.3302,
+      "step": 2292
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7988087330918787e-05,
+      "loss": 2.4393,
+      "step": 2293
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798613852512361e-05,
+      "loss": 2.9481,
+      "step": 2294
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7984188881619563e-05,
+      "loss": 3.8173,
+      "step": 2295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798223840061116e-05,
+      "loss": 2.1355,
+      "step": 2296
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7980287082302996e-05,
+      "loss": 2.92,
+      "step": 2297
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7978334926899748e-05,
+      "loss": 2.9197,
+      "step": 2298
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79763819346062e-05,
+      "loss": 2.5725,
+      "step": 2299
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797442810562721e-05,
+      "loss": 2.545,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7972473440167716e-05,
+      "loss": 2.8274,
+      "step": 2301
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7970517938432766e-05,
+      "loss": 2.9497,
+      "step": 2302
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796856160062747e-05,
+      "loss": 3.0518,
+      "step": 2303
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796660442695705e-05,
+      "loss": 1.8119,
+      "step": 2304
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79646464176268e-05,
+      "loss": 2.5839,
+      "step": 2305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7962687572842102e-05,
+      "loss": 3.0416,
+      "step": 2306
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7960727892808437e-05,
+      "loss": 4.1294,
+      "step": 2307
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795876737773136e-05,
+      "loss": 2.1473,
+      "step": 2308
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795680602781652e-05,
+      "loss": 2.6833,
+      "step": 2309
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7954843843269665e-05,
+      "loss": 2.0732,
+      "step": 2310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7952880824296605e-05,
+      "loss": 3.5035,
+      "step": 2311
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795091697110326e-05,
+      "loss": 2.1706,
+      "step": 2312
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794895228389562e-05,
+      "loss": 3.0771,
+      "step": 2313
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7946986762879785e-05,
+      "loss": 1.9708,
+      "step": 2314
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794502040826192e-05,
+      "loss": 1.8891,
+      "step": 2315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943053220248284e-05,
+      "loss": 2.4559,
+      "step": 2316
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794108519904523e-05,
+      "loss": 2.6807,
+      "step": 2317
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79391163448592e-05,
+      "loss": 3.4054,
+      "step": 2318
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7937146657896708e-05,
+      "loss": 2.5457,
+      "step": 2319
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793517613836437e-05,
+      "loss": 2.9807,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7933204786468883e-05,
+      "loss": 1.6879,
+      "step": 2321
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7931232602417035e-05,
+      "loss": 2.9388,
+      "step": 2322
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7929259586415694e-05,
+      "loss": 3.04,
+      "step": 2323
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7927285738671825e-05,
+      "loss": 2.6208,
+      "step": 2324
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7925311059392472e-05,
+      "loss": 4.0946,
+      "step": 2325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7923335548784773e-05,
+      "loss": 1.7912,
+      "step": 2326
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792135920705595e-05,
+      "loss": 2.6407,
+      "step": 2327
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7919382034413306e-05,
+      "loss": 3.6177,
+      "step": 2328
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917404031064245e-05,
+      "loss": 2.7222,
+      "step": 2329
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7915425197216246e-05,
+      "loss": 3.2918,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791344553307688e-05,
+      "loss": 2.2949,
+      "step": 2331
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7911465038853805e-05,
+      "loss": 2.4355,
+      "step": 2332
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7909483714754765e-05,
+      "loss": 2.8931,
+      "step": 2333
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7907501560987594e-05,
+      "loss": 2.8894,
+      "step": 2334
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7905518577760207e-05,
+      "loss": 2.3982,
+      "step": 2335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7903534765280616e-05,
+      "loss": 3.1147,
+      "step": 2336
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7901550123756906e-05,
+      "loss": 3.3218,
+      "step": 2337
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789956465339726e-05,
+      "loss": 2.8882,
+      "step": 2338
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7897578354409945e-05,
+      "loss": 4.1336,
+      "step": 2339
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7895591227003316e-05,
+      "loss": 3.4993,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789360327138581e-05,
+      "loss": 2.968,
+      "step": 2341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789161448776596e-05,
+      "loss": 3.7899,
+      "step": 2342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788962487635237e-05,
+      "loss": 3.4104,
+      "step": 2343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7887634437353754e-05,
+      "loss": 3.2256,
+      "step": 2344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788564317097889e-05,
+      "loss": 2.9813,
+      "step": 2345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7883651077436656e-05,
+      "loss": 3.4088,
+      "step": 2346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881658156936012e-05,
+      "loss": 1.7077,
+      "step": 2347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7879664409686007e-05,
+      "loss": 2.3321,
+      "step": 2348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787766983589578e-05,
+      "loss": 2.7523,
+      "step": 2349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7875674435774546e-05,
+      "loss": 3.2911,
+      "step": 2350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7873678209531615e-05,
+      "loss": 3.1161,
+      "step": 2351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871681157376382e-05,
+      "loss": 2.856,
+      "step": 2352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7869683279518334e-05,
+      "loss": 2.0361,
+      "step": 2353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7867684576167028e-05,
+      "loss": 2.6729,
+      "step": 2354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786568504753213e-05,
+      "loss": 2.3689,
+      "step": 2355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7863684693823375e-05,
+      "loss": 3.0685,
+      "step": 2356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7861683515250588e-05,
+      "loss": 3.0197,
+      "step": 2357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7859681512023694e-05,
+      "loss": 1.5543,
+      "step": 2358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7857678684352684e-05,
+      "loss": 3.9166,
+      "step": 2359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7855675032447648e-05,
+      "loss": 1.8697,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7853670556518762e-05,
+      "loss": 3.4525,
+      "step": 2361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7851665256776284e-05,
+      "loss": 3.174,
+      "step": 2362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784965913343056e-05,
+      "loss": 2.5892,
+      "step": 2363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7847652186692025e-05,
+      "loss": 2.3628,
+      "step": 2364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78456444167712e-05,
+      "loss": 2.9769,
+      "step": 2365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7843635823878687e-05,
+      "loss": 2.8375,
+      "step": 2366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7841626408225177e-05,
+      "loss": 2.0055,
+      "step": 2367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7839616170021452e-05,
+      "loss": 2.4158,
+      "step": 2368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783760510947838e-05,
+      "loss": 3.1063,
+      "step": 2369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835593226806902e-05,
+      "loss": 2.4299,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7833580522218062e-05,
+      "loss": 2.621,
+      "step": 2371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7831566995922983e-05,
+      "loss": 3.4832,
+      "step": 2372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7829552648132872e-05,
+      "loss": 3.8798,
+      "step": 2373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827537479059026e-05,
+      "loss": 2.0922,
+      "step": 2374
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7825521488912833e-05,
+      "loss": 3.1049,
+      "step": 2375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782350467790575e-05,
+      "loss": 2.8678,
+      "step": 2376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7821487046249334e-05,
+      "loss": 2.6239,
+      "step": 2377
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7819468594155235e-05,
+      "loss": 2.4772,
+      "step": 2378
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7817449321835166e-05,
+      "loss": 3.416,
+      "step": 2379
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7815429229500946e-05,
+      "loss": 2.759,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781340831736447e-05,
+      "loss": 2.607,
+      "step": 2381
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811386585637728e-05,
+      "loss": 3.3519,
+      "step": 2382
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780936403453279e-05,
+      "loss": 2.0629,
+      "step": 2383
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78073406642618e-05,
+      "loss": 3.6876,
+      "step": 2384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7805316475037016e-05,
+      "loss": 2.6126,
+      "step": 2385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780329146707076e-05,
+      "loss": 2.4176,
+      "step": 2386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7801265640575442e-05,
+      "loss": 3.0531,
+      "step": 2387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779923899576357e-05,
+      "loss": 2.7276,
+      "step": 2388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779721153284772e-05,
+      "loss": 2.6292,
+      "step": 2389
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7795183252040568e-05,
+      "loss": 3.2845,
+      "step": 2390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7793154153554875e-05,
+      "loss": 3.0219,
+      "step": 2391
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7791124237603477e-05,
+      "loss": 4.0503,
+      "step": 2392
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7789093504399307e-05,
+      "loss": 3.8095,
+      "step": 2393
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778706195415538e-05,
+      "loss": 2.9374,
+      "step": 2394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7785029587084793e-05,
+      "loss": 2.3186,
+      "step": 2395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782996403400737e-05,
+      "loss": 3.6558,
+      "step": 2396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7780962403316477e-05,
+      "loss": 3.3107,
+      "step": 2397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7778927587045373e-05,
+      "loss": 3.5654,
+      "step": 2398
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777689195480087e-05,
+      "loss": 4.1006,
+      "step": 2399
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7774855506796497e-05,
+      "loss": 2.9496,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777281824324586e-05,
+      "loss": 2.9754,
+      "step": 2401
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770780164362665e-05,
+      "loss": 2.0333,
+      "step": 2402
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77687412703607e-05,
+      "loss": 3.2588,
+      "step": 2403
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776670156145383e-05,
+      "loss": 2.1732,
+      "step": 2404
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7764661037856013e-05,
+      "loss": 2.1023,
+      "step": 2405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776261969978129e-05,
+      "loss": 2.3177,
+      "step": 2406
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7760577547443785e-05,
+      "loss": 3.2187,
+      "step": 2407
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775853458105772e-05,
+      "loss": 3.6122,
+      "step": 2408
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7756490800837377e-05,
+      "loss": 2.2516,
+      "step": 2409
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7754446206997152e-05,
+      "loss": 3.2504,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7752400799751514e-05,
+      "loss": 4.1093,
+      "step": 2411
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750354579315004e-05,
+      "loss": 2.0175,
+      "step": 2412
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7748307545902275e-05,
+      "loss": 3.4948,
+      "step": 2413
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774625969972804e-05,
+      "loss": 3.3541,
+      "step": 2414
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774421104100712e-05,
+      "loss": 2.987,
+      "step": 2415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77421615699544e-05,
+      "loss": 3.413,
+      "step": 2416
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7740111286784864e-05,
+      "loss": 2.9105,
+      "step": 2417
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773806019171358e-05,
+      "loss": 3.5625,
+      "step": 2418
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7736008284955693e-05,
+      "loss": 2.2465,
+      "step": 2419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7733955566726438e-05,
+      "loss": 3.4742,
+      "step": 2420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731902037241143e-05,
+      "loss": 2.5917,
+      "step": 2421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772984769671521e-05,
+      "loss": 3.0644,
+      "step": 2422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7727792545364123e-05,
+      "loss": 2.3121,
+      "step": 2423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772573658340347e-05,
+      "loss": 2.2538,
+      "step": 2424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7723679811048904e-05,
+      "loss": 3.056,
+      "step": 2425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772162222851617e-05,
+      "loss": 2.3264,
+      "step": 2426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7719563836021107e-05,
+      "loss": 1.9557,
+      "step": 2427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7717504633779618e-05,
+      "loss": 1.4873,
+      "step": 2428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7715444622007715e-05,
+      "loss": 2.8825,
+      "step": 2429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771338380092148e-05,
+      "loss": 2.7387,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771132217073708e-05,
+      "loss": 2.252,
+      "step": 2431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7709259731670774e-05,
+      "loss": 1.6248,
+      "step": 2432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7707196483938903e-05,
+      "loss": 1.9592,
+      "step": 2433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7705132427757895e-05,
+      "loss": 3.2402,
+      "step": 2434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7703067563344252e-05,
+      "loss": 2.4266,
+      "step": 2435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770100189091457e-05,
+      "loss": 2.2452,
+      "step": 2436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7698935410685537e-05,
+      "loss": 2.4663,
+      "step": 2437
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769686812287391e-05,
+      "loss": 2.8977,
+      "step": 2438
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7694800027696536e-05,
+      "loss": 2.2916,
+      "step": 2439
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7692731125370355e-05,
+      "loss": 2.3149,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769066141611238e-05,
+      "loss": 2.6863,
+      "step": 2441
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7688590900139715e-05,
+      "loss": 3.6752,
+      "step": 2442
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768651957766955e-05,
+      "loss": 2.7198,
+      "step": 2443
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7684447448919156e-05,
+      "loss": 3.2202,
+      "step": 2444
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768237451410589e-05,
+      "loss": 3.2772,
+      "step": 2445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768030077344719e-05,
+      "loss": 2.8041,
+      "step": 2446
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7678226227160586e-05,
+      "loss": 1.9727,
+      "step": 2447
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676150875463688e-05,
+      "loss": 1.7783,
+      "step": 2448
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7674074718574187e-05,
+      "loss": 2.9678,
+      "step": 2449
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767199775670986e-05,
+      "loss": 2.6864,
+      "step": 2450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7669919990088583e-05,
+      "loss": 1.8146,
+      "step": 2451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7667841418928292e-05,
+      "loss": 0.9755,
+      "step": 2452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7665762043447022e-05,
+      "loss": 2.4918,
+      "step": 2453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663681863862895e-05,
+      "loss": 3.0245,
+      "step": 2454
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7661600880394113e-05,
+      "loss": 1.6951,
+      "step": 2455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765951909325895e-05,
+      "loss": 3.6218,
+      "step": 2456
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7657436502675785e-05,
+      "loss": 2.5093,
+      "step": 2457
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7655353108863068e-05,
+      "loss": 1.4139,
+      "step": 2458
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7653268912039346e-05,
+      "loss": 3.0369,
+      "step": 2459
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7651183912423228e-05,
+      "loss": 3.2175,
+      "step": 2460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764909811023343e-05,
+      "loss": 2.7303,
+      "step": 2461
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7647011505688745e-05,
+      "loss": 1.9238,
+      "step": 2462
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7644924099008038e-05,
+      "loss": 2.2726,
+      "step": 2463
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764283589041028e-05,
+      "loss": 2.5444,
+      "step": 2464
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7640746880114505e-05,
+      "loss": 3.4077,
+      "step": 2465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7638657068339845e-05,
+      "loss": 3.4458,
+      "step": 2466
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763656645530551e-05,
+      "loss": 2.2355,
+      "step": 2467
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7634475041230796e-05,
+      "loss": 2.3672,
+      "step": 2468
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7632382826335082e-05,
+      "loss": 2.0685,
+      "step": 2469
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7630289810837836e-05,
+      "loss": 1.7406,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7628195994958596e-05,
+      "loss": 2.71,
+      "step": 2471
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7626101378917004e-05,
+      "loss": 2.5845,
+      "step": 2472
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762400596293277e-05,
+      "loss": 3.3578,
+      "step": 2473
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76219097472257e-05,
+      "loss": 3.9087,
+      "step": 2474
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7619812732015664e-05,
+      "loss": 1.9512,
+      "step": 2475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761771491752264e-05,
+      "loss": 2.7271,
+      "step": 2476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7615616303966675e-05,
+      "loss": 3.6414,
+      "step": 2477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7613516891567907e-05,
+      "loss": 2.7114,
+      "step": 2478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761141668054655e-05,
+      "loss": 2.8192,
+      "step": 2479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7609315671122912e-05,
+      "loss": 2.0276,
+      "step": 2480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7607213863517376e-05,
+      "loss": 1.9235,
+      "step": 2481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7605111257950408e-05,
+      "loss": 2.6124,
+      "step": 2482
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760300785464257e-05,
+      "loss": 3.8735,
+      "step": 2483
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760090365381449e-05,
+      "loss": 2.0557,
+      "step": 2484
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75987986556869e-05,
+      "loss": 2.4154,
+      "step": 2485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7596692860480593e-05,
+      "loss": 2.8496,
+      "step": 2486
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759458626841646e-05,
+      "loss": 1.6392,
+      "step": 2487
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759247887971548e-05,
+      "loss": 2.4654,
+      "step": 2488
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75903706945987e-05,
+      "loss": 3.1213,
+      "step": 2489
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758826171328727e-05,
+      "loss": 3.32,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75861519360024e-05,
+      "loss": 3.5047,
+      "step": 2491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7584041362965397e-05,
+      "loss": 2.1123,
+      "step": 2492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7581929994397654e-05,
+      "loss": 2.9257,
+      "step": 2493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7579817830520644e-05,
+      "loss": 2.8472,
+      "step": 2494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7577704871555924e-05,
+      "loss": 3.1484,
+      "step": 2495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7575591117725132e-05,
+      "loss": 3.4182,
+      "step": 2496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757347656924999e-05,
+      "loss": 2.6701,
+      "step": 2497
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7571361226352305e-05,
+      "loss": 1.2065,
+      "step": 2498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756924508925397e-05,
+      "loss": 1.8491,
+      "step": 2499
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7567128158176955e-05,
+      "loss": 2.4526,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7565010433343316e-05,
+      "loss": 2.42,
+      "step": 2501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756289191497519e-05,
+      "loss": 3.1551,
+      "step": 2502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75607726032948e-05,
+      "loss": 1.9116,
+      "step": 2503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7558652498524464e-05,
+      "loss": 3.0439,
+      "step": 2504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7556531600886554e-05,
+      "loss": 2.3009,
+      "step": 2505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7554409910603553e-05,
+      "loss": 2.8374,
+      "step": 2506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7552287427898015e-05,
+      "loss": 2.5899,
+      "step": 2507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7550164152992573e-05,
+      "loss": 3.0648,
+      "step": 2508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7548040086109957e-05,
+      "loss": 1.7902,
+      "step": 2509
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7545915227472967e-05,
+      "loss": 2.9726,
+      "step": 2510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7543789577304487e-05,
+      "loss": 2.535,
+      "step": 2511
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7541663135827493e-05,
+      "loss": 3.1805,
+      "step": 2512
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753953590326504e-05,
+      "loss": 1.884,
+      "step": 2513
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7537407879840266e-05,
+      "loss": 2.7759,
+      "step": 2514
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753527906577638e-05,
+      "loss": 2.9255,
+      "step": 2515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75331494612967e-05,
+      "loss": 2.7895,
+      "step": 2516
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75310190666246e-05,
+      "loss": 2.5574,
+      "step": 2517
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752888788198355e-05,
+      "loss": 1.9327,
+      "step": 2518
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752675590759711e-05,
+      "loss": 2.7486,
+      "step": 2519
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7524623143688905e-05,
+      "loss": 1.0859,
+      "step": 2520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7522489590482653e-05,
+      "loss": 2.3582,
+      "step": 2521
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7520355248202157e-05,
+      "loss": 3.5227,
+      "step": 2522
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75182201170713e-05,
+      "loss": 2.0568,
+      "step": 2523
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7516084197314044e-05,
+      "loss": 3.4574,
+      "step": 2524
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7513947489154443e-05,
+      "loss": 1.7321,
+      "step": 2525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751180999281662e-05,
+      "loss": 3.1323,
+      "step": 2526
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750967170852479e-05,
+      "loss": 2.8928,
+      "step": 2527
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7507532636503256e-05,
+      "loss": 2.2463,
+      "step": 2528
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7505392776976392e-05,
+      "loss": 2.5871,
+      "step": 2529
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503252130168657e-05,
+      "loss": 1.9161,
+      "step": 2530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7501110696304598e-05,
+      "loss": 2.9905,
+      "step": 2531
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749896847560884e-05,
+      "loss": 2.0764,
+      "step": 2532
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7496825468306092e-05,
+      "loss": 2.4555,
+      "step": 2533
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7494681674621148e-05,
+      "loss": 2.4391,
+      "step": 2534
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749253709477888e-05,
+      "loss": 1.3553,
+      "step": 2535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7490391729004242e-05,
+      "loss": 3.0366,
+      "step": 2536
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748824557752228e-05,
+      "loss": 2.327,
+      "step": 2537
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7486098640558105e-05,
+      "loss": 2.7302,
+      "step": 2538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7483950918336933e-05,
+      "loss": 3.8678,
+      "step": 2539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748180241108404e-05,
+      "loss": 3.8541,
+      "step": 2540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74796531190248e-05,
+      "loss": 3.1668,
+      "step": 2541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747750304238466e-05,
+      "loss": 3.2601,
+      "step": 2542
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747535218138916e-05,
+      "loss": 2.1719,
+      "step": 2543
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7473200536263905e-05,
+      "loss": 2.7498,
+      "step": 2544
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74710481072346e-05,
+      "loss": 3.6835,
+      "step": 2545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746889489452702e-05,
+      "loss": 1.9157,
+      "step": 2546
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7466740898367033e-05,
+      "loss": 3.7441,
+      "step": 2547
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746458611898058e-05,
+      "loss": 2.5543,
+      "step": 2548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7462430556593687e-05,
+      "loss": 3.0355,
+      "step": 2549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7460274211432463e-05,
+      "loss": 2.9683,
+      "step": 2550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7458117083723096e-05,
+      "loss": 2.3715,
+      "step": 2551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7455959173691863e-05,
+      "loss": 2.6522,
+      "step": 2552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7453800481565118e-05,
+      "loss": 2.5295,
+      "step": 2553
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7451641007569296e-05,
+      "loss": 2.9815,
+      "step": 2554
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7449480751930915e-05,
+      "loss": 2.5112,
+      "step": 2555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744731971487658e-05,
+      "loss": 2.8697,
+      "step": 2556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7445157896632967e-05,
+      "loss": 3.3654,
+      "step": 2557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442995297426846e-05,
+      "loss": 2.2374,
+      "step": 2558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7440831917485064e-05,
+      "loss": 3.3786,
+      "step": 2559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7438667757034547e-05,
+      "loss": 3.0059,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7436502816302304e-05,
+      "loss": 2.5643,
+      "step": 2561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743433709551543e-05,
+      "loss": 2.5721,
+      "step": 2562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7432170594901102e-05,
+      "loss": 3.3879,
+      "step": 2563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743000331468657e-05,
+      "loss": 2.5229,
+      "step": 2564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7427835255099173e-05,
+      "loss": 2.775,
+      "step": 2565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7425666416366335e-05,
+      "loss": 3.4912,
+      "step": 2566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742349679871555e-05,
+      "loss": 2.3666,
+      "step": 2567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7421326402374406e-05,
+      "loss": 1.715,
+      "step": 2568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7419155227570564e-05,
+      "loss": 2.4952,
+      "step": 2569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7416983274531777e-05,
+      "loss": 1.8761,
+      "step": 2570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7414810543485862e-05,
+      "loss": 2.4556,
+      "step": 2571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7412637034660735e-05,
+      "loss": 2.7902,
+      "step": 2572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7410462748284387e-05,
+      "loss": 2.1684,
+      "step": 2573
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740828768458489e-05,
+      "loss": 2.6609,
+      "step": 2574
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74061118437904e-05,
+      "loss": 1.6642,
+      "step": 2575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740393522612915e-05,
+      "loss": 2.2919,
+      "step": 2576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401757831829454e-05,
+      "loss": 3.3057,
+      "step": 2577
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399579661119713e-05,
+      "loss": 2.1247,
+      "step": 2578
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7397400714228414e-05,
+      "loss": 2.1722,
+      "step": 2579
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739522099138411e-05,
+      "loss": 3.4199,
+      "step": 2580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7393040492815445e-05,
+      "loss": 1.997,
+      "step": 2581
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7390859218751145e-05,
+      "loss": 2.5574,
+      "step": 2582
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7388677169420015e-05,
+      "loss": 1.8846,
+      "step": 2583
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386494345050944e-05,
+      "loss": 2.899,
+      "step": 2584
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7384310745872896e-05,
+      "loss": 2.6851,
+      "step": 2585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738212637211492e-05,
+      "loss": 3.0314,
+      "step": 2586
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7379941224006152e-05,
+      "loss": 3.6789,
+      "step": 2587
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73777553017758e-05,
+      "loss": 0.8301,
+      "step": 2588
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737556860565316e-05,
+      "loss": 3.0171,
+      "step": 2589
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7373381135867605e-05,
+      "loss": 2.733,
+      "step": 2590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7371192892648587e-05,
+      "loss": 2.8054,
+      "step": 2591
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7369003876225644e-05,
+      "loss": 1.7867,
+      "step": 2592
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7366814086828395e-05,
+      "loss": 2.4782,
+      "step": 2593
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7364623524686542e-05,
+      "loss": 2.2624,
+      "step": 2594
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7362432190029862e-05,
+      "loss": 1.5429,
+      "step": 2595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7360240083088213e-05,
+      "loss": 3.3046,
+      "step": 2596
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7358047204091542e-05,
+      "loss": 4.0524,
+      "step": 2597
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7355853553269865e-05,
+      "loss": 3.2551,
+      "step": 2598
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735365913085329e-05,
+      "loss": 2.5364,
+      "step": 2599
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7351463937072008e-05,
+      "loss": 2.8572,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734926797215627e-05,
+      "loss": 1.6568,
+      "step": 2601
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7347071236336435e-05,
+      "loss": 2.7283,
+      "step": 2602
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7344873729842927e-05,
+      "loss": 1.5468,
+      "step": 2603
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734267545290625e-05,
+      "loss": 3.2139,
+      "step": 2604
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7340476405757e-05,
+      "loss": 3.1418,
+      "step": 2605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733827658862584e-05,
+      "loss": 2.5937,
+      "step": 2606
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7336076001743525e-05,
+      "loss": 3.1803,
+      "step": 2607
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7333874645340886e-05,
+      "loss": 3.4297,
+      "step": 2608
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7331672519648834e-05,
+      "loss": 3.5347,
+      "step": 2609
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732946962489836e-05,
+      "loss": 3.224,
+      "step": 2610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732726596132054e-05,
+      "loss": 2.8723,
+      "step": 2611
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325061529146528e-05,
+      "loss": 1.9302,
+      "step": 2612
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732285632860756e-05,
+      "loss": 2.2045,
+      "step": 2613
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732065035993495e-05,
+      "loss": 2.8779,
+      "step": 2614
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7318443623360092e-05,
+      "loss": 3.4596,
+      "step": 2615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7316236119114466e-05,
+      "loss": 4.4967,
+      "step": 2616
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731402784742962e-05,
+      "loss": 3.5269,
+      "step": 2617
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7311818808537206e-05,
+      "loss": 2.3804,
+      "step": 2618
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7309609002668932e-05,
+      "loss": 1.8943,
+      "step": 2619
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7307398430056595e-05,
+      "loss": 3.0807,
+      "step": 2620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7305187090932083e-05,
+      "loss": 2.3727,
+      "step": 2621
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7302974985527346e-05,
+      "loss": 2.8902,
+      "step": 2622
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7300762114074428e-05,
+      "loss": 3.0085,
+      "step": 2623
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7298548476805446e-05,
+      "loss": 2.754,
+      "step": 2624
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7296334073952606e-05,
+      "loss": 2.2124,
+      "step": 2625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294118905748184e-05,
+      "loss": 2.563,
+      "step": 2626
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729190297242454e-05,
+      "loss": 2.3574,
+      "step": 2627
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7289686274214116e-05,
+      "loss": 3.0219,
+      "step": 2628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7287468811349437e-05,
+      "loss": 3.4061,
+      "step": 2629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72852505840631e-05,
+      "loss": 3.4892,
+      "step": 2630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7283031592587792e-05,
+      "loss": 2.7381,
+      "step": 2631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7280811837156268e-05,
+      "loss": 2.6669,
+      "step": 2632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7278591318001376e-05,
+      "loss": 2.9662,
+      "step": 2633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7276370035356037e-05,
+      "loss": 2.6906,
+      "step": 2634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7274147989453246e-05,
+      "loss": 2.8496,
+      "step": 2635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7271925180526094e-05,
+      "loss": 1.444,
+      "step": 2636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7269701608807744e-05,
+      "loss": 2.5499,
+      "step": 2637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7267477274531432e-05,
+      "loss": 3.3,
+      "step": 2638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7265252177930483e-05,
+      "loss": 3.1938,
+      "step": 2639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72630263192383e-05,
+      "loss": 2.3709,
+      "step": 2640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7260799698688367e-05,
+      "loss": 3.5529,
+      "step": 2641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725857231651424e-05,
+      "loss": 2.2536,
+      "step": 2642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725634417294957e-05,
+      "loss": 3.5918,
+      "step": 2643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7254115268228073e-05,
+      "loss": 2.978,
+      "step": 2644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7251885602583547e-05,
+      "loss": 2.4878,
+      "step": 2645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7249655176249883e-05,
+      "loss": 2.6702,
+      "step": 2646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7247423989461036e-05,
+      "loss": 2.795,
+      "step": 2647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724519204245105e-05,
+      "loss": 3.5431,
+      "step": 2648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724295933545404e-05,
+      "loss": 3.8468,
+      "step": 2649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7240725868704218e-05,
+      "loss": 2.6586,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7238491642435855e-05,
+      "loss": 2.0369,
+      "step": 2651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723625665688331e-05,
+      "loss": 2.2992,
+      "step": 2652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723402091228103e-05,
+      "loss": 4.963,
+      "step": 2653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723178440886353e-05,
+      "loss": 0.9788,
+      "step": 2654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722954714686541e-05,
+      "loss": 3.0153,
+      "step": 2655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7227309126521347e-05,
+      "loss": 2.3569,
+      "step": 2656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72250703480661e-05,
+      "loss": 3.8665,
+      "step": 2657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7222830811734502e-05,
+      "loss": 2.7188,
+      "step": 2658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722059051776148e-05,
+      "loss": 3.3205,
+      "step": 2659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7218349466382024e-05,
+      "loss": 3.2562,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7216107657831206e-05,
+      "loss": 1.5787,
+      "step": 2661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7213865092344188e-05,
+      "loss": 3.1062,
+      "step": 2662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72116217701562e-05,
+      "loss": 2.6093,
+      "step": 2663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7209377691502565e-05,
+      "loss": 2.6827,
+      "step": 2664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7207132856618668e-05,
+      "loss": 1.1576,
+      "step": 2665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7204887265739978e-05,
+      "loss": 3.2139,
+      "step": 2666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7202640919102057e-05,
+      "loss": 3.4651,
+      "step": 2667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720039381694053e-05,
+      "loss": 3.5641,
+      "step": 2668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7198145959491113e-05,
+      "loss": 3.617,
+      "step": 2669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719589734698959e-05,
+      "loss": 4.6634,
+      "step": 2670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719364797967183e-05,
+      "loss": 4.079,
+      "step": 2671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7191397857773787e-05,
+      "loss": 1.5915,
+      "step": 2672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7189146981531486e-05,
+      "loss": 2.8876,
+      "step": 2673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718689535118103e-05,
+      "loss": 3.1884,
+      "step": 2674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718464296695861e-05,
+      "loss": 1.6932,
+      "step": 2675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7182389829100484e-05,
+      "loss": 3.6684,
+      "step": 2676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7180135937843005e-05,
+      "loss": 2.3917,
+      "step": 2677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7177881293422586e-05,
+      "loss": 4.0302,
+      "step": 2678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7175625896075732e-05,
+      "loss": 3.1206,
+      "step": 2679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173369746039026e-05,
+      "loss": 2.5474,
+      "step": 2680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7171112843549126e-05,
+      "loss": 2.2892,
+      "step": 2681
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716885518884277e-05,
+      "loss": 2.9558,
+      "step": 2682
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7166596782156778e-05,
+      "loss": 2.6404,
+      "step": 2683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7164337623728044e-05,
+      "loss": 3.9857,
+      "step": 2684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7162077713793547e-05,
+      "loss": 3.4162,
+      "step": 2685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7159817052590332e-05,
+      "loss": 2.1778,
+      "step": 2686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7157555640355546e-05,
+      "loss": 2.9515,
+      "step": 2687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7155293477326385e-05,
+      "loss": 2.8111,
+      "step": 2688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715303056374015e-05,
+      "loss": 1.943,
+      "step": 2689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7150766899834205e-05,
+      "loss": 3.0271,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7148502485846005e-05,
+      "loss": 2.8938,
+      "step": 2691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714623732201307e-05,
+      "loss": 3.0343,
+      "step": 2692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7143971408573e-05,
+      "loss": 3.6404,
+      "step": 2693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7141704745763494e-05,
+      "loss": 4.4049,
+      "step": 2694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7139437333822303e-05,
+      "loss": 3.1387,
+      "step": 2695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713716917298727e-05,
+      "loss": 2.0144,
+      "step": 2696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713490026349631e-05,
+      "loss": 2.1792,
+      "step": 2697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7132630605587433e-05,
+      "loss": 3.9202,
+      "step": 2698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713036019949871e-05,
+      "loss": 1.8763,
+      "step": 2699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7128089045468294e-05,
+      "loss": 2.8905,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712581714373442e-05,
+      "loss": 2.353,
+      "step": 2701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71235444945354e-05,
+      "loss": 1.6014,
+      "step": 2702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712127109810962e-05,
+      "loss": 2.6517,
+      "step": 2703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7118996954695553e-05,
+      "loss": 2.6805,
+      "step": 2704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711672206453175e-05,
+      "loss": 2.4194,
+      "step": 2705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711444642785683e-05,
+      "loss": 1.59,
+      "step": 2706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7112170044909497e-05,
+      "loss": 2.0229,
+      "step": 2707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7109892915928535e-05,
+      "loss": 1.6556,
+      "step": 2708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7107615041152807e-05,
+      "loss": 2.276,
+      "step": 2709
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7105336420821247e-05,
+      "loss": 3.5117,
+      "step": 2710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7103057055172875e-05,
+      "loss": 3.0372,
+      "step": 2711
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7100776944446783e-05,
+      "loss": 2.2405,
+      "step": 2712
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7098496088882143e-05,
+      "loss": 3.9136,
+      "step": 2713
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709621448871821e-05,
+      "loss": 2.7349,
+      "step": 2714
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709393214419431e-05,
+      "loss": 2.8056,
+      "step": 2715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709164905554986e-05,
+      "loss": 2.9628,
+      "step": 2716
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708936522302433e-05,
+      "loss": 2.1639,
+      "step": 2717
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7087080646857293e-05,
+      "loss": 3.612,
+      "step": 2718
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7084795327288387e-05,
+      "loss": 2.6439,
+      "step": 2719
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7082509264557333e-05,
+      "loss": 4.3281,
+      "step": 2720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7080222458903928e-05,
+      "loss": 1.8155,
+      "step": 2721
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707793491056805e-05,
+      "loss": 1.6148,
+      "step": 2722
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7075646619789646e-05,
+      "loss": 2.7205,
+      "step": 2723
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7073357586808753e-05,
+      "loss": 3.5665,
+      "step": 2724
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 2.6012,
+      "step": 2725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7068777295200007e-05,
+      "loss": 3.3175,
+      "step": 2726
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70664860370526e-05,
+      "loss": 2.7444,
+      "step": 2727
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706419403766361e-05,
+      "loss": 2.1967,
+      "step": 2728
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706190129727345e-05,
+      "loss": 2.5357,
+      "step": 2729
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705960781612262e-05,
+      "loss": 2.8341,
+      "step": 2730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7057313594451696e-05,
+      "loss": 2.2356,
+      "step": 2731
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7055018632501326e-05,
+      "loss": 2.8566,
+      "step": 2732
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7052722930512246e-05,
+      "loss": 3.6059,
+      "step": 2733
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7050426488725264e-05,
+      "loss": 2.4883,
+      "step": 2734
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7048129307381266e-05,
+      "loss": 3.6924,
+      "step": 2735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7045831386721213e-05,
+      "loss": 2.2049,
+      "step": 2736
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704353272698615e-05,
+      "loss": 3.2776,
+      "step": 2737
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7041233328417194e-05,
+      "loss": 3.1036,
+      "step": 2738
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703893319125554e-05,
+      "loss": 2.6155,
+      "step": 2739
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7036632315742464e-05,
+      "loss": 3.1114,
+      "step": 2740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7034330702119314e-05,
+      "loss": 2.465,
+      "step": 2741
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7032028350627525e-05,
+      "loss": 3.7688,
+      "step": 2742
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7029725261508594e-05,
+      "loss": 2.813,
+      "step": 2743
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7027421435004114e-05,
+      "loss": 3.553,
+      "step": 2744
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7025116871355737e-05,
+      "loss": 2.896,
+      "step": 2745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7022811570805206e-05,
+      "loss": 2.2906,
+      "step": 2746
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7020505533594332e-05,
+      "loss": 2.7504,
+      "step": 2747
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7018198759965018e-05,
+      "loss": 3.4528,
+      "step": 2748
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701589125015922e-05,
+      "loss": 2.4566,
+      "step": 2749
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7013583004418994e-05,
+      "loss": 4.2715,
+      "step": 2750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701127402298646e-05,
+      "loss": 1.3081,
+      "step": 2751
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7008964306103823e-05,
+      "loss": 1.954,
+      "step": 2752
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7006653854013362e-05,
+      "loss": 3.0624,
+      "step": 2753
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7004342666957426e-05,
+      "loss": 3.4684,
+      "step": 2754
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7002030745178455e-05,
+      "loss": 2.4437,
+      "step": 2755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6999718088918956e-05,
+      "loss": 1.919,
+      "step": 2756
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6997404698421517e-05,
+      "loss": 2.2447,
+      "step": 2757
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69950905739288e-05,
+      "loss": 2.5618,
+      "step": 2758
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6992775715683544e-05,
+      "loss": 2.5143,
+      "step": 2759
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6990460123928577e-05,
+      "loss": 2.3399,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6988143798906783e-05,
+      "loss": 1.4719,
+      "step": 2761
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698582674086114e-05,
+      "loss": 3.445,
+      "step": 2762
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6983508950034687e-05,
+      "loss": 3.5664,
+      "step": 2763
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698119042667056e-05,
+      "loss": 3.9088,
+      "step": 2764
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6978871171011963e-05,
+      "loss": 3.7458,
+      "step": 2765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6976551183302164e-05,
+      "loss": 2.8952,
+      "step": 2766
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6974230463784528e-05,
+      "loss": 2.3766,
+      "step": 2767
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6971909012702483e-05,
+      "loss": 1.9304,
+      "step": 2768
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696958683029954e-05,
+      "loss": 2.1414,
+      "step": 2769
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696726391681929e-05,
+      "loss": 2.581,
+      "step": 2770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6964940272505387e-05,
+      "loss": 2.9657,
+      "step": 2771
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6962615897601573e-05,
+      "loss": 3.7656,
+      "step": 2772
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960290792351668e-05,
+      "loss": 2.4832,
+      "step": 2773
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6957964956999563e-05,
+      "loss": 3.0384,
+      "step": 2774
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695563839178923e-05,
+      "loss": 2.0187,
+      "step": 2775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6953311096964706e-05,
+      "loss": 2.301,
+      "step": 2776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6950983072770123e-05,
+      "loss": 2.5241,
+      "step": 2777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948654319449674e-05,
+      "loss": 2.4034,
+      "step": 2778
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6946324837247636e-05,
+      "loss": 1.2609,
+      "step": 2779
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6943994626408365e-05,
+      "loss": 1.9101,
+      "step": 2780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694166368717628e-05,
+      "loss": 2.9983,
+      "step": 2781
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69393320197959e-05,
+      "loss": 3.1803,
+      "step": 2782
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6936999624511793e-05,
+      "loss": 2.3394,
+      "step": 2783
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6934666501568618e-05,
+      "loss": 3.8201,
+      "step": 2784
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6932332651211115e-05,
+      "loss": 4.0909,
+      "step": 2785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692999807368409e-05,
+      "loss": 3.7812,
+      "step": 2786
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692766276923243e-05,
+      "loss": 2.0088,
+      "step": 2787
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69253267381011e-05,
+      "loss": 2.9176,
+      "step": 2788
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6922989980535135e-05,
+      "loss": 2.6835,
+      "step": 2789
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692065249677965e-05,
+      "loss": 3.2097,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691831428707984e-05,
+      "loss": 2.6213,
+      "step": 2791
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6915975351680968e-05,
+      "loss": 2.9782,
+      "step": 2792
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6913635690828382e-05,
+      "loss": 2.3137,
+      "step": 2793
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6911295304767497e-05,
+      "loss": 3.0273,
+      "step": 2794
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6908954193743816e-05,
+      "loss": 2.7285,
+      "step": 2795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69066123580029e-05,
+      "loss": 4.0047,
+      "step": 2796
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6904269797790404e-05,
+      "loss": 3.3993,
+      "step": 2797
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6901926513352052e-05,
+      "loss": 1.9899,
+      "step": 2798
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6899582504933637e-05,
+      "loss": 2.7457,
+      "step": 2799
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6897237772781046e-05,
+      "loss": 2.4851,
+      "step": 2800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6894892317140216e-05,
+      "loss": 2.4987,
+      "step": 2801
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6892546138257187e-05,
+      "loss": 1.8052,
+      "step": 2802
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6890199236378056e-05,
+      "loss": 2.3206,
+      "step": 2803
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6887851611749005e-05,
+      "loss": 2.534,
+      "step": 2804
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6885503264616282e-05,
+      "loss": 1.6939,
+      "step": 2805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688315419522623e-05,
+      "loss": 3.2224,
+      "step": 2806
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6880804403825245e-05,
+      "loss": 1.8519,
+      "step": 2807
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6878453890659815e-05,
+      "loss": 2.2479,
+      "step": 2808
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6876102655976492e-05,
+      "loss": 2.5763,
+      "step": 2809
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6873750700021917e-05,
+      "loss": 1.7941,
+      "step": 2810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6871398023042793e-05,
+      "loss": 2.4734,
+      "step": 2811
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686904462528591e-05,
+      "loss": 3.3772,
+      "step": 2812
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6866690506998125e-05,
+      "loss": 4.4677,
+      "step": 2813
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6864335668426373e-05,
+      "loss": 2.7778,
+      "step": 2814
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686198010981767e-05,
+      "loss": 2.9093,
+      "step": 2815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68596238314191e-05,
+      "loss": 2.7688,
+      "step": 2816
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6857266833477827e-05,
+      "loss": 3.8266,
+      "step": 2817
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685490911624109e-05,
+      "loss": 1.8981,
+      "step": 2818
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68525506799562e-05,
+      "loss": 2.5671,
+      "step": 2819
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6850191524870548e-05,
+      "loss": 2.0105,
+      "step": 2820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6847831651231598e-05,
+      "loss": 3.4356,
+      "step": 2821
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684547105928689e-05,
+      "loss": 3.0203,
+      "step": 2822
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6843109749284037e-05,
+      "loss": 1.811,
+      "step": 2823
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6840747721470733e-05,
+      "loss": 3.7411,
+      "step": 2824
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6838384976094738e-05,
+      "loss": 2.0322,
+      "step": 2825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68360215134039e-05,
+      "loss": 3.0447,
+      "step": 2826
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6833657333646132e-05,
+      "loss": 3.1019,
+      "step": 2827
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6831292437069425e-05,
+      "loss": 3.9231,
+      "step": 2828
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828926823921845e-05,
+      "loss": 3.9111,
+      "step": 2829
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682656049445154e-05,
+      "loss": 2.6848,
+      "step": 2830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6824193448906718e-05,
+      "loss": 1.6741,
+      "step": 2831
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6821825687535675e-05,
+      "loss": 3.2249,
+      "step": 2832
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681945721058678e-05,
+      "loss": 2.0919,
+      "step": 2833
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817088018308477e-05,
+      "loss": 2.6338,
+      "step": 2834
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6814718110949274e-05,
+      "loss": 2.6348,
+      "step": 2835
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6812347488757774e-05,
+      "loss": 2.6755,
+      "step": 2836
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6809976151982635e-05,
+      "loss": 3.1902,
+      "step": 2837
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6807604100872604e-05,
+      "loss": 2.9699,
+      "step": 2838
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6805231335676505e-05,
+      "loss": 3.7464,
+      "step": 2839
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6802857856643214e-05,
+      "loss": 2.427,
+      "step": 2840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680048366402171e-05,
+      "loss": 3.7181,
+      "step": 2841
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6798108758061036e-05,
+      "loss": 3.0152,
+      "step": 2842
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.67957331390103e-05,
+      "loss": 2.0669,
+      "step": 2843
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6793356807118695e-05,
+      "loss": 3.6954,
+      "step": 2844
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6790979762635497e-05,
+      "loss": 2.3926,
+      "step": 2845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6788602005810035e-05,
+      "loss": 1.5796,
+      "step": 2846
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678622353689173e-05,
+      "loss": 1.5516,
+      "step": 2847
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6783844356130073e-05,
+      "loss": 1.879,
+      "step": 2848
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6781464463774628e-05,
+      "loss": 3.817,
+      "step": 2849
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6779083860075032e-05,
+      "loss": 3.2539,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776702545281004e-05,
+      "loss": 2.2389,
+      "step": 2851
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677432051964233e-05,
+      "loss": 2.5674,
+      "step": 2852
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6771937783408873e-05,
+      "loss": 2.647,
+      "step": 2853
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6769554336830577e-05,
+      "loss": 1.8677,
+      "step": 2854
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6767170180157442e-05,
+      "loss": 3.0771,
+      "step": 2855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6764785313639568e-05,
+      "loss": 3.0189,
+      "step": 2856
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676239973752711e-05,
+      "loss": 1.84,
+      "step": 2857
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6760013452070304e-05,
+      "loss": 2.9446,
+      "step": 2858
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675762645751946e-05,
+      "loss": 2.017,
+      "step": 2859
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6755238754124965e-05,
+      "loss": 2.2196,
+      "step": 2860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675285034213728e-05,
+      "loss": 3.2652,
+      "step": 2861
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675046122180693e-05,
+      "loss": 2.39,
+      "step": 2862
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6748071393384527e-05,
+      "loss": 1.905,
+      "step": 2863
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6745680857120757e-05,
+      "loss": 2.675,
+      "step": 2864
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674328961326637e-05,
+      "loss": 3.597,
+      "step": 2865
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6740897662072203e-05,
+      "loss": 1.4598,
+      "step": 2866
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6738505003789157e-05,
+      "loss": 3.6597,
+      "step": 2867
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6736111638668203e-05,
+      "loss": 3.0535,
+      "step": 2868
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673371756696041e-05,
+      "loss": 2.8308,
+      "step": 2869
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6731322788916892e-05,
+      "loss": 3.0203,
+      "step": 2870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6728927304788854e-05,
+      "loss": 2.6005,
+      "step": 2871
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6726531114827572e-05,
+      "loss": 2.6297,
+      "step": 2872
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6724134219284397e-05,
+      "loss": 2.2607,
+      "step": 2873
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672173661841075e-05,
+      "loss": 2.76,
+      "step": 2874
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6719338312458123e-05,
+      "loss": 2.4003,
+      "step": 2875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6716939301678098e-05,
+      "loss": 2.1594,
+      "step": 2876
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6714539586322316e-05,
+      "loss": 2.2487,
+      "step": 2877
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671213916664249e-05,
+      "loss": 3.494,
+      "step": 2878
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670973804289042e-05,
+      "loss": 3.3548,
+      "step": 2879
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6707336215317968e-05,
+      "loss": 1.4225,
+      "step": 2880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6704933684177077e-05,
+      "loss": 3.945,
+      "step": 2881
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6702530449719763e-05,
+      "loss": 2.6966,
+      "step": 2882
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670012651219811e-05,
+      "loss": 3.3428,
+      "step": 2883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6697721871864286e-05,
+      "loss": 3.9912,
+      "step": 2884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6695316528970517e-05,
+      "loss": 2.0146,
+      "step": 2885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6692910483769124e-05,
+      "loss": 3.716,
+      "step": 2886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669050373651248e-05,
+      "loss": 1.6866,
+      "step": 2887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6688096287453048e-05,
+      "loss": 3.6052,
+      "step": 2888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6685688136843355e-05,
+      "loss": 1.8982,
+      "step": 2889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6683279284936004e-05,
+      "loss": 2.4773,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6680869731983674e-05,
+      "loss": 1.4796,
+      "step": 2891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6678459478239116e-05,
+      "loss": 2.9884,
+      "step": 2892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6676048523955156e-05,
+      "loss": 3.2806,
+      "step": 2893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667363686938469e-05,
+      "loss": 1.7802,
+      "step": 2894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6671224514780692e-05,
+      "loss": 3.6406,
+      "step": 2895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6668811460396202e-05,
+      "loss": 3.0137,
+      "step": 2896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6666397706484342e-05,
+      "loss": 3.7938,
+      "step": 2897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66639832532983e-05,
+      "loss": 2.4073,
+      "step": 2898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6661568101091345e-05,
+      "loss": 1.8372,
+      "step": 2899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665915225011681e-05,
+      "loss": 2.1065,
+      "step": 2900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6656735700628115e-05,
+      "loss": 3.439,
+      "step": 2901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665431845287874e-05,
+      "loss": 3.8449,
+      "step": 2902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665190050712224e-05,
+      "loss": 2.447,
+      "step": 2903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664948186361225e-05,
+      "loss": 2.4716,
+      "step": 2904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6647062522602474e-05,
+      "loss": 2.7349,
+      "step": 2905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6644642484346688e-05,
+      "loss": 2.6352,
+      "step": 2906
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664222174909875e-05,
+      "loss": 2.5709,
+      "step": 2907
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663980031711257e-05,
+      "loss": 3.3841,
+      "step": 2908
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637378188642156e-05,
+      "loss": 2.9728,
+      "step": 2909
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6634955363941573e-05,
+      "loss": 2.7654,
+      "step": 2910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663253184326497e-05,
+      "loss": 2.9571,
+      "step": 2911
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6630107626866558e-05,
+      "loss": 1.6845,
+      "step": 2912
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6627682715000627e-05,
+      "loss": 2.4153,
+      "step": 2913
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662525710792154e-05,
+      "loss": 3.3205,
+      "step": 2914
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662283080588373e-05,
+      "loss": 2.8633,
+      "step": 2915
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6620403809141707e-05,
+      "loss": 3.0502,
+      "step": 2916
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6617976117950046e-05,
+      "loss": 2.9657,
+      "step": 2917
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661554773256341e-05,
+      "loss": 3.4017,
+      "step": 2918
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661311865323652e-05,
+      "loss": 3.6087,
+      "step": 2919
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6610688880224178e-05,
+      "loss": 2.0956,
+      "step": 2920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660825841378125e-05,
+      "loss": 2.6115,
+      "step": 2921
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660582725416268e-05,
+      "loss": 2.947,
+      "step": 2922
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6603395401623498e-05,
+      "loss": 2.7683,
+      "step": 2923
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6600962856418782e-05,
+      "loss": 3.2662,
+      "step": 2924
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65985296188037e-05,
+      "loss": 1.7588,
+      "step": 2925
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6596095689033484e-05,
+      "loss": 2.5876,
+      "step": 2926
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6593661067363446e-05,
+      "loss": 2.5683,
+      "step": 2927
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6591225754048963e-05,
+      "loss": 3.3236,
+      "step": 2928
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588789749345487e-05,
+      "loss": 3.4908,
+      "step": 2929
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6586353053508548e-05,
+      "loss": 2.5677,
+      "step": 2930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658391566679374e-05,
+      "loss": 3.3796,
+      "step": 2931
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6581477589456737e-05,
+      "loss": 2.4238,
+      "step": 2932
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6579038821753282e-05,
+      "loss": 2.0007,
+      "step": 2933
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6576599363939185e-05,
+      "loss": 2.9464,
+      "step": 2934
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657415921627034e-05,
+      "loss": 3.0194,
+      "step": 2935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6571718379002705e-05,
+      "loss": 1.8231,
+      "step": 2936
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656927685239231e-05,
+      "loss": 2.426,
+      "step": 2937
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6566834636695264e-05,
+      "loss": 4.1042,
+      "step": 2938
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6564391732167743e-05,
+      "loss": 2.7346,
+      "step": 2939
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6561948139065997e-05,
+      "loss": 2.357,
+      "step": 2940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6559503857646342e-05,
+      "loss": 3.0968,
+      "step": 2941
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655705888816518e-05,
+      "loss": 3.1137,
+      "step": 2942
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655461323087897e-05,
+      "loss": 3.0566,
+      "step": 2943
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6552166886044253e-05,
+      "loss": 2.3921,
+      "step": 2944
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654971985391764e-05,
+      "loss": 3.9459,
+      "step": 2945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6547272134755813e-05,
+      "loss": 2.9948,
+      "step": 2946
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6544823728815528e-05,
+      "loss": 2.9321,
+      "step": 2947
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6542374636353605e-05,
+      "loss": 2.9897,
+      "step": 2948
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6539924857626947e-05,
+      "loss": 3.0019,
+      "step": 2949
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6537474392892527e-05,
+      "loss": 1.5642,
+      "step": 2950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6535023242407385e-05,
+      "loss": 3.3938,
+      "step": 2951
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653257140642863e-05,
+      "loss": 2.8014,
+      "step": 2952
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6530118885213458e-05,
+      "loss": 2.2407,
+      "step": 2953
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652766567901912e-05,
+      "loss": 3.3596,
+      "step": 2954
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6525211788102946e-05,
+      "loss": 2.0487,
+      "step": 2955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6522757212722346e-05,
+      "loss": 1.811,
+      "step": 2956
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6520301953134782e-05,
+      "loss": 3.4267,
+      "step": 2957
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517846009597804e-05,
+      "loss": 3.2915,
+      "step": 2958
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6515389382369034e-05,
+      "loss": 2.4073,
+      "step": 2959
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6512932071706153e-05,
+      "loss": 2.4386,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651047407786693e-05,
+      "loss": 3.6551,
+      "step": 2961
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650801540110919e-05,
+      "loss": 2.3476,
+      "step": 2962
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6505556041690836e-05,
+      "loss": 2.8111,
+      "step": 2963
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650309599986985e-05,
+      "loss": 3.0677,
+      "step": 2964
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6500635275904274e-05,
+      "loss": 3.3422,
+      "step": 2965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6498173870052227e-05,
+      "loss": 2.5633,
+      "step": 2966
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6495711782571902e-05,
+      "loss": 2.096,
+      "step": 2967
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6493249013721558e-05,
+      "loss": 2.6387,
+      "step": 2968
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649078556375953e-05,
+      "loss": 2.8762,
+      "step": 2969
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6488321432944218e-05,
+      "loss": 1.4313,
+      "step": 2970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6485856621534102e-05,
+      "loss": 2.8921,
+      "step": 2971
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6483391129787725e-05,
+      "loss": 2.3673,
+      "step": 2972
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648092495796372e-05,
+      "loss": 2.6814,
+      "step": 2973
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6478458106320755e-05,
+      "loss": 3.8843,
+      "step": 2974
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6475990575117603e-05,
+      "loss": 1.5982,
+      "step": 2975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64735223646131e-05,
+      "loss": 4.7018,
+      "step": 2976
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647105347506614e-05,
+      "loss": 2.9264,
+      "step": 2977
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646858390673571e-05,
+      "loss": 3.0553,
+      "step": 2978
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6466113659880845e-05,
+      "loss": 2.1867,
+      "step": 2979
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646364273476067e-05,
+      "loss": 1.4939,
+      "step": 2980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646117113163437e-05,
+      "loss": 3.1314,
+      "step": 2981
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6458698850761203e-05,
+      "loss": 1.8655,
+      "step": 2982
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6456225892400505e-05,
+      "loss": 3.3414,
+      "step": 2983
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453752256811676e-05,
+      "loss": 2.7747,
+      "step": 2984
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6451277944254186e-05,
+      "loss": 3.3619,
+      "step": 2985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644880295498758e-05,
+      "loss": 2.6136,
+      "step": 2986
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6446327289271478e-05,
+      "loss": 3.0377,
+      "step": 2987
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644385094736556e-05,
+      "loss": 2.4313,
+      "step": 2988
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6441373929529583e-05,
+      "loss": 3.5206,
+      "step": 2989
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6438896236023374e-05,
+      "loss": 2.564,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643641786710684e-05,
+      "loss": 2.9902,
+      "step": 2991
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6433938823039942e-05,
+      "loss": 3.4567,
+      "step": 2992
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6431459104082715e-05,
+      "loss": 3.3982,
+      "step": 2993
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6428978710495286e-05,
+      "loss": 1.9004,
+      "step": 2994
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6426497642537826e-05,
+      "loss": 2.6245,
+      "step": 2995
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642401590047059e-05,
+      "loss": 2.7133,
+      "step": 2996
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64215334845539e-05,
+      "loss": 1.9812,
+      "step": 2997
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6419050395048147e-05,
+      "loss": 3.092,
+      "step": 2998
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6416566632213803e-05,
+      "loss": 3.2327,
+      "step": 2999
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6414082196311402e-05,
+      "loss": 2.4704,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6411597087601542e-05,
+      "loss": 3.8956,
+      "step": 3001
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640911130634491e-05,
+      "loss": 2.4866,
+      "step": 3002
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6406624852802243e-05,
+      "loss": 1.4221,
+      "step": 3003
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6404137727234366e-05,
+      "loss": 2.0613,
+      "step": 3004
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640164992990216e-05,
+      "loss": 3.2618,
+      "step": 3005
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6399161461066593e-05,
+      "loss": 3.2653,
+      "step": 3006
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6396672320988684e-05,
+      "loss": 1.0742,
+      "step": 3007
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639418250992954e-05,
+      "loss": 2.3112,
+      "step": 3008
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6391692028150323e-05,
+      "loss": 3.9598,
+      "step": 3009
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638920087591228e-05,
+      "loss": 2.7054,
+      "step": 3010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6386709053476715e-05,
+      "loss": 2.3565,
+      "step": 3011
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6384216561105014e-05,
+      "loss": 1.8739,
+      "step": 3012
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6381723399058623e-05,
+      "loss": 2.5376,
+      "step": 3013
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6379229567599072e-05,
+      "loss": 2.8747,
+      "step": 3014
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637673506698794e-05,
+      "loss": 3.5068,
+      "step": 3015
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63742398974869e-05,
+      "loss": 2.7542,
+      "step": 3016
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6371744059357675e-05,
+      "loss": 3.3016,
+      "step": 3017
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636924755286207e-05,
+      "loss": 3.68,
+      "step": 3018
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636675037826196e-05,
+      "loss": 3.1072,
+      "step": 3019
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6364252535819284e-05,
+      "loss": 3.1072,
+      "step": 3020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636175402579605e-05,
+      "loss": 3.108,
+      "step": 3021
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6359254848454352e-05,
+      "loss": 2.1265,
+      "step": 3022
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635675500405633e-05,
+      "loss": 2.4589,
+      "step": 3023
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635425449286421e-05,
+      "loss": 3.1453,
+      "step": 3024
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6351753315140285e-05,
+      "loss": 2.5897,
+      "step": 3025
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6349251471146917e-05,
+      "loss": 2.3335,
+      "step": 3026
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6346748961146542e-05,
+      "loss": 2.8558,
+      "step": 3027
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6344245785401653e-05,
+      "loss": 2.6386,
+      "step": 3028
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6341741944174824e-05,
+      "loss": 2.7363,
+      "step": 3029
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63392374377287e-05,
+      "loss": 2.444,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633673226632599e-05,
+      "loss": 3.2118,
+      "step": 3031
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6334226430229475e-05,
+      "loss": 2.4113,
+      "step": 3032
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6331719929702004e-05,
+      "loss": 1.6366,
+      "step": 3033
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6329212765006503e-05,
+      "loss": 2.267,
+      "step": 3034
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6326704936405953e-05,
+      "loss": 2.6744,
+      "step": 3035
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632419644416342e-05,
+      "loss": 1.8645,
+      "step": 3036
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6321687288542035e-05,
+      "loss": 3.2939,
+      "step": 3037
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631917746980499e-05,
+      "loss": 3.8084,
+      "step": 3038
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6316666988215558e-05,
+      "loss": 2.3836,
+      "step": 3039
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6314155844037074e-05,
+      "loss": 3.4749,
+      "step": 3040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6311644037532947e-05,
+      "loss": 2.682,
+      "step": 3041
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630913156896665e-05,
+      "loss": 2.5281,
+      "step": 3042
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6306618438601737e-05,
+      "loss": 2.3547,
+      "step": 3043
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6304104646701818e-05,
+      "loss": 1.7864,
+      "step": 3044
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6301590193530585e-05,
+      "loss": 1.6312,
+      "step": 3045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6299075079351784e-05,
+      "loss": 1.9473,
+      "step": 3046
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6296559304429242e-05,
+      "loss": 2.6823,
+      "step": 3047
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629404286902685e-05,
+      "loss": 2.9277,
+      "step": 3048
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6291525773408576e-05,
+      "loss": 3.5057,
+      "step": 3049
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6289008017838447e-05,
+      "loss": 2.7038,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6286489602580564e-05,
+      "loss": 2.3778,
+      "step": 3051
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62839705278991e-05,
+      "loss": 2.7407,
+      "step": 3052
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628145079405829e-05,
+      "loss": 3.1623,
+      "step": 3053
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6278930401322445e-05,
+      "loss": 2.4619,
+      "step": 3054
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6276409349955945e-05,
+      "loss": 2.3986,
+      "step": 3055
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627388764022323e-05,
+      "loss": 2.7688,
+      "step": 3056
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6271365272388827e-05,
+      "loss": 2.0884,
+      "step": 3057
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6268842246717307e-05,
+      "loss": 2.2291,
+      "step": 3058
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626631856347333e-05,
+      "loss": 2.7464,
+      "step": 3059
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626379422292162e-05,
+      "loss": 2.8779,
+      "step": 3060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6261269225326968e-05,
+      "loss": 3.0933,
+      "step": 3061
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6258743570954235e-05,
+      "loss": 2.5505,
+      "step": 3062
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6256217260068346e-05,
+      "loss": 3.0712,
+      "step": 3063
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6253690292934303e-05,
+      "loss": 1.7312,
+      "step": 3064
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6251162669817172e-05,
+      "loss": 2.6203,
+      "step": 3065
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624863439098209e-05,
+      "loss": 4.7102,
+      "step": 3066
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6246105456694263e-05,
+      "loss": 3.3561,
+      "step": 3067
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624357586721896e-05,
+      "loss": 2.4026,
+      "step": 3068
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6241045622821526e-05,
+      "loss": 3.1071,
+      "step": 3069
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6238514723767372e-05,
+      "loss": 1.2903,
+      "step": 3070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623598317032198e-05,
+      "loss": 3.4686,
+      "step": 3071
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6233450962750895e-05,
+      "loss": 2.4551,
+      "step": 3072
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6230918101319733e-05,
+      "loss": 3.6699,
+      "step": 3073
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6228384586294178e-05,
+      "loss": 2.862,
+      "step": 3074
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622585041793999e-05,
+      "loss": 2.4288,
+      "step": 3075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622331559652299e-05,
+      "loss": 2.4919,
+      "step": 3076
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6220780122309065e-05,
+      "loss": 3.9095,
+      "step": 3077
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6218243995564177e-05,
+      "loss": 3.0693,
+      "step": 3078
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621570721655435e-05,
+      "loss": 2.1301,
+      "step": 3079
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6213169785545688e-05,
+      "loss": 2.0327,
+      "step": 3080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6210631702804353e-05,
+      "loss": 2.9658,
+      "step": 3081
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6208092968596574e-05,
+      "loss": 2.7535,
+      "step": 3082
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6205553583188653e-05,
+      "loss": 3.3312,
+      "step": 3083
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6203013546846967e-05,
+      "loss": 2.2301,
+      "step": 3084
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6200472859837946e-05,
+      "loss": 3.1134,
+      "step": 3085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61979315224281e-05,
+      "loss": 2.7563,
+      "step": 3086
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6195389534884004e-05,
+      "loss": 3.0222,
+      "step": 3087
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61928468974723e-05,
+      "loss": 2.2514,
+      "step": 3088
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6190303610459696e-05,
+      "loss": 2.8191,
+      "step": 3089
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6187759674112972e-05,
+      "loss": 2.8857,
+      "step": 3090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618521508869898e-05,
+      "loss": 2.5812,
+      "step": 3091
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618266985448463e-05,
+      "loss": 2.9117,
+      "step": 3092
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6180123971736906e-05,
+      "loss": 2.5388,
+      "step": 3093
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6177577440722863e-05,
+      "loss": 1.6483,
+      "step": 3094
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6175030261709615e-05,
+      "loss": 4.2971,
+      "step": 3095
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6172482434964353e-05,
+      "loss": 3.086,
+      "step": 3096
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616993396075433e-05,
+      "loss": 3.1129,
+      "step": 3097
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6167384839346872e-05,
+      "loss": 3.5943,
+      "step": 3098
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6164835071009364e-05,
+      "loss": 2.812,
+      "step": 3099
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6162284656009276e-05,
+      "loss": 3.2786,
+      "step": 3100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615973359461412e-05,
+      "loss": 1.9146,
+      "step": 3101
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6157181887091503e-05,
+      "loss": 2.8743,
+      "step": 3102
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6154629533709082e-05,
+      "loss": 3.3485,
+      "step": 3103
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6152076534734585e-05,
+      "loss": 3.2736,
+      "step": 3104
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6149522890435815e-05,
+      "loss": 2.4893,
+      "step": 3105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6146968601080633e-05,
+      "loss": 2.2809,
+      "step": 3106
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614441366693697e-05,
+      "loss": 1.6376,
+      "step": 3107
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6141858088272838e-05,
+      "loss": 2.4515,
+      "step": 3108
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6139301865356292e-05,
+      "loss": 1.4839,
+      "step": 3109
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6136744998455477e-05,
+      "loss": 3.1506,
+      "step": 3110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6134187487838594e-05,
+      "loss": 2.949,
+      "step": 3111
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6131629333773908e-05,
+      "loss": 2.8496,
+      "step": 3112
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6129070536529767e-05,
+      "loss": 3.2058,
+      "step": 3113
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612651109637457e-05,
+      "loss": 2.3341,
+      "step": 3114
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6123951013576796e-05,
+      "loss": 2.2586,
+      "step": 3115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612139028840498e-05,
+      "loss": 2.6058,
+      "step": 3116
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6118828921127736e-05,
+      "loss": 3.335,
+      "step": 3117
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6116266912013734e-05,
+      "loss": 3.2043,
+      "step": 3118
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611370426133172e-05,
+      "loss": 2.8932,
+      "step": 3119
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6111140969350504e-05,
+      "loss": 2.2868,
+      "step": 3120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610857703633896e-05,
+      "loss": 2.5403,
+      "step": 3121
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6106012462566042e-05,
+      "loss": 1.9891,
+      "step": 3122
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6103447248300754e-05,
+      "loss": 2.4178,
+      "step": 3123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610088139381217e-05,
+      "loss": 2.8851,
+      "step": 3124
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6098314899369446e-05,
+      "loss": 3.0979,
+      "step": 3125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6095747765241797e-05,
+      "loss": 2.9325,
+      "step": 3126
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6093179991698493e-05,
+      "loss": 3.1161,
+      "step": 3127
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609061157900889e-05,
+      "loss": 2.3304,
+      "step": 3128
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60880425274424e-05,
+      "loss": 2.996,
+      "step": 3129
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6085472837268504e-05,
+      "loss": 2.5536,
+      "step": 3130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608290250875675e-05,
+      "loss": 4.0913,
+      "step": 3131
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6080331542176754e-05,
+      "loss": 1.0576,
+      "step": 3132
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60777599377982e-05,
+      "loss": 2.4829,
+      "step": 3133
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6075187695890837e-05,
+      "loss": 1.55,
+      "step": 3134
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607261481672448e-05,
+      "loss": 1.9679,
+      "step": 3135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6070041300569014e-05,
+      "loss": 3.5046,
+      "step": 3136
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6067467147694382e-05,
+      "loss": 2.0562,
+      "step": 3137
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6064892358370608e-05,
+      "loss": 2.8122,
+      "step": 3138
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6062316932867777e-05,
+      "loss": 3.3156,
+      "step": 3139
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6059740871456035e-05,
+      "loss": 3.6547,
+      "step": 3140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6057164174405603e-05,
+      "loss": 1.8694,
+      "step": 3141
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6054586841986754e-05,
+      "loss": 3.0361,
+      "step": 3142
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605200887446985e-05,
+      "loss": 3.1933,
+      "step": 3143
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60494302721253e-05,
+      "loss": 3.5799,
+      "step": 3144
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6046851035223594e-05,
+      "loss": 2.9354,
+      "step": 3145
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6044271164035278e-05,
+      "loss": 2.3907,
+      "step": 3146
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604169065883097e-05,
+      "loss": 4.2489,
+      "step": 3147
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603910951988135e-05,
+      "loss": 3.2558,
+      "step": 3148
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6036527747457172e-05,
+      "loss": 2.8119,
+      "step": 3149
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603394534182925e-05,
+      "loss": 2.2485,
+      "step": 3150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6031362303268464e-05,
+      "loss": 2.5801,
+      "step": 3151
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602877863204576e-05,
+      "loss": 3.057,
+      "step": 3152
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6026194328432166e-05,
+      "loss": 1.9863,
+      "step": 3153
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6023609392698753e-05,
+      "loss": 2.7159,
+      "step": 3154
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6021023825116672e-05,
+      "loss": 2.2239,
+      "step": 3155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6018437625957135e-05,
+      "loss": 2.1539,
+      "step": 3156
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601585079549142e-05,
+      "loss": 1.5413,
+      "step": 3157
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601326333399088e-05,
+      "loss": 3.013,
+      "step": 3158
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601067524172693e-05,
+      "loss": 2.935,
+      "step": 3159
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6008086518971037e-05,
+      "loss": 2.5153,
+      "step": 3160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6005497165994754e-05,
+      "loss": 1.6376,
+      "step": 3161
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6002907183069693e-05,
+      "loss": 2.5956,
+      "step": 3162
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6000316570467525e-05,
+      "loss": 2.4801,
+      "step": 3163
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599772532846e-05,
+      "loss": 2.8222,
+      "step": 3164
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599513345731892e-05,
+      "loss": 2.4322,
+      "step": 3165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599254095731617e-05,
+      "loss": 3.111,
+      "step": 3166
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5989947828723682e-05,
+      "loss": 3.837,
+      "step": 3167
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598735407181347e-05,
+      "loss": 1.2698,
+      "step": 3168
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5984759686857602e-05,
+      "loss": 2.8029,
+      "step": 3169
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598216467412822e-05,
+      "loss": 3.5696,
+      "step": 3170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5979569033897527e-05,
+      "loss": 2.5476,
+      "step": 3171
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5976972766437796e-05,
+      "loss": 3.606,
+      "step": 3172
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597437587202136e-05,
+      "loss": 2.1357,
+      "step": 3173
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5971778350920622e-05,
+      "loss": 3.3717,
+      "step": 3174
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5969180203408052e-05,
+      "loss": 2.1097,
+      "step": 3175
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596658142975618e-05,
+      "loss": 3.5333,
+      "step": 3176
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596398203023762e-05,
+      "loss": 2.4509,
+      "step": 3177
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596138200512501e-05,
+      "loss": 2.7335,
+      "step": 3178
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59587813546911e-05,
+      "loss": 2.9127,
+      "step": 3179
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5956180079208684e-05,
+      "loss": 3.7115,
+      "step": 3180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5953578178950617e-05,
+      "loss": 2.9031,
+      "step": 3181
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5950975654189832e-05,
+      "loss": 2.1768,
+      "step": 3182
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5948372505199325e-05,
+      "loss": 2.6008,
+      "step": 3183
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5945768732252144e-05,
+      "loss": 2.1799,
+      "step": 3184
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5943164335621418e-05,
+      "loss": 2.684,
+      "step": 3185
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5940559315580338e-05,
+      "loss": 2.2687,
+      "step": 3186
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5937953672402157e-05,
+      "loss": 2.7476,
+      "step": 3187
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5935347406360192e-05,
+      "loss": 2.3544,
+      "step": 3188
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5932740517727835e-05,
+      "loss": 2.8102,
+      "step": 3189
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593013300677853e-05,
+      "loss": 2.6231,
+      "step": 3190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5927524873785797e-05,
+      "loss": 2.5521,
+      "step": 3191
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5924916119023214e-05,
+      "loss": 3.9373,
+      "step": 3192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5922306742764426e-05,
+      "loss": 2.6608,
+      "step": 3193
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5919696745283154e-05,
+      "loss": 1.9037,
+      "step": 3194
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591708612685316e-05,
+      "loss": 2.2622,
+      "step": 3195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5914474887748297e-05,
+      "loss": 3.7917,
+      "step": 3196
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591186302824247e-05,
+      "loss": 2.9056,
+      "step": 3197
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5909250548609644e-05,
+      "loss": 2.5477,
+      "step": 3198
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5906637449123864e-05,
+      "loss": 2.3746,
+      "step": 3199
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5904023730059227e-05,
+      "loss": 3.3252,
+      "step": 3200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5901409391689906e-05,
+      "loss": 2.3344,
+      "step": 3201
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5898794434290124e-05,
+      "loss": 2.5217,
+      "step": 3202
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5896178858134187e-05,
+      "loss": 1.7293,
+      "step": 3203
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589356266349645e-05,
+      "loss": 1.616,
+      "step": 3204
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5890945850651347e-05,
+      "loss": 2.7424,
+      "step": 3205
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5888328419873363e-05,
+      "loss": 4.0414,
+      "step": 3206
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5885710371437056e-05,
+      "loss": 3.3272,
+      "step": 3207
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5883091705617045e-05,
+      "loss": 2.7885,
+      "step": 3208
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5880472422688023e-05,
+      "loss": 2.9821,
+      "step": 3209
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 2.967,
+      "step": 3210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5875232006601993e-05,
+      "loss": 3.5846,
+      "step": 3211
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5872610873994685e-05,
+      "loss": 2.6493,
+      "step": 3212
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586998912537775e-05,
+      "loss": 3.0338,
+      "step": 3213
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5867366761026198e-05,
+      "loss": 3.2391,
+      "step": 3214
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586474378121511e-05,
+      "loss": 2.8583,
+      "step": 3215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5862120186219614e-05,
+      "loss": 2.4881,
+      "step": 3216
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5859495976314918e-05,
+      "loss": 2.9945,
+      "step": 3217
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585687115177629e-05,
+      "loss": 1.6988,
+      "step": 3218
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585424571287906e-05,
+      "loss": 3.2572,
+      "step": 3219
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5851619659898623e-05,
+      "loss": 2.908,
+      "step": 3220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5848992993110445e-05,
+      "loss": 2.094,
+      "step": 3221
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584636571279005e-05,
+      "loss": 2.7443,
+      "step": 3222
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5843737819213017e-05,
+      "loss": 3.2021,
+      "step": 3223
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5841109312655017e-05,
+      "loss": 2.8528,
+      "step": 3224
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5838480193391753e-05,
+      "loss": 2.4072,
+      "step": 3225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583585046169902e-05,
+      "loss": 3.1729,
+      "step": 3226
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5833220117852655e-05,
+      "loss": 2.6248,
+      "step": 3227
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5830589162128574e-05,
+      "loss": 1.7362,
+      "step": 3228
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582795759480275e-05,
+      "loss": 1.9922,
+      "step": 3229
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582532541615122e-05,
+      "loss": 2.5836,
+      "step": 3230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5822692626450096e-05,
+      "loss": 3.5879,
+      "step": 3231
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582005922597553e-05,
+      "loss": 2.5796,
+      "step": 3232
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5817425215003773e-05,
+      "loss": 3.4784,
+      "step": 3233
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581479059381111e-05,
+      "loss": 2.7534,
+      "step": 3234
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5812155362673895e-05,
+      "loss": 2.9071,
+      "step": 3235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580951952186856e-05,
+      "loss": 3.1104,
+      "step": 3236
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5806883071671587e-05,
+      "loss": 3.261,
+      "step": 3237
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5804246012359535e-05,
+      "loss": 2.406,
+      "step": 3238
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5801608344209012e-05,
+      "loss": 3.6233,
+      "step": 3239
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.57989700674967e-05,
+      "loss": 1.8895,
+      "step": 3240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579633118249934e-05,
+      "loss": 2.7913,
+      "step": 3241
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5793691689493743e-05,
+      "loss": 2.6436,
+      "step": 3242
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5791051588756777e-05,
+      "loss": 2.6863,
+      "step": 3243
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578841088056538e-05,
+      "loss": 2.5684,
+      "step": 3244
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5785769565196543e-05,
+      "loss": 3.6118,
+      "step": 3245
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578312764292733e-05,
+      "loss": 2.7375,
+      "step": 3246
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5780485114034868e-05,
+      "loss": 3.1397,
+      "step": 3247
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5777841978796348e-05,
+      "loss": 1.8537,
+      "step": 3248
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5775198237489016e-05,
+      "loss": 3.8641,
+      "step": 3249
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5772553890390196e-05,
+      "loss": 3.4871,
+      "step": 3250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5769908937777262e-05,
+      "loss": 2.3486,
+      "step": 3251
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5767263379927663e-05,
+      "loss": 1.7145,
+      "step": 3252
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57646172171189e-05,
+      "loss": 2.1051,
+      "step": 3253
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576197044962854e-05,
+      "loss": 2.5011,
+      "step": 3254
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5759323077734233e-05,
+      "loss": 3.4457,
+      "step": 3255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5756675101713657e-05,
+      "loss": 2.9012,
+      "step": 3256
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575402652184458e-05,
+      "loss": 3.4996,
+      "step": 3257
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575137733840483e-05,
+      "loss": 2.814,
+      "step": 3258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574872755167229e-05,
+      "loss": 2.3771,
+      "step": 3259
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5746077161924905e-05,
+      "loss": 3.0195,
+      "step": 3260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57434261694407e-05,
+      "loss": 1.8368,
+      "step": 3261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5740774574497747e-05,
+      "loss": 1.1048,
+      "step": 3262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5738122377374182e-05,
+      "loss": 3.1066,
+      "step": 3263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573546957834821e-05,
+      "loss": 2.0354,
+      "step": 3264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5732816177698097e-05,
+      "loss": 3.7531,
+      "step": 3265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573016217570218e-05,
+      "loss": 2.7558,
+      "step": 3266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5727507572638837e-05,
+      "loss": 2.7009,
+      "step": 3267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572485236878654e-05,
+      "loss": 3.3604,
+      "step": 3268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572219656442379e-05,
+      "loss": 2.9276,
+      "step": 3269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5719540159829185e-05,
+      "loss": 3.6719,
+      "step": 3270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5716883155281365e-05,
+      "loss": 3.3896,
+      "step": 3271
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5714225551059027e-05,
+      "loss": 3.1755,
+      "step": 3272
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5711567347440955e-05,
+      "loss": 2.6132,
+      "step": 3273
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5708908544705973e-05,
+      "loss": 3.5559,
+      "step": 3274
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5706249143132982e-05,
+      "loss": 1.6712,
+      "step": 3275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570358914300094e-05,
+      "loss": 1.5038,
+      "step": 3276
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5700928544588867e-05,
+      "loss": 2.0907,
+      "step": 3277
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5698267348175852e-05,
+      "loss": 3.3343,
+      "step": 3278
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5695605554041035e-05,
+      "loss": 3.4161,
+      "step": 3279
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5692943162463628e-05,
+      "loss": 1.7565,
+      "step": 3280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5690280173722907e-05,
+      "loss": 2.7859,
+      "step": 3281
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5687616588098207e-05,
+      "loss": 3.6292,
+      "step": 3282
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5684952405868923e-05,
+      "loss": 2.7596,
+      "step": 3283
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682287627314513e-05,
+      "loss": 2.744,
+      "step": 3284
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5679622252714507e-05,
+      "loss": 2.5011,
+      "step": 3285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5676956282348488e-05,
+      "loss": 3.0154,
+      "step": 3286
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56742897164961e-05,
+      "loss": 1.6486,
+      "step": 3287
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5671622555437055e-05,
+      "loss": 3.5323,
+      "step": 3288
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566895479945113e-05,
+      "loss": 2.9856,
+      "step": 3289
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5666286448818152e-05,
+      "loss": 3.4095,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5663617503818026e-05,
+      "loss": 3.1199,
+      "step": 3291
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566094796473071e-05,
+      "loss": 4.4608,
+      "step": 3292
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5658277831836227e-05,
+      "loss": 1.8119,
+      "step": 3293
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565560710541466e-05,
+      "loss": 3.2057,
+      "step": 3294
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565293578574615e-05,
+      "loss": 1.8934,
+      "step": 3295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565026387311092e-05,
+      "loss": 3.3205,
+      "step": 3296
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564759136778923e-05,
+      "loss": 3.5452,
+      "step": 3297
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5644918270061418e-05,
+      "loss": 2.2406,
+      "step": 3298
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5642244580207877e-05,
+      "loss": 1.5007,
+      "step": 3299
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5639570298509067e-05,
+      "loss": 2.2456,
+      "step": 3300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5636895425245507e-05,
+      "loss": 2.5268,
+      "step": 3301
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5634219960697778e-05,
+      "loss": 2.9463,
+      "step": 3302
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563154390514652e-05,
+      "loss": 2.9375,
+      "step": 3303
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562886725887245e-05,
+      "loss": 3.1979,
+      "step": 3304
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5626190022156328e-05,
+      "loss": 2.4801,
+      "step": 3305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5623512195278984e-05,
+      "loss": 2.42,
+      "step": 3306
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5620833778521306e-05,
+      "loss": 2.8452,
+      "step": 3307
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5618154772164257e-05,
+      "loss": 3.7231,
+      "step": 3308
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5615475176488843e-05,
+      "loss": 2.5156,
+      "step": 3309
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5612794991776147e-05,
+      "loss": 1.482,
+      "step": 3310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561011421830731e-05,
+      "loss": 3.2087,
+      "step": 3311
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5607432856363523e-05,
+      "loss": 2.5781,
+      "step": 3312
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5604750906226056e-05,
+      "loss": 3.0747,
+      "step": 3313
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5602068368176233e-05,
+      "loss": 3.3334,
+      "step": 3314
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5599385242495437e-05,
+      "loss": 2.5972,
+      "step": 3315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559670152946512e-05,
+      "loss": 4.2861,
+      "step": 3316
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559401722936678e-05,
+      "loss": 3.9473,
+      "step": 3317
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5591332342482002e-05,
+      "loss": 3.2667,
+      "step": 3318
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558864686909241e-05,
+      "loss": 3.0301,
+      "step": 3319
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5585960809479698e-05,
+      "loss": 3.5535,
+      "step": 3320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558327416392562e-05,
+      "loss": 2.1484,
+      "step": 3321
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5580586932712e-05,
+      "loss": 2.1225,
+      "step": 3322
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5577899116120708e-05,
+      "loss": 2.7106,
+      "step": 3323
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5575210714433687e-05,
+      "loss": 2.1001,
+      "step": 3324
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5572521727932937e-05,
+      "loss": 1.8051,
+      "step": 3325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556983215690052e-05,
+      "loss": 2.155,
+      "step": 3326
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556714200161856e-05,
+      "loss": 2.873,
+      "step": 3327
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5564451262369247e-05,
+      "loss": 1.7199,
+      "step": 3328
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5561759939434818e-05,
+      "loss": 2.6716,
+      "step": 3329
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5559068033097583e-05,
+      "loss": 3.1207,
+      "step": 3330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5556375543639912e-05,
+      "loss": 2.9917,
+      "step": 3331
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5553682471344237e-05,
+      "loss": 2.3244,
+      "step": 3332
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5550988816493043e-05,
+      "loss": 1.9005,
+      "step": 3333
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554829457936889e-05,
+      "loss": 3.3453,
+      "step": 3334
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554559976025438e-05,
+      "loss": 2.9615,
+      "step": 3335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5542904359432198e-05,
+      "loss": 2.4878,
+      "step": 3336
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5540208377185072e-05,
+      "loss": 2.9305,
+      "step": 3337
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55375118137958e-05,
+      "loss": 2.4191,
+      "step": 3338
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553481466954724e-05,
+      "loss": 2.5485,
+      "step": 3339
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5532116944722308e-05,
+      "loss": 2.4004,
+      "step": 3340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5529418639603987e-05,
+      "loss": 2.8885,
+      "step": 3341
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552671975447531e-05,
+      "loss": 4.2198,
+      "step": 3342
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5524020289619383e-05,
+      "loss": 2.2829,
+      "step": 3343
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5521320245319364e-05,
+      "loss": 2.8654,
+      "step": 3344
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5518619621858474e-05,
+      "loss": 1.9819,
+      "step": 3345
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551591841952e-05,
+      "loss": 2.2181,
+      "step": 3346
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5513216638587286e-05,
+      "loss": 2.1617,
+      "step": 3347
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5510514279343736e-05,
+      "loss": 2.4089,
+      "step": 3348
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5507811342072807e-05,
+      "loss": 3.1046,
+      "step": 3349
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5505107827058038e-05,
+      "loss": 3.3408,
+      "step": 3350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5502403734583002e-05,
+      "loss": 1.327,
+      "step": 3351
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5499699064931354e-05,
+      "loss": 3.2943,
+      "step": 3352
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54969938183868e-05,
+      "loss": 2.3544,
+      "step": 3353
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5494287995233107e-05,
+      "loss": 2.4889,
+      "step": 3354
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5491581595754102e-05,
+      "loss": 2.1681,
+      "step": 3355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5488874620233674e-05,
+      "loss": 3.5985,
+      "step": 3356
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5486167068955774e-05,
+      "loss": 2.8999,
+      "step": 3357
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5483458942204407e-05,
+      "loss": 2.5599,
+      "step": 3358
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5480750240263653e-05,
+      "loss": 2.3843,
+      "step": 3359
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547804096341763e-05,
+      "loss": 3.3454,
+      "step": 3360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5475331111950538e-05,
+      "loss": 1.9499,
+      "step": 3361
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5472620686146625e-05,
+      "loss": 1.5524,
+      "step": 3362
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5469909686290197e-05,
+      "loss": 2.3161,
+      "step": 3363
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5467198112665632e-05,
+      "loss": 3.4293,
+      "step": 3364
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546448596555736e-05,
+      "loss": 2.8553,
+      "step": 3365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5461773245249866e-05,
+      "loss": 3.2039,
+      "step": 3366
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5459059952027716e-05,
+      "loss": 3.0747,
+      "step": 3367
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5456346086175508e-05,
+      "loss": 3.7407,
+      "step": 3368
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545363164797792e-05,
+      "loss": 2.9112,
+      "step": 3369
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5450916637719683e-05,
+      "loss": 2.202,
+      "step": 3370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544820105568559e-05,
+      "loss": 2.7018,
+      "step": 3371
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5445484902160494e-05,
+      "loss": 2.6152,
+      "step": 3372
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5442768177429302e-05,
+      "loss": 2.4904,
+      "step": 3373
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544005088177699e-05,
+      "loss": 1.4159,
+      "step": 3374
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5437333015488586e-05,
+      "loss": 3.7065,
+      "step": 3375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543461457884919e-05,
+      "loss": 2.6193,
+      "step": 3376
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5431895572143942e-05,
+      "loss": 2.2835,
+      "step": 3377
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542917599565806e-05,
+      "loss": 2.7421,
+      "step": 3378
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5426455849676814e-05,
+      "loss": 2.225,
+      "step": 3379
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5423735134485537e-05,
+      "loss": 2.1351,
+      "step": 3380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5421013850369612e-05,
+      "loss": 2.7957,
+      "step": 3381
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5418291997614497e-05,
+      "loss": 0.7328,
+      "step": 3382
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5415569576505695e-05,
+      "loss": 2.0111,
+      "step": 3383
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541284658732878e-05,
+      "loss": 2.5409,
+      "step": 3384
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5410123030369387e-05,
+      "loss": 2.4675,
+      "step": 3385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5407398905913188e-05,
+      "loss": 3.4231,
+      "step": 3386
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5404674214245946e-05,
+      "loss": 2.3925,
+      "step": 3387
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540194895565346e-05,
+      "loss": 2.811,
+      "step": 3388
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5399223130421603e-05,
+      "loss": 3.5217,
+      "step": 3389
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5396496738836292e-05,
+      "loss": 2.9965,
+      "step": 3390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5393769781183525e-05,
+      "loss": 3.5374,
+      "step": 3391
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5391042257749338e-05,
+      "loss": 2.9058,
+      "step": 3392
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5388314168819838e-05,
+      "loss": 3.4491,
+      "step": 3393
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5385585514681193e-05,
+      "loss": 2.0615,
+      "step": 3394
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5382856295619622e-05,
+      "loss": 1.6176,
+      "step": 3395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5380126511921404e-05,
+      "loss": 1.5558,
+      "step": 3396
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537739616387289e-05,
+      "loss": 2.3919,
+      "step": 3397
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5374665251760474e-05,
+      "loss": 2.3502,
+      "step": 3398
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5371933775870617e-05,
+      "loss": 2.4582,
+      "step": 3399
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536920173648984e-05,
+      "loss": 3.5802,
+      "step": 3400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536646913390472e-05,
+      "loss": 2.3822,
+      "step": 3401
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5363735968401892e-05,
+      "loss": 1.9648,
+      "step": 3402
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5361002240268058e-05,
+      "loss": 2.5131,
+      "step": 3403
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5358267949789968e-05,
+      "loss": 3.6075,
+      "step": 3404
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535553309725444e-05,
+      "loss": 2.3584,
+      "step": 3405
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5352797682948345e-05,
+      "loss": 2.5798,
+      "step": 3406
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5350061707158616e-05,
+      "loss": 2.6776,
+      "step": 3407
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5347325170172246e-05,
+      "loss": 2.2627,
+      "step": 3408
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534458807227628e-05,
+      "loss": 3.2696,
+      "step": 3409
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5341850413757834e-05,
+      "loss": 2.4213,
+      "step": 3410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5339112194904066e-05,
+      "loss": 1.522,
+      "step": 3411
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533637341600221e-05,
+      "loss": 2.1335,
+      "step": 3412
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5333634077339558e-05,
+      "loss": 2.4559,
+      "step": 3413
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5330894179203436e-05,
+      "loss": 2.5418,
+      "step": 3414
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532815372188126e-05,
+      "loss": 2.7895,
+      "step": 3415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532541270566049e-05,
+      "loss": 2.522,
+      "step": 3416
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5322671130828635e-05,
+      "loss": 2.4146,
+      "step": 3417
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531992899767329e-05,
+      "loss": 2.1413,
+      "step": 3418
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5317186306482082e-05,
+      "loss": 2.7795,
+      "step": 3419
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5314443057542703e-05,
+      "loss": 3.6286,
+      "step": 3420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5311699251142916e-05,
+      "loss": 2.6743,
+      "step": 3421
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530895488757053e-05,
+      "loss": 2.2446,
+      "step": 3422
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530620996711342e-05,
+      "loss": 3.0712,
+      "step": 3423
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5303464490059506e-05,
+      "loss": 2.0864,
+      "step": 3424
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530071845669678e-05,
+      "loss": 2.5123,
+      "step": 3425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5297971867313293e-05,
+      "loss": 2.8547,
+      "step": 3426
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5295224722197145e-05,
+      "loss": 2.5377,
+      "step": 3427
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5292477021636498e-05,
+      "loss": 1.4691,
+      "step": 3428
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5289728765919575e-05,
+      "loss": 1.502,
+      "step": 3429
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5286979955334655e-05,
+      "loss": 3.1794,
+      "step": 3430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5284230590170074e-05,
+      "loss": 3.9862,
+      "step": 3431
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528148067071423e-05,
+      "loss": 2.8274,
+      "step": 3432
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5278730197255578e-05,
+      "loss": 3.7897,
+      "step": 3433
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5275979170082627e-05,
+      "loss": 3.3509,
+      "step": 3434
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5273227589483945e-05,
+      "loss": 1.3648,
+      "step": 3435
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5270475455748165e-05,
+      "loss": 1.6631,
+      "step": 3436
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526772276916397e-05,
+      "loss": 3.0024,
+      "step": 3437
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5264969530020105e-05,
+      "loss": 4.0576,
+      "step": 3438
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526221573860537e-05,
+      "loss": 3.6097,
+      "step": 3439
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5259461395208628e-05,
+      "loss": 2.7353,
+      "step": 3440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5256706500118798e-05,
+      "loss": 3.5547,
+      "step": 3441
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5253951053624852e-05,
+      "loss": 2.7098,
+      "step": 3442
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5251195056015822e-05,
+      "loss": 2.927,
+      "step": 3443
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5248438507580806e-05,
+      "loss": 1.976,
+      "step": 3444
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5245681408608946e-05,
+      "loss": 2.9865,
+      "step": 3445
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5242923759389454e-05,
+      "loss": 2.3923,
+      "step": 3446
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524016556021159e-05,
+      "loss": 1.9987,
+      "step": 3447
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5237406811364682e-05,
+      "loss": 3.4725,
+      "step": 3448
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5234647513138106e-05,
+      "loss": 2.2032,
+      "step": 3449
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52318876658213e-05,
+      "loss": 2.4988,
+      "step": 3450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5229127269703763e-05,
+      "loss": 3.5195,
+      "step": 3451
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5226366325075042e-05,
+      "loss": 2.4379,
+      "step": 3452
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522360483222475e-05,
+      "loss": 2.7698,
+      "step": 3453
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5220842791442558e-05,
+      "loss": 2.7389,
+      "step": 3454
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5218080203018181e-05,
+      "loss": 0.9928,
+      "step": 3455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5215317067241415e-05,
+      "loss": 2.2111,
+      "step": 3456
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5212553384402088e-05,
+      "loss": 2.9419,
+      "step": 3457
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5209789154790107e-05,
+      "loss": 2.7207,
+      "step": 3458
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5207024378695423e-05,
+      "loss": 3.0205,
+      "step": 3459
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5204259056408046e-05,
+      "loss": 1.8854,
+      "step": 3460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520149318821805e-05,
+      "loss": 2.5529,
+      "step": 3461
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519872677441556e-05,
+      "loss": 2.4516,
+      "step": 3462
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5195959815290758e-05,
+      "loss": 3.5659,
+      "step": 3463
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5193192311133884e-05,
+      "loss": 2.7978,
+      "step": 3464
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5190424262235241e-05,
+      "loss": 2.1449,
+      "step": 3465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5187655668885187e-05,
+      "loss": 2.1771,
+      "step": 3466
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5184886531374123e-05,
+      "loss": 2.3882,
+      "step": 3467
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5182116849992528e-05,
+      "loss": 3.3804,
+      "step": 3468
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5179346625030929e-05,
+      "loss": 3.1883,
+      "step": 3469
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5176575856779904e-05,
+      "loss": 3.3945,
+      "step": 3470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5173804545530097e-05,
+      "loss": 2.7835,
+      "step": 3471
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5171032691572207e-05,
+      "loss": 2.9153,
+      "step": 3472
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5168260295196988e-05,
+      "loss": 2.6585,
+      "step": 3473
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5165487356695247e-05,
+      "loss": 2.0611,
+      "step": 3474
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516271387635786e-05,
+      "loss": 3.0626,
+      "step": 3475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5159939854475743e-05,
+      "loss": 1.9294,
+      "step": 3476
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515716529133989e-05,
+      "loss": 2.5351,
+      "step": 3477
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5154390187241328e-05,
+      "loss": 2.3818,
+      "step": 3478
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515161454247116e-05,
+      "loss": 2.6264,
+      "step": 3479
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5148838357320537e-05,
+      "loss": 4.0455,
+      "step": 3480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5146061632080665e-05,
+      "loss": 2.0705,
+      "step": 3481
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5143284367042809e-05,
+      "loss": 2.6713,
+      "step": 3482
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5140506562498297e-05,
+      "loss": 2.2945,
+      "step": 3483
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5137728218738504e-05,
+      "loss": 2.953,
+      "step": 3484
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5134949336054866e-05,
+      "loss": 2.182,
+      "step": 3485
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5132169914738875e-05,
+      "loss": 2.6608,
+      "step": 3486
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5129389955082077e-05,
+      "loss": 2.9173,
+      "step": 3487
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512660945737608e-05,
+      "loss": 2.4396,
+      "step": 3488
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5123828421912545e-05,
+      "loss": 2.5266,
+      "step": 3489
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512104684898319e-05,
+      "loss": 2.1947,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5118264738879785e-05,
+      "loss": 2.2062,
+      "step": 3491
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5115482091894164e-05,
+      "loss": 2.192,
+      "step": 3492
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5112698908318213e-05,
+      "loss": 2.8276,
+      "step": 3493
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5109915188443877e-05,
+      "loss": 1.3079,
+      "step": 3494
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5107130932563151e-05,
+      "loss": 3.6476,
+      "step": 3495
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5104346140968096e-05,
+      "loss": 3.3612,
+      "step": 3496
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5101560813950822e-05,
+      "loss": 3.3028,
+      "step": 3497
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5098774951803492e-05,
+      "loss": 2.5325,
+      "step": 3498
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5095988554818335e-05,
+      "loss": 4.1594,
+      "step": 3499
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5093201623287631e-05,
+      "loss": 3.0968,
+      "step": 3500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5090414157503715e-05,
+      "loss": 3.0816,
+      "step": 3501
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5087626157758979e-05,
+      "loss": 3.1677,
+      "step": 3502
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5084837624345869e-05,
+      "loss": 3.4583,
+      "step": 3503
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5082048557556892e-05,
+      "loss": 3.7937,
+      "step": 3504
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507925895768461e-05,
+      "loss": 3.0102,
+      "step": 3505
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5076468825021637e-05,
+      "loss": 3.3023,
+      "step": 3506
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5073678159860646e-05,
+      "loss": 2.02,
+      "step": 3507
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507088696249436e-05,
+      "loss": 2.7751,
+      "step": 3508
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068095233215569e-05,
+      "loss": 2.4186,
+      "step": 3509
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5065302972317108e-05,
+      "loss": 3.5834,
+      "step": 3510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5062510180091874e-05,
+      "loss": 2.5676,
+      "step": 3511
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505971685683282e-05,
+      "loss": 2.6836,
+      "step": 3512
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505692300283295e-05,
+      "loss": 3.0592,
+      "step": 3513
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5054128618385324e-05,
+      "loss": 3.5692,
+      "step": 3514
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5051333703783069e-05,
+      "loss": 2.2456,
+      "step": 3515
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5048538259319347e-05,
+      "loss": 4.6569,
+      "step": 3516
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045742285287395e-05,
+      "loss": 2.6952,
+      "step": 3517
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5042945781980494e-05,
+      "loss": 2.5306,
+      "step": 3518
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5040148749691983e-05,
+      "loss": 2.858,
+      "step": 3519
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5037351188715265e-05,
+      "loss": 2.7081,
+      "step": 3520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5034553099343785e-05,
+      "loss": 2.6215,
+      "step": 3521
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5031754481871053e-05,
+      "loss": 3.2692,
+      "step": 3522
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5028955336590625e-05,
+      "loss": 3.2483,
+      "step": 3523
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5026155663796123e-05,
+      "loss": 2.8198,
+      "step": 3524
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5023355463781221e-05,
+      "loss": 2.7919,
+      "step": 3525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5020554736839645e-05,
+      "loss": 1.9876,
+      "step": 3526
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5017753483265173e-05,
+      "loss": 4.0886,
+      "step": 3527
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5014951703351655e-05,
+      "loss": 3.9963,
+      "step": 3528
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5012149397392977e-05,
+      "loss": 2.9203,
+      "step": 3529
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5009346565683088e-05,
+      "loss": 2.6229,
+      "step": 3530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5006543208515993e-05,
+      "loss": 2.8565,
+      "step": 3531
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.500373932618575e-05,
+      "loss": 2.096,
+      "step": 3532
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.500093491898648e-05,
+      "loss": 3.1447,
+      "step": 3533
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4998129987212344e-05,
+      "loss": 2.8458,
+      "step": 3534
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4995324531157569e-05,
+      "loss": 3.1548,
+      "step": 3535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4992518551116436e-05,
+      "loss": 2.4789,
+      "step": 3536
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4989712047383276e-05,
+      "loss": 2.9516,
+      "step": 3537
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4986905020252482e-05,
+      "loss": 2.6917,
+      "step": 3538
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4984097470018496e-05,
+      "loss": 2.7637,
+      "step": 3539
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981289396975818e-05,
+      "loss": 2.0428,
+      "step": 3540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4978480801419e-05,
+      "loss": 2.9629,
+      "step": 3541
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497567168364265e-05,
+      "loss": 4.0901,
+      "step": 3542
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4972862043941436e-05,
+      "loss": 1.849,
+      "step": 3543
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4970051882610073e-05,
+      "loss": 3.7625,
+      "step": 3544
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4967241199943332e-05,
+      "loss": 2.5814,
+      "step": 3545
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4964429996236045e-05,
+      "loss": 2.0232,
+      "step": 3546
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4961618271783091e-05,
+      "loss": 3.0673,
+      "step": 3547
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4958806026879411e-05,
+      "loss": 0.9826,
+      "step": 3548
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4955993261819988e-05,
+      "loss": 1.3833,
+      "step": 3549
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4953179976899878e-05,
+      "loss": 1.4546,
+      "step": 3550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4950366172414175e-05,
+      "loss": 2.7492,
+      "step": 3551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4947551848658036e-05,
+      "loss": 3.2115,
+      "step": 3552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4944737005926668e-05,
+      "loss": 2.8425,
+      "step": 3553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4941921644515338e-05,
+      "loss": 1.6244,
+      "step": 3554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4939105764719369e-05,
+      "loss": 2.8499,
+      "step": 3555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4936289366834123e-05,
+      "loss": 3.7317,
+      "step": 3556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4933472451155034e-05,
+      "loss": 2.3769,
+      "step": 3557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4930655017977583e-05,
+      "loss": 3.4593,
+      "step": 3558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4927837067597301e-05,
+      "loss": 1.0481,
+      "step": 3559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4925018600309784e-05,
+      "loss": 3.8083,
+      "step": 3560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4922199616410675e-05,
+      "loss": 1.6758,
+      "step": 3561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4919380116195665e-05,
+      "loss": 3.1416,
+      "step": 3562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4916560099960519e-05,
+      "loss": 3.0156,
+      "step": 3563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913739568001034e-05,
+      "loss": 3.0689,
+      "step": 3564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4910918520613074e-05,
+      "loss": 3.0334,
+      "step": 3565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4908096958092553e-05,
+      "loss": 3.0544,
+      "step": 3566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4905274880735441e-05,
+      "loss": 3.1937,
+      "step": 3567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4902452288837761e-05,
+      "loss": 1.7068,
+      "step": 3568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4899629182695587e-05,
+      "loss": 1.7991,
+      "step": 3569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4896805562605052e-05,
+      "loss": 2.5944,
+      "step": 3570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489398142886234e-05,
+      "loss": 3.2053,
+      "step": 3571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4891156781763692e-05,
+      "loss": 3.662,
+      "step": 3572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4888331621605397e-05,
+      "loss": 2.206,
+      "step": 3573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4885505948683801e-05,
+      "loss": 2.9338,
+      "step": 3574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4882679763295307e-05,
+      "loss": 2.2491,
+      "step": 3575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4879853065736366e-05,
+      "loss": 3.085,
+      "step": 3576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4877025856303486e-05,
+      "loss": 2.6895,
+      "step": 3577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4874198135293232e-05,
+      "loss": 3.0341,
+      "step": 3578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4871369903002211e-05,
+      "loss": 4.0934,
+      "step": 3579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4868541159727097e-05,
+      "loss": 3.1444,
+      "step": 3580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4865711905764613e-05,
+      "loss": 2.1825,
+      "step": 3581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486288214141153e-05,
+      "loss": 2.8206,
+      "step": 3582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4860051866964682e-05,
+      "loss": 3.1325,
+      "step": 3583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485722108272095e-05,
+      "loss": 2.412,
+      "step": 3584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4854389788977266e-05,
+      "loss": 2.1278,
+      "step": 3585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4851557986030626e-05,
+      "loss": 3.1681,
+      "step": 3586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4848725674178069e-05,
+      "loss": 2.9996,
+      "step": 3587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4845892853716692e-05,
+      "loss": 1.8685,
+      "step": 3588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4843059524943644e-05,
+      "loss": 2.2472,
+      "step": 3589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4840225688156132e-05,
+      "loss": 3.8895,
+      "step": 3590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4837391343651409e-05,
+      "loss": 3.3319,
+      "step": 3591
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4834556491726781e-05,
+      "loss": 1.5254,
+      "step": 3592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483172113267962e-05,
+      "loss": 1.9006,
+      "step": 3593
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4828885266807335e-05,
+      "loss": 2.5612,
+      "step": 3594
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4826048894407396e-05,
+      "loss": 4.2983,
+      "step": 3595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482321201577733e-05,
+      "loss": 3.3585,
+      "step": 3596
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4820374631214704e-05,
+      "loss": 2.2584,
+      "step": 3597
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4817536741017153e-05,
+      "loss": 3.9187,
+      "step": 3598
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4814698345482359e-05,
+      "loss": 3.446,
+      "step": 3599
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4811859444908053e-05,
+      "loss": 2.227,
+      "step": 3600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809020039592024e-05,
+      "loss": 3.1755,
+      "step": 3601
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4806180129832108e-05,
+      "loss": 2.259,
+      "step": 3602
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4803339715926206e-05,
+      "loss": 2.9878,
+      "step": 3603
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4800498798172263e-05,
+      "loss": 3.5952,
+      "step": 3604
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4797657376868273e-05,
+      "loss": 3.9106,
+      "step": 3605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479481545231229e-05,
+      "loss": 2.1613,
+      "step": 3606
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4791973024802422e-05,
+      "loss": 3.0485,
+      "step": 3607
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478913009463682e-05,
+      "loss": 3.6614,
+      "step": 3608
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4786286662113704e-05,
+      "loss": 1.3385,
+      "step": 3609
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4783442727531328e-05,
+      "loss": 1.711,
+      "step": 3610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4780598291188011e-05,
+      "loss": 2.9284,
+      "step": 3611
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4777753353382121e-05,
+      "loss": 3.1597,
+      "step": 3612
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4774907914412076e-05,
+      "loss": 2.9795,
+      "step": 3613
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4772061974576353e-05,
+      "loss": 4.1373,
+      "step": 3614
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4769215534173476e-05,
+      "loss": 3.1685,
+      "step": 3615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4766368593502028e-05,
+      "loss": 2.075,
+      "step": 3616
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4763521152860632e-05,
+      "loss": 2.6199,
+      "step": 3617
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760673212547975e-05,
+      "loss": 3.4711,
+      "step": 3618
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4757824772862797e-05,
+      "loss": 2.3835,
+      "step": 3619
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4754975834103877e-05,
+      "loss": 2.0708,
+      "step": 3620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4752126396570064e-05,
+      "loss": 2.6751,
+      "step": 3621
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4749276460560247e-05,
+      "loss": 2.981,
+      "step": 3622
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474642602637337e-05,
+      "loss": 2.385,
+      "step": 3623
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474357509430843e-05,
+      "loss": 2.1663,
+      "step": 3624
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4740723664664483e-05,
+      "loss": 1.4756,
+      "step": 3625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473787173774062e-05,
+      "loss": 1.7244,
+      "step": 3626
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4735019313836008e-05,
+      "loss": 2.1958,
+      "step": 3627
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473216639324984e-05,
+      "loss": 2.3612,
+      "step": 3628
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4729312976281385e-05,
+      "loss": 2.8263,
+      "step": 3629
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4726459063229946e-05,
+      "loss": 2.6446,
+      "step": 3630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4723604654394885e-05,
+      "loss": 2.2817,
+      "step": 3631
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.472074975007562e-05,
+      "loss": 3.5555,
+      "step": 3632
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4717894350571616e-05,
+      "loss": 3.7353,
+      "step": 3633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4715038456182394e-05,
+      "loss": 3.9372,
+      "step": 3634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4712182067207516e-05,
+      "loss": 2.8226,
+      "step": 3635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4709325183946613e-05,
+      "loss": 3.5006,
+      "step": 3636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4706467806699354e-05,
+      "loss": 2.8157,
+      "step": 3637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4703609935765463e-05,
+      "loss": 2.6331,
+      "step": 3638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4700751571444724e-05,
+      "loss": 3.0908,
+      "step": 3639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4697892714036959e-05,
+      "loss": 2.6437,
+      "step": 3640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4695033363842053e-05,
+      "loss": 3.4085,
+      "step": 3641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692173521159936e-05,
+      "loss": 1.7721,
+      "step": 3642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4689313186290594e-05,
+      "loss": 3.2601,
+      "step": 3643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4686452359534067e-05,
+      "loss": 3.2379,
+      "step": 3644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4683591041190433e-05,
+      "loss": 2.6984,
+      "step": 3645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4680729231559835e-05,
+      "loss": 3.4569,
+      "step": 3646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467786693094247e-05,
+      "loss": 2.3854,
+      "step": 3647
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467500413963857e-05,
+      "loss": 1.9695,
+      "step": 3648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4672140857948436e-05,
+      "loss": 2.9448,
+      "step": 3649
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4669277086172406e-05,
+      "loss": 3.5803,
+      "step": 3650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4666412824610883e-05,
+      "loss": 2.5699,
+      "step": 3651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4663548073564316e-05,
+      "loss": 3.1825,
+      "step": 3652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4660682833333195e-05,
+      "loss": 2.1709,
+      "step": 3653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4657817104218075e-05,
+      "loss": 2.7618,
+      "step": 3654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4654950886519563e-05,
+      "loss": 2.6972,
+      "step": 3655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4652084180538304e-05,
+      "loss": 4.3704,
+      "step": 3656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4649216986575007e-05,
+      "loss": 3.2881,
+      "step": 3657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4646349304930426e-05,
+      "loss": 2.4381,
+      "step": 3658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4643481135905367e-05,
+      "loss": 3.4411,
+      "step": 3659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4640612479800686e-05,
+      "loss": 2.658,
+      "step": 3660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4637743336917294e-05,
+      "loss": 2.3485,
+      "step": 3661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4634873707556154e-05,
+      "loss": 2.8074,
+      "step": 3662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4632003592018269e-05,
+      "loss": 3.1663,
+      "step": 3663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4629132990604706e-05,
+      "loss": 2.2092,
+      "step": 3664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4626261903616579e-05,
+      "loss": 3.8998,
+      "step": 3665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4623390331355048e-05,
+      "loss": 2.3305,
+      "step": 3666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.462051827412133e-05,
+      "loss": 1.3489,
+      "step": 3667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4617645732216686e-05,
+      "loss": 2.5601,
+      "step": 3668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4614772705942437e-05,
+      "loss": 2.2927,
+      "step": 3669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611899195599952e-05,
+      "loss": 3.0389,
+      "step": 3670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4609025201490643e-05,
+      "loss": 2.2331,
+      "step": 3671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4606150723915984e-05,
+      "loss": 1.923,
+      "step": 3672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4603275763177491e-05,
+      "loss": 2.665,
+      "step": 3673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4600400319576734e-05,
+      "loss": 2.8248,
+      "step": 3674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4597524393415336e-05,
+      "loss": 3.2217,
+      "step": 3675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4594647984994966e-05,
+      "loss": 3.0187,
+      "step": 3676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4591771094617345e-05,
+      "loss": 2.406,
+      "step": 3677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4588893722584247e-05,
+      "loss": 4.1383,
+      "step": 3678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4586015869197499e-05,
+      "loss": 3.1102,
+      "step": 3679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4583137534758968e-05,
+      "loss": 1.7964,
+      "step": 3680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4580258719570584e-05,
+      "loss": 3.4185,
+      "step": 3681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4577379423934313e-05,
+      "loss": 3.4897,
+      "step": 3682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457449964815219e-05,
+      "loss": 3.4107,
+      "step": 3683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4571619392526279e-05,
+      "loss": 2.8513,
+      "step": 3684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4568738657358715e-05,
+      "loss": 2.9347,
+      "step": 3685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4565857442951668e-05,
+      "loss": 2.8526,
+      "step": 3686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4562975749607366e-05,
+      "loss": 2.5804,
+      "step": 3687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456009357762809e-05,
+      "loss": 3.3445,
+      "step": 3688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4557210927316157e-05,
+      "loss": 2.5679,
+      "step": 3689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455432779897395e-05,
+      "loss": 2.2625,
+      "step": 3690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4551444192903895e-05,
+      "loss": 2.0529,
+      "step": 3691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4548560109408465e-05,
+      "loss": 3.2917,
+      "step": 3692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4545675548790192e-05,
+      "loss": 3.6771,
+      "step": 3693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542790511351652e-05,
+      "loss": 3.6805,
+      "step": 3694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4539904997395468e-05,
+      "loss": 2.3794,
+      "step": 3695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4537019007224324e-05,
+      "loss": 2.3975,
+      "step": 3696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453413254114094e-05,
+      "loss": 2.9303,
+      "step": 3697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4531245599448099e-05,
+      "loss": 3.2844,
+      "step": 3698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4528358182448621e-05,
+      "loss": 3.2072,
+      "step": 3699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4525470290445392e-05,
+      "loss": 2.7969,
+      "step": 3700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4522581923741329e-05,
+      "loss": 3.6907,
+      "step": 3701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4519693082639413e-05,
+      "loss": 3.2838,
+      "step": 3702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4516803767442666e-05,
+      "loss": 1.7982,
+      "step": 3703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4513913978454169e-05,
+      "loss": 2.2649,
+      "step": 3704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4511023715977048e-05,
+      "loss": 2.7797,
+      "step": 3705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4508132980314473e-05,
+      "loss": 2.471,
+      "step": 3706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4505241771769672e-05,
+      "loss": 1.8702,
+      "step": 3707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4502350090645919e-05,
+      "loss": 3.1209,
+      "step": 3708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4499457937246537e-05,
+      "loss": 2.3067,
+      "step": 3709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4496565311874902e-05,
+      "loss": 2.1346,
+      "step": 3710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4493672214834432e-05,
+      "loss": 2.5664,
+      "step": 3711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4490778646428601e-05,
+      "loss": 3.5373,
+      "step": 3712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4487884606960934e-05,
+      "loss": 2.3295,
+      "step": 3713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4484990096735002e-05,
+      "loss": 1.9087,
+      "step": 3714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4482095116054421e-05,
+      "loss": 1.5615,
+      "step": 3715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4479199665222869e-05,
+      "loss": 3.1004,
+      "step": 3716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447630374454406e-05,
+      "loss": 2.0839,
+      "step": 3717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4473407354321763e-05,
+      "loss": 3.23,
+      "step": 3718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4470510494859796e-05,
+      "loss": 2.2424,
+      "step": 3719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4467613166462024e-05,
+      "loss": 1.7999,
+      "step": 3720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464715369432368e-05,
+      "loss": 2.437,
+      "step": 3721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4461817104074788e-05,
+      "loss": 1.9381,
+      "step": 3722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4458918370693306e-05,
+      "loss": 2.2324,
+      "step": 3723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.445601916959198e-05,
+      "loss": 1.9636,
+      "step": 3724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4453119501074924e-05,
+      "loss": 2.7016,
+      "step": 3725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44502193654463e-05,
+      "loss": 2.5685,
+      "step": 3726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4447318763010319e-05,
+      "loss": 1.9297,
+      "step": 3727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4444417694071242e-05,
+      "loss": 2.206,
+      "step": 3728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4441516158933374e-05,
+      "loss": 2.5634,
+      "step": 3729
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4438614157901073e-05,
+      "loss": 3.6546,
+      "step": 3730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4435711691278747e-05,
+      "loss": 2.8197,
+      "step": 3731
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4432808759370853e-05,
+      "loss": 3.0475,
+      "step": 3732
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4429905362481896e-05,
+      "loss": 2.8893,
+      "step": 3733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4427001500916422e-05,
+      "loss": 3.2671,
+      "step": 3734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4424097174979038e-05,
+      "loss": 1.8511,
+      "step": 3735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4421192384974396e-05,
+      "loss": 2.9435,
+      "step": 3736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4418287131207189e-05,
+      "loss": 1.3647,
+      "step": 3737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4415381413982168e-05,
+      "loss": 3.4231,
+      "step": 3738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441247523360413e-05,
+      "loss": 3.096,
+      "step": 3739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409568590377918e-05,
+      "loss": 4.3576,
+      "step": 3740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4406661484608429e-05,
+      "loss": 1.815,
+      "step": 3741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44037539166006e-05,
+      "loss": 2.2759,
+      "step": 3742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4400845886659423e-05,
+      "loss": 3.5906,
+      "step": 3743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439793739508994e-05,
+      "loss": 2.7073,
+      "step": 3744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4395028442197231e-05,
+      "loss": 2.1735,
+      "step": 3745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4392119028286443e-05,
+      "loss": 2.8934,
+      "step": 3746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4389209153662748e-05,
+      "loss": 4.2861,
+      "step": 3747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4386298818631388e-05,
+      "loss": 2.0409,
+      "step": 3748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4383388023497635e-05,
+      "loss": 2.3738,
+      "step": 3749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4380476768566825e-05,
+      "loss": 1.9312,
+      "step": 3750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4377565054144332e-05,
+      "loss": 2.5769,
+      "step": 3751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.437465288053558e-05,
+      "loss": 2.453,
+      "step": 3752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4371740248046046e-05,
+      "loss": 2.9983,
+      "step": 3753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436882715698125e-05,
+      "loss": 2.9489,
+      "step": 3754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4365913607646762e-05,
+      "loss": 2.5475,
+      "step": 3755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4362999600348198e-05,
+      "loss": 1.9152,
+      "step": 3756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4360085135391224e-05,
+      "loss": 2.5287,
+      "step": 3757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4357170213081556e-05,
+      "loss": 2.0051,
+      "step": 3758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4354254833724958e-05,
+      "loss": 2.959,
+      "step": 3759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4351338997627233e-05,
+      "loss": 3.2971,
+      "step": 3760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4348422705094245e-05,
+      "loss": 1.5363,
+      "step": 3761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4345505956431896e-05,
+      "loss": 2.004,
+      "step": 3762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4342588751946141e-05,
+      "loss": 2.3323,
+      "step": 3763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433967109194298e-05,
+      "loss": 1.7797,
+      "step": 3764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433675297672846e-05,
+      "loss": 2.5969,
+      "step": 3765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4333834406608682e-05,
+      "loss": 1.3802,
+      "step": 3766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4330915381889788e-05,
+      "loss": 3.1116,
+      "step": 3767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4327995902877972e-05,
+      "loss": 4.4656,
+      "step": 3768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4325075969879473e-05,
+      "loss": 3.0468,
+      "step": 3769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4322155583200577e-05,
+      "loss": 2.8874,
+      "step": 3770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4319234743147619e-05,
+      "loss": 1.5687,
+      "step": 3771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4316313450026986e-05,
+      "loss": 2.8665,
+      "step": 3772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43133917041451e-05,
+      "loss": 3.3893,
+      "step": 3773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4310469505808447e-05,
+      "loss": 1.9108,
+      "step": 3774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4307546855323549e-05,
+      "loss": 2.1424,
+      "step": 3775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304623752996974e-05,
+      "loss": 3.2611,
+      "step": 3776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430170019913535e-05,
+      "loss": 3.1222,
+      "step": 3777
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4298776194045337e-05,
+      "loss": 2.2703,
+      "step": 3778
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4295851738033655e-05,
+      "loss": 3.5227,
+      "step": 3779
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429292683140706e-05,
+      "loss": 2.6966,
+      "step": 3780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4290001474472366e-05,
+      "loss": 2.8285,
+      "step": 3781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428707566753643e-05,
+      "loss": 3.2459,
+      "step": 3782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428414941090615e-05,
+      "loss": 1.6402,
+      "step": 3783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428122270488848e-05,
+      "loss": 2.8421,
+      "step": 3784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4278295549790419e-05,
+      "loss": 2.8559,
+      "step": 3785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4275367945919013e-05,
+      "loss": 1.9444,
+      "step": 3786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427243989358135e-05,
+      "loss": 2.82,
+      "step": 3787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4269511393084572e-05,
+      "loss": 2.3034,
+      "step": 3788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4266582444735866e-05,
+      "loss": 1.904,
+      "step": 3789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4263653048842461e-05,
+      "loss": 2.7839,
+      "step": 3790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426072320571164e-05,
+      "loss": 3.2418,
+      "step": 3791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4257792915650728e-05,
+      "loss": 3.7648,
+      "step": 3792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254862178967098e-05,
+      "loss": 3.5465,
+      "step": 3793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4251930995968179e-05,
+      "loss": 3.1224,
+      "step": 3794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424899936696143e-05,
+      "loss": 1.6317,
+      "step": 3795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4246067292254367e-05,
+      "loss": 2.6654,
+      "step": 3796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4243134772154554e-05,
+      "loss": 3.3102,
+      "step": 3797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4240201806969594e-05,
+      "loss": 3.6603,
+      "step": 3798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4237268397007145e-05,
+      "loss": 2.3303,
+      "step": 3799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4234334542574906e-05,
+      "loss": 1.7509,
+      "step": 3800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4231400243980623e-05,
+      "loss": 1.877,
+      "step": 3801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4228465501532096e-05,
+      "loss": 2.9956,
+      "step": 3802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422553031553716e-05,
+      "loss": 4.0915,
+      "step": 3803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4222594686303707e-05,
+      "loss": 4.6203,
+      "step": 3804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4219658614139674e-05,
+      "loss": 1.8476,
+      "step": 3805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421672209935303e-05,
+      "loss": 2.4888,
+      "step": 3806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4213785142251813e-05,
+      "loss": 1.4921,
+      "step": 3807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4210847743144087e-05,
+      "loss": 2.3742,
+      "step": 3808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4207909902337978e-05,
+      "loss": 3.6537,
+      "step": 3809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4204971620141648e-05,
+      "loss": 2.5274,
+      "step": 3810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420203289686331e-05,
+      "loss": 2.6161,
+      "step": 3811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4199093732811227e-05,
+      "loss": 2.7231,
+      "step": 3812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4196154128293698e-05,
+      "loss": 3.2595,
+      "step": 3813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4193214083619075e-05,
+      "loss": 1.9733,
+      "step": 3814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4190273599095761e-05,
+      "loss": 3.0857,
+      "step": 3815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4187332675032189e-05,
+      "loss": 3.1661,
+      "step": 3816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4184391311736856e-05,
+      "loss": 3.6602,
+      "step": 3817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4181449509518292e-05,
+      "loss": 3.1023,
+      "step": 3818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4178507268685082e-05,
+      "loss": 3.8626,
+      "step": 3819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4175564589545853e-05,
+      "loss": 2.2847,
+      "step": 3820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417262147240928e-05,
+      "loss": 2.6299,
+      "step": 3821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416967791758408e-05,
+      "loss": 1.9808,
+      "step": 3822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166733925379017e-05,
+      "loss": 1.427,
+      "step": 3823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4163789496102902e-05,
+      "loss": 2.781,
+      "step": 3824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4160844630064596e-05,
+      "loss": 2.7662,
+      "step": 3825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4157899327572999e-05,
+      "loss": 2.4891,
+      "step": 3826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154953588937059e-05,
+      "loss": 3.2999,
+      "step": 3827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4152007414465771e-05,
+      "loss": 2.3873,
+      "step": 3828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4149060804468178e-05,
+      "loss": 2.036,
+      "step": 3829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4146113759253362e-05,
+      "loss": 1.7652,
+      "step": 3830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4143166279130454e-05,
+      "loss": 2.7005,
+      "step": 3831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4140218364408634e-05,
+      "loss": 2.403,
+      "step": 3832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4137270015397122e-05,
+      "loss": 3.2623,
+      "step": 3833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4134321232405185e-05,
+      "loss": 1.8396,
+      "step": 3834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4131372015742141e-05,
+      "loss": 2.7748,
+      "step": 3835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4128422365717346e-05,
+      "loss": 3.3703,
+      "step": 3836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4125472282640204e-05,
+      "loss": 2.011,
+      "step": 3837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4122521766820172e-05,
+      "loss": 2.7059,
+      "step": 3838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4119570818566735e-05,
+      "loss": 2.1401,
+      "step": 3839
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411661943818944e-05,
+      "loss": 2.4085,
+      "step": 3840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4113667625997872e-05,
+      "loss": 2.3924,
+      "step": 3841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4110715382301661e-05,
+      "loss": 2.2693,
+      "step": 3842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4107762707410486e-05,
+      "loss": 3.347,
+      "step": 3843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4104809601634069e-05,
+      "loss": 2.8399,
+      "step": 3844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4101856065282174e-05,
+      "loss": 2.6192,
+      "step": 3845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4098902098664614e-05,
+      "loss": 3.3854,
+      "step": 3846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4095947702091248e-05,
+      "loss": 3.4898,
+      "step": 3847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.409299287587198e-05,
+      "loss": 3.2817,
+      "step": 3848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4090037620316752e-05,
+      "loss": 3.4592,
+      "step": 3849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4087081935735565e-05,
+      "loss": 3.2649,
+      "step": 3850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4084125822438447e-05,
+      "loss": 2.8311,
+      "step": 3851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4081169280735488e-05,
+      "loss": 2.1764,
+      "step": 3852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407821231093681e-05,
+      "loss": 2.632,
+      "step": 3853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407525491335259e-05,
+      "loss": 3.526,
+      "step": 3854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4072297088293043e-05,
+      "loss": 2.8817,
+      "step": 3855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4069338836068434e-05,
+      "loss": 3.2875,
+      "step": 3856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4066380156989063e-05,
+      "loss": 3.0093,
+      "step": 3857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406342105136529e-05,
+      "loss": 2.6099,
+      "step": 3858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4060461519507506e-05,
+      "loss": 2.8443,
+      "step": 3859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4057501561726157e-05,
+      "loss": 2.2828,
+      "step": 3860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405454117833172e-05,
+      "loss": 1.8988,
+      "step": 3861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4051580369634734e-05,
+      "loss": 3.2086,
+      "step": 3862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4048619135945769e-05,
+      "loss": 2.7587,
+      "step": 3863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404565747757545e-05,
+      "loss": 2.7483,
+      "step": 3864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4042695394834435e-05,
+      "loss": 2.1673,
+      "step": 3865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403973288803344e-05,
+      "loss": 3.7769,
+      "step": 3866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4036769957483212e-05,
+      "loss": 2.762,
+      "step": 3867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403380660349455e-05,
+      "loss": 2.2304,
+      "step": 3868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4030842826378297e-05,
+      "loss": 2.0397,
+      "step": 3869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4027878626445339e-05,
+      "loss": 2.4983,
+      "step": 3870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4024914004006607e-05,
+      "loss": 2.8854,
+      "step": 3871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4021948959373075e-05,
+      "loss": 2.6259,
+      "step": 3872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4018983492855767e-05,
+      "loss": 2.4402,
+      "step": 3873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016017604765742e-05,
+      "loss": 1.1609,
+      "step": 3874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4013051295414108e-05,
+      "loss": 3.0337,
+      "step": 3875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4010084565112018e-05,
+      "loss": 2.4776,
+      "step": 3876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400711741417067e-05,
+      "loss": 2.2055,
+      "step": 3877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4004149842901305e-05,
+      "loss": 4.0288,
+      "step": 3878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4001181851615203e-05,
+      "loss": 3.1404,
+      "step": 3879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3998213440623691e-05,
+      "loss": 1.9887,
+      "step": 3880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399524461023815e-05,
+      "loss": 3.6302,
+      "step": 3881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3992275360769989e-05,
+      "loss": 1.8016,
+      "step": 3882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3989305692530675e-05,
+      "loss": 2.7281,
+      "step": 3883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3986335605831707e-05,
+      "loss": 2.4658,
+      "step": 3884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3983365100984633e-05,
+      "loss": 3.0914,
+      "step": 3885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3980394178301051e-05,
+      "loss": 1.7002,
+      "step": 3886
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977422838092592e-05,
+      "loss": 1.5698,
+      "step": 3887
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3974451080670934e-05,
+      "loss": 2.8781,
+      "step": 3888
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3971478906347806e-05,
+      "loss": 2.8416,
+      "step": 3889
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3968506315434973e-05,
+      "loss": 2.5929,
+      "step": 3890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3965533308244248e-05,
+      "loss": 3.9047,
+      "step": 3891
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3962559885087482e-05,
+      "loss": 2.7174,
+      "step": 3892
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3959586046276572e-05,
+      "loss": 2.0764,
+      "step": 3893
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3956611792123469e-05,
+      "loss": 1.9748,
+      "step": 3894
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3953637122940147e-05,
+      "loss": 2.5287,
+      "step": 3895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3950662039038643e-05,
+      "loss": 2.6532,
+      "step": 3896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3947686540731024e-05,
+      "loss": 1.6661,
+      "step": 3897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3944710628329409e-05,
+      "loss": 2.6509,
+      "step": 3898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394173430214596e-05,
+      "loss": 2.3377,
+      "step": 3899
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3938757562492873e-05,
+      "loss": 2.4229,
+      "step": 3900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39357804096824e-05,
+      "loss": 3.5186,
+      "step": 3901
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393280284402683e-05,
+      "loss": 2.0736,
+      "step": 3902
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392982486583849e-05,
+      "loss": 3.4559,
+      "step": 3903
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3926846475429767e-05,
+      "loss": 0.8285,
+      "step": 3904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3923867673113067e-05,
+      "loss": 2.9389,
+      "step": 3905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3920888459200863e-05,
+      "loss": 3.5394,
+      "step": 3906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3917908834005656e-05,
+      "loss": 1.3627,
+      "step": 3907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3914928797839996e-05,
+      "loss": 1.5848,
+      "step": 3908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3911948351016475e-05,
+      "loss": 2.3571,
+      "step": 3909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390896749384773e-05,
+      "loss": 3.6649,
+      "step": 3910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3905986226646435e-05,
+      "loss": 2.338,
+      "step": 3911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3903004549725313e-05,
+      "loss": 2.5615,
+      "step": 3912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3900022463397127e-05,
+      "loss": 4.0789,
+      "step": 3913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3897039967974688e-05,
+      "loss": 2.8019,
+      "step": 3914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894057063770841e-05,
+      "loss": 3.1929,
+      "step": 3915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3891073751098481e-05,
+      "loss": 1.8697,
+      "step": 3916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3888090030270545e-05,
+      "loss": 2.2037,
+      "step": 3917
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3885105901600006e-05,
+      "loss": 3.9043,
+      "step": 3918
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3882121365399894e-05,
+      "loss": 2.8348,
+      "step": 3919
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3879136421983265e-05,
+      "loss": 3.3223,
+      "step": 3920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3876151071663233e-05,
+      "loss": 2.6677,
+      "step": 3921
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3873165314752941e-05,
+      "loss": 3.6639,
+      "step": 3922
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3870179151565581e-05,
+      "loss": 3.4539,
+      "step": 3923
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3867192582414393e-05,
+      "loss": 2.4492,
+      "step": 3924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3864205607612648e-05,
+      "loss": 3.5689,
+      "step": 3925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3861218227473674e-05,
+      "loss": 2.3209,
+      "step": 3926
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3858230442310828e-05,
+      "loss": 2.852,
+      "step": 3927
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3855242252437511e-05,
+      "loss": 2.1542,
+      "step": 3928
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3852253658167178e-05,
+      "loss": 2.7483,
+      "step": 3929
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3849264659813314e-05,
+      "loss": 2.9699,
+      "step": 3930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384627525768945e-05,
+      "loss": 3.3122,
+      "step": 3931
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3843285452109166e-05,
+      "loss": 2.3995,
+      "step": 3932
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3840295243386073e-05,
+      "loss": 3.7607,
+      "step": 3933
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3837304631833832e-05,
+      "loss": 3.0574,
+      "step": 3934
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3834313617766146e-05,
+      "loss": 1.6994,
+      "step": 3935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3831322201496757e-05,
+      "loss": 2.3451,
+      "step": 3936
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3828330383339453e-05,
+      "loss": 1.2747,
+      "step": 3937
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3825338163608055e-05,
+      "loss": 1.5231,
+      "step": 3938
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3822345542616443e-05,
+      "loss": 2.8119,
+      "step": 3939
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3819352520678519e-05,
+      "loss": 3.061,
+      "step": 3940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816359098108243e-05,
+      "loss": 1.5624,
+      "step": 3941
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.381336527521961e-05,
+      "loss": 2.6587,
+      "step": 3942
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3810371052326656e-05,
+      "loss": 3.8531,
+      "step": 3943
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3807376429743467e-05,
+      "loss": 0.8311,
+      "step": 3944
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.380438140778416e-05,
+      "loss": 3.2191,
+      "step": 3945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38013859867629e-05,
+      "loss": 2.0626,
+      "step": 3946
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3798390166993892e-05,
+      "loss": 2.3393,
+      "step": 3947
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3795393948791382e-05,
+      "loss": 2.1412,
+      "step": 3948
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3792397332469667e-05,
+      "loss": 1.6951,
+      "step": 3949
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378940031834307e-05,
+      "loss": 3.2293,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3786402906725966e-05,
+      "loss": 2.3991,
+      "step": 3951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3783405097932772e-05,
+      "loss": 1.6585,
+      "step": 3952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378040689227794e-05,
+      "loss": 2.6889,
+      "step": 3953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377740829007597e-05,
+      "loss": 2.8909,
+      "step": 3954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3774409291641407e-05,
+      "loss": 2.6248,
+      "step": 3955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3771409897288823e-05,
+      "loss": 2.3864,
+      "step": 3956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3768410107332846e-05,
+      "loss": 3.468,
+      "step": 3957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3765409922088137e-05,
+      "loss": 2.9747,
+      "step": 3958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3762409341869404e-05,
+      "loss": 4.1147,
+      "step": 3959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3759408366991391e-05,
+      "loss": 2.544,
+      "step": 3960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375640699776889e-05,
+      "loss": 2.9492,
+      "step": 3961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375340523451673e-05,
+      "loss": 3.2098,
+      "step": 3962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375040307754978e-05,
+      "loss": 3.0006,
+      "step": 3963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3747400527182952e-05,
+      "loss": 1.9992,
+      "step": 3964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3744397583731204e-05,
+      "loss": 4.2722,
+      "step": 3965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3741394247509524e-05,
+      "loss": 2.2074,
+      "step": 3966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3738390518832955e-05,
+      "loss": 2.6354,
+      "step": 3967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373538639801657e-05,
+      "loss": 3.1622,
+      "step": 3968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3732381885375488e-05,
+      "loss": 2.6254,
+      "step": 3969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3729376981224869e-05,
+      "loss": 3.4196,
+      "step": 3970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726371685879911e-05,
+      "loss": 2.7321,
+      "step": 3971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3723365999655859e-05,
+      "loss": 2.218,
+      "step": 3972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3720359922867996e-05,
+      "loss": 2.6382,
+      "step": 3973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3717353455831643e-05,
+      "loss": 4.1806,
+      "step": 3974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3714346598862168e-05,
+      "loss": 2.74,
+      "step": 3975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3711339352274969e-05,
+      "loss": 3.192,
+      "step": 3976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3708331716385497e-05,
+      "loss": 2.3584,
+      "step": 3977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.370532369150924e-05,
+      "loss": 3.3954,
+      "step": 3978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3702315277961724e-05,
+      "loss": 2.2172,
+      "step": 3979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3699306476058523e-05,
+      "loss": 2.2204,
+      "step": 3980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3696297286115237e-05,
+      "loss": 3.3134,
+      "step": 3981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3693287708447523e-05,
+      "loss": 3.3776,
+      "step": 3982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3690277743371065e-05,
+      "loss": 2.1765,
+      "step": 3983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3687267391201604e-05,
+      "loss": 2.1779,
+      "step": 3984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3684256652254906e-05,
+      "loss": 2.6036,
+      "step": 3985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3681245526846782e-05,
+      "loss": 2.6875,
+      "step": 3986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3678234015293089e-05,
+      "loss": 3.295,
+      "step": 3987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3675222117909716e-05,
+      "loss": 2.6945,
+      "step": 3988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3672209835012602e-05,
+      "loss": 1.9947,
+      "step": 3989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3669197166917723e-05,
+      "loss": 2.7219,
+      "step": 3990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3666184113941086e-05,
+      "loss": 2.269,
+      "step": 3991
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366317067639875e-05,
+      "loss": 2.4665,
+      "step": 3992
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3660156854606819e-05,
+      "loss": 2.2943,
+      "step": 3993
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3657142648881414e-05,
+      "loss": 2.9592,
+      "step": 3994
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365412805953872e-05,
+      "loss": 3.6214,
+      "step": 3995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3651113086894951e-05,
+      "loss": 2.1907,
+      "step": 3996
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3648097731266368e-05,
+      "loss": 2.9655,
+      "step": 3997
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3645081992969262e-05,
+      "loss": 2.4981,
+      "step": 3998
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3642065872319971e-05,
+      "loss": 1.4537,
+      "step": 3999
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3639049369634878e-05,
+      "loss": 2.4,
+      "step": 4000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363603248523039e-05,
+      "loss": 1.7265,
+      "step": 4001
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3633015219422974e-05,
+      "loss": 2.3497,
+      "step": 4002
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.362999757252912e-05,
+      "loss": 1.4202,
+      "step": 4003
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3626979544865369e-05,
+      "loss": 2.5425,
+      "step": 4004
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3623961136748296e-05,
+      "loss": 3.5722,
+      "step": 4005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3620942348494518e-05,
+      "loss": 2.6536,
+      "step": 4006
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3617923180420696e-05,
+      "loss": 2.274,
+      "step": 4007
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3614903632843523e-05,
+      "loss": 2.2875,
+      "step": 4008
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3611883706079735e-05,
+      "loss": 2.6565,
+      "step": 4009
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3608863400446113e-05,
+      "loss": 2.1311,
+      "step": 4010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3605842716259465e-05,
+      "loss": 2.1584,
+      "step": 4011
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3602821653836654e-05,
+      "loss": 2.497,
+      "step": 4012
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3599800213494571e-05,
+      "loss": 3.2175,
+      "step": 4013
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3596778395550154e-05,
+      "loss": 2.0342,
+      "step": 4014
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3593756200320373e-05,
+      "loss": 2.2011,
+      "step": 4015
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3590733628122253e-05,
+      "loss": 1.8901,
+      "step": 4016
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3587710679272833e-05,
+      "loss": 1.904,
+      "step": 4017
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3584687354089222e-05,
+      "loss": 3.5042,
+      "step": 4018
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3581663652888536e-05,
+      "loss": 3.1161,
+      "step": 4019
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357863957598796e-05,
+      "loss": 3.7657,
+      "step": 4020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3575615123704703e-05,
+      "loss": 2.6986,
+      "step": 4021
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3572590296356012e-05,
+      "loss": 2.2217,
+      "step": 4022
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356956509425918e-05,
+      "loss": 2.6674,
+      "step": 4023
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3566539517731536e-05,
+      "loss": 2.8075,
+      "step": 4024
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356351356709045e-05,
+      "loss": 1.7528,
+      "step": 4025
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3560487242653331e-05,
+      "loss": 3.5439,
+      "step": 4026
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3557460544737624e-05,
+      "loss": 2.7154,
+      "step": 4027
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3554433473660818e-05,
+      "loss": 2.6589,
+      "step": 4028
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3551406029740435e-05,
+      "loss": 2.6313,
+      "step": 4029
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3548378213294042e-05,
+      "loss": 2.7835,
+      "step": 4030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.354535002463924e-05,
+      "loss": 3.5955,
+      "step": 4031
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.354232146409368e-05,
+      "loss": 2.6063,
+      "step": 4032
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3539292531975038e-05,
+      "loss": 3.0682,
+      "step": 4033
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3536263228601036e-05,
+      "loss": 2.51,
+      "step": 4034
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3533233554289433e-05,
+      "loss": 2.9806,
+      "step": 4035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353020350935803e-05,
+      "loss": 2.6244,
+      "step": 4036
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3527173094124659e-05,
+      "loss": 1.858,
+      "step": 4037
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3524142308907205e-05,
+      "loss": 0.9621,
+      "step": 4038
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3521111154023573e-05,
+      "loss": 2.1485,
+      "step": 4039
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3518079629791725e-05,
+      "loss": 1.4319,
+      "step": 4040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3515047736529656e-05,
+      "loss": 2.6013,
+      "step": 4041
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3512015474555389e-05,
+      "loss": 2.7146,
+      "step": 4042
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3508982844187003e-05,
+      "loss": 2.5902,
+      "step": 4043
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3505949845742599e-05,
+      "loss": 4.9539,
+      "step": 4044
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3502916479540327e-05,
+      "loss": 3.0014,
+      "step": 4045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3499882745898376e-05,
+      "loss": 2.9571,
+      "step": 4046
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3496848645134967e-05,
+      "loss": 3.3634,
+      "step": 4047
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3493814177568365e-05,
+      "loss": 4.241,
+      "step": 4048
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.349077934351687e-05,
+      "loss": 2.8224,
+      "step": 4049
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3487744143298822e-05,
+      "loss": 3.706,
+      "step": 4050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3484708577232606e-05,
+      "loss": 3.4504,
+      "step": 4051
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3481672645636627e-05,
+      "loss": 2.7443,
+      "step": 4052
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347863634882935e-05,
+      "loss": 2.4018,
+      "step": 4053
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3475599687129265e-05,
+      "loss": 2.9574,
+      "step": 4054
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3472562660854902e-05,
+      "loss": 2.8611,
+      "step": 4055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3469525270324835e-05,
+      "loss": 2.9369,
+      "step": 4056
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3466487515857668e-05,
+      "loss": 1.0689,
+      "step": 4057
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3463449397772045e-05,
+      "loss": 3.5295,
+      "step": 4058
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3460410916386661e-05,
+      "loss": 1.961,
+      "step": 4059
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345737207202023e-05,
+      "loss": 2.8817,
+      "step": 4060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3454332864991515e-05,
+      "loss": 2.3059,
+      "step": 4061
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3451293295619312e-05,
+      "loss": 2.8043,
+      "step": 4062
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3448253364222462e-05,
+      "loss": 2.0804,
+      "step": 4063
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3445213071119841e-05,
+      "loss": 2.3644,
+      "step": 4064
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3442172416630355e-05,
+      "loss": 3.2877,
+      "step": 4065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3439131401072958e-05,
+      "loss": 2.7914,
+      "step": 4066
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3436090024766638e-05,
+      "loss": 2.1424,
+      "step": 4067
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3433048288030424e-05,
+      "loss": 3.3763,
+      "step": 4068
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3430006191183378e-05,
+      "loss": 3.0978,
+      "step": 4069
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3426963734544601e-05,
+      "loss": 2.2493,
+      "step": 4070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3423920918433234e-05,
+      "loss": 2.4145,
+      "step": 4071
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342087774316845e-05,
+      "loss": 2.0705,
+      "step": 4072
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3417834209069472e-05,
+      "loss": 1.7206,
+      "step": 4073
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3414790316455546e-05,
+      "loss": 1.3244,
+      "step": 4074
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3411746065645961e-05,
+      "loss": 2.1948,
+      "step": 4075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3408701456960052e-05,
+      "loss": 1.7613,
+      "step": 4076
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340565649071718e-05,
+      "loss": 2.8281,
+      "step": 4077
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3402611167236748e-05,
+      "loss": 3.2285,
+      "step": 4078
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33995654868382e-05,
+      "loss": 2.3978,
+      "step": 4079
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3396519449841006e-05,
+      "loss": 1.8061,
+      "step": 4080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393473056564685e-05,
+      "loss": 3.1117,
+      "step": 4081
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3390426307328794e-05,
+      "loss": 2.9524,
+      "step": 4082
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3387379202452917e-05,
+      "loss": 2.9322,
+      "step": 4083
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338433174225668e-05,
+      "loss": 3.277,
+      "step": 4084
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3381283927059751e-05,
+      "loss": 2.7206,
+      "step": 4085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3378235757181834e-05,
+      "loss": 2.3219,
+      "step": 4086
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3375187232942666e-05,
+      "loss": 1.9142,
+      "step": 4087
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3372138354662018e-05,
+      "loss": 1.6462,
+      "step": 4088
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336908912265971e-05,
+      "loss": 4.5298,
+      "step": 4089
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3366039537255589e-05,
+      "loss": 3.3467,
+      "step": 4090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3362989598769543e-05,
+      "loss": 3.1936,
+      "step": 4091
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3359939307521494e-05,
+      "loss": 2.5303,
+      "step": 4092
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3356888663831406e-05,
+      "loss": 3.7746,
+      "step": 4093
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335383766801928e-05,
+      "loss": 2.6782,
+      "step": 4094
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3350786320405145e-05,
+      "loss": 2.9103,
+      "step": 4095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3347734621309076e-05,
+      "loss": 2.521,
+      "step": 4096
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3344682571051187e-05,
+      "loss": 2.4869,
+      "step": 4097
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3341630169951616e-05,
+      "loss": 2.6919,
+      "step": 4098
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3338577418330552e-05,
+      "loss": 2.9981,
+      "step": 4099
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3335524316508208e-05,
+      "loss": 2.9833,
+      "step": 4100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3332470864804845e-05,
+      "loss": 3.9292,
+      "step": 4101
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3329417063540756e-05,
+      "loss": 2.5812,
+      "step": 4102
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3326362913036266e-05,
+      "loss": 3.2107,
+      "step": 4103
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3323308413611748e-05,
+      "loss": 3.276,
+      "step": 4104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3320253565587602e-05,
+      "loss": 2.7774,
+      "step": 4105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3317198369284264e-05,
+      "loss": 2.4085,
+      "step": 4106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3314142825022213e-05,
+      "loss": 1.811,
+      "step": 4107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3311086933121961e-05,
+      "loss": 2.5842,
+      "step": 4108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330803069390406e-05,
+      "loss": 2.4394,
+      "step": 4109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3304974107689088e-05,
+      "loss": 2.7063,
+      "step": 4110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3301917174797673e-05,
+      "loss": 2.8553,
+      "step": 4111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3298859895550473e-05,
+      "loss": 2.9238,
+      "step": 4112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3295802270268177e-05,
+      "loss": 4.4395,
+      "step": 4113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.329274429927152e-05,
+      "loss": 3.4551,
+      "step": 4114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328968598288127e-05,
+      "loss": 3.0799,
+      "step": 4115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3286627321418229e-05,
+      "loss": 2.4501,
+      "step": 4116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3283568315203232e-05,
+      "loss": 3.4632,
+      "step": 4117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3280508964557162e-05,
+      "loss": 2.4357,
+      "step": 4118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3277449269800924e-05,
+      "loss": 3.0953,
+      "step": 4119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3274389231255466e-05,
+      "loss": 3.1464,
+      "step": 4120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.327132884924178e-05,
+      "loss": 3.1514,
+      "step": 4121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3268268124080881e-05,
+      "loss": 2.8368,
+      "step": 4122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326520705609382e-05,
+      "loss": 2.8129,
+      "step": 4123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3262145645601693e-05,
+      "loss": 2.8745,
+      "step": 4124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3259083892925633e-05,
+      "loss": 2.585,
+      "step": 4125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3256021798386795e-05,
+      "loss": 3.8321,
+      "step": 4126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3252959362306381e-05,
+      "loss": 2.1506,
+      "step": 4127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3249896585005628e-05,
+      "loss": 3.2693,
+      "step": 4128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3246833466805807e-05,
+      "loss": 2.5443,
+      "step": 4129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3243770008028225e-05,
+      "loss": 1.8885,
+      "step": 4130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3240706208994226e-05,
+      "loss": 2.4593,
+      "step": 4131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3237642070025183e-05,
+      "loss": 3.496,
+      "step": 4132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3234577591442515e-05,
+      "loss": 2.2524,
+      "step": 4133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3231512773567667e-05,
+      "loss": 2.9292,
+      "step": 4134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3228447616722128e-05,
+      "loss": 3.0053,
+      "step": 4135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.322538212122742e-05,
+      "loss": 2.2573,
+      "step": 4136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3222316287405095e-05,
+      "loss": 1.8438,
+      "step": 4137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3219250115576745e-05,
+      "loss": 2.1955,
+      "step": 4138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3216183606064e-05,
+      "loss": 2.7032,
+      "step": 4139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3213116759188525e-05,
+      "loss": 3.7426,
+      "step": 4140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321004957527201e-05,
+      "loss": 2.9071,
+      "step": 4141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3206982054636195e-05,
+      "loss": 2.1418,
+      "step": 4142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3203914197602847e-05,
+      "loss": 3.4107,
+      "step": 4143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.320084600449377e-05,
+      "loss": 2.094,
+      "step": 4144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.31977774756308e-05,
+      "loss": 1.7342,
+      "step": 4145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3194708611335816e-05,
+      "loss": 2.0292,
+      "step": 4146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3191639411930726e-05,
+      "loss": 2.6754,
+      "step": 4147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3188569877737474e-05,
+      "loss": 2.4195,
+      "step": 4148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3185500009078038e-05,
+      "loss": 2.3132,
+      "step": 4149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3182429806274442e-05,
+      "loss": 2.7813,
+      "step": 4150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3179359269648723e-05,
+      "loss": 1.313,
+      "step": 4151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3176288399522975e-05,
+      "loss": 3.4559,
+      "step": 4152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173217196219317e-05,
+      "loss": 1.7992,
+      "step": 4153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3170145660059898e-05,
+      "loss": 3.9103,
+      "step": 4154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3167073791366915e-05,
+      "loss": 3.0932,
+      "step": 4155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3164001590462592e-05,
+      "loss": 2.6562,
+      "step": 4156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3160929057669183e-05,
+      "loss": 3.2014,
+      "step": 4157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3157856193308988e-05,
+      "loss": 2.7969,
+      "step": 4158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3154782997704336e-05,
+      "loss": 2.8077,
+      "step": 4159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3151709471177589e-05,
+      "loss": 2.1286,
+      "step": 4160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3148635614051143e-05,
+      "loss": 1.9286,
+      "step": 4161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3145561426647438e-05,
+      "loss": 2.3868,
+      "step": 4162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3142486909288934e-05,
+      "loss": 2.1988,
+      "step": 4163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3139412062298141e-05,
+      "loss": 2.0501,
+      "step": 4164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3136336885997591e-05,
+      "loss": 2.3016,
+      "step": 4165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3133261380709861e-05,
+      "loss": 1.3307,
+      "step": 4166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3130185546757552e-05,
+      "loss": 3.1208,
+      "step": 4167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312710938446331e-05,
+      "loss": 1.9424,
+      "step": 4168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3124032894149803e-05,
+      "loss": 2.2257,
+      "step": 4169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120956076139746e-05,
+      "loss": 2.7768,
+      "step": 4170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3117878930755881e-05,
+      "loss": 2.9069,
+      "step": 4171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3114801458320988e-05,
+      "loss": 2.7679,
+      "step": 4172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3111723659157875e-05,
+      "loss": 2.0457,
+      "step": 4173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3108645533589394e-05,
+      "loss": 3.1562,
+      "step": 4174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3105567081938423e-05,
+      "loss": 2.2143,
+      "step": 4175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3102488304527883e-05,
+      "loss": 4.0327,
+      "step": 4176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3099409201680715e-05,
+      "loss": 2.6295,
+      "step": 4177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309632977371991e-05,
+      "loss": 2.373,
+      "step": 4178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3093250020968477e-05,
+      "loss": 3.5419,
+      "step": 4179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 2.6011,
+      "step": 4180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308708954238599e-05,
+      "loss": 2.4132,
+      "step": 4181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3084008817201137e-05,
+      "loss": 3.0307,
+      "step": 4182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3080927768518074e-05,
+      "loss": 3.0142,
+      "step": 4183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3077846396659986e-05,
+      "loss": 3.0205,
+      "step": 4184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3074764701950095e-05,
+      "loss": 2.5878,
+      "step": 4185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3071682684711661e-05,
+      "loss": 3.5224,
+      "step": 4186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3068600345267965e-05,
+      "loss": 2.3289,
+      "step": 4187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3065517683942339e-05,
+      "loss": 3.0705,
+      "step": 4188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3062434701058134e-05,
+      "loss": 2.9481,
+      "step": 4189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.305935139693874e-05,
+      "loss": 1.9771,
+      "step": 4190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3056267771907583e-05,
+      "loss": 2.4528,
+      "step": 4191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3053183826288124e-05,
+      "loss": 3.9685,
+      "step": 4192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.305009956040385e-05,
+      "loss": 2.5626,
+      "step": 4193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304701497457829e-05,
+      "loss": 0.8877,
+      "step": 4194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3043930069134998e-05,
+      "loss": 2.7703,
+      "step": 4195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040844844397573e-05,
+      "loss": 3.4839,
+      "step": 4196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3037759300689632e-05,
+      "loss": 3.3659,
+      "step": 4197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3034673438334841e-05,
+      "loss": 2.9666,
+      "step": 4198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3031587257656892e-05,
+      "loss": 2.3628,
+      "step": 4199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3028500758979507e-05,
+      "loss": 1.5914,
+      "step": 4200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3025413942626453e-05,
+      "loss": 3.3612,
+      "step": 4201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3022326808921513e-05,
+      "loss": 2.8712,
+      "step": 4202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301923935818852e-05,
+      "loss": 2.9326,
+      "step": 4203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016151590751332e-05,
+      "loss": 3.1,
+      "step": 4204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3013063506933838e-05,
+      "loss": 2.2387,
+      "step": 4205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300997510705997e-05,
+      "loss": 2.1093,
+      "step": 4206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3006886391453679e-05,
+      "loss": 2.6526,
+      "step": 4207
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3003797360438961e-05,
+      "loss": 2.5893,
+      "step": 4208
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3000708014339842e-05,
+      "loss": 2.809,
+      "step": 4209
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299761835348038e-05,
+      "loss": 2.6546,
+      "step": 4210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2994528378184663e-05,
+      "loss": 2.6554,
+      "step": 4211
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2991438088776818e-05,
+      "loss": 2.0679,
+      "step": 4212
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2988347485581003e-05,
+      "loss": 2.477,
+      "step": 4213
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2985256568921404e-05,
+      "loss": 2.6602,
+      "step": 4214
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2982165339122248e-05,
+      "loss": 3.1791,
+      "step": 4215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2979073796507786e-05,
+      "loss": 2.7642,
+      "step": 4216
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297598194140231e-05,
+      "loss": 3.9382,
+      "step": 4217
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297288977413014e-05,
+      "loss": 2.8871,
+      "step": 4218
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969797295015632e-05,
+      "loss": 2.5337,
+      "step": 4219
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.296670450438317e-05,
+      "loss": 2.336,
+      "step": 4220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2963611402557174e-05,
+      "loss": 1.2403,
+      "step": 4221
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2960517989862096e-05,
+      "loss": 3.0778,
+      "step": 4222
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2957424266622422e-05,
+      "loss": 1.7072,
+      "step": 4223
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2954330233162669e-05,
+      "loss": 1.2907,
+      "step": 4224
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2951235889807386e-05,
+      "loss": 3.1941,
+      "step": 4225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2948141236881154e-05,
+      "loss": 2.5956,
+      "step": 4226
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2945046274708588e-05,
+      "loss": 2.2773,
+      "step": 4227
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2941951003614337e-05,
+      "loss": 2.8883,
+      "step": 4228
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2938855423923081e-05,
+      "loss": 3.2401,
+      "step": 4229
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2935759535959528e-05,
+      "loss": 2.8882,
+      "step": 4230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2932663340048429e-05,
+      "loss": 3.2331,
+      "step": 4231
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2929566836514556e-05,
+      "loss": 2.7827,
+      "step": 4232
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2926470025682715e-05,
+      "loss": 3.41,
+      "step": 4233
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2923372907877752e-05,
+      "loss": 2.9969,
+      "step": 4234
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2920275483424538e-05,
+      "loss": 2.2991,
+      "step": 4235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291717775264798e-05,
+      "loss": 3.2863,
+      "step": 4236
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2914079715873016e-05,
+      "loss": 3.3884,
+      "step": 4237
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2910981373424614e-05,
+      "loss": 2.2197,
+      "step": 4238
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2907882725627776e-05,
+      "loss": 3.2533,
+      "step": 4239
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2904783772807534e-05,
+      "loss": 3.8166,
+      "step": 4240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2901684515288956e-05,
+      "loss": 1.4667,
+      "step": 4241
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2898584953397145e-05,
+      "loss": 2.4845,
+      "step": 4242
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.289548508745722e-05,
+      "loss": 3.3498,
+      "step": 4243
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2892384917794347e-05,
+      "loss": 1.3938,
+      "step": 4244
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2889284444733722e-05,
+      "loss": 1.5264,
+      "step": 4245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2886183668600567e-05,
+      "loss": 2.973,
+      "step": 4246
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2883082589720145e-05,
+      "loss": 2.8025,
+      "step": 4247
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2879981208417735e-05,
+      "loss": 2.6969,
+      "step": 4248
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2876879525018664e-05,
+      "loss": 2.3156,
+      "step": 4249
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2873777539848284e-05,
+      "loss": 2.6043,
+      "step": 4250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2870675253231977e-05,
+      "loss": 2.5387,
+      "step": 4251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2867572665495156e-05,
+      "loss": 2.3821,
+      "step": 4252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2864469776963275e-05,
+      "loss": 2.2994,
+      "step": 4253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286136658796181e-05,
+      "loss": 2.0981,
+      "step": 4254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2858263098816265e-05,
+      "loss": 1.9356,
+      "step": 4255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285515930985219e-05,
+      "loss": 1.7745,
+      "step": 4256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2852055221395156e-05,
+      "loss": 2.4191,
+      "step": 4257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2848950833770764e-05,
+      "loss": 1.8914,
+      "step": 4258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284584614730465e-05,
+      "loss": 3.4788,
+      "step": 4259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2842741162322487e-05,
+      "loss": 2.8661,
+      "step": 4260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2839635879149967e-05,
+      "loss": 2.6194,
+      "step": 4261
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2836530298112824e-05,
+      "loss": 2.2099,
+      "step": 4262
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2833424419536815e-05,
+      "loss": 1.636,
+      "step": 4263
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2830318243747736e-05,
+      "loss": 1.6688,
+      "step": 4264
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.282721177107141e-05,
+      "loss": 1.59,
+      "step": 4265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2824105001833691e-05,
+      "loss": 1.7359,
+      "step": 4266
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2820997936360464e-05,
+      "loss": 2.2232,
+      "step": 4267
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2817890574977648e-05,
+      "loss": 2.4698,
+      "step": 4268
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2814782918011183e-05,
+      "loss": 3.9303,
+      "step": 4269
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2811674965787058e-05,
+      "loss": 1.8923,
+      "step": 4270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2808566718631279e-05,
+      "loss": 3.2296,
+      "step": 4271
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2805458176869885e-05,
+      "loss": 1.724,
+      "step": 4272
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2802349340828948e-05,
+      "loss": 2.2515,
+      "step": 4273
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2799240210834572e-05,
+      "loss": 3.0119,
+      "step": 4274
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.279613078721289e-05,
+      "loss": 2.1322,
+      "step": 4275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2793021070290065e-05,
+      "loss": 2.797,
+      "step": 4276
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2789911060392295e-05,
+      "loss": 3.6488,
+      "step": 4277
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2786800757845802e-05,
+      "loss": 1.1416,
+      "step": 4278
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2783690162976839e-05,
+      "loss": 3.5558,
+      "step": 4279
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2780579276111702e-05,
+      "loss": 1.5711,
+      "step": 4280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2777468097576701e-05,
+      "loss": 2.8658,
+      "step": 4281
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2774356627698193e-05,
+      "loss": 2.2104,
+      "step": 4282
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2771244866802546e-05,
+      "loss": 2.5584,
+      "step": 4283
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2768132815216174e-05,
+      "loss": 2.57,
+      "step": 4284
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.276502047326552e-05,
+      "loss": 3.0889,
+      "step": 4285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2761907841277048e-05,
+      "loss": 1.9939,
+      "step": 4286
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2758794919577264e-05,
+      "loss": 2.3023,
+      "step": 4287
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2755681708492696e-05,
+      "loss": 3.1931,
+      "step": 4288
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2752568208349905e-05,
+      "loss": 1.8589,
+      "step": 4289
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2749454419475486e-05,
+      "loss": 1.5538,
+      "step": 4290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274634034219606e-05,
+      "loss": 3.004,
+      "step": 4291
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2743225976838277e-05,
+      "loss": 2.7875,
+      "step": 4292
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2740111323728818e-05,
+      "loss": 3.6057,
+      "step": 4293
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2736996383194403e-05,
+      "loss": 2.0055,
+      "step": 4294
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273388115556177e-05,
+      "loss": 1.803,
+      "step": 4295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730765641157689e-05,
+      "loss": 3.5741,
+      "step": 4296
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2727649840308965e-05,
+      "loss": 3.1291,
+      "step": 4297
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2724533753342433e-05,
+      "loss": 2.8383,
+      "step": 4298
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2721417380584957e-05,
+      "loss": 2.6876,
+      "step": 4299
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2718300722363431e-05,
+      "loss": 2.3757,
+      "step": 4300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2715183779004772e-05,
+      "loss": 2.6013,
+      "step": 4301
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2712066550835936e-05,
+      "loss": 3.2251,
+      "step": 4302
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2708949038183907e-05,
+      "loss": 3.0442,
+      "step": 4303
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2705831241375695e-05,
+      "loss": 3.3872,
+      "step": 4304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2702713160738344e-05,
+      "loss": 2.7475,
+      "step": 4305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2699594796598927e-05,
+      "loss": 3.356,
+      "step": 4306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2696476149284545e-05,
+      "loss": 1.9369,
+      "step": 4307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2693357219122331e-05,
+      "loss": 3.2038,
+      "step": 4308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.269023800643944e-05,
+      "loss": 2.0924,
+      "step": 4309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2687118511563075e-05,
+      "loss": 2.7571,
+      "step": 4310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2683998734820444e-05,
+      "loss": 2.7576,
+      "step": 4311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2680878676538804e-05,
+      "loss": 2.0889,
+      "step": 4312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2677758337045436e-05,
+      "loss": 2.1188,
+      "step": 4313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2674637716667641e-05,
+      "loss": 1.9422,
+      "step": 4314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2671516815732767e-05,
+      "loss": 2.9394,
+      "step": 4315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2668395634568175e-05,
+      "loss": 3.3689,
+      "step": 4316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2665274173501268e-05,
+      "loss": 3.6388,
+      "step": 4317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266215243285947e-05,
+      "loss": 3.335,
+      "step": 4318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2659030412970236e-05,
+      "loss": 2.9399,
+      "step": 4319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2655908114161053e-05,
+      "loss": 2.1839,
+      "step": 4320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2652785536759438e-05,
+      "loss": 3.4777,
+      "step": 4321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2649662681092926e-05,
+      "loss": 4.0655,
+      "step": 4322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26465395474891e-05,
+      "loss": 3.2256,
+      "step": 4323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2643416136275557e-05,
+      "loss": 3.2282,
+      "step": 4324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2640292447779932e-05,
+      "loss": 3.4568,
+      "step": 4325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2637168482329882e-05,
+      "loss": 3.2309,
+      "step": 4326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2634044240253098e-05,
+      "loss": 1.8423,
+      "step": 4327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2630919721877299e-05,
+      "loss": 3.4586,
+      "step": 4328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262779492753023e-05,
+      "loss": 1.9331,
+      "step": 4329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2624669857539669e-05,
+      "loss": 1.9699,
+      "step": 4330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2621544512233422e-05,
+      "loss": 2.7306,
+      "step": 4331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261841889193932e-05,
+      "loss": 3.1003,
+      "step": 4332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2615292996985234e-05,
+      "loss": 3.0835,
+      "step": 4333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2612166827699049e-05,
+      "loss": 2.0594,
+      "step": 4334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2609040384408685e-05,
+      "loss": 2.5165,
+      "step": 4335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2605913667442096e-05,
+      "loss": 3.3443,
+      "step": 4336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2602786677127256e-05,
+      "loss": 2.5979,
+      "step": 4337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2599659413792176e-05,
+      "loss": 3.0802,
+      "step": 4338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2596531877764887e-05,
+      "loss": 3.4772,
+      "step": 4339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2593404069373452e-05,
+      "loss": 3.3457,
+      "step": 4340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2590275988945967e-05,
+      "loss": 2.7043,
+      "step": 4341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2587147636810558e-05,
+      "loss": 2.3131,
+      "step": 4342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2584019013295365e-05,
+      "loss": 2.6439,
+      "step": 4343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2580890118728572e-05,
+      "loss": 2.3295,
+      "step": 4344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2577760953438382e-05,
+      "loss": 1.5237,
+      "step": 4345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2574631517753036e-05,
+      "loss": 3.3867,
+      "step": 4346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2571501812000789e-05,
+      "loss": 3.6182,
+      "step": 4347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2568371836509936e-05,
+      "loss": 3.2324,
+      "step": 4348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2565241591608798e-05,
+      "loss": 2.1487,
+      "step": 4349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2562111077625723e-05,
+      "loss": 2.0633,
+      "step": 4350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.255898029488909e-05,
+      "loss": 3.0716,
+      "step": 4351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2555849243727298e-05,
+      "loss": 2.2344,
+      "step": 4352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2552717924468783e-05,
+      "loss": 3.7782,
+      "step": 4353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2549586337442006e-05,
+      "loss": 2.7008,
+      "step": 4354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2546454482975454e-05,
+      "loss": 3.6048,
+      "step": 4355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2543322361397648e-05,
+      "loss": 3.3418,
+      "step": 4356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2540189973037127e-05,
+      "loss": 2.2982,
+      "step": 4357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2537057318222468e-05,
+      "loss": 1.9096,
+      "step": 4358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2533924397282268e-05,
+      "loss": 2.3488,
+      "step": 4359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2530791210545163e-05,
+      "loss": 2.0256,
+      "step": 4360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2527657758339807e-05,
+      "loss": 3.2837,
+      "step": 4361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2524524040994881e-05,
+      "loss": 1.9227,
+      "step": 4362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2521390058839097e-05,
+      "loss": 2.1557,
+      "step": 4363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518255812201203e-05,
+      "loss": 2.358,
+      "step": 4364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251512130140996e-05,
+      "loss": 2.191,
+      "step": 4365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2511986526794166e-05,
+      "loss": 2.9047,
+      "step": 4366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2508851488682642e-05,
+      "loss": 1.7607,
+      "step": 4367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2505716187404242e-05,
+      "loss": 2.354,
+      "step": 4368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2502580623287846e-05,
+      "loss": 2.8982,
+      "step": 4369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2499444796662354e-05,
+      "loss": 2.3362,
+      "step": 4370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2496308707856706e-05,
+      "loss": 2.7925,
+      "step": 4371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2493172357199856e-05,
+      "loss": 2.8736,
+      "step": 4372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2490035745020798e-05,
+      "loss": 3.6878,
+      "step": 4373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486898871648552e-05,
+      "loss": 2.2667,
+      "step": 4374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.248376173741215e-05,
+      "loss": 2.5536,
+      "step": 4375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2480624342640673e-05,
+      "loss": 2.0699,
+      "step": 4376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2477486687663212e-05,
+      "loss": 1.8946,
+      "step": 4377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2474348772808897e-05,
+      "loss": 2.5959,
+      "step": 4378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2471210598406882e-05,
+      "loss": 3.0459,
+      "step": 4379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2468072164786342e-05,
+      "loss": 2.4603,
+      "step": 4380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2464933472276488e-05,
+      "loss": 2.7287,
+      "step": 4381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2461794521206554e-05,
+      "loss": 2.1294,
+      "step": 4382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2458655311905798e-05,
+      "loss": 3.5971,
+      "step": 4383
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2455515844703512e-05,
+      "loss": 3.655,
+      "step": 4384
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2452376119929009e-05,
+      "loss": 2.679,
+      "step": 4385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2449236137911634e-05,
+      "loss": 3.5238,
+      "step": 4386
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2446095898980755e-05,
+      "loss": 3.2768,
+      "step": 4387
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2442955403465768e-05,
+      "loss": 2.8468,
+      "step": 4388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.24398146516961e-05,
+      "loss": 2.7282,
+      "step": 4389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2436673644001196e-05,
+      "loss": 2.7639,
+      "step": 4390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2433532380710535e-05,
+      "loss": 3.7512,
+      "step": 4391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2430390862153625e-05,
+      "loss": 1.6872,
+      "step": 4392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2427249088659992e-05,
+      "loss": 3.1654,
+      "step": 4393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2424107060559194e-05,
+      "loss": 2.2971,
+      "step": 4394
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2420964778180815e-05,
+      "loss": 3.155,
+      "step": 4395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2417822241854466e-05,
+      "loss": 2.4996,
+      "step": 4396
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241467945190979e-05,
+      "loss": 2.5551,
+      "step": 4397
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2411536408676443e-05,
+      "loss": 1.1093,
+      "step": 4398
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2408393112484123e-05,
+      "loss": 3.3535,
+      "step": 4399
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2405249563662539e-05,
+      "loss": 3.7288,
+      "step": 4400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240210576254144e-05,
+      "loss": 2.6465,
+      "step": 4401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2398961709450593e-05,
+      "loss": 3.1006,
+      "step": 4402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2395817404719797e-05,
+      "loss": 3.2429,
+      "step": 4403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392672848678877e-05,
+      "loss": 2.076,
+      "step": 4404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2389528041657679e-05,
+      "loss": 3.9665,
+      "step": 4405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2386382983986079e-05,
+      "loss": 2.5513,
+      "step": 4406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2383237675993983e-05,
+      "loss": 2.6182,
+      "step": 4407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238009211801131e-05,
+      "loss": 2.6435,
+      "step": 4408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2376946310368023e-05,
+      "loss": 2.1026,
+      "step": 4409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.23738002533941e-05,
+      "loss": 3.1431,
+      "step": 4410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2370653947419548e-05,
+      "loss": 2.2971,
+      "step": 4411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2367507392774398e-05,
+      "loss": 2.4238,
+      "step": 4412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2364360589788711e-05,
+      "loss": 1.8566,
+      "step": 4413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236121353879257e-05,
+      "loss": 1.8724,
+      "step": 4414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2358066240116092e-05,
+      "loss": 3.0186,
+      "step": 4415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2354918694089406e-05,
+      "loss": 3.5615,
+      "step": 4416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235177090104268e-05,
+      "loss": 2.4313,
+      "step": 4417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.23486228613061e-05,
+      "loss": 1.6526,
+      "step": 4418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2345474575209884e-05,
+      "loss": 3.4342,
+      "step": 4419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2342326043084268e-05,
+      "loss": 3.44,
+      "step": 4420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233917726525952e-05,
+      "loss": 3.4468,
+      "step": 4421
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2336028242065938e-05,
+      "loss": 3.4855,
+      "step": 4422
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2332878973833834e-05,
+      "loss": 3.9563,
+      "step": 4423
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2329729460893552e-05,
+      "loss": 3.1401,
+      "step": 4424
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2326579703575464e-05,
+      "loss": 3.0278,
+      "step": 4425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232342970220996e-05,
+      "loss": 2.2173,
+      "step": 4426
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2320279457127467e-05,
+      "loss": 3.5466,
+      "step": 4427
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2317128968658424e-05,
+      "loss": 2.0506,
+      "step": 4428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2313978237133308e-05,
+      "loss": 3.1365,
+      "step": 4429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2310827262882614e-05,
+      "loss": 2.289,
+      "step": 4430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2307676046236866e-05,
+      "loss": 3.1304,
+      "step": 4431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2304524587526609e-05,
+      "loss": 1.4645,
+      "step": 4432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.230137288708242e-05,
+      "loss": 2.0287,
+      "step": 4433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2298220945234896e-05,
+      "loss": 2.7704,
+      "step": 4434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2295068762314661e-05,
+      "loss": 3.3415,
+      "step": 4435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2291916338652365e-05,
+      "loss": 1.3991,
+      "step": 4436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2288763674578681e-05,
+      "loss": 2.6813,
+      "step": 4437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2285610770424311e-05,
+      "loss": 2.2377,
+      "step": 4438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228245762651998e-05,
+      "loss": 2.6064,
+      "step": 4439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2279304243196438e-05,
+      "loss": 3.0363,
+      "step": 4440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2276150620784454e-05,
+      "loss": 2.1893,
+      "step": 4441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2272996759614837e-05,
+      "loss": 2.5211,
+      "step": 4442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2269842660018412e-05,
+      "loss": 1.7665,
+      "step": 4443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2266688322326024e-05,
+      "loss": 2.6124,
+      "step": 4444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2263533746868552e-05,
+      "loss": 2.7674,
+      "step": 4445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2260378933976895e-05,
+      "loss": 2.2224,
+      "step": 4446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2257223883981978e-05,
+      "loss": 2.0811,
+      "step": 4447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225406859721475e-05,
+      "loss": 2.198,
+      "step": 4448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225091307400619e-05,
+      "loss": 2.9263,
+      "step": 4449
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2247757314687296e-05,
+      "loss": 2.4008,
+      "step": 4450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.224460131958909e-05,
+      "loss": 3.2673,
+      "step": 4451
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2241445089042623e-05,
+      "loss": 3.5756,
+      "step": 4452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2238288623378971e-05,
+      "loss": 2.6873,
+      "step": 4453
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223513192292923e-05,
+      "loss": 2.9401,
+      "step": 4454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2231974988024522e-05,
+      "loss": 2.2014,
+      "step": 4455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2228817818995998e-05,
+      "loss": 3.8998,
+      "step": 4456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2225660416174827e-05,
+      "loss": 2.4661,
+      "step": 4457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222250277989221e-05,
+      "loss": 3.1271,
+      "step": 4458
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2219344910479368e-05,
+      "loss": 1.464,
+      "step": 4459
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2216186808267544e-05,
+      "loss": 3.8002,
+      "step": 4460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221302847358801e-05,
+      "loss": 2.8787,
+      "step": 4461
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2209869906772058e-05,
+      "loss": 3.1592,
+      "step": 4462
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2206711108151011e-05,
+      "loss": 2.6522,
+      "step": 4463
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2203552078056209e-05,
+      "loss": 2.1887,
+      "step": 4464
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2200392816819022e-05,
+      "loss": 2.618,
+      "step": 4465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2197233324770839e-05,
+      "loss": 2.534,
+      "step": 4466
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219407360224308e-05,
+      "loss": 3.3621,
+      "step": 4467
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2190913649567185e-05,
+      "loss": 3.0891,
+      "step": 4468
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2187753467074613e-05,
+      "loss": 2.9299,
+      "step": 4469
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2184593055096853e-05,
+      "loss": 3.2328,
+      "step": 4470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2181432413965428e-05,
+      "loss": 3.8143,
+      "step": 4471
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2178271544011864e-05,
+      "loss": 2.6699,
+      "step": 4472
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2175110445567722e-05,
+      "loss": 3.4358,
+      "step": 4473
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2171949118964592e-05,
+      "loss": 2.973,
+      "step": 4474
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2168787564534078e-05,
+      "loss": 3.4448,
+      "step": 4475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2165625782607817e-05,
+      "loss": 2.9448,
+      "step": 4476
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2162463773517462e-05,
+      "loss": 2.5664,
+      "step": 4477
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2159301537594691e-05,
+      "loss": 2.6223,
+      "step": 4478
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2156139075171212e-05,
+      "loss": 2.24,
+      "step": 4479
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215297638657875e-05,
+      "loss": 3.3993,
+      "step": 4480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2149813472149059e-05,
+      "loss": 4.1765,
+      "step": 4481
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2146650332213911e-05,
+      "loss": 2.8808,
+      "step": 4482
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2143486967105107e-05,
+      "loss": 1.6999,
+      "step": 4483
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2140323377154467e-05,
+      "loss": 3.618,
+      "step": 4484
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2137159562693839e-05,
+      "loss": 2.9083,
+      "step": 4485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2133995524055094e-05,
+      "loss": 2.5296,
+      "step": 4486
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2130831261570117e-05,
+      "loss": 3.6213,
+      "step": 4487
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2127666775570837e-05,
+      "loss": 2.6927,
+      "step": 4488
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2124502066389181e-05,
+      "loss": 2.6136,
+      "step": 4489
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2121337134357121e-05,
+      "loss": 2.2053,
+      "step": 4490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2118171979806642e-05,
+      "loss": 2.4635,
+      "step": 4491
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.211500660306975e-05,
+      "loss": 2.6644,
+      "step": 4492
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2111841004478485e-05,
+      "loss": 1.8937,
+      "step": 4493
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2108675184364898e-05,
+      "loss": 3.6826,
+      "step": 4494
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2105509143061072e-05,
+      "loss": 2.6939,
+      "step": 4495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210234288089911e-05,
+      "loss": 2.0161,
+      "step": 4496
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2099176398211134e-05,
+      "loss": 4.0513,
+      "step": 4497
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2096009695329298e-05,
+      "loss": 3.2094,
+      "step": 4498
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2092842772585773e-05,
+      "loss": 2.4528,
+      "step": 4499
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2089675630312755e-05,
+      "loss": 2.3648,
+      "step": 4500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2086508268842459e-05,
+      "loss": 2.3007,
+      "step": 4501
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208334068850713e-05,
+      "loss": 3.0695,
+      "step": 4502
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2080172889639033e-05,
+      "loss": 1.9555,
+      "step": 4503
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2077004872570454e-05,
+      "loss": 2.6734,
+      "step": 4504
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2073836637633705e-05,
+      "loss": 2.534,
+      "step": 4505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2070668185161116e-05,
+      "loss": 2.9943,
+      "step": 4506
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2067499515485046e-05,
+      "loss": 2.5872,
+      "step": 4507
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206433062893787e-05,
+      "loss": 1.2335,
+      "step": 4508
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2061161525851991e-05,
+      "loss": 2.2495,
+      "step": 4509
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2057992206559837e-05,
+      "loss": 3.7081,
+      "step": 4510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2054822671393852e-05,
+      "loss": 2.5158,
+      "step": 4511
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051652920686505e-05,
+      "loss": 3.3554,
+      "step": 4512
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2048482954770289e-05,
+      "loss": 2.8973,
+      "step": 4513
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204531277397772e-05,
+      "loss": 3.1887,
+      "step": 4514
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204214237864133e-05,
+      "loss": 3.9436,
+      "step": 4515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2038971769093685e-05,
+      "loss": 1.8118,
+      "step": 4516
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2035800945667365e-05,
+      "loss": 2.1624,
+      "step": 4517
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2032629908694969e-05,
+      "loss": 3.1673,
+      "step": 4518
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2029458658509135e-05,
+      "loss": 2.1466,
+      "step": 4519
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2026287195442503e-05,
+      "loss": 3.0387,
+      "step": 4520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2023115519827751e-05,
+      "loss": 2.4758,
+      "step": 4521
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201994363199757e-05,
+      "loss": 2.4656,
+      "step": 4522
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2016771532284675e-05,
+      "loss": 2.5732,
+      "step": 4523
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201359922102181e-05,
+      "loss": 1.6961,
+      "step": 4524
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2010426698541728e-05,
+      "loss": 3.6203,
+      "step": 4525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2007253965177219e-05,
+      "loss": 2.5688,
+      "step": 4526
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2004081021261083e-05,
+      "loss": 2.8786,
+      "step": 4527
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200090786712615e-05,
+      "loss": 3.0357,
+      "step": 4528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199773450310527e-05,
+      "loss": 2.8969,
+      "step": 4529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199456092953131e-05,
+      "loss": 1.38,
+      "step": 4530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1991387146737165e-05,
+      "loss": 2.9399,
+      "step": 4531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1988213155055754e-05,
+      "loss": 3.3289,
+      "step": 4532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1985038954820008e-05,
+      "loss": 2.4201,
+      "step": 4533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1981864546362891e-05,
+      "loss": 3.1148,
+      "step": 4534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.197868993001738e-05,
+      "loss": 3.7646,
+      "step": 4535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1975515106116472e-05,
+      "loss": 2.9168,
+      "step": 4536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1972340074993207e-05,
+      "loss": 3.2012,
+      "step": 4537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1969164836980618e-05,
+      "loss": 2.2667,
+      "step": 4538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196598939241178e-05,
+      "loss": 2.8231,
+      "step": 4539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1962813741619777e-05,
+      "loss": 3.2899,
+      "step": 4540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.195963788493772e-05,
+      "loss": 2.6995,
+      "step": 4541
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1956461822698748e-05,
+      "loss": 3.1664,
+      "step": 4542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.195328555523601e-05,
+      "loss": 2.1904,
+      "step": 4543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1950109082882681e-05,
+      "loss": 2.2489,
+      "step": 4544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.194693240597196e-05,
+      "loss": 3.8815,
+      "step": 4545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1943755524837067e-05,
+      "loss": 2.1331,
+      "step": 4546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1940578439811241e-05,
+      "loss": 2.7477,
+      "step": 4547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.193740115122774e-05,
+      "loss": 3.5966,
+      "step": 4548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1934223659419855e-05,
+      "loss": 3.7099,
+      "step": 4549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1931045964720882e-05,
+      "loss": 3.261,
+      "step": 4550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1927868067464147e-05,
+      "loss": 3.1787,
+      "step": 4551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1924689967983006e-05,
+      "loss": 2.3387,
+      "step": 4552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1921511666610812e-05,
+      "loss": 3.3974,
+      "step": 4553
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1918333163680968e-05,
+      "loss": 2.9471,
+      "step": 4554
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1915154459526876e-05,
+      "loss": 2.7152,
+      "step": 4555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.191197555448197e-05,
+      "loss": 3.7047,
+      "step": 4556
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1908796448879706e-05,
+      "loss": 3.4476,
+      "step": 4557
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.190561714305355e-05,
+      "loss": 1.8347,
+      "step": 4558
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1902437637337e-05,
+      "loss": 1.9786,
+      "step": 4559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189925793206357e-05,
+      "loss": 1.7647,
+      "step": 4560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18960780275668e-05,
+      "loss": 2.1198,
+      "step": 4561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1892897924180242e-05,
+      "loss": 3.3676,
+      "step": 4562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188971762223748e-05,
+      "loss": 2.0764,
+      "step": 4563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1886537122072106e-05,
+      "loss": 3.2066,
+      "step": 4564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188335642401775e-05,
+      "loss": 2.1532,
+      "step": 4565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188017552840804e-05,
+      "loss": 3.2354,
+      "step": 4566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1876994435576648e-05,
+      "loss": 3.8205,
+      "step": 4567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187381314585725e-05,
+      "loss": 3.4026,
+      "step": 4568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1870631659583547e-05,
+      "loss": 2.3317,
+      "step": 4569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1867449977089264e-05,
+      "loss": 2.9193,
+      "step": 4570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1864268098708148e-05,
+      "loss": 2.7297,
+      "step": 4571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1861086024773963e-05,
+      "loss": 2.7601,
+      "step": 4572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1857903755620488e-05,
+      "loss": 2.9447,
+      "step": 4573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185472129158153e-05,
+      "loss": 1.5036,
+      "step": 4574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1851538632990922e-05,
+      "loss": 2.9151,
+      "step": 4575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1848355780182502e-05,
+      "loss": 2.5067,
+      "step": 4576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184517273349014e-05,
+      "loss": 2.9777,
+      "step": 4577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184198949324772e-05,
+      "loss": 2.6072,
+      "step": 4578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838806059789151e-05,
+      "loss": 3.5092,
+      "step": 4579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1835622433448361e-05,
+      "loss": 2.874,
+      "step": 4580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1832438614559298e-05,
+      "loss": 2.7412,
+      "step": 4581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1829254603455931e-05,
+      "loss": 3.2735,
+      "step": 4582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1826070400472242e-05,
+      "loss": 1.2431,
+      "step": 4583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1822886005942244e-05,
+      "loss": 2.6611,
+      "step": 4584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181970142019997e-05,
+      "loss": 2.1751,
+      "step": 4585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1816516643579457e-05,
+      "loss": 1.6848,
+      "step": 4586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181333167641478e-05,
+      "loss": 4.2813,
+      "step": 4587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1810146519040023e-05,
+      "loss": 2.5598,
+      "step": 4588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1806961171789297e-05,
+      "loss": 3.7901,
+      "step": 4589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1803775634996735e-05,
+      "loss": 2.6277,
+      "step": 4590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1800589908996479e-05,
+      "loss": 2.2797,
+      "step": 4591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1797403994122698e-05,
+      "loss": 2.6739,
+      "step": 4592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179421789070958e-05,
+      "loss": 2.7699,
+      "step": 4593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179103159909133e-05,
+      "loss": 3.3482,
+      "step": 4594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1787845119602184e-05,
+      "loss": 3.0284,
+      "step": 4595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178465845257638e-05,
+      "loss": 4.3266,
+      "step": 4596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1781471598348185e-05,
+      "loss": 2.4283,
+      "step": 4597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1778284557251887e-05,
+      "loss": 3.0695,
+      "step": 4598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1775097329621793e-05,
+      "loss": 2.9469,
+      "step": 4599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177190991579223e-05,
+      "loss": 3.7766,
+      "step": 4600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1768722316097539e-05,
+      "loss": 1.7769,
+      "step": 4601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1765534530872085e-05,
+      "loss": 1.7397,
+      "step": 4602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1762346560450253e-05,
+      "loss": 2.3653,
+      "step": 4603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1759158405166446e-05,
+      "loss": 3.2297,
+      "step": 4604
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1755970065355087e-05,
+      "loss": 2.9859,
+      "step": 4605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1752781541350617e-05,
+      "loss": 2.1394,
+      "step": 4606
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.17495928334875e-05,
+      "loss": 3.7096,
+      "step": 4607
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1746403942100215e-05,
+      "loss": 2.1264,
+      "step": 4608
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174321486752326e-05,
+      "loss": 4.9303,
+      "step": 4609
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174002561009116e-05,
+      "loss": 3.6575,
+      "step": 4610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1736836170138446e-05,
+      "loss": 1.7443,
+      "step": 4611
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1733646547999678e-05,
+      "loss": 3.0329,
+      "step": 4612
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1730456744009435e-05,
+      "loss": 2.6237,
+      "step": 4613
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172726675850231e-05,
+      "loss": 2.749,
+      "step": 4614
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1724076591812919e-05,
+      "loss": 2.6238,
+      "step": 4615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1720886244275893e-05,
+      "loss": 2.4611,
+      "step": 4616
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1717695716225887e-05,
+      "loss": 1.996,
+      "step": 4617
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1714505007997576e-05,
+      "loss": 2.8223,
+      "step": 4618
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1711314119925644e-05,
+      "loss": 0.8283,
+      "step": 4619
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1708123052344803e-05,
+      "loss": 2.9956,
+      "step": 4620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1704931805589785e-05,
+      "loss": 2.8348,
+      "step": 4621
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.170174037999533e-05,
+      "loss": 3.0109,
+      "step": 4622
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1698548775896208e-05,
+      "loss": 3.8302,
+      "step": 4623
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1695356993627203e-05,
+      "loss": 2.9925,
+      "step": 4624
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1692165033523117e-05,
+      "loss": 2.4858,
+      "step": 4625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1688972895918776e-05,
+      "loss": 2.8973,
+      "step": 4626
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1685780581149011e-05,
+      "loss": 3.1882,
+      "step": 4627
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1682588089548692e-05,
+      "loss": 2.4767,
+      "step": 4628
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1679395421452688e-05,
+      "loss": 2.4377,
+      "step": 4629
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1676202577195901e-05,
+      "loss": 2.6417,
+      "step": 4630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1673009557113244e-05,
+      "loss": 3.0781,
+      "step": 4631
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1669816361539647e-05,
+      "loss": 2.0869,
+      "step": 4632
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1666622990810064e-05,
+      "loss": 2.705,
+      "step": 4633
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1663429445259464e-05,
+      "loss": 2.9208,
+      "step": 4634
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1660235725222835e-05,
+      "loss": 2.8793,
+      "step": 4635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1657041831035186e-05,
+      "loss": 2.6065,
+      "step": 4636
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1653847763031538e-05,
+      "loss": 1.8322,
+      "step": 4637
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1650653521546937e-05,
+      "loss": 2.2401,
+      "step": 4638
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1647459106916438e-05,
+      "loss": 2.8453,
+      "step": 4639
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164426451947513e-05,
+      "loss": 2.3014,
+      "step": 4640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1641069759558101e-05,
+      "loss": 3.1833,
+      "step": 4641
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1637874827500472e-05,
+      "loss": 2.9032,
+      "step": 4642
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1634679723637379e-05,
+      "loss": 2.1227,
+      "step": 4643
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1631484448303964e-05,
+      "loss": 2.9647,
+      "step": 4644
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1628289001835405e-05,
+      "loss": 2.4934,
+      "step": 4645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1625093384566887e-05,
+      "loss": 3.1145,
+      "step": 4646
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1621897596833616e-05,
+      "loss": 3.0091,
+      "step": 4647
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1618701638970815e-05,
+      "loss": 2.3772,
+      "step": 4648
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1615505511313723e-05,
+      "loss": 2.9961,
+      "step": 4649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1612309214197599e-05,
+      "loss": 3.8299,
+      "step": 4650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1609112747957724e-05,
+      "loss": 4.1637,
+      "step": 4651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1605916112929388e-05,
+      "loss": 1.853,
+      "step": 4652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1602719309447905e-05,
+      "loss": 2.0392,
+      "step": 4653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159952233784861e-05,
+      "loss": 4.029,
+      "step": 4654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1596325198466841e-05,
+      "loss": 3.2737,
+      "step": 4655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1593127891637968e-05,
+      "loss": 3.6783,
+      "step": 4656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1589930417697372e-05,
+      "loss": 2.2466,
+      "step": 4657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1586732776980456e-05,
+      "loss": 3.061,
+      "step": 4658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1583534969822634e-05,
+      "loss": 1.914,
+      "step": 4659
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1580336996559343e-05,
+      "loss": 2.354,
+      "step": 4660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157713885752604e-05,
+      "loss": 3.5656,
+      "step": 4661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1573940553058186e-05,
+      "loss": 1.5278,
+      "step": 4662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1570742083491273e-05,
+      "loss": 3.6729,
+      "step": 4663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156754344916081e-05,
+      "loss": 2.8785,
+      "step": 4664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156434465040231e-05,
+      "loss": 3.6077,
+      "step": 4665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156114568755132e-05,
+      "loss": 4.1344,
+      "step": 4666
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1557946560943389e-05,
+      "loss": 3.7578,
+      "step": 4667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1554747270914098e-05,
+      "loss": 4.3608,
+      "step": 4668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155154781779903e-05,
+      "loss": 3.0586,
+      "step": 4669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1548348201933799e-05,
+      "loss": 1.3619,
+      "step": 4670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1545148423654029e-05,
+      "loss": 3.3637,
+      "step": 4671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1541948483295358e-05,
+      "loss": 3.7324,
+      "step": 4672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1538748381193447e-05,
+      "loss": 2.5426,
+      "step": 4673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1535548117683977e-05,
+      "loss": 2.3681,
+      "step": 4674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1532347693102632e-05,
+      "loss": 3.5329,
+      "step": 4675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1529147107785129e-05,
+      "loss": 1.9778,
+      "step": 4676
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1525946362067188e-05,
+      "loss": 4.8176,
+      "step": 4677
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1522745456284557e-05,
+      "loss": 3.08,
+      "step": 4678
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1519544390772996e-05,
+      "loss": 2.1256,
+      "step": 4679
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151634316586828e-05,
+      "loss": 2.1758,
+      "step": 4680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1513141781906202e-05,
+      "loss": 2.7654,
+      "step": 4681
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150994023922258e-05,
+      "loss": 3.0089,
+      "step": 4682
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1506738538153228e-05,
+      "loss": 3.5973,
+      "step": 4683
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1503536679034e-05,
+      "loss": 2.1318,
+      "step": 4684
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150033466220075e-05,
+      "loss": 2.5546,
+      "step": 4685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1497132487989357e-05,
+      "loss": 1.9271,
+      "step": 4686
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1493930156735716e-05,
+      "loss": 2.399,
+      "step": 4687
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1490727668775735e-05,
+      "loss": 1.5205,
+      "step": 4688
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148752502444534e-05,
+      "loss": 2.8605,
+      "step": 4689
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1484322224080474e-05,
+      "loss": 2.802,
+      "step": 4690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1481119268017094e-05,
+      "loss": 2.3131,
+      "step": 4691
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.147791615659118e-05,
+      "loss": 2.6951,
+      "step": 4692
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1474712890138716e-05,
+      "loss": 1.576,
+      "step": 4693
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1471509468995719e-05,
+      "loss": 3.1614,
+      "step": 4694
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1468305893498204e-05,
+      "loss": 3.3287,
+      "step": 4695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1465102163982218e-05,
+      "loss": 2.6343,
+      "step": 4696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1461898280783815e-05,
+      "loss": 3.2975,
+      "step": 4697
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1458694244239067e-05,
+      "loss": 2.1867,
+      "step": 4698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1455490054684063e-05,
+      "loss": 2.3058,
+      "step": 4699
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1452285712454905e-05,
+      "loss": 3.1585,
+      "step": 4700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449081217887716e-05,
+      "loss": 2.5633,
+      "step": 4701
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1445876571318637e-05,
+      "loss": 3.0107,
+      "step": 4702
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.144267177308381e-05,
+      "loss": 1.2526,
+      "step": 4703
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1439466823519414e-05,
+      "loss": 1.8446,
+      "step": 4704
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1436261722961627e-05,
+      "loss": 3.1357,
+      "step": 4705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1433056471746654e-05,
+      "loss": 2.1852,
+      "step": 4706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1429851070210706e-05,
+      "loss": 3.4326,
+      "step": 4707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1426645518690015e-05,
+      "loss": 2.6887,
+      "step": 4708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1423439817520837e-05,
+      "loss": 3.3799,
+      "step": 4709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1420233967039423e-05,
+      "loss": 3.7309,
+      "step": 4710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1417027967582058e-05,
+      "loss": 1.2681,
+      "step": 4711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1413821819485035e-05,
+      "loss": 2.911,
+      "step": 4712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1410615523084664e-05,
+      "loss": 2.2767,
+      "step": 4713
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1407409078717275e-05,
+      "loss": 4.2532,
+      "step": 4714
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1404202486719205e-05,
+      "loss": 2.5513,
+      "step": 4715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.140099574742681e-05,
+      "loss": 1.2331,
+      "step": 4716
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1397788861176466e-05,
+      "loss": 2.004,
+      "step": 4717
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1394581828304555e-05,
+      "loss": 2.6001,
+      "step": 4718
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1391374649147482e-05,
+      "loss": 1.3207,
+      "step": 4719
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.138816732404167e-05,
+      "loss": 3.6589,
+      "step": 4720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1384959853323546e-05,
+      "loss": 2.5164,
+      "step": 4721
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1381752237329557e-05,
+      "loss": 1.8875,
+      "step": 4722
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1378544476396176e-05,
+      "loss": 1.9771,
+      "step": 4723
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1375336570859877e-05,
+      "loss": 3.0392,
+      "step": 4724
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1372128521057155e-05,
+      "loss": 1.6631,
+      "step": 4725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1368920327324517e-05,
+      "loss": 2.0621,
+      "step": 4726
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1365711989998494e-05,
+      "loss": 2.3278,
+      "step": 4727
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136250350941562e-05,
+      "loss": 2.6264,
+      "step": 4728
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1359294885912449e-05,
+      "loss": 3.7538,
+      "step": 4729
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1356086119825553e-05,
+      "loss": 1.8351,
+      "step": 4730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135287721149152e-05,
+      "loss": 3.9558,
+      "step": 4731
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1349668161246945e-05,
+      "loss": 3.3883,
+      "step": 4732
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1346458969428445e-05,
+      "loss": 3.4168,
+      "step": 4733
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1343249636372646e-05,
+      "loss": 2.2287,
+      "step": 4734
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1340040162416197e-05,
+      "loss": 1.8134,
+      "step": 4735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1336830547895752e-05,
+      "loss": 2.4599,
+      "step": 4736
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133362079314799e-05,
+      "loss": 2.7747,
+      "step": 4737
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1330410898509594e-05,
+      "loss": 2.1497,
+      "step": 4738
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.132720086431727e-05,
+      "loss": 2.1272,
+      "step": 4739
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1323990690907734e-05,
+      "loss": 2.1176,
+      "step": 4740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1320780378617717e-05,
+      "loss": 2.2513,
+      "step": 4741
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1317569927783971e-05,
+      "loss": 2.743,
+      "step": 4742
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1314359338743257e-05,
+      "loss": 2.4488,
+      "step": 4743
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1311148611832346e-05,
+      "loss": 3.1382,
+      "step": 4744
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1307937747388034e-05,
+      "loss": 3.1911,
+      "step": 4745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1304726745747119e-05,
+      "loss": 3.8604,
+      "step": 4746
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1301515607246422e-05,
+      "loss": 2.3654,
+      "step": 4747
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.129830433222278e-05,
+      "loss": 2.9706,
+      "step": 4748
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.129509292101304e-05,
+      "loss": 2.191,
+      "step": 4749
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1291881373954066e-05,
+      "loss": 2.6034,
+      "step": 4750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1288669691382728e-05,
+      "loss": 2.3298,
+      "step": 4751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128545787363592e-05,
+      "loss": 2.887,
+      "step": 4752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1282245921050552e-05,
+      "loss": 2.6794,
+      "step": 4753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1279033833963532e-05,
+      "loss": 3.8237,
+      "step": 4754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1275821612711803e-05,
+      "loss": 3.2947,
+      "step": 4755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1272609257632305e-05,
+      "loss": 1.8135,
+      "step": 4756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269396769062002e-05,
+      "loss": 2.4975,
+      "step": 4757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1266184147337873e-05,
+      "loss": 3.1936,
+      "step": 4758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12629713927969e-05,
+      "loss": 1.6745,
+      "step": 4759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1259758505776092e-05,
+      "loss": 1.5576,
+      "step": 4760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1256545486612462e-05,
+      "loss": 2.999,
+      "step": 4761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1253332335643043e-05,
+      "loss": 3.9456,
+      "step": 4762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125011905320488e-05,
+      "loss": 2.8113,
+      "step": 4763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1246905639635029e-05,
+      "loss": 3.5842,
+      "step": 4764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1243692095270565e-05,
+      "loss": 2.5657,
+      "step": 4765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1240478420448568e-05,
+      "loss": 2.3606,
+      "step": 4766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1237264615506145e-05,
+      "loss": 2.557,
+      "step": 4767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1234050680780407e-05,
+      "loss": 2.2039,
+      "step": 4768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1230836616608479e-05,
+      "loss": 1.1295,
+      "step": 4769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1227622423327501e-05,
+      "loss": 2.5431,
+      "step": 4770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1224408101274631e-05,
+      "loss": 3.9146,
+      "step": 4771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1221193650787032e-05,
+      "loss": 3.9697,
+      "step": 4772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1217979072201887e-05,
+      "loss": 1.8817,
+      "step": 4773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1214764365856393e-05,
+      "loss": 2.4766,
+      "step": 4774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1211549532087749e-05,
+      "loss": 0.7471,
+      "step": 4775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1208334571233186e-05,
+      "loss": 2.3521,
+      "step": 4776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1205119483629933e-05,
+      "loss": 3.3724,
+      "step": 4777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1201904269615242e-05,
+      "loss": 2.5721,
+      "step": 4778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1198688929526369e-05,
+      "loss": 1.6774,
+      "step": 4779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.119547346370059e-05,
+      "loss": 1.6871,
+      "step": 4780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1192257872475197e-05,
+      "loss": 2.2547,
+      "step": 4781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1189042156187482e-05,
+      "loss": 1.7813,
+      "step": 4782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1185826315174766e-05,
+      "loss": 3.308,
+      "step": 4783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118261034977437e-05,
+      "loss": 1.9025,
+      "step": 4784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1179394260323639e-05,
+      "loss": 2.7415,
+      "step": 4785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1176178047159926e-05,
+      "loss": 1.1203,
+      "step": 4786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1172961710620591e-05,
+      "loss": 2.504,
+      "step": 4787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.116974525104302e-05,
+      "loss": 2.5922,
+      "step": 4788
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11665286687646e-05,
+      "loss": 3.5144,
+      "step": 4789
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1163311964122733e-05,
+      "loss": 1.9084,
+      "step": 4790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1160095137454845e-05,
+      "loss": 1.8639,
+      "step": 4791
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1156878189098357e-05,
+      "loss": 3.001,
+      "step": 4792
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.115366111939072e-05,
+      "loss": 2.8988,
+      "step": 4793
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1150443928669383e-05,
+      "loss": 2.2894,
+      "step": 4794
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114722661727182e-05,
+      "loss": 3.123,
+      "step": 4795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114400918553551e-05,
+      "loss": 2.9434,
+      "step": 4796
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1140791633797943e-05,
+      "loss": 3.2233,
+      "step": 4797
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113757396239663e-05,
+      "loss": 1.9702,
+      "step": 4798
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1134356171669085e-05,
+      "loss": 3.1622,
+      "step": 4799
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1131138261952845e-05,
+      "loss": 3.8872,
+      "step": 4800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.112792023358545e-05,
+      "loss": 4.7023,
+      "step": 4801
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1124702086904455e-05,
+      "loss": 2.7862,
+      "step": 4802
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1121483822247434e-05,
+      "loss": 2.5968,
+      "step": 4803
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1118265439951968e-05,
+      "loss": 1.7656,
+      "step": 4804
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1115046940355643e-05,
+      "loss": 3.2449,
+      "step": 4805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111182832379607e-05,
+      "loss": 3.7779,
+      "step": 4806
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108609590610867e-05,
+      "loss": 2.831,
+      "step": 4807
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.110539074113766e-05,
+      "loss": 2.1614,
+      "step": 4808
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102171775714097e-05,
+      "loss": 2.2091,
+      "step": 4809
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.109895269467783e-05,
+      "loss": 2.3,
+      "step": 4810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1095733498366527e-05,
+      "loss": 3.0441,
+      "step": 4811
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1092514187117865e-05,
+      "loss": 3.1785,
+      "step": 4812
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1089294761269534e-05,
+      "loss": 1.9823,
+      "step": 4813
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086075221159237e-05,
+      "loss": 3.0078,
+      "step": 4814
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1082855567124693e-05,
+      "loss": 2.8372,
+      "step": 4815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1079635799503625e-05,
+      "loss": 3.907,
+      "step": 4816
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1076415918633771e-05,
+      "loss": 2.0664,
+      "step": 4817
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1073195924852882e-05,
+      "loss": 3.603,
+      "step": 4818
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1069975818498724e-05,
+      "loss": 2.4794,
+      "step": 4819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1066755599909065e-05,
+      "loss": 1.211,
+      "step": 4820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10635352694217e-05,
+      "loss": 2.3714,
+      "step": 4821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1060314827374417e-05,
+      "loss": 2.9797,
+      "step": 4822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1057094274105028e-05,
+      "loss": 2.8827,
+      "step": 4823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1053873609951362e-05,
+      "loss": 2.4946,
+      "step": 4824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105065283525124e-05,
+      "loss": 3.1959,
+      "step": 4825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1047431950342517e-05,
+      "loss": 3.6143,
+      "step": 4826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1044210955563037e-05,
+      "loss": 2.7416,
+      "step": 4827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1040989851250678e-05,
+      "loss": 0.9008,
+      "step": 4828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1037768637743316e-05,
+      "loss": 2.038,
+      "step": 4829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1034547315378838e-05,
+      "loss": 2.5824,
+      "step": 4830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1031325884495152e-05,
+      "loss": 2.793,
+      "step": 4831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1028104345430161e-05,
+      "loss": 2.2537,
+      "step": 4832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10248826985218e-05,
+      "loss": 3.1331,
+      "step": 4833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1021660944108e-05,
+      "loss": 2.3622,
+      "step": 4834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.101843908252671e-05,
+      "loss": 2.3888,
+      "step": 4835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1015217114115884e-05,
+      "loss": 3.1209,
+      "step": 4836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1011995039213494e-05,
+      "loss": 2.7796,
+      "step": 4837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1008772858157524e-05,
+      "loss": 3.2019,
+      "step": 4838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1005550571285964e-05,
+      "loss": 2.5015,
+      "step": 4839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1002328178936813e-05,
+      "loss": 1.8034,
+      "step": 4840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0999105681448087e-05,
+      "loss": 2.142,
+      "step": 4841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0995883079157817e-05,
+      "loss": 3.1617,
+      "step": 4842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099266037240403e-05,
+      "loss": 2.9425,
+      "step": 4843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0989437561524776e-05,
+      "loss": 3.6567,
+      "step": 4844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0986214646858115e-05,
+      "loss": 2.9564,
+      "step": 4845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0982991628742116e-05,
+      "loss": 2.3515,
+      "step": 4846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.097976850751486e-05,
+      "loss": 2.7976,
+      "step": 4847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.097654528351443e-05,
+      "loss": 3.1034,
+      "step": 4848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0973321957078935e-05,
+      "loss": 2.3084,
+      "step": 4849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0970098528546482e-05,
+      "loss": 1.7319,
+      "step": 4850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0966874998255197e-05,
+      "loss": 1.878,
+      "step": 4851
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0963651366543214e-05,
+      "loss": 1.5807,
+      "step": 4852
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0960427633748672e-05,
+      "loss": 2.7424,
+      "step": 4853
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0957203800209729e-05,
+      "loss": 3.3828,
+      "step": 4854
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0953979866264549e-05,
+      "loss": 2.37,
+      "step": 4855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.095075583225131e-05,
+      "loss": 2.7032,
+      "step": 4856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0947531698508198e-05,
+      "loss": 2.2914,
+      "step": 4857
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0944307465373405e-05,
+      "loss": 2.9101,
+      "step": 4858
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0941083133185146e-05,
+      "loss": 1.3292,
+      "step": 4859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0937858702281631e-05,
+      "loss": 2.7667,
+      "step": 4860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093463417300109e-05,
+      "loss": 3.2556,
+      "step": 4861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0931409545681764e-05,
+      "loss": 3.081,
+      "step": 4862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0928184820661897e-05,
+      "loss": 2.7185,
+      "step": 4863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0924959998279754e-05,
+      "loss": 3.1264,
+      "step": 4864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0921735078873599e-05,
+      "loss": 1.83,
+      "step": 4865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091851006278171e-05,
+      "loss": 2.3807,
+      "step": 4866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0915284950342381e-05,
+      "loss": 2.701,
+      "step": 4867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0912059741893908e-05,
+      "loss": 3.5193,
+      "step": 4868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.09088344377746e-05,
+      "loss": 2.6582,
+      "step": 4869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090560903832278e-05,
+      "loss": 2.8466,
+      "step": 4870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0902383543876773e-05,
+      "loss": 1.8356,
+      "step": 4871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089915795477492e-05,
+      "loss": 2.1228,
+      "step": 4872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089593227135557e-05,
+      "loss": 3.479,
+      "step": 4873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0892706493957083e-05,
+      "loss": 3.1823,
+      "step": 4874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088948062291783e-05,
+      "loss": 3.1916,
+      "step": 4875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0886254658576186e-05,
+      "loss": 1.8651,
+      "step": 4876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0883028601270545e-05,
+      "loss": 1.2807,
+      "step": 4877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0879802451339298e-05,
+      "loss": 2.8144,
+      "step": 4878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0876576209120857e-05,
+      "loss": 2.8367,
+      "step": 4879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.087334987495364e-05,
+      "loss": 3.7815,
+      "step": 4880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0870123449176075e-05,
+      "loss": 2.3962,
+      "step": 4881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0866896932126602e-05,
+      "loss": 2.2969,
+      "step": 4882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086367032414366e-05,
+      "loss": 3.0407,
+      "step": 4883
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0860443625565712e-05,
+      "loss": 3.3678,
+      "step": 4884
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0857216836731221e-05,
+      "loss": 3.1005,
+      "step": 4885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085398995797866e-05,
+      "loss": 3.8596,
+      "step": 4886
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085076298964652e-05,
+      "loss": 4.1358,
+      "step": 4887
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0847535932073288e-05,
+      "loss": 2.1483,
+      "step": 4888
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084430878559747e-05,
+      "loss": 1.5849,
+      "step": 4889
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0841081550557577e-05,
+      "loss": 2.71,
+      "step": 4890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0837854227292138e-05,
+      "loss": 2.481,
+      "step": 4891
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0834626816139678e-05,
+      "loss": 3.0567,
+      "step": 4892
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0831399317438736e-05,
+      "loss": 2.9011,
+      "step": 4893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0828171731527863e-05,
+      "loss": 3.5122,
+      "step": 4894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0824944058745623e-05,
+      "loss": 1.6423,
+      "step": 4895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0821716299430577e-05,
+      "loss": 0.994,
+      "step": 4896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0818488453921308e-05,
+      "loss": 3.0766,
+      "step": 4897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0815260522556394e-05,
+      "loss": 2.1189,
+      "step": 4898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.081203250567444e-05,
+      "loss": 2.0599,
+      "step": 4899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0808804403614044e-05,
+      "loss": 2.2518,
+      "step": 4900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0805576216713818e-05,
+      "loss": 3.6702,
+      "step": 4901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0802347945312386e-05,
+      "loss": 2.3884,
+      "step": 4902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0799119589748378e-05,
+      "loss": 1.7318,
+      "step": 4903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0795891150360435e-05,
+      "loss": 2.264,
+      "step": 4904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0792662627487207e-05,
+      "loss": 2.224,
+      "step": 4905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0789434021467344e-05,
+      "loss": 2.9126,
+      "step": 4906
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0786205332639517e-05,
+      "loss": 2.718,
+      "step": 4907
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0782976561342398e-05,
+      "loss": 2.2618,
+      "step": 4908
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0779747707914672e-05,
+      "loss": 3.3174,
+      "step": 4909
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0776518772695035e-05,
+      "loss": 1.4693,
+      "step": 4910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0773289756022178e-05,
+      "loss": 2.2511,
+      "step": 4911
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0770060658234815e-05,
+      "loss": 2.0043,
+      "step": 4912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0766831479671664e-05,
+      "loss": 3.9884,
+      "step": 4913
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.076360222067145e-05,
+      "loss": 3.0632,
+      "step": 4914
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0760372881572904e-05,
+      "loss": 2.182,
+      "step": 4915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0757143462714777e-05,
+      "loss": 2.4012,
+      "step": 4916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075391396443581e-05,
+      "loss": 1.9467,
+      "step": 4917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075068438707477e-05,
+      "loss": 1.8133,
+      "step": 4918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0747454730970422e-05,
+      "loss": 3.1941,
+      "step": 4919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0744224996461541e-05,
+      "loss": 2.8621,
+      "step": 4920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0740995183886911e-05,
+      "loss": 1.8283,
+      "step": 4921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0737765293585324e-05,
+      "loss": 2.4556,
+      "step": 4922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0734535325895587e-05,
+      "loss": 1.9424,
+      "step": 4923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0731305281156499e-05,
+      "loss": 2.3919,
+      "step": 4924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0728075159706881e-05,
+      "loss": 2.7164,
+      "step": 4925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.072484496188556e-05,
+      "loss": 2.2715,
+      "step": 4926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0721614688031366e-05,
+      "loss": 2.3236,
+      "step": 4927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0718384338483141e-05,
+      "loss": 1.3224,
+      "step": 4928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071515391357973e-05,
+      "loss": 3.4798,
+      "step": 4929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0711923413659995e-05,
+      "loss": 3.0656,
+      "step": 4930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07086928390628e-05,
+      "loss": 4.0718,
+      "step": 4931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0705462190127011e-05,
+      "loss": 3.0028,
+      "step": 4932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0702231467191516e-05,
+      "loss": 3.2265,
+      "step": 4933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0699000670595194e-05,
+      "loss": 3.7455,
+      "step": 4934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.069576980067695e-05,
+      "loss": 2.9084,
+      "step": 4935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0692538857775685e-05,
+      "loss": 1.1752,
+      "step": 4936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0689307842230305e-05,
+      "loss": 2.0233,
+      "step": 4937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0686076754379734e-05,
+      "loss": 2.6161,
+      "step": 4938
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0682845594562892e-05,
+      "loss": 2.2845,
+      "step": 4939
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0679614363118718e-05,
+      "loss": 2.5401,
+      "step": 4940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0676383060386153e-05,
+      "loss": 1.1564,
+      "step": 4941
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0673151686704141e-05,
+      "loss": 2.8779,
+      "step": 4942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0669920242411642e-05,
+      "loss": 3.2679,
+      "step": 4943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066668872784762e-05,
+      "loss": 1.9618,
+      "step": 4944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0663457143351044e-05,
+      "loss": 1.2424,
+      "step": 4945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0660225489260894e-05,
+      "loss": 1.938,
+      "step": 4946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0656993765916153e-05,
+      "loss": 2.8253,
+      "step": 4947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0653761973655819e-05,
+      "loss": 1.7498,
+      "step": 4948
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0650530112818884e-05,
+      "loss": 1.5318,
+      "step": 4949
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0647298183744359e-05,
+      "loss": 3.0894,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.064406618677126e-05,
+      "loss": 2.4387,
+      "step": 4951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0640834122238606e-05,
+      "loss": 1.9945,
+      "step": 4952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0637601990485426e-05,
+      "loss": 3.302,
+      "step": 4953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0634369791850761e-05,
+      "loss": 1.0539,
+      "step": 4954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0631137526673647e-05,
+      "loss": 3.7002,
+      "step": 4955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0627905195293135e-05,
+      "loss": 2.8919,
+      "step": 4956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0624672798048283e-05,
+      "loss": 2.9732,
+      "step": 4957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0621440335278152e-05,
+      "loss": 2.9014,
+      "step": 4958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0618207807321817e-05,
+      "loss": 1.4105,
+      "step": 4959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.061497521451835e-05,
+      "loss": 2.4847,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0611742557206837e-05,
+      "loss": 2.5165,
+      "step": 4961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0608509835726368e-05,
+      "loss": 2.9808,
+      "step": 4962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0605277050416043e-05,
+      "loss": 2.4841,
+      "step": 4963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0602044201614965e-05,
+      "loss": 3.5279,
+      "step": 4964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0598811289662243e-05,
+      "loss": 2.8642,
+      "step": 4965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0595578314896999e-05,
+      "loss": 3.1783,
+      "step": 4966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059234527765835e-05,
+      "loss": 3.0834,
+      "step": 4967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0589112178285432e-05,
+      "loss": 2.0958,
+      "step": 4968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058587901711738e-05,
+      "loss": 2.9413,
+      "step": 4969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0582645794493337e-05,
+      "loss": 3.1825,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.057941251075246e-05,
+      "loss": 2.8437,
+      "step": 4971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0576179166233895e-05,
+      "loss": 1.8856,
+      "step": 4972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0572945761276809e-05,
+      "loss": 2.9461,
+      "step": 4973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0569712296220375e-05,
+      "loss": 2.6143,
+      "step": 4974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0566478771403763e-05,
+      "loss": 2.0891,
+      "step": 4975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056324518716616e-05,
+      "loss": 2.4583,
+      "step": 4976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0560011543846749e-05,
+      "loss": 2.383,
+      "step": 4977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0556777841784725e-05,
+      "loss": 2.4224,
+      "step": 4978
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055354408131929e-05,
+      "loss": 2.0824,
+      "step": 4979
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055031026278965e-05,
+      "loss": 1.8213,
+      "step": 4980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0547076386535017e-05,
+      "loss": 2.7596,
+      "step": 4981
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054384245289461e-05,
+      "loss": 2.2891,
+      "step": 4982
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0540608462207655e-05,
+      "loss": 2.9258,
+      "step": 4983
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0537374414813384e-05,
+      "loss": 2.563,
+      "step": 4984
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0534140311051026e-05,
+      "loss": 2.679,
+      "step": 4985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053090615125983e-05,
+      "loss": 1.8575,
+      "step": 4986
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0527671935779042e-05,
+      "loss": 2.4577,
+      "step": 4987
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0524437664947918e-05,
+      "loss": 2.9293,
+      "step": 4988
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0521203339105719e-05,
+      "loss": 2.5494,
+      "step": 4989
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0517968958591705e-05,
+      "loss": 4.0986,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0514734523745151e-05,
+      "loss": 2.5937,
+      "step": 4991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051150003490534e-05,
+      "loss": 2.2319,
+      "step": 4992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0508265492411545e-05,
+      "loss": 1.8214,
+      "step": 4993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0505030896603059e-05,
+      "loss": 3.0833,
+      "step": 4994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0501796247819176e-05,
+      "loss": 4.1462,
+      "step": 4995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0498561546399194e-05,
+      "loss": 4.5691,
+      "step": 4996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0495326792682421e-05,
+      "loss": 2.2331,
+      "step": 4997
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0492091987008167e-05,
+      "loss": 2.285,
+      "step": 4998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0488857129715748e-05,
+      "loss": 1.8633,
+      "step": 4999
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0485622221144485e-05,
+      "loss": 2.1828,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0482387261633703e-05,
+      "loss": 3.003,
+      "step": 5001
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0479152251522737e-05,
+      "loss": 3.4965,
+      "step": 5002
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0475917191150924e-05,
+      "loss": 1.9539,
+      "step": 5003
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0472682080857606e-05,
+      "loss": 2.2721,
+      "step": 5004
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046944692098213e-05,
+      "loss": 3.921,
+      "step": 5005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0466211711863851e-05,
+      "loss": 2.98,
+      "step": 5006
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046297645384213e-05,
+      "loss": 3.2067,
+      "step": 5007
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0459741147256325e-05,
+      "loss": 1.3955,
+      "step": 5008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0456505792445812e-05,
+      "loss": 3.6952,
+      "step": 5009
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0453270389749956e-05,
+      "loss": 2.8298,
+      "step": 5010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.045003493950814e-05,
+      "loss": 1.3028,
+      "step": 5011
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044679944205975e-05,
+      "loss": 2.5903,
+      "step": 5012
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044356389774417e-05,
+      "loss": 3.2273,
+      "step": 5013
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0440328306900791e-05,
+      "loss": 2.5092,
+      "step": 5014
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0437092669869025e-05,
+      "loss": 2.8213,
+      "step": 5015
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043385698698826e-05,
+      "loss": 2.2267,
+      "step": 5016
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0430621258597914e-05,
+      "loss": 2.9381,
+      "step": 5017
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0427385485037398e-05,
+      "loss": 2.7612,
+      "step": 5018
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0424149666646124e-05,
+      "loss": 1.0744,
+      "step": 5019
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0420913803763522e-05,
+      "loss": 2.4476,
+      "step": 5020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0417677896729015e-05,
+      "loss": 3.002,
+      "step": 5021
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0414441945882033e-05,
+      "loss": 4.0345,
+      "step": 5022
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0411205951562015e-05,
+      "loss": 2.8873,
+      "step": 5023
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04079699141084e-05,
+      "loss": 3.2202,
+      "step": 5024
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0404733833860639e-05,
+      "loss": 2.8229,
+      "step": 5025
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0401497711158173e-05,
+      "loss": 2.1326,
+      "step": 5026
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0398261546340464e-05,
+      "loss": 3.0728,
+      "step": 5027
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0395025339746965e-05,
+      "loss": 3.0373,
+      "step": 5028
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0391789091717142e-05,
+      "loss": 1.8961,
+      "step": 5029
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0388552802590461e-05,
+      "loss": 2.4917,
+      "step": 5030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0385316472706393e-05,
+      "loss": 2.2422,
+      "step": 5031
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0382080102404417e-05,
+      "loss": 2.3913,
+      "step": 5032
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037884369202401e-05,
+      "loss": 3.3685,
+      "step": 5033
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037560724190466e-05,
+      "loss": 3.1165,
+      "step": 5034
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0372370752385854e-05,
+      "loss": 1.9887,
+      "step": 5035
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0369134223807082e-05,
+      "loss": 2.8385,
+      "step": 5036
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0365897656507846e-05,
+      "loss": 2.7059,
+      "step": 5037
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0362661050827643e-05,
+      "loss": 1.7735,
+      "step": 5038
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0359424407105976e-05,
+      "loss": 3.4088,
+      "step": 5039
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0356187725682359e-05,
+      "loss": 2.4401,
+      "step": 5040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0352951006896302e-05,
+      "loss": 4.389,
+      "step": 5041
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0349714251087323e-05,
+      "loss": 3.3717,
+      "step": 5042
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0346477458594944e-05,
+      "loss": 1.4592,
+      "step": 5043
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0343240629758683e-05,
+      "loss": 2.1366,
+      "step": 5044
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0340003764918078e-05,
+      "loss": 2.728,
+      "step": 5045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0336766864412655e-05,
+      "loss": 2.9342,
+      "step": 5046
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0333529928581948e-05,
+      "loss": 2.4405,
+      "step": 5047
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0330292957765502e-05,
+      "loss": 2.2203,
+      "step": 5048
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0327055952302856e-05,
+      "loss": 2.4693,
+      "step": 5049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0323818912533561e-05,
+      "loss": 3.3023,
+      "step": 5050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0320581838797163e-05,
+      "loss": 1.5241,
+      "step": 5051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0317344731433217e-05,
+      "loss": 3.1613,
+      "step": 5052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0314107590781284e-05,
+      "loss": 3.5575,
+      "step": 5053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0310870417180922e-05,
+      "loss": 2.1901,
+      "step": 5054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0307633210971697e-05,
+      "loss": 1.8737,
+      "step": 5055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0304395972493172e-05,
+      "loss": 0.8571,
+      "step": 5056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0301158702084926e-05,
+      "loss": 1.8402,
+      "step": 5057
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0297921400086528e-05,
+      "loss": 3.0294,
+      "step": 5058
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0294684066837557e-05,
+      "loss": 3.0118,
+      "step": 5059
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0291446702677598e-05,
+      "loss": 3.4624,
+      "step": 5060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0288209307946232e-05,
+      "loss": 2.7072,
+      "step": 5061
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0284971882983043e-05,
+      "loss": 2.9061,
+      "step": 5062
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0281734428127633e-05,
+      "loss": 2.7248,
+      "step": 5063
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0278496943719585e-05,
+      "loss": 2.92,
+      "step": 5064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0275259430098502e-05,
+      "loss": 2.387,
+      "step": 5065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0272021887603977e-05,
+      "loss": 1.8673,
+      "step": 5066
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0268784316575622e-05,
+      "loss": 3.0356,
+      "step": 5067
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0265546717353041e-05,
+      "loss": 2.2362,
+      "step": 5068
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0262309090275841e-05,
+      "loss": 2.6364,
+      "step": 5069
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0259071435683636e-05,
+      "loss": 2.2144,
+      "step": 5070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0255833753916038e-05,
+      "loss": 3.6043,
+      "step": 5071
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0252596045312666e-05,
+      "loss": 2.6191,
+      "step": 5072
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0249358310213144e-05,
+      "loss": 2.1482,
+      "step": 5073
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0246120548957091e-05,
+      "loss": 3.1616,
+      "step": 5074
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0242882761884132e-05,
+      "loss": 2.4313,
+      "step": 5075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.02396449493339e-05,
+      "loss": 2.555,
+      "step": 5076
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0236407111646027e-05,
+      "loss": 2.9685,
+      "step": 5077
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0233169249160145e-05,
+      "loss": 2.7609,
+      "step": 5078
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0229931362215891e-05,
+      "loss": 3.34,
+      "step": 5079
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.02266934511529e-05,
+      "loss": 2.6274,
+      "step": 5080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0223455516310825e-05,
+      "loss": 3.2038,
+      "step": 5081
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0220217558029298e-05,
+      "loss": 2.4662,
+      "step": 5082
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0216979576647972e-05,
+      "loss": 2.7517,
+      "step": 5083
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0213741572506497e-05,
+      "loss": 2.7755,
+      "step": 5084
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0210503545944522e-05,
+      "loss": 2.2809,
+      "step": 5085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0207265497301704e-05,
+      "loss": 2.7654,
+      "step": 5086
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0204027426917695e-05,
+      "loss": 2.4088,
+      "step": 5087
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0200789335132157e-05,
+      "loss": 2.3917,
+      "step": 5088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0197551222284749e-05,
+      "loss": 2.5386,
+      "step": 5089
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0194313088715135e-05,
+      "loss": 2.6262,
+      "step": 5090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019107493476298e-05,
+      "loss": 2.9,
+      "step": 5091
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0187836760767954e-05,
+      "loss": 2.674,
+      "step": 5092
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0184598567069719e-05,
+      "loss": 2.4719,
+      "step": 5093
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0181360354007952e-05,
+      "loss": 2.8361,
+      "step": 5094
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0178122121922324e-05,
+      "loss": 2.6983,
+      "step": 5095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0174883871152517e-05,
+      "loss": 2.4232,
+      "step": 5096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.01716456020382e-05,
+      "loss": 2.5263,
+      "step": 5097
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0168407314919057e-05,
+      "loss": 2.8686,
+      "step": 5098
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0165169010134769e-05,
+      "loss": 2.2312,
+      "step": 5099
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0161930688025018e-05,
+      "loss": 1.7949,
+      "step": 5100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0158692348929486e-05,
+      "loss": 1.418,
+      "step": 5101
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0155453993187867e-05,
+      "loss": 1.6479,
+      "step": 5102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0152215621139845e-05,
+      "loss": 2.6548,
+      "step": 5103
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014897723312511e-05,
+      "loss": 3.0505,
+      "step": 5104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0145738829483354e-05,
+      "loss": 1.3244,
+      "step": 5105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0142500410554274e-05,
+      "loss": 2.9032,
+      "step": 5106
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0139261976677562e-05,
+      "loss": 1.8568,
+      "step": 5107
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.013602352819291e-05,
+      "loss": 2.06,
+      "step": 5108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0132785065440027e-05,
+      "loss": 2.3929,
+      "step": 5109
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0129546588758605e-05,
+      "loss": 1.7214,
+      "step": 5110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0126308098488346e-05,
+      "loss": 4.0027,
+      "step": 5111
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0123069594968952e-05,
+      "loss": 2.118,
+      "step": 5112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0119831078540133e-05,
+      "loss": 3.4033,
+      "step": 5113
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0116592549541588e-05,
+      "loss": 1.6693,
+      "step": 5114
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0113354008313025e-05,
+      "loss": 3.216,
+      "step": 5115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0110115455194157e-05,
+      "loss": 3.1452,
+      "step": 5116
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0106876890524683e-05,
+      "loss": 2.071,
+      "step": 5117
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0103638314644322e-05,
+      "loss": 2.8968,
+      "step": 5118
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.010039972789278e-05,
+      "loss": 2.3866,
+      "step": 5119
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0097161130609774e-05,
+      "loss": 2.9015,
+      "step": 5120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0093922523135021e-05,
+      "loss": 2.3987,
+      "step": 5121
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0090683905808225e-05,
+      "loss": 1.4606,
+      "step": 5122
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.008744527896911e-05,
+      "loss": 2.7658,
+      "step": 5123
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0084206642957393e-05,
+      "loss": 3.6817,
+      "step": 5124
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0080967998112787e-05,
+      "loss": 3.1385,
+      "step": 5125
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0077729344775018e-05,
+      "loss": 3.0275,
+      "step": 5126
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0074490683283797e-05,
+      "loss": 2.3316,
+      "step": 5127
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0071252013978852e-05,
+      "loss": 2.0455,
+      "step": 5128
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0068013337199898e-05,
+      "loss": 3.493,
+      "step": 5129
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0064774653286662e-05,
+      "loss": 3.3599,
+      "step": 5130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0061535962578867e-05,
+      "loss": 2.9428,
+      "step": 5131
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0058297265416234e-05,
+      "loss": 3.6972,
+      "step": 5132
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0055058562138489e-05,
+      "loss": 2.2001,
+      "step": 5133
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0051819853085358e-05,
+      "loss": 2.2131,
+      "step": 5134
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0048581138596563e-05,
+      "loss": 2.8869,
+      "step": 5135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0045342419011832e-05,
+      "loss": 2.6046,
+      "step": 5136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004210369467089e-05,
+      "loss": 2.6879,
+      "step": 5137
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0038864965913469e-05,
+      "loss": 2.7153,
+      "step": 5138
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0035626233079291e-05,
+      "loss": 3.7795,
+      "step": 5139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003238749650809e-05,
+      "loss": 1.5746,
+      "step": 5140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0029148756539586e-05,
+      "loss": 1.8814,
+      "step": 5141
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0025910013513519e-05,
+      "loss": 2.6538,
+      "step": 5142
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0022671267769605e-05,
+      "loss": 2.3164,
+      "step": 5143
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0019432519647585e-05,
+      "loss": 2.9077,
+      "step": 5144
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001619376948718e-05,
+      "loss": 1.9168,
+      "step": 5145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0012955017628127e-05,
+      "loss": 2.4862,
+      "step": 5146
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0009716264410153e-05,
+      "loss": 1.6643,
+      "step": 5147
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0006477510172984e-05,
+      "loss": 1.8944,
+      "step": 5148
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.000323875525636e-05,
+      "loss": 3.0264,
+      "step": 5149
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 1.7725,
+      "step": 5150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.996761244743643e-06,
+      "loss": 2.9747,
+      "step": 5151
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993522489827016e-06,
+      "loss": 2.7489,
+      "step": 5152
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99028373558985e-06,
+      "loss": 3.2115,
+      "step": 5153
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987044982371875e-06,
+      "loss": 3.7009,
+      "step": 5154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98380623051282e-06,
+      "loss": 1.7147,
+      "step": 5155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.980567480352417e-06,
+      "loss": 2.9046,
+      "step": 5156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.977328732230394e-06,
+      "loss": 3.8823,
+      "step": 5157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.974089986486488e-06,
+      "loss": 3.4832,
+      "step": 5158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.970851243460416e-06,
+      "loss": 2.4655,
+      "step": 5159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.967612503491915e-06,
+      "loss": 1.98,
+      "step": 5160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96437376692071e-06,
+      "loss": 2.0657,
+      "step": 5161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.961135034086533e-06,
+      "loss": 4.1095,
+      "step": 5162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.957896305329112e-06,
+      "loss": 2.711,
+      "step": 5163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.954657580988171e-06,
+      "loss": 2.0865,
+      "step": 5164
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95141886140344e-06,
+      "loss": 3.9225,
+      "step": 5165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.948180146914644e-06,
+      "loss": 3.1001,
+      "step": 5166
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.944941437861511e-06,
+      "loss": 3.6141,
+      "step": 5167
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.941702734583771e-06,
+      "loss": 2.5586,
+      "step": 5168
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.938464037421138e-06,
+      "loss": 3.2567,
+      "step": 5169
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.935225346713341e-06,
+      "loss": 3.5879,
+      "step": 5170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931986662800106e-06,
+      "loss": 2.1595,
+      "step": 5171
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.928747986021153e-06,
+      "loss": 2.5942,
+      "step": 5172
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925509316716208e-06,
+      "loss": 2.3948,
+      "step": 5173
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.922270655224989e-06,
+      "loss": 1.6328,
+      "step": 5174
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.919032001887215e-06,
+      "loss": 1.6951,
+      "step": 5175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91579335704261e-06,
+      "loss": 2.1745,
+      "step": 5176
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912554721030892e-06,
+      "loss": 2.7315,
+      "step": 5177
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90931609419178e-06,
+      "loss": 3.6791,
+      "step": 5178
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.906077476864986e-06,
+      "loss": 4.9894,
+      "step": 5179
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90283886939023e-06,
+      "loss": 4.1082,
+      "step": 5180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.899600272107223e-06,
+      "loss": 3.4232,
+      "step": 5181
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.896361685355681e-06,
+      "loss": 2.524,
+      "step": 5182
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.893123109475319e-06,
+      "loss": 2.1672,
+      "step": 5183
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88988454480585e-06,
+      "loss": 2.6054,
+      "step": 5184
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.886645991686977e-06,
+      "loss": 2.4317,
+      "step": 5185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883407450458415e-06,
+      "loss": 2.3831,
+      "step": 5186
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.880168921459869e-06,
+      "loss": 3.4452,
+      "step": 5187
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.876930405031047e-06,
+      "loss": 2.2023,
+      "step": 5188
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.873691901511657e-06,
+      "loss": 2.9041,
+      "step": 5189
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.870453411241399e-06,
+      "loss": 2.6048,
+      "step": 5190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.867214934559975e-06,
+      "loss": 1.7893,
+      "step": 5191
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86397647180709e-06,
+      "loss": 3.1738,
+      "step": 5192
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86073802332244e-06,
+      "loss": 2.4596,
+      "step": 5193
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.857499589445731e-06,
+      "loss": 2.9837,
+      "step": 5194
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.854261170516648e-06,
+      "loss": 2.2938,
+      "step": 5195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.851022766874892e-06,
+      "loss": 2.4819,
+      "step": 5196
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.847784378860158e-06,
+      "loss": 3.0314,
+      "step": 5197
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844546006812135e-06,
+      "loss": 1.9831,
+      "step": 5198
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.841307651070516e-06,
+      "loss": 2.6388,
+      "step": 5199
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.838069311974986e-06,
+      "loss": 2.9019,
+      "step": 5200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.834830989865233e-06,
+      "loss": 1.5185,
+      "step": 5201
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.831592685080945e-06,
+      "loss": 2.7632,
+      "step": 5202
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.828354397961801e-06,
+      "loss": 1.9257,
+      "step": 5203
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.825116128847488e-06,
+      "loss": 2.2769,
+      "step": 5204
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.821877878077678e-06,
+      "loss": 3.4979,
+      "step": 5205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.818639645992052e-06,
+      "loss": 2.9114,
+      "step": 5206
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.815401432930285e-06,
+      "loss": 3.098,
+      "step": 5207
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.812163239232051e-06,
+      "loss": 3.1383,
+      "step": 5208
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.808925065237023e-06,
+      "loss": 1.8686,
+      "step": 5209
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.805686911284867e-06,
+      "loss": 3.2761,
+      "step": 5210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802448777715253e-06,
+      "loss": 3.4174,
+      "step": 5211
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.799210664867844e-06,
+      "loss": 3.2783,
+      "step": 5212
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.795972573082305e-06,
+      "loss": 2.039,
+      "step": 5213
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.792734502698301e-06,
+      "loss": 3.1946,
+      "step": 5214
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789496454055482e-06,
+      "loss": 1.8794,
+      "step": 5215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.786258427493505e-06,
+      "loss": 2.5969,
+      "step": 5216
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.78302042335203e-06,
+      "loss": 2.8723,
+      "step": 5217
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.779782441970702e-06,
+      "loss": 1.8864,
+      "step": 5218
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.776544483689182e-06,
+      "loss": 1.6201,
+      "step": 5219
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.773306548847102e-06,
+      "loss": 3.4688,
+      "step": 5220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.770068637784114e-06,
+      "loss": 2.3605,
+      "step": 5221
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.766830750839857e-06,
+      "loss": 3.7287,
+      "step": 5222
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.763592888353975e-06,
+      "loss": 2.4886,
+      "step": 5223
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.760355050666102e-06,
+      "loss": 2.8095,
+      "step": 5224
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.757117238115871e-06,
+      "loss": 2.7251,
+      "step": 5225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.753879451042914e-06,
+      "loss": 3.2645,
+      "step": 5226
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.750641689786859e-06,
+      "loss": 2.1261,
+      "step": 5227
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.747403954687334e-06,
+      "loss": 2.7864,
+      "step": 5228
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.744166246083967e-06,
+      "loss": 2.2141,
+      "step": 5229
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.740928564316369e-06,
+      "loss": 1.9196,
+      "step": 5230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.737690909724162e-06,
+      "loss": 1.8233,
+      "step": 5231
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.734453282646962e-06,
+      "loss": 2.8751,
+      "step": 5232
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.731215683424378e-06,
+      "loss": 3.7544,
+      "step": 5233
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.727978112396026e-06,
+      "loss": 1.2156,
+      "step": 5234
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.724740569901503e-06,
+      "loss": 1.9392,
+      "step": 5235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.721503056280418e-06,
+      "loss": 2.798,
+      "step": 5236
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.71826557187237e-06,
+      "loss": 2.3756,
+      "step": 5237
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.715028117016955e-06,
+      "loss": 2.0655,
+      "step": 5238
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.711790692053773e-06,
+      "loss": 2.9593,
+      "step": 5239
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.708553297322407e-06,
+      "loss": 2.0571,
+      "step": 5240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.705315933162445e-06,
+      "loss": 2.9355,
+      "step": 5241
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.702078599913475e-06,
+      "loss": 1.5843,
+      "step": 5242
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.698841297915077e-06,
+      "loss": 1.8548,
+      "step": 5243
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.69560402750683e-06,
+      "loss": 2.5613,
+      "step": 5244
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.692366789028308e-06,
+      "loss": 3.4892,
+      "step": 5245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.689129582819081e-06,
+      "loss": 4.4012,
+      "step": 5246
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.685892409218718e-06,
+      "loss": 2.281,
+      "step": 5247
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.682655268566783e-06,
+      "loss": 3.005,
+      "step": 5248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.679418161202838e-06,
+      "loss": 2.8934,
+      "step": 5249
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.676181087466444e-06,
+      "loss": 3.1929,
+      "step": 5250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.672944047697147e-06,
+      "loss": 2.5831,
+      "step": 5251
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.669707042234502e-06,
+      "loss": 3.3111,
+      "step": 5252
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.666470071418053e-06,
+      "loss": 2.3515,
+      "step": 5253
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663233135587347e-06,
+      "loss": 2.4529,
+      "step": 5254
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.659996235081926e-06,
+      "loss": 3.3909,
+      "step": 5255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.656759370241318e-06,
+      "loss": 2.8821,
+      "step": 5256
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.65352254140506e-06,
+      "loss": 3.1212,
+      "step": 5257
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.650285748912678e-06,
+      "loss": 2.8633,
+      "step": 5258
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6470489931037e-06,
+      "loss": 2.5376,
+      "step": 5259
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.643812274317644e-06,
+      "loss": 2.6231,
+      "step": 5260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.640575592894027e-06,
+      "loss": 2.7568,
+      "step": 5261
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.637338949172362e-06,
+      "loss": 3.2147,
+      "step": 5262
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.634102343492157e-06,
+      "loss": 2.4767,
+      "step": 5263
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.630865776192918e-06,
+      "loss": 3.6255,
+      "step": 5264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.627629247614151e-06,
+      "loss": 4.3433,
+      "step": 5265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.624392758095344e-06,
+      "loss": 3.7088,
+      "step": 5266
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.621156307975992e-06,
+      "loss": 1.3703,
+      "step": 5267
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.617919897595586e-06,
+      "loss": 2.5581,
+      "step": 5268
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.614683527293608e-06,
+      "loss": 2.8336,
+      "step": 5269
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.611447197409544e-06,
+      "loss": 3.2505,
+      "step": 5270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.608210908282862e-06,
+      "loss": 2.8026,
+      "step": 5271
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.604974660253039e-06,
+      "loss": 3.2734,
+      "step": 5272
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.601738453659537e-06,
+      "loss": 1.1823,
+      "step": 5273
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.598502288841827e-06,
+      "loss": 1.5295,
+      "step": 5274
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.595266166139366e-06,
+      "loss": 2.7922,
+      "step": 5275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.592030085891602e-06,
+      "loss": 2.7151,
+      "step": 5276
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588794048437988e-06,
+      "loss": 2.2961,
+      "step": 5277
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58555805411797e-06,
+      "loss": 2.0202,
+      "step": 5278
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.582322103270989e-06,
+      "loss": 2.2485,
+      "step": 5279
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.579086196236483e-06,
+      "loss": 3.1112,
+      "step": 5280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.57585033335388e-06,
+      "loss": 2.0763,
+      "step": 5281
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.572614514962607e-06,
+      "loss": 3.2213,
+      "step": 5282
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.569378741402087e-06,
+      "loss": 3.6538,
+      "step": 5283
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.56614301301174e-06,
+      "loss": 3.2727,
+      "step": 5284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.562907330130981e-06,
+      "loss": 3.3342,
+      "step": 5285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.55967169309921e-06,
+      "loss": 3.2979,
+      "step": 5286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.556436102255834e-06,
+      "loss": 3.5945,
+      "step": 5287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.553200557940254e-06,
+      "loss": 2.2469,
+      "step": 5288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.549965060491862e-06,
+      "loss": 3.5346,
+      "step": 5289
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54672961025005e-06,
+      "loss": 2.8829,
+      "step": 5290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.543494207554193e-06,
+      "loss": 1.9336,
+      "step": 5291
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.540258852743676e-06,
+      "loss": 2.7361,
+      "step": 5292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.537023546157873e-06,
+      "loss": 3.0943,
+      "step": 5293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.533788288136149e-06,
+      "loss": 2.6893,
+      "step": 5294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.530553079017872e-06,
+      "loss": 3.9153,
+      "step": 5295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.527317919142398e-06,
+      "loss": 3.1628,
+      "step": 5296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.52408280884908e-06,
+      "loss": 2.9316,
+      "step": 5297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.520847748477266e-06,
+      "loss": 2.2821,
+      "step": 5298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.517612738366299e-06,
+      "loss": 2.577,
+      "step": 5299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.514377778855521e-06,
+      "loss": 2.8783,
+      "step": 5300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.511142870284257e-06,
+      "loss": 2.7351,
+      "step": 5301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507908012991835e-06,
+      "loss": 3.0054,
+      "step": 5302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50467320731758e-06,
+      "loss": 3.4878,
+      "step": 5303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.501438453600808e-06,
+      "loss": 2.4184,
+      "step": 5304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.498203752180827e-06,
+      "loss": 3.3855,
+      "step": 5305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.494969103396946e-06,
+      "loss": 3.0751,
+      "step": 5306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49173450758846e-06,
+      "loss": 2.9839,
+      "step": 5307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.488499965094664e-06,
+      "loss": 2.6581,
+      "step": 5308
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.485265476254849e-06,
+      "loss": 2.9175,
+      "step": 5309
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.482031041408296e-06,
+      "loss": 2.3912,
+      "step": 5310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.478796660894288e-06,
+      "loss": 2.4491,
+      "step": 5311
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.475562335052086e-06,
+      "loss": 2.889,
+      "step": 5312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47232806422096e-06,
+      "loss": 3.1844,
+      "step": 5313
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.469093848740172e-06,
+      "loss": 3.1125,
+      "step": 5314
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.465859688948977e-06,
+      "loss": 2.9474,
+      "step": 5315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.462625585186621e-06,
+      "loss": 3.1883,
+      "step": 5316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.459391537792347e-06,
+      "loss": 2.6376,
+      "step": 5317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45615754710539e-06,
+      "loss": 3.2997,
+      "step": 5318
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.452923613464984e-06,
+      "loss": 2.7144,
+      "step": 5319
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.449689737210352e-06,
+      "loss": 2.8189,
+      "step": 5320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.446455918680715e-06,
+      "loss": 3.5534,
+      "step": 5321
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.443222158215278e-06,
+      "loss": 3.3478,
+      "step": 5322
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.439988456153254e-06,
+      "loss": 2.5898,
+      "step": 5323
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.436754812833843e-06,
+      "loss": 2.2662,
+      "step": 5324
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.433521228596237e-06,
+      "loss": 4.1274,
+      "step": 5325
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43028770377963e-06,
+      "loss": 3.1257,
+      "step": 5326
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.427054238723193e-06,
+      "loss": 1.8756,
+      "step": 5327
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.423820833766108e-06,
+      "loss": 2.2611,
+      "step": 5328
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.420587489247544e-06,
+      "loss": 0.938,
+      "step": 5329
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.417354205506663e-06,
+      "loss": 3.4392,
+      "step": 5330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.414120982882622e-06,
+      "loss": 1.7598,
+      "step": 5331
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.410887821714571e-06,
+      "loss": 3.3838,
+      "step": 5332
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407654722341653e-06,
+      "loss": 1.9683,
+      "step": 5333
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.404421685103004e-06,
+      "loss": 2.2837,
+      "step": 5334
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.401188710337757e-06,
+      "loss": 2.4147,
+      "step": 5335
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.39795579838504e-06,
+      "loss": 3.3394,
+      "step": 5336
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.394722949583962e-06,
+      "loss": 2.8339,
+      "step": 5337
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.391490164273635e-06,
+      "loss": 2.4352,
+      "step": 5338
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388257442793167e-06,
+      "loss": 3.2252,
+      "step": 5339
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.385024785481653e-06,
+      "loss": 2.4488,
+      "step": 5340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.381792192678188e-06,
+      "loss": 2.2327,
+      "step": 5341
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37855966472185e-06,
+      "loss": 1.3425,
+      "step": 5342
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37532720195172e-06,
+      "loss": 2.6403,
+      "step": 5343
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.372094804706867e-06,
+      "loss": 0.8699,
+      "step": 5344
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.368862473326355e-06,
+      "loss": 3.2069,
+      "step": 5345
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.365630208149244e-06,
+      "loss": 3.5533,
+      "step": 5346
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.362398009514575e-06,
+      "loss": 2.3752,
+      "step": 5347
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.359165877761396e-06,
+      "loss": 1.4921,
+      "step": 5348
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.355933813228743e-06,
+      "loss": 3.3733,
+      "step": 5349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.352701816255643e-06,
+      "loss": 1.4745,
+      "step": 5350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.349469887181122e-06,
+      "loss": 3.2825,
+      "step": 5351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.346238026344186e-06,
+      "loss": 1.3826,
+      "step": 5352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.343006234083848e-06,
+      "loss": 2.9704,
+      "step": 5353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.339774510739107e-06,
+      "loss": 2.4611,
+      "step": 5354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.336542856648958e-06,
+      "loss": 3.4225,
+      "step": 5355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.333311272152385e-06,
+      "loss": 2.1273,
+      "step": 5356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.33007975758836e-06,
+      "loss": 3.2872,
+      "step": 5357
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32684831329586e-06,
+      "loss": 3.339,
+      "step": 5358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32361693961385e-06,
+      "loss": 2.7041,
+      "step": 5359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.320385636881283e-06,
+      "loss": 3.0149,
+      "step": 5360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.317154405437113e-06,
+      "loss": 3.1967,
+      "step": 5361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.313923245620271e-06,
+      "loss": 2.9929,
+      "step": 5362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.310692157769699e-06,
+      "loss": 2.3688,
+      "step": 5363
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.307461142224318e-06,
+      "loss": 3.1082,
+      "step": 5364
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30423019932305e-06,
+      "loss": 2.7842,
+      "step": 5365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.300999329404805e-06,
+      "loss": 3.3232,
+      "step": 5366
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29776853280849e-06,
+      "loss": 2.2838,
+      "step": 5367
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29453780987299e-06,
+      "loss": 2.415,
+      "step": 5368
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.291307160937205e-06,
+      "loss": 3.5445,
+      "step": 5369
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.288076586340005e-06,
+      "loss": 3.919,
+      "step": 5370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28484608642027e-06,
+      "loss": 2.822,
+      "step": 5371
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.281615661516866e-06,
+      "loss": 2.5998,
+      "step": 5372
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.278385311968638e-06,
+      "loss": 3.5941,
+      "step": 5373
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.275155038114442e-06,
+      "loss": 1.2906,
+      "step": 5374
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27192484029312e-06,
+      "loss": 2.0505,
+      "step": 5375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.268694718843503e-06,
+      "loss": 3.2948,
+      "step": 5376
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.265464674104418e-06,
+      "loss": 2.1674,
+      "step": 5377
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.262234706414677e-06,
+      "loss": 2.7108,
+      "step": 5378
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.259004816113092e-06,
+      "loss": 2.9583,
+      "step": 5379
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255775003538462e-06,
+      "loss": 1.9058,
+      "step": 5380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25254526902958e-06,
+      "loss": 2.3997,
+      "step": 5381
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.249315612925235e-06,
+      "loss": 4.1908,
+      "step": 5382
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.246086035564194e-06,
+      "loss": 2.0371,
+      "step": 5383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.242856537285227e-06,
+      "loss": 2.5845,
+      "step": 5384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.239627118427098e-06,
+      "loss": 4.0978,
+      "step": 5385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.236397779328554e-06,
+      "loss": 3.1833,
+      "step": 5386
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23316852032834e-06,
+      "loss": 3.4899,
+      "step": 5387
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.229939341765188e-06,
+      "loss": 2.151,
+      "step": 5388
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.226710243977825e-06,
+      "loss": 2.8194,
+      "step": 5389
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22348122730497e-06,
+      "loss": 3.4594,
+      "step": 5390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.220252292085328e-06,
+      "loss": 2.6758,
+      "step": 5391
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.217023438657606e-06,
+      "loss": 2.2471,
+      "step": 5392
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.213794667360488e-06,
+      "loss": 2.7481,
+      "step": 5393
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.21056597853266e-06,
+      "loss": 3.3171,
+      "step": 5394
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.207337372512797e-06,
+      "loss": 2.3126,
+      "step": 5395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.204108849639565e-06,
+      "loss": 2.2916,
+      "step": 5396
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.200880410251627e-06,
+      "loss": 2.3841,
+      "step": 5397
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.197652054687619e-06,
+      "loss": 2.7532,
+      "step": 5398
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.194423783286185e-06,
+      "loss": 2.444,
+      "step": 5399
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.19119559638596e-06,
+      "loss": 3.2955,
+      "step": 5400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.187967494325562e-06,
+      "loss": 2.2619,
+      "step": 5401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.184739477443608e-06,
+      "loss": 3.6363,
+      "step": 5402
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.181511546078696e-06,
+      "loss": 2.8861,
+      "step": 5403
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.178283700569424e-06,
+      "loss": 1.9638,
+      "step": 5404
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.17505594125438e-06,
+      "loss": 2.8694,
+      "step": 5405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.171828268472137e-06,
+      "loss": 2.9348,
+      "step": 5406
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16860068256127e-06,
+      "loss": 3.1038,
+      "step": 5407
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.165373183860329e-06,
+      "loss": 2.3672,
+      "step": 5408
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.162145772707867e-06,
+      "loss": 3.6853,
+      "step": 5409
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.158918449442425e-06,
+      "loss": 1.5146,
+      "step": 5410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.155691214402533e-06,
+      "loss": 2.5354,
+      "step": 5411
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.152464067926717e-06,
+      "loss": 2.6194,
+      "step": 5412
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.149237010353485e-06,
+      "loss": 3.1995,
+      "step": 5413
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.146010042021341e-06,
+      "loss": 2.807,
+      "step": 5414
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.142783163268782e-06,
+      "loss": 2.2502,
+      "step": 5415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.139556374434288e-06,
+      "loss": 2.1785,
+      "step": 5416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.136329675856344e-06,
+      "loss": 3.4808,
+      "step": 5417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.133103067873403e-06,
+      "loss": 2.2939,
+      "step": 5418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.129876550823926e-06,
+      "loss": 3.3971,
+      "step": 5419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.126650125046361e-06,
+      "loss": 3.2517,
+      "step": 5420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.123423790879146e-06,
+      "loss": 2.5009,
+      "step": 5421
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.120197548660707e-06,
+      "loss": 2.7358,
+      "step": 5422
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.116971398729462e-06,
+      "loss": 2.6198,
+      "step": 5423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.113745341423816e-06,
+      "loss": 2.9897,
+      "step": 5424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.110519377082174e-06,
+      "loss": 3.0831,
+      "step": 5425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107293506042918e-06,
+      "loss": 4.082,
+      "step": 5426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10406772864443e-06,
+      "loss": 1.9204,
+      "step": 5427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.100842045225084e-06,
+      "loss": 2.5204,
+      "step": 5428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09761645612323e-06,
+      "loss": 3.0609,
+      "step": 5429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.094390961677223e-06,
+      "loss": 2.5789,
+      "step": 5430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.091165562225401e-06,
+      "loss": 3.2441,
+      "step": 5431
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.087940258106093e-06,
+      "loss": 3.9768,
+      "step": 5432
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.084715049657622e-06,
+      "loss": 2.1375,
+      "step": 5433
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.081489937218293e-06,
+      "loss": 2.3733,
+      "step": 5434
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.078264921126405e-06,
+      "loss": 3.0306,
+      "step": 5435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.075040001720247e-06,
+      "loss": 2.3667,
+      "step": 5436
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.071815179338101e-06,
+      "loss": 2.56,
+      "step": 5437
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06859045431824e-06,
+      "loss": 2.1996,
+      "step": 5438
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.065365826998913e-06,
+      "loss": 3.2489,
+      "step": 5439
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.062141297718372e-06,
+      "loss": 3.1969,
+      "step": 5440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.058916866814857e-06,
+      "loss": 3.6686,
+      "step": 5441
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.055692534626595e-06,
+      "loss": 1.5004,
+      "step": 5442
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.052468301491809e-06,
+      "loss": 3.6656,
+      "step": 5443
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.049244167748694e-06,
+      "loss": 2.575,
+      "step": 5444
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.046020133735455e-06,
+      "loss": 2.3846,
+      "step": 5445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.042796199790274e-06,
+      "loss": 3.6866,
+      "step": 5446
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.039572366251333e-06,
+      "loss": 2.5652,
+      "step": 5447
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.036348633456791e-06,
+      "loss": 4.1148,
+      "step": 5448
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.033125001744807e-06,
+      "loss": 2.0646,
+      "step": 5449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02990147145352e-06,
+      "loss": 1.9092,
+      "step": 5450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.026678042921068e-06,
+      "loss": 3.3196,
+      "step": 5451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.023454716485572e-06,
+      "loss": 1.1642,
+      "step": 5452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.020231492485146e-06,
+      "loss": 2.2221,
+      "step": 5453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.017008371257887e-06,
+      "loss": 3.3442,
+      "step": 5454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.013785353141887e-06,
+      "loss": 2.4663,
+      "step": 5455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.010562438475225e-06,
+      "loss": 1.8718,
+      "step": 5456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.007339627595972e-06,
+      "loss": 2.211,
+      "step": 5457
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.004116920842188e-06,
+      "loss": 3.3057,
+      "step": 5458
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.000894318551914e-06,
+      "loss": 2.953,
+      "step": 5459
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.99767182106319e-06,
+      "loss": 2.3143,
+      "step": 5460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.994449428714041e-06,
+      "loss": 3.0792,
+      "step": 5461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.991227141842477e-06,
+      "loss": 2.4798,
+      "step": 5462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988004960786508e-06,
+      "loss": 2.1235,
+      "step": 5463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984782885884119e-06,
+      "loss": 2.2121,
+      "step": 5464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.981560917473292e-06,
+      "loss": 3.1873,
+      "step": 5465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.978339055892001e-06,
+      "loss": 2.971,
+      "step": 5466
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9751173014782e-06,
+      "loss": 2.6557,
+      "step": 5467
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.971895654569842e-06,
+      "loss": 2.8498,
+      "step": 5468
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.968674115504853e-06,
+      "loss": 3.2369,
+      "step": 5469
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.965452684621164e-06,
+      "loss": 2.7685,
+      "step": 5470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.962231362256688e-06,
+      "loss": 2.5159,
+      "step": 5471
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.959010148749324e-06,
+      "loss": 4.4584,
+      "step": 5472
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.955789044436966e-06,
+      "loss": 3.4671,
+      "step": 5473
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.952568049657488e-06,
+      "loss": 2.2877,
+      "step": 5474
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.949347164748761e-06,
+      "loss": 2.7397,
+      "step": 5475
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94612639004864e-06,
+      "loss": 1.8617,
+      "step": 5476
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94290572589497e-06,
+      "loss": 2.1573,
+      "step": 5477
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.939685172625588e-06,
+      "loss": 2.5479,
+      "step": 5478
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.936464730578305e-06,
+      "loss": 3.8025,
+      "step": 5479
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.933244400090937e-06,
+      "loss": 3.4059,
+      "step": 5480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93002418150128e-06,
+      "loss": 2.5818,
+      "step": 5481
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92680407514712e-06,
+      "loss": 3.9915,
+      "step": 5482
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.923584081366234e-06,
+      "loss": 2.9258,
+      "step": 5483
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92036420049638e-06,
+      "loss": 3.3997,
+      "step": 5484
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91714443287531e-06,
+      "loss": 1.8656,
+      "step": 5485
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913924778840764e-06,
+      "loss": 2.9785,
+      "step": 5486
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.910705238730467e-06,
+      "loss": 2.8727,
+      "step": 5487
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.907485812882137e-06,
+      "loss": 2.5976,
+      "step": 5488
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.904266501633478e-06,
+      "loss": 2.1195,
+      "step": 5489
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.901047305322172e-06,
+      "loss": 3.0714,
+      "step": 5490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897828224285905e-06,
+      "loss": 3.4068,
+      "step": 5491
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89460925886234e-06,
+      "loss": 1.0519,
+      "step": 5492
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.891390409389133e-06,
+      "loss": 3.0055,
+      "step": 5493
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.888171676203933e-06,
+      "loss": 3.8154,
+      "step": 5494
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88495305964436e-06,
+      "loss": 2.9502,
+      "step": 5495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.881734560048037e-06,
+      "loss": 1.8031,
+      "step": 5496
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.878516177752566e-06,
+      "loss": 2.8936,
+      "step": 5497
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.875297913095544e-06,
+      "loss": 3.1179,
+      "step": 5498
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.872079766414554e-06,
+      "loss": 2.5754,
+      "step": 5499
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.868861738047158e-06,
+      "loss": 2.1768,
+      "step": 5500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.865643828330917e-06,
+      "loss": 3.1569,
+      "step": 5501
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.862426037603373e-06,
+      "loss": 1.7371,
+      "step": 5502
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.859208366202059e-06,
+      "loss": 3.2147,
+      "step": 5503
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.855990814464497e-06,
+      "loss": 2.7852,
+      "step": 5504
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.852773382728184e-06,
+      "loss": 2.7145,
+      "step": 5505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.849556071330618e-06,
+      "loss": 2.0817,
+      "step": 5506
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.846338880609282e-06,
+      "loss": 3.8893,
+      "step": 5507
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.843121810901643e-06,
+      "loss": 2.2388,
+      "step": 5508
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.839904862545159e-06,
+      "loss": 3.2717,
+      "step": 5509
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.836688035877268e-06,
+      "loss": 2.4745,
+      "step": 5510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.833471331235404e-06,
+      "loss": 2.8487,
+      "step": 5511
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.830254748956983e-06,
+      "loss": 3.6441,
+      "step": 5512
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.827038289379409e-06,
+      "loss": 2.8321,
+      "step": 5513
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.823821952840079e-06,
+      "loss": 2.7903,
+      "step": 5514
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.820605739676363e-06,
+      "loss": 2.7462,
+      "step": 5515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.817389650225631e-06,
+      "loss": 2.6708,
+      "step": 5516
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.814173684825237e-06,
+      "loss": 3.2414,
+      "step": 5517
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81095784381252e-06,
+      "loss": 1.6985,
+      "step": 5518
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.807742127524808e-06,
+      "loss": 3.8549,
+      "step": 5519
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.804526536299413e-06,
+      "loss": 3.5424,
+      "step": 5520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.801311070473634e-06,
+      "loss": 2.1168,
+      "step": 5521
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79809573038476e-06,
+      "loss": 3.1224,
+      "step": 5522
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.794880516370066e-06,
+      "loss": 1.0389,
+      "step": 5523
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79166542876682e-06,
+      "loss": 2.2124,
+      "step": 5524
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.788450467912254e-06,
+      "loss": 4.0084,
+      "step": 5525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.785235634143612e-06,
+      "loss": 2.8496,
+      "step": 5526
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.782020927798114e-06,
+      "loss": 3.413,
+      "step": 5527
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.778806349212968e-06,
+      "loss": 4.0888,
+      "step": 5528
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.775591898725374e-06,
+      "loss": 2.1727,
+      "step": 5529
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.772377576672502e-06,
+      "loss": 2.0589,
+      "step": 5530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.769163383391525e-06,
+      "loss": 2.8133,
+      "step": 5531
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.765949319219595e-06,
+      "loss": 2.4598,
+      "step": 5532
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.762735384493856e-06,
+      "loss": 2.6453,
+      "step": 5533
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.759521579551434e-06,
+      "loss": 3.8288,
+      "step": 5534
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75630790472944e-06,
+      "loss": 4.1106,
+      "step": 5535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.753094360364973e-06,
+      "loss": 2.5674,
+      "step": 5536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.749880946795122e-06,
+      "loss": 1.196,
+      "step": 5537
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.746667664356957e-06,
+      "loss": 2.9948,
+      "step": 5538
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.743454513387541e-06,
+      "loss": 2.6441,
+      "step": 5539
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.740241494223911e-06,
+      "loss": 1.8584,
+      "step": 5540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.737028607203101e-06,
+      "loss": 4.169,
+      "step": 5541
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.73381585266213e-06,
+      "loss": 2.7623,
+      "step": 5542
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.730603230938e-06,
+      "loss": 2.345,
+      "step": 5543
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.727390742367698e-06,
+      "loss": 1.9038,
+      "step": 5544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.724178387288202e-06,
+      "loss": 2.8805,
+      "step": 5545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.720966166036471e-06,
+      "loss": 1.8779,
+      "step": 5546
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.717754078949451e-06,
+      "loss": 2.3317,
+      "step": 5547
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.71454212636408e-06,
+      "loss": 1.4746,
+      "step": 5548
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.711330308617274e-06,
+      "loss": 2.4275,
+      "step": 5549
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.708118626045939e-06,
+      "loss": 2.2634,
+      "step": 5550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.704907078986962e-06,
+      "loss": 2.1174,
+      "step": 5551
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.701695667777221e-06,
+      "loss": 1.8567,
+      "step": 5552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69848439275358e-06,
+      "loss": 3.8995,
+      "step": 5553
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.695273254252884e-06,
+      "loss": 3.4656,
+      "step": 5554
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.692062252611973e-06,
+      "loss": 2.3538,
+      "step": 5555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.688851388167658e-06,
+      "loss": 3.1548,
+      "step": 5556
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.685640661256746e-06,
+      "loss": 3.6231,
+      "step": 5557
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.682430072216029e-06,
+      "loss": 2.4463,
+      "step": 5558
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.679219621382283e-06,
+      "loss": 2.5772,
+      "step": 5559
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.676009309092273e-06,
+      "loss": 2.4571,
+      "step": 5560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.672799135682735e-06,
+      "loss": 3.1287,
+      "step": 5561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66958910149041e-06,
+      "loss": 2.1514,
+      "step": 5562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.666379206852012e-06,
+      "loss": 2.6003,
+      "step": 5563
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.663169452104248e-06,
+      "loss": 2.3063,
+      "step": 5564
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.659959837583808e-06,
+      "loss": 1.7129,
+      "step": 5565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.656750363627356e-06,
+      "loss": 2.0141,
+      "step": 5566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.653541030571559e-06,
+      "loss": 2.1321,
+      "step": 5567
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.650331838753057e-06,
+      "loss": 2.6065,
+      "step": 5568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.647122788508482e-06,
+      "loss": 3.4016,
+      "step": 5569
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.643913880174449e-06,
+      "loss": 2.266,
+      "step": 5570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.640705114087554e-06,
+      "loss": 1.06,
+      "step": 5571
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.637496490584385e-06,
+      "loss": 1.9473,
+      "step": 5572
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63428801000151e-06,
+      "loss": 2.1151,
+      "step": 5573
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.631079672675483e-06,
+      "loss": 2.2789,
+      "step": 5574
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62787147894285e-06,
+      "loss": 2.6552,
+      "step": 5575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.624663429140128e-06,
+      "loss": 3.2672,
+      "step": 5576
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.621455523603827e-06,
+      "loss": 2.8926,
+      "step": 5577
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.618247762670445e-06,
+      "loss": 2.7303,
+      "step": 5578
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61504014667646e-06,
+      "loss": 1.3547,
+      "step": 5579
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.611832675958335e-06,
+      "loss": 1.6683,
+      "step": 5580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60862535085252e-06,
+      "loss": 2.1583,
+      "step": 5581
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.605418171695447e-06,
+      "loss": 1.6006,
+      "step": 5582
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.602211138823537e-06,
+      "loss": 2.5024,
+      "step": 5583
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.599004252573191e-06,
+      "loss": 3.402,
+      "step": 5584
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.595797513280799e-06,
+      "loss": 2.767,
+      "step": 5585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.592590921282728e-06,
+      "loss": 2.4274,
+      "step": 5586
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.589384476915337e-06,
+      "loss": 2.4923,
+      "step": 5587
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.586178180514968e-06,
+      "loss": 1.4211,
+      "step": 5588
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.582972032417946e-06,
+      "loss": 3.1155,
+      "step": 5589
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.579766032960582e-06,
+      "loss": 3.1355,
+      "step": 5590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.576560182479168e-06,
+      "loss": 2.3643,
+      "step": 5591
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.573354481309986e-06,
+      "loss": 2.87,
+      "step": 5592
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.570148929789297e-06,
+      "loss": 3.1893,
+      "step": 5593
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.566943528253348e-06,
+      "loss": 2.4372,
+      "step": 5594
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.563738277038376e-06,
+      "loss": 3.7518,
+      "step": 5595
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.560533176480588e-06,
+      "loss": 3.0875,
+      "step": 5596
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.557328226916191e-06,
+      "loss": 2.8571,
+      "step": 5597
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.554123428681367e-06,
+      "loss": 1.535,
+      "step": 5598
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.550918782112284e-06,
+      "loss": 2.9572,
+      "step": 5599
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5477142875451e-06,
+      "loss": 1.7061,
+      "step": 5600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.544509945315943e-06,
+      "loss": 1.5372,
+      "step": 5601
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.541305755760938e-06,
+      "loss": 2.4308,
+      "step": 5602
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.538101719216189e-06,
+      "loss": 2.1776,
+      "step": 5603
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.534897836017784e-06,
+      "loss": 3.5573,
+      "step": 5604
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.531694106501796e-06,
+      "loss": 1.1513,
+      "step": 5605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.528490531004285e-06,
+      "loss": 2.5956,
+      "step": 5606
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.525287109861285e-06,
+      "loss": 3.0701,
+      "step": 5607
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522083843408823e-06,
+      "loss": 2.0198,
+      "step": 5608
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.518880731982908e-06,
+      "loss": 1.2699,
+      "step": 5609
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.515677775919528e-06,
+      "loss": 1.8778,
+      "step": 5610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.512474975554665e-06,
+      "loss": 3.2769,
+      "step": 5611
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50927233122427e-06,
+      "loss": 1.608,
+      "step": 5612
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.506069843264288e-06,
+      "loss": 2.4609,
+      "step": 5613
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502867512010645e-06,
+      "loss": 1.8961,
+      "step": 5614
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.499665337799254e-06,
+      "loss": 2.0559,
+      "step": 5615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.496463320966004e-06,
+      "loss": 2.7437,
+      "step": 5616
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.493261461846775e-06,
+      "loss": 2.2167,
+      "step": 5617
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.490059760777425e-06,
+      "loss": 3.1959,
+      "step": 5618
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.486858218093798e-06,
+      "loss": 2.1881,
+      "step": 5619
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48365683413172e-06,
+      "loss": 1.921,
+      "step": 5620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.480455609227009e-06,
+      "loss": 2.1671,
+      "step": 5621
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.477254543715446e-06,
+      "loss": 3.2987,
+      "step": 5622
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.474053637932816e-06,
+      "loss": 2.4504,
+      "step": 5623
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.470852892214875e-06,
+      "loss": 2.98,
+      "step": 5624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46765230689737e-06,
+      "loss": 3.2162,
+      "step": 5625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.464451882316028e-06,
+      "loss": 2.6112,
+      "step": 5626
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.461251618806554e-06,
+      "loss": 1.5759,
+      "step": 5627
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.458051516704644e-06,
+      "loss": 3.0219,
+      "step": 5628
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.454851576345975e-06,
+      "loss": 2.6991,
+      "step": 5629
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.451651798066203e-06,
+      "loss": 3.6187,
+      "step": 5630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.448452182200974e-06,
+      "loss": 1.6455,
+      "step": 5631
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.445252729085907e-06,
+      "loss": 2.4395,
+      "step": 5632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.442053439056613e-06,
+      "loss": 1.437,
+      "step": 5633
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.438854312448683e-06,
+      "loss": 2.3694,
+      "step": 5634
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43565534959769e-06,
+      "loss": 2.831,
+      "step": 5635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.432456550839196e-06,
+      "loss": 3.1602,
+      "step": 5636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.429257916508729e-06,
+      "loss": 3.5306,
+      "step": 5637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.426059446941817e-06,
+      "loss": 2.2387,
+      "step": 5638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.422861142473964e-06,
+      "loss": 1.9524,
+      "step": 5639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.419663003440657e-06,
+      "loss": 2.5416,
+      "step": 5640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.416465030177369e-06,
+      "loss": 2.6407,
+      "step": 5641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.413267223019548e-06,
+      "loss": 2.6903,
+      "step": 5642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41006958230263e-06,
+      "loss": 1.0307,
+      "step": 5643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.406872108362034e-06,
+      "loss": 3.3085,
+      "step": 5644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40367480153316e-06,
+      "loss": 2.0222,
+      "step": 5645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.400477662151397e-06,
+      "loss": 2.5408,
+      "step": 5646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.397280690552096e-06,
+      "loss": 2.516,
+      "step": 5647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.394083887070614e-06,
+      "loss": 2.5947,
+      "step": 5648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390887252042279e-06,
+      "loss": 2.8851,
+      "step": 5649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.387690785802403e-06,
+      "loss": 2.622,
+      "step": 5650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.384494488686282e-06,
+      "loss": 3.1738,
+      "step": 5651
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38129836102919e-06,
+      "loss": 2.6026,
+      "step": 5652
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.378102403166388e-06,
+      "loss": 3.3535,
+      "step": 5653
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.374906615433114e-06,
+      "loss": 2.3433,
+      "step": 5654
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.371710998164595e-06,
+      "loss": 2.7983,
+      "step": 5655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36851555169604e-06,
+      "loss": 2.6251,
+      "step": 5656
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.365320276362628e-06,
+      "loss": 2.6394,
+      "step": 5657
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36212517249953e-06,
+      "loss": 3.7425,
+      "step": 5658
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358930240441902e-06,
+      "loss": 3.6118,
+      "step": 5659
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.355735480524874e-06,
+      "loss": 2.7144,
+      "step": 5660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.352540893083565e-06,
+      "loss": 2.7478,
+      "step": 5661
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.349346478453068e-06,
+      "loss": 3.2821,
+      "step": 5662
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.346152236968465e-06,
+      "loss": 3.0519,
+      "step": 5663
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.342958168964816e-06,
+      "loss": 2.6947,
+      "step": 5664
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.339764274777165e-06,
+      "loss": 2.7148,
+      "step": 5665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.336570554740538e-06,
+      "loss": 1.7469,
+      "step": 5666
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.333377009189939e-06,
+      "loss": 2.657,
+      "step": 5667
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.330183638460356e-06,
+      "loss": 3.3542,
+      "step": 5668
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32699044288676e-06,
+      "loss": 2.9676,
+      "step": 5669
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3237974228041e-06,
+      "loss": 3.4427,
+      "step": 5670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.320604578547312e-06,
+      "loss": 2.2532,
+      "step": 5671
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.317411910451313e-06,
+      "loss": 2.5847,
+      "step": 5672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31421941885099e-06,
+      "loss": 2.6788,
+      "step": 5673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.311027104081228e-06,
+      "loss": 1.9339,
+      "step": 5674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.307834966476885e-06,
+      "loss": 2.3126,
+      "step": 5675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.304643006372797e-06,
+      "loss": 1.8589,
+      "step": 5676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.301451224103796e-06,
+      "loss": 2.4109,
+      "step": 5677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29825962000467e-06,
+      "loss": 3.4596,
+      "step": 5678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.295068194410218e-06,
+      "loss": 1.6386,
+      "step": 5679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.291876947655197e-06,
+      "loss": 1.4367,
+      "step": 5680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.288685880074357e-06,
+      "loss": 3.6558,
+      "step": 5681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.285494992002429e-06,
+      "loss": 2.4251,
+      "step": 5682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.282304283774116e-06,
+      "loss": 1.9794,
+      "step": 5683
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.27911375572411e-06,
+      "loss": 3.2837,
+      "step": 5684
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.275923408187086e-06,
+      "loss": 2.2742,
+      "step": 5685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272733241497694e-06,
+      "loss": 2.8438,
+      "step": 5686
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.26954325599057e-06,
+      "loss": 3.8151,
+      "step": 5687
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.266353452000326e-06,
+      "loss": 2.2269,
+      "step": 5688
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.263163829861558e-06,
+      "loss": 2.0812,
+      "step": 5689
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.259974389908842e-06,
+      "loss": 3.0223,
+      "step": 5690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.256785132476738e-06,
+      "loss": 3.0365,
+      "step": 5691
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.253596057899788e-06,
+      "loss": 2.7045,
+      "step": 5692
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.250407166512502e-06,
+      "loss": 2.1214,
+      "step": 5693
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247218458649385e-06,
+      "loss": 2.2658,
+      "step": 5694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.244029934644916e-06,
+      "loss": 2.2578,
+      "step": 5695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.240841594833554e-06,
+      "loss": 3.0146,
+      "step": 5696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23765343954975e-06,
+      "loss": 3.7315,
+      "step": 5697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.234465469127919e-06,
+      "loss": 3.0561,
+      "step": 5698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.231277683902466e-06,
+      "loss": 2.4767,
+      "step": 5699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.228090084207773e-06,
+      "loss": 3.0316,
+      "step": 5700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.224902670378207e-06,
+      "loss": 2.2679,
+      "step": 5701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221715442748117e-06,
+      "loss": 2.0578,
+      "step": 5702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21852840165182e-06,
+      "loss": 2.3338,
+      "step": 5703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.215341547423624e-06,
+      "loss": 2.3639,
+      "step": 5704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.212154880397817e-06,
+      "loss": 2.7765,
+      "step": 5705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.208968400908668e-06,
+      "loss": 2.9211,
+      "step": 5706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.205782109290424e-06,
+      "loss": 2.7986,
+      "step": 5707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202596005877307e-06,
+      "loss": 1.6292,
+      "step": 5708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.199410091003523e-06,
+      "loss": 1.7931,
+      "step": 5709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.196224365003267e-06,
+      "loss": 2.6243,
+      "step": 5710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.193038828210703e-06,
+      "loss": 2.7958,
+      "step": 5711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.189853480959982e-06,
+      "loss": 2.4557,
+      "step": 5712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.186668323585226e-06,
+      "loss": 2.9015,
+      "step": 5713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183483356420547e-06,
+      "loss": 2.1065,
+      "step": 5714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.180298579800034e-06,
+      "loss": 2.569,
+      "step": 5715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.177113994057756e-06,
+      "loss": 1.9834,
+      "step": 5716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.173929599527761e-06,
+      "loss": 2.5601,
+      "step": 5717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.170745396544072e-06,
+      "loss": 2.3991,
+      "step": 5718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.167561385440704e-06,
+      "loss": 3.0827,
+      "step": 5719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16437756655164e-06,
+      "loss": 3.4662,
+      "step": 5720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16119394021085e-06,
+      "loss": 2.7847,
+      "step": 5721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.158010506752283e-06,
+      "loss": 2.5602,
+      "step": 5722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.154827266509865e-06,
+      "loss": 3.081,
+      "step": 5723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1516442198175e-06,
+      "loss": 2.0641,
+      "step": 5724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.148461367009081e-06,
+      "loss": 1.8165,
+      "step": 5725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14527870841847e-06,
+      "loss": 3.5662,
+      "step": 5726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142096244379514e-06,
+      "loss": 1.1633,
+      "step": 5727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.138913975226044e-06,
+      "loss": 1.9202,
+      "step": 5728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.135731901291854e-06,
+      "loss": 2.3924,
+      "step": 5729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.132550022910737e-06,
+      "loss": 2.7685,
+      "step": 5730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.129368340416454e-06,
+      "loss": 3.3682,
+      "step": 5731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.126186854142752e-06,
+      "loss": 1.945,
+      "step": 5732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.123005564423356e-06,
+      "loss": 3.1443,
+      "step": 5733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.119824471591962e-06,
+      "loss": 3.0073,
+      "step": 5734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.116643575982254e-06,
+      "loss": 2.5748,
+      "step": 5735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.113462877927893e-06,
+      "loss": 3.3695,
+      "step": 5736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.110282377762522e-06,
+      "loss": 2.5716,
+      "step": 5737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10710207581976e-06,
+      "loss": 3.2731,
+      "step": 5738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103921972433205e-06,
+      "loss": 1.2964,
+      "step": 5739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.100742067936432e-06,
+      "loss": 2.9353,
+      "step": 5740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.097562362663002e-06,
+      "loss": 2.8133,
+      "step": 5741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094382856946452e-06,
+      "loss": 3.3825,
+      "step": 5742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.0912035511203e-06,
+      "loss": 2.1088,
+      "step": 5743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.088024445518033e-06,
+      "loss": 3.3348,
+      "step": 5744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.084845540473127e-06,
+      "loss": 3.599,
+      "step": 5745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.081666836319034e-06,
+      "loss": 2.4926,
+      "step": 5746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.078488333389188e-06,
+      "loss": 1.676,
+      "step": 5747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.075310032017e-06,
+      "loss": 3.0117,
+      "step": 5748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.072131932535855e-06,
+      "loss": 2.6494,
+      "step": 5749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.068954035279121e-06,
+      "loss": 2.5646,
+      "step": 5750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.065776340580148e-06,
+      "loss": 1.5312,
+      "step": 5751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.062598848772261e-06,
+      "loss": 3.211,
+      "step": 5752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.059421560188765e-06,
+      "loss": 3.7849,
+      "step": 5753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.056244475162938e-06,
+      "loss": 2.4938,
+      "step": 5754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053067594028044e-06,
+      "loss": 3.1313,
+      "step": 5755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.049890917117322e-06,
+      "loss": 2.6859,
+      "step": 5756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.046714444763994e-06,
+      "loss": 3.0984,
+      "step": 5757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.043538177301256e-06,
+      "loss": 0.9774,
+      "step": 5758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.040362115062282e-06,
+      "loss": 2.0434,
+      "step": 5759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.037186258380226e-06,
+      "loss": 3.1225,
+      "step": 5760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.034010607588223e-06,
+      "loss": 3.2948,
+      "step": 5761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.030835163019382e-06,
+      "loss": 2.7138,
+      "step": 5762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.027659925006798e-06,
+      "loss": 2.9961,
+      "step": 5763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02448489388353e-06,
+      "loss": 2.7909,
+      "step": 5764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.021310069982624e-06,
+      "loss": 3.4333,
+      "step": 5765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.018135453637112e-06,
+      "loss": 2.932,
+      "step": 5766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.014961045179992e-06,
+      "loss": 3.7483,
+      "step": 5767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01178684494425e-06,
+      "loss": 1.4834,
+      "step": 5768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.008612853262837e-06,
+      "loss": 2.8236,
+      "step": 5769
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.005439070468692e-06,
+      "loss": 3.6564,
+      "step": 5770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002265496894733e-06,
+      "loss": 2.6532,
+      "step": 5771
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999092132873851e-06,
+      "loss": 2.9695,
+      "step": 5772
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.995918978738918e-06,
+      "loss": 1.9341,
+      "step": 5773
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.992746034822783e-06,
+      "loss": 2.7757,
+      "step": 5774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.989573301458274e-06,
+      "loss": 2.1205,
+      "step": 5775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.986400778978192e-06,
+      "loss": 1.9822,
+      "step": 5776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.983228467715325e-06,
+      "loss": 1.5723,
+      "step": 5777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.980056368002435e-06,
+      "loss": 3.0311,
+      "step": 5778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.976884480172254e-06,
+      "loss": 2.8543,
+      "step": 5779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9737128045575e-06,
+      "loss": 3.5187,
+      "step": 5780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97054134149087e-06,
+      "loss": 2.0189,
+      "step": 5781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.967370091305033e-06,
+      "loss": 3.6602,
+      "step": 5782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.964199054332642e-06,
+      "loss": 2.0902,
+      "step": 5783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96102823090632e-06,
+      "loss": 2.7622,
+      "step": 5784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.957857621358674e-06,
+      "loss": 2.2436,
+      "step": 5785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.954687226022285e-06,
+      "loss": 2.4471,
+      "step": 5786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.951517045229713e-06,
+      "loss": 1.6881,
+      "step": 5787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948347079313494e-06,
+      "loss": 2.9468,
+      "step": 5788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.945177328606153e-06,
+      "loss": 2.5149,
+      "step": 5789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.942007793440165e-06,
+      "loss": 2.1669,
+      "step": 5790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93883847414801e-06,
+      "loss": 2.4348,
+      "step": 5791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.935669371062132e-06,
+      "loss": 2.9782,
+      "step": 5792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932500484514957e-06,
+      "loss": 2.3034,
+      "step": 5793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.929331814838889e-06,
+      "loss": 1.4863,
+      "step": 5794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9261633623663e-06,
+      "loss": 3.1848,
+      "step": 5795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922995127429547e-06,
+      "loss": 2.6763,
+      "step": 5796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.919827110360968e-06,
+      "loss": 2.0717,
+      "step": 5797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.916659311492871e-06,
+      "loss": 2.9795,
+      "step": 5798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.913491731157546e-06,
+      "loss": 2.2128,
+      "step": 5799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91032436968725e-06,
+      "loss": 3.3647,
+      "step": 5800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90715722741423e-06,
+      "loss": 2.9082,
+      "step": 5801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.903990304670704e-06,
+      "loss": 2.501,
+      "step": 5802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.900823601788867e-06,
+      "loss": 3.0829,
+      "step": 5803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.897657119100896e-06,
+      "loss": 3.8,
+      "step": 5804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.894490856938931e-06,
+      "loss": 3.4249,
+      "step": 5805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.891324815635104e-06,
+      "loss": 3.12,
+      "step": 5806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.888158995521518e-06,
+      "loss": 3.1706,
+      "step": 5807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88499339693025e-06,
+      "loss": 2.8146,
+      "step": 5808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.881828020193362e-06,
+      "loss": 3.5416,
+      "step": 5809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.87866286564288e-06,
+      "loss": 2.4827,
+      "step": 5810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.87549793361082e-06,
+      "loss": 3.2815,
+      "step": 5811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.872333224429166e-06,
+      "loss": 3.1889,
+      "step": 5812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.869168738429883e-06,
+      "loss": 3.0642,
+      "step": 5813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.866004475944913e-06,
+      "loss": 2.6193,
+      "step": 5814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.862840437306165e-06,
+      "loss": 2.9663,
+      "step": 5815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.859676622845535e-06,
+      "loss": 2.0095,
+      "step": 5816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.856513032894896e-06,
+      "loss": 1.8822,
+      "step": 5817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85334966778609e-06,
+      "loss": 2.4329,
+      "step": 5818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.850186527850944e-06,
+      "loss": 2.9275,
+      "step": 5819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.847023613421251e-06,
+      "loss": 3.6435,
+      "step": 5820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84386092482879e-06,
+      "loss": 2.3364,
+      "step": 5821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84069846240531e-06,
+      "loss": 1.8679,
+      "step": 5822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83753622648254e-06,
+      "loss": 3.1605,
+      "step": 5823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.834374217392188e-06,
+      "loss": 2.6805,
+      "step": 5824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.831212435465925e-06,
+      "loss": 2.0488,
+      "step": 5825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.828050881035412e-06,
+      "loss": 3.5828,
+      "step": 5826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.82488955443228e-06,
+      "loss": 2.9905,
+      "step": 5827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.82172845598814e-06,
+      "loss": 2.6307,
+      "step": 5828
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.818567586034578e-06,
+      "loss": 1.8856,
+      "step": 5829
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.815406944903148e-06,
+      "loss": 2.8196,
+      "step": 5830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.812246532925392e-06,
+      "loss": 2.6122,
+      "step": 5831
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80908635043282e-06,
+      "loss": 2.9973,
+      "step": 5832
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.805926397756921e-06,
+      "loss": 2.5627,
+      "step": 5833
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.802766675229166e-06,
+      "loss": 2.5032,
+      "step": 5834
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.799607183180981e-06,
+      "loss": 2.3088,
+      "step": 5835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.796447921943793e-06,
+      "loss": 2.1321,
+      "step": 5836
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79328889184899e-06,
+      "loss": 2.2988,
+      "step": 5837
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.790130093227943e-06,
+      "loss": 3.2949,
+      "step": 5838
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.786971526411996e-06,
+      "loss": 2.8376,
+      "step": 5839
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78381319173246e-06,
+      "loss": 3.2057,
+      "step": 5840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.780655089520634e-06,
+      "loss": 1.2551,
+      "step": 5841
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777497220107791e-06,
+      "loss": 3.0052,
+      "step": 5842
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.774339583825173e-06,
+      "loss": 2.8662,
+      "step": 5843
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.771182181004005e-06,
+      "loss": 2.3474,
+      "step": 5844
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.768025011975481e-06,
+      "loss": 2.4879,
+      "step": 5845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.764868077070772e-06,
+      "loss": 3.3402,
+      "step": 5846
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.761711376621032e-06,
+      "loss": 2.2163,
+      "step": 5847
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.758554910957378e-06,
+      "loss": 3.1622,
+      "step": 5848
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75539868041091e-06,
+      "loss": 3.3157,
+      "step": 5849
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.752242685312709e-06,
+      "loss": 2.385,
+      "step": 5850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.749086925993814e-06,
+      "loss": 3.8779,
+      "step": 5851
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.745931402785252e-06,
+      "loss": 1.9877,
+      "step": 5852
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.742776116018024e-06,
+      "loss": 2.2337,
+      "step": 5853
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.739621066023108e-06,
+      "loss": 2.8013,
+      "step": 5854
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.736466253131451e-06,
+      "loss": 3.6333,
+      "step": 5855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.733311677673979e-06,
+      "loss": 2.9687,
+      "step": 5856
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.730157339981591e-06,
+      "loss": 2.8648,
+      "step": 5857
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.727003240385163e-06,
+      "loss": 3.1208,
+      "step": 5858
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.723849379215545e-06,
+      "loss": 3.1738,
+      "step": 5859
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.720695756803569e-06,
+      "loss": 3.0409,
+      "step": 5860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.717542373480025e-06,
+      "loss": 3.1012,
+      "step": 5861
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.714389229575692e-06,
+      "loss": 2.9356,
+      "step": 5862
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71123632542132e-06,
+      "loss": 2.0765,
+      "step": 5863
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.708083661347637e-06,
+      "loss": 1.6668,
+      "step": 5864
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.704931237685342e-06,
+      "loss": 1.8533,
+      "step": 5865
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.701779054765107e-06,
+      "loss": 3.645,
+      "step": 5866
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.698627112917583e-06,
+      "loss": 2.2267,
+      "step": 5867
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.695475412473393e-06,
+      "loss": 2.6689,
+      "step": 5868
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.692323953763136e-06,
+      "loss": 3.0121,
+      "step": 5869
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.689172737117389e-06,
+      "loss": 2.3224,
+      "step": 5870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.686021762866695e-06,
+      "loss": 3.4611,
+      "step": 5871
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.682871031341579e-06,
+      "loss": 2.6733,
+      "step": 5872
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.679720542872537e-06,
+      "loss": 3.5771,
+      "step": 5873
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.676570297790042e-06,
+      "loss": 2.656,
+      "step": 5874
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.673420296424541e-06,
+      "loss": 2.0432,
+      "step": 5875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670270539106452e-06,
+      "loss": 2.6064,
+      "step": 5876
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.667121026166168e-06,
+      "loss": 1.7011,
+      "step": 5877
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.663971757934064e-06,
+      "loss": 2.0669,
+      "step": 5878
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660822734740478e-06,
+      "loss": 2.4955,
+      "step": 5879
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.657673956915735e-06,
+      "loss": 1.7151,
+      "step": 5880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.654525424790121e-06,
+      "loss": 2.0037,
+      "step": 5881
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651377138693903e-06,
+      "loss": 1.5037,
+      "step": 5882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.648229098957323e-06,
+      "loss": 2.1756,
+      "step": 5883
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.645081305910596e-06,
+      "loss": 2.9159,
+      "step": 5884
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.641933759883913e-06,
+      "loss": 1.0467,
+      "step": 5885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.638786461207433e-06,
+      "loss": 2.2118,
+      "step": 5886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.635639410211292e-06,
+      "loss": 2.5455,
+      "step": 5887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.632492607225604e-06,
+      "loss": 2.2825,
+      "step": 5888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.629346052580455e-06,
+      "loss": 2.6772,
+      "step": 5889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6261997466059035e-06,
+      "loss": 3.0301,
+      "step": 5890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.62305368963198e-06,
+      "loss": 2.5887,
+      "step": 5891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.619907881988692e-06,
+      "loss": 3.2188,
+      "step": 5892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6167623240060205e-06,
+      "loss": 1.9574,
+      "step": 5893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.613617016013921e-06,
+      "loss": 3.5676,
+      "step": 5894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.610471958342326e-06,
+      "loss": 3.3906,
+      "step": 5895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607327151321127e-06,
+      "loss": 3.0271,
+      "step": 5896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.604182595280205e-06,
+      "loss": 2.2499,
+      "step": 5897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.60103829054941e-06,
+      "loss": 2.2466,
+      "step": 5898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597894237458564e-06,
+      "loss": 2.7444,
+      "step": 5899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.594750436337467e-06,
+      "loss": 2.766,
+      "step": 5900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.591606887515883e-06,
+      "loss": 2.2104,
+      "step": 5901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.588463591323559e-06,
+      "loss": 2.0489,
+      "step": 5902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.585320548090213e-06,
+      "loss": 2.905,
+      "step": 5903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.582177758145532e-06,
+      "loss": 2.5306,
+      "step": 5904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.579035221819188e-06,
+      "loss": 2.134,
+      "step": 5905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.57589293944081e-06,
+      "loss": 3.8259,
+      "step": 5906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.572750911340011e-06,
+      "loss": 1.9,
+      "step": 5907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.569609137846376e-06,
+      "loss": 3.2458,
+      "step": 5908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.566467619289464e-06,
+      "loss": 1.9898,
+      "step": 5909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5633263559988035e-06,
+      "loss": 2.939,
+      "step": 5910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.560185348303904e-06,
+      "loss": 2.8303,
+      "step": 5911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.557044596534234e-06,
+      "loss": 1.7701,
+      "step": 5912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.553904101019247e-06,
+      "loss": 3.2974,
+      "step": 5913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.550763862088369e-06,
+      "loss": 1.3698,
+      "step": 5914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.547623880070992e-06,
+      "loss": 3.3717,
+      "step": 5915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.544484155296492e-06,
+      "loss": 3.098,
+      "step": 5916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.541344688094204e-06,
+      "loss": 2.8757,
+      "step": 5917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.538205478793448e-06,
+      "loss": 1.62,
+      "step": 5918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.535066527723512e-06,
+      "loss": 3.827,
+      "step": 5919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.531927835213657e-06,
+      "loss": 2.6976,
+      "step": 5920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.528789401593122e-06,
+      "loss": 3.0595,
+      "step": 5921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.525651227191105e-06,
+      "loss": 1.412,
+      "step": 5922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.52251331233679e-06,
+      "loss": 3.6452,
+      "step": 5923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.519375657359331e-06,
+      "loss": 2.9731,
+      "step": 5924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.516238262587851e-06,
+      "loss": 4.0443,
+      "step": 5925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513101128351454e-06,
+      "loss": 3.9579,
+      "step": 5926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.509964254979204e-06,
+      "loss": 2.2373,
+      "step": 5927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.506827642800146e-06,
+      "loss": 3.0022,
+      "step": 5928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.503691292143298e-06,
+      "loss": 2.5047,
+      "step": 5929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.500555203337647e-06,
+      "loss": 2.8252,
+      "step": 5930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.49741937671216e-06,
+      "loss": 2.0557,
+      "step": 5931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.49428381259576e-06,
+      "loss": 2.7068,
+      "step": 5932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.491148511317361e-06,
+      "loss": 2.4918,
+      "step": 5933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.488013473205838e-06,
+      "loss": 1.7801,
+      "step": 5934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.48487869859004e-06,
+      "loss": 2.5854,
+      "step": 5935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4817441877988005e-06,
+      "loss": 2.0346,
+      "step": 5936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4786099411609035e-06,
+      "loss": 3.6182,
+      "step": 5937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.475475959005123e-06,
+      "loss": 3.4073,
+      "step": 5938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.472342241660197e-06,
+      "loss": 3.1136,
+      "step": 5939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.469208789454838e-06,
+      "loss": 2.8921,
+      "step": 5940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.466075602717733e-06,
+      "loss": 3.3599,
+      "step": 5941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.462942681777537e-06,
+      "loss": 2.6644,
+      "step": 5942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.459810026962877e-06,
+      "loss": 3.0863,
+      "step": 5943
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.456677638602355e-06,
+      "loss": 2.3011,
+      "step": 5944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4535455170245476e-06,
+      "loss": 2.8127,
+      "step": 5945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.450413662557998e-06,
+      "loss": 2.5829,
+      "step": 5946
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.44728207553122e-06,
+      "loss": 2.4814,
+      "step": 5947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.444150756272704e-06,
+      "loss": 2.9843,
+      "step": 5948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.441019705110912e-06,
+      "loss": 2.2752,
+      "step": 5949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4378889223742766e-06,
+      "loss": 2.4889,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.434758408391205e-06,
+      "loss": 2.4023,
+      "step": 5951
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.431628163490067e-06,
+      "loss": 3.7731,
+      "step": 5952
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428498187999215e-06,
+      "loss": 3.0175,
+      "step": 5953
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4253684822469684e-06,
+      "loss": 3.9207,
+      "step": 5954
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.422239046561619e-06,
+      "loss": 3.0571,
+      "step": 5955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.419109881271434e-06,
+      "loss": 1.6972,
+      "step": 5956
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.415980986704638e-06,
+      "loss": 2.7878,
+      "step": 5957
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4128523631894464e-06,
+      "loss": 2.4129,
+      "step": 5958
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4097240110540334e-06,
+      "loss": 3.1584,
+      "step": 5959
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.40659593062655e-06,
+      "loss": 1.1398,
+      "step": 5960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.40346812223512e-06,
+      "loss": 3.5875,
+      "step": 5961
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.40034058620783e-06,
+      "loss": 3.0569,
+      "step": 5962
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.397213322872747e-06,
+      "loss": 3.5847,
+      "step": 5963
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.394086332557907e-06,
+      "loss": 2.5584,
+      "step": 5964
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.390959615591315e-06,
+      "loss": 2.7363,
+      "step": 5965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.387833172300958e-06,
+      "loss": 2.8678,
+      "step": 5966
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3847070030147704e-06,
+      "loss": 3.2332,
+      "step": 5967
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.38158110806068e-06,
+      "loss": 1.9272,
+      "step": 5968
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.378455487766581e-06,
+      "loss": 2.596,
+      "step": 5969
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.375330142460331e-06,
+      "loss": 3.126,
+      "step": 5970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3722050724697705e-06,
+      "loss": 2.0858,
+      "step": 5971
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3690802781227056e-06,
+      "loss": 2.205,
+      "step": 5972
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.365955759746906e-06,
+      "loss": 3.4239,
+      "step": 5973
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.362831517670121e-06,
+      "loss": 3.7266,
+      "step": 5974
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35970755222007e-06,
+      "loss": 3.5982,
+      "step": 5975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356583863724442e-06,
+      "loss": 3.3305,
+      "step": 5976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.353460452510903e-06,
+      "loss": 3.144,
+      "step": 5977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.350337318907075e-06,
+      "loss": 2.0302,
+      "step": 5978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.347214463240566e-06,
+      "loss": 2.4338,
+      "step": 5979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.344091885838949e-06,
+      "loss": 2.9718,
+      "step": 5980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.340969587029765e-06,
+      "loss": 2.3456,
+      "step": 5981
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.337847567140534e-06,
+      "loss": 3.0873,
+      "step": 5982
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.334725826498735e-06,
+      "loss": 1.7485,
+      "step": 5983
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.331604365431826e-06,
+      "loss": 2.751,
+      "step": 5984
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.328483184267236e-06,
+      "loss": 2.0202,
+      "step": 5985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.32536228333236e-06,
+      "loss": 3.7712,
+      "step": 5986
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.32224166295457e-06,
+      "loss": 2.2639,
+      "step": 5987
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.319121323461198e-06,
+      "loss": 1.7173,
+      "step": 5988
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.316001265179557e-06,
+      "loss": 2.9332,
+      "step": 5989
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.312881488436928e-06,
+      "loss": 2.8197,
+      "step": 5990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.309761993560559e-06,
+      "loss": 3.3382,
+      "step": 5991
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3066427808776754e-06,
+      "loss": 2.4167,
+      "step": 5992
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3035238507154595e-06,
+      "loss": 3.0417,
+      "step": 5993
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.300405203401076e-06,
+      "loss": 3.1038,
+      "step": 5994
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.297286839261659e-06,
+      "loss": 3.8277,
+      "step": 5995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.294168758624307e-06,
+      "loss": 1.3787,
+      "step": 5996
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2910509618160974e-06,
+      "loss": 1.8788,
+      "step": 5997
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.287933449164068e-06,
+      "loss": 2.4696,
+      "step": 5998
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2848162209952325e-06,
+      "loss": 2.0088,
+      "step": 5999
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2816992776365714e-06,
+      "loss": 4.1055,
+      "step": 6000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2785826194150426e-06,
+      "loss": 1.8591,
+      "step": 6001
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.27546624665757e-06,
+      "loss": 1.5812,
+      "step": 6002
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.272350159691039e-06,
+      "loss": 3.3854,
+      "step": 6003
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.269234358842314e-06,
+      "loss": 3.3329,
+      "step": 6004
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2661188444382345e-06,
+      "loss": 3.6798,
+      "step": 6005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.263003616805599e-06,
+      "loss": 3.2744,
+      "step": 6006
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.259888676271185e-06,
+      "loss": 1.7539,
+      "step": 6007
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.256774023161728e-06,
+      "loss": 2.3182,
+      "step": 6008
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.253659657803945e-06,
+      "loss": 3.4087,
+      "step": 6009
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.250545580524515e-06,
+      "loss": 2.1318,
+      "step": 6010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.247431791650096e-06,
+      "loss": 1.7551,
+      "step": 6011
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.244318291507308e-06,
+      "loss": 2.8508,
+      "step": 6012
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24120508042274e-06,
+      "loss": 2.9846,
+      "step": 6013
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2380921587229536e-06,
+      "loss": 3.1903,
+      "step": 6014
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.234979526734482e-06,
+      "loss": 2.3041,
+      "step": 6015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.231867184783826e-06,
+      "loss": 2.8165,
+      "step": 6016
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.228755133197459e-06,
+      "loss": 3.7255,
+      "step": 6017
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.225643372301812e-06,
+      "loss": 3.0033,
+      "step": 6018
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.222531902423299e-06,
+      "loss": 3.363,
+      "step": 6019
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.219420723888301e-06,
+      "loss": 1.1537,
+      "step": 6020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.216309837023161e-06,
+      "loss": 3.9885,
+      "step": 6021
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.213199242154203e-06,
+      "loss": 4.4737,
+      "step": 6022
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.210088939607709e-06,
+      "loss": 2.0527,
+      "step": 6023
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2069789297099355e-06,
+      "loss": 2.4758,
+      "step": 6024
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.203869212787112e-06,
+      "loss": 3.2402,
+      "step": 6025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2007597891654275e-06,
+      "loss": 2.896,
+      "step": 6026
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.197650659171056e-06,
+      "loss": 1.808,
+      "step": 6027
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.194541823130119e-06,
+      "loss": 2.1637,
+      "step": 6028
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.191433281368725e-06,
+      "loss": 2.3268,
+      "step": 6029
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.188325034212944e-06,
+      "loss": 2.448,
+      "step": 6030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.185217081988818e-06,
+      "loss": 2.9942,
+      "step": 6031
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.182109425022357e-06,
+      "loss": 2.2451,
+      "step": 6032
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.179002063639539e-06,
+      "loss": 3.928,
+      "step": 6033
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.175894998166311e-06,
+      "loss": 1.0056,
+      "step": 6034
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1727882289285915e-06,
+      "loss": 2.6359,
+      "step": 6035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.169681756252265e-06,
+      "loss": 3.6175,
+      "step": 6036
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.166575580463184e-06,
+      "loss": 3.6613,
+      "step": 6037
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.163469701887182e-06,
+      "loss": 2.6447,
+      "step": 6038
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.160364120850034e-06,
+      "loss": 1.7095,
+      "step": 6039
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.157258837677514e-06,
+      "loss": 4.0858,
+      "step": 6040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.154153852695349e-06,
+      "loss": 2.7725,
+      "step": 6041
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.151049166229237e-06,
+      "loss": 2.1616,
+      "step": 6042
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1479447786048495e-06,
+      "loss": 3.6379,
+      "step": 6043
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.144840690147812e-06,
+      "loss": 2.6299,
+      "step": 6044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1417369011837355e-06,
+      "loss": 2.6632,
+      "step": 6045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.138633412038194e-06,
+      "loss": 1.8464,
+      "step": 6046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.135530223036726e-06,
+      "loss": 1.951,
+      "step": 6047
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.132427334504846e-06,
+      "loss": 2.3782,
+      "step": 6048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.129324746768027e-06,
+      "loss": 3.6242,
+      "step": 6049
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.126222460151719e-06,
+      "loss": 4.3144,
+      "step": 6050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.123120474981338e-06,
+      "loss": 2.818,
+      "step": 6051
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.120018791582266e-06,
+      "loss": 2.7946,
+      "step": 6052
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.116917410279862e-06,
+      "loss": 1.5653,
+      "step": 6053
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.113816331399435e-06,
+      "loss": 3.1236,
+      "step": 6054
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.110715555266281e-06,
+      "loss": 2.6999,
+      "step": 6055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.107615082205654e-06,
+      "loss": 1.853,
+      "step": 6056
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.104514912542784e-06,
+      "loss": 3.3036,
+      "step": 6057
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1014150466028605e-06,
+      "loss": 2.794,
+      "step": 6058
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0983154847110445e-06,
+      "loss": 1.8419,
+      "step": 6059
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.095216227192467e-06,
+      "loss": 2.2197,
+      "step": 6060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.092117274372227e-06,
+      "loss": 4.2661,
+      "step": 6061
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.089018626575388e-06,
+      "loss": 3.3082,
+      "step": 6062
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.085920284126989e-06,
+      "loss": 0.9702,
+      "step": 6063
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.082822247352024e-06,
+      "loss": 2.0186,
+      "step": 6064
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0797245165754654e-06,
+      "loss": 2.5884,
+      "step": 6065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.076627092122251e-06,
+      "loss": 2.8243,
+      "step": 6066
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.073529974317288e-06,
+      "loss": 2.6841,
+      "step": 6067
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.070433163485451e-06,
+      "loss": 3.6907,
+      "step": 6068
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.067336659951576e-06,
+      "loss": 3.4121,
+      "step": 6069
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.064240464040472e-06,
+      "loss": 3.1794,
+      "step": 6070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0611445760769216e-06,
+      "loss": 2.547,
+      "step": 6071
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0580489963856646e-06,
+      "loss": 2.9462,
+      "step": 6072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.054953725291418e-06,
+      "loss": 2.9241,
+      "step": 6073
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.05185876311885e-06,
+      "loss": 1.2669,
+      "step": 6074
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.048764110192618e-06,
+      "loss": 3.3268,
+      "step": 6075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.045669766837333e-06,
+      "loss": 3.4997,
+      "step": 6076
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0425757333775794e-06,
+      "loss": 3.9283,
+      "step": 6077
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.039482010137908e-06,
+      "loss": 1.9041,
+      "step": 6078
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.036388597442829e-06,
+      "loss": 2.2935,
+      "step": 6079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.033295495616834e-06,
+      "loss": 2.6298,
+      "step": 6080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0302027049843706e-06,
+      "loss": 3.3015,
+      "step": 6081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027110225869862e-06,
+      "loss": 2.8297,
+      "step": 6082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.024018058597694e-06,
+      "loss": 2.9435,
+      "step": 6083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.020926203492218e-06,
+      "loss": 3.0372,
+      "step": 6084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.017834660877756e-06,
+      "loss": 3.98,
+      "step": 6085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0147434310785975e-06,
+      "loss": 2.1162,
+      "step": 6086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.011652514418998e-06,
+      "loss": 2.4784,
+      "step": 6087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.008561911223186e-06,
+      "loss": 2.1542,
+      "step": 6088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.005471621815341e-06,
+      "loss": 3.094,
+      "step": 6089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.002381646519625e-06,
+      "loss": 3.0332,
+      "step": 6090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9992919856601615e-06,
+      "loss": 1.8592,
+      "step": 6091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9962026395610416e-06,
+      "loss": 3.2582,
+      "step": 6092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.993113608546325e-06,
+      "loss": 3.3374,
+      "step": 6093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.990024892940036e-06,
+      "loss": 2.9668,
+      "step": 6094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.986936493066165e-06,
+      "loss": 3.604,
+      "step": 6095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983848409248672e-06,
+      "loss": 3.0507,
+      "step": 6096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.980760641811481e-06,
+      "loss": 2.7017,
+      "step": 6097
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.977673191078487e-06,
+      "loss": 2.0204,
+      "step": 6098
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.974586057373552e-06,
+      "loss": 3.8585,
+      "step": 6099
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971499241020495e-06,
+      "loss": 2.2037,
+      "step": 6100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.968412742343112e-06,
+      "loss": 2.0099,
+      "step": 6101
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.96532656166516e-06,
+      "loss": 3.499,
+      "step": 6102
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.962240699310369e-06,
+      "loss": 1.3271,
+      "step": 6103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.959155155602433e-06,
+      "loss": 1.6227,
+      "step": 6104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.956069930865005e-06,
+      "loss": 1.8147,
+      "step": 6105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9529850254217135e-06,
+      "loss": 2.0556,
+      "step": 6106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.949900439596151e-06,
+      "loss": 2.6168,
+      "step": 6107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.946816173711878e-06,
+      "loss": 2.9564,
+      "step": 6108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9437322280924195e-06,
+      "loss": 1.8979,
+      "step": 6109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.940648603061263e-06,
+      "loss": 3.2575,
+      "step": 6110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.93756529894187e-06,
+      "loss": 2.862,
+      "step": 6111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.934482316057663e-06,
+      "loss": 2.7807,
+      "step": 6112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.931399654732034e-06,
+      "loss": 2.3449,
+      "step": 6113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.928317315288344e-06,
+      "loss": 2.2398,
+      "step": 6114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.925235298049906e-06,
+      "loss": 2.4577,
+      "step": 6115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.922153603340016e-06,
+      "loss": 1.8508,
+      "step": 6116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.919072231481929e-06,
+      "loss": 2.5594,
+      "step": 6117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.915991182798865e-06,
+      "loss": 1.59,
+      "step": 6118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.912910457614015e-06,
+      "loss": 2.0337,
+      "step": 6119
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 2.3173,
+      "step": 6120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906749979031525e-06,
+      "loss": 2.4141,
+      "step": 6121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.903670226280095e-06,
+      "loss": 1.3859,
+      "step": 6122
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.900590798319287e-06,
+      "loss": 3.2604,
+      "step": 6123
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.897511695472124e-06,
+      "loss": 2.6863,
+      "step": 6124
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894432918061579e-06,
+      "loss": 3.1096,
+      "step": 6125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.891354466410609e-06,
+      "loss": 1.9833,
+      "step": 6126
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.888276340842127e-06,
+      "loss": 2.5334,
+      "step": 6127
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.885198541679016e-06,
+      "loss": 2.4159,
+      "step": 6128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.882121069244123e-06,
+      "loss": 2.3553,
+      "step": 6129
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8790439238602576e-06,
+      "loss": 2.1513,
+      "step": 6130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8759671058501995e-06,
+      "loss": 1.9032,
+      "step": 6131
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.872890615536694e-06,
+      "loss": 2.3424,
+      "step": 6132
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8698144532424484e-06,
+      "loss": 4.3106,
+      "step": 6133
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.866738619290143e-06,
+      "loss": 2.4453,
+      "step": 6134
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.863663114002411e-06,
+      "loss": 2.203,
+      "step": 6135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.860587937701862e-06,
+      "loss": 2.0151,
+      "step": 6136
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.857513090711068e-06,
+      "loss": 2.0898,
+      "step": 6137
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8544385733525665e-06,
+      "loss": 1.5474,
+      "step": 6138
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.851364385948861e-06,
+      "loss": 3.9971,
+      "step": 6139
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.848290528822417e-06,
+      "loss": 1.2831,
+      "step": 6140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.845217002295668e-06,
+      "loss": 3.7668,
+      "step": 6141
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.842143806691013e-06,
+      "loss": 2.7762,
+      "step": 6142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.839070942330817e-06,
+      "loss": 1.9415,
+      "step": 6143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.835998409537412e-06,
+      "loss": 3.8587,
+      "step": 6144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8329262086330864e-06,
+      "loss": 2.7096,
+      "step": 6145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.829854339940103e-06,
+      "loss": 2.0926,
+      "step": 6146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.826782803780687e-06,
+      "loss": 2.2335,
+      "step": 6147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.823711600477025e-06,
+      "loss": 2.4069,
+      "step": 6148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.820640730351281e-06,
+      "loss": 2.9612,
+      "step": 6149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8175701937255645e-06,
+      "loss": 2.1365,
+      "step": 6150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.814499990921963e-06,
+      "loss": 3.1661,
+      "step": 6151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.811430122262529e-06,
+      "loss": 3.7391,
+      "step": 6152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.808360588069276e-06,
+      "loss": 2.8683,
+      "step": 6153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.805291388664186e-06,
+      "loss": 2.2264,
+      "step": 6154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.802222524369202e-06,
+      "loss": 2.2205,
+      "step": 6155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.799153995506234e-06,
+      "loss": 3.9427,
+      "step": 6156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.796085802397154e-06,
+      "loss": 2.8357,
+      "step": 6157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.793017945363804e-06,
+      "loss": 2.706,
+      "step": 6158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.789950424727989e-06,
+      "loss": 3.2115,
+      "step": 6159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.786883240811479e-06,
+      "loss": 3.2442,
+      "step": 6160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7838163939360026e-06,
+      "loss": 2.5965,
+      "step": 6161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.780749884423257e-06,
+      "loss": 2.9134,
+      "step": 6162
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.777683712594909e-06,
+      "loss": 1.8785,
+      "step": 6163
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.774617878772582e-06,
+      "loss": 1.4629,
+      "step": 6164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.771552383277875e-06,
+      "loss": 2.3767,
+      "step": 6165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.768487226432336e-06,
+      "loss": 1.547,
+      "step": 6166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.76542240855749e-06,
+      "loss": 2.2484,
+      "step": 6167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.76235792997482e-06,
+      "loss": 3.0784,
+      "step": 6168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.759293791005777e-06,
+      "loss": 3.2752,
+      "step": 6169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.756229991971779e-06,
+      "loss": 4.0528,
+      "step": 6170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7531665331941955e-06,
+      "loss": 2.9694,
+      "step": 6171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.750103414994374e-06,
+      "loss": 1.9406,
+      "step": 6172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.747040637693622e-06,
+      "loss": 3.7393,
+      "step": 6173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.743978201613206e-06,
+      "loss": 2.9446,
+      "step": 6174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.740916107074372e-06,
+      "loss": 2.2457,
+      "step": 6175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.737854354398308e-06,
+      "loss": 2.0575,
+      "step": 6176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.734792943906184e-06,
+      "loss": 2.2398,
+      "step": 6177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.731731875919123e-06,
+      "loss": 3.043,
+      "step": 6178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.728671150758222e-06,
+      "loss": 3.5766,
+      "step": 6179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.725610768744535e-06,
+      "loss": 3.3889,
+      "step": 6180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.72255073019908e-06,
+      "loss": 3.5771,
+      "step": 6181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719491035442842e-06,
+      "loss": 3.002,
+      "step": 6182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.716431684796769e-06,
+      "loss": 2.9235,
+      "step": 6183
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.713372678581773e-06,
+      "loss": 2.0738,
+      "step": 6184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.710314017118734e-06,
+      "loss": 3.3614,
+      "step": 6185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.707255700728482e-06,
+      "loss": 4.458,
+      "step": 6186
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.704197729731825e-06,
+      "loss": 2.2675,
+      "step": 6187
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7011401044495304e-06,
+      "loss": 2.7649,
+      "step": 6188
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6980828252023275e-06,
+      "loss": 3.1473,
+      "step": 6189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695025892310913e-06,
+      "loss": 1.8593,
+      "step": 6190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.691969306095943e-06,
+      "loss": 2.8487,
+      "step": 6191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.68891306687804e-06,
+      "loss": 2.0601,
+      "step": 6192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.685857174977788e-06,
+      "loss": 1.5375,
+      "step": 6193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.682801630715737e-06,
+      "loss": 3.3122,
+      "step": 6194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6797464344124045e-06,
+      "loss": 1.2996,
+      "step": 6195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.676691586388255e-06,
+      "loss": 3.7989,
+      "step": 6196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6736370869637355e-06,
+      "loss": 2.0252,
+      "step": 6197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.670582936459249e-06,
+      "loss": 3.1471,
+      "step": 6198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.667529135195158e-06,
+      "loss": 2.5008,
+      "step": 6199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.664475683491797e-06,
+      "loss": 2.6815,
+      "step": 6200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.661422581669453e-06,
+      "loss": 1.703,
+      "step": 6201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6583698300483864e-06,
+      "loss": 3.2324,
+      "step": 6202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6553174289488156e-06,
+      "loss": 2.4661,
+      "step": 6203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652265378690923e-06,
+      "loss": 2.3781,
+      "step": 6204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.649213679594859e-06,
+      "loss": 2.7527,
+      "step": 6205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.646162331980726e-06,
+      "loss": 2.429,
+      "step": 6206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.643111336168597e-06,
+      "loss": 1.4733,
+      "step": 6207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6400606924785095e-06,
+      "loss": 2.2053,
+      "step": 6208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.637010401230459e-06,
+      "loss": 3.3786,
+      "step": 6209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.633960462744415e-06,
+      "loss": 3.9288,
+      "step": 6210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6309108773402945e-06,
+      "loss": 2.4508,
+      "step": 6211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.627861645337984e-06,
+      "loss": 2.8958,
+      "step": 6212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.624812767057338e-06,
+      "loss": 2.221,
+      "step": 6213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.621764242818167e-06,
+      "loss": 2.9881,
+      "step": 6214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618716072940248e-06,
+      "loss": 2.0608,
+      "step": 6215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.615668257743322e-06,
+      "loss": 1.2366,
+      "step": 6216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.612620797547087e-06,
+      "loss": 2.4431,
+      "step": 6217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.609573692671209e-06,
+      "loss": 2.5069,
+      "step": 6218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.606526943435314e-06,
+      "loss": 2.2122,
+      "step": 6219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.603480550158995e-06,
+      "loss": 2.646,
+      "step": 6220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.600434513161807e-06,
+      "loss": 3.6853,
+      "step": 6221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.597388832763254e-06,
+      "loss": 3.1581,
+      "step": 6222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.594343509282821e-06,
+      "loss": 2.2819,
+      "step": 6223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.591298543039949e-06,
+      "loss": 2.7943,
+      "step": 6224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.588253934354039e-06,
+      "loss": 1.8377,
+      "step": 6225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.585209683544459e-06,
+      "loss": 2.7244,
+      "step": 6226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.582165790930533e-06,
+      "loss": 2.8422,
+      "step": 6227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.579122256831551e-06,
+      "loss": 1.8608,
+      "step": 6228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.576079081566771e-06,
+      "loss": 1.981,
+      "step": 6229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5730362654554015e-06,
+      "loss": 1.7137,
+      "step": 6230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.569993808816628e-06,
+      "loss": 3.6571,
+      "step": 6231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.566951711969581e-06,
+      "loss": 2.7438,
+      "step": 6232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.563909975233365e-06,
+      "loss": 2.861,
+      "step": 6233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.560868598927046e-06,
+      "loss": 3.0441,
+      "step": 6234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5578275833696485e-06,
+      "loss": 2.5157,
+      "step": 6235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.554786928880165e-06,
+      "loss": 2.1109,
+      "step": 6236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.551746635777541e-06,
+      "loss": 2.0187,
+      "step": 6237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.54870670438069e-06,
+      "loss": 2.5324,
+      "step": 6238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.545667135008489e-06,
+      "loss": 3.0859,
+      "step": 6239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.542627927979772e-06,
+      "loss": 1.4549,
+      "step": 6240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5395890836133445e-06,
+      "loss": 2.0726,
+      "step": 6241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5365506022279564e-06,
+      "loss": 0.8718,
+      "step": 6242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.533512484142335e-06,
+      "loss": 2.5359,
+      "step": 6243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.530474729675167e-06,
+      "loss": 3.0886,
+      "step": 6244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.527437339145097e-06,
+      "loss": 2.8727,
+      "step": 6245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.52440031287074e-06,
+      "loss": 3.3432,
+      "step": 6246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.521363651170654e-06,
+      "loss": 4.5228,
+      "step": 6247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.518327354363374e-06,
+      "loss": 3.5925,
+      "step": 6248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515291422767398e-06,
+      "loss": 2.4795,
+      "step": 6249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5122558567011775e-06,
+      "loss": 1.379,
+      "step": 6250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.509220656483133e-06,
+      "loss": 2.7526,
+      "step": 6251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.506185822431639e-06,
+      "loss": 2.5551,
+      "step": 6252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.503151354865036e-06,
+      "loss": 2.8907,
+      "step": 6253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.500117254101627e-06,
+      "loss": 2.1163,
+      "step": 6254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.497083520459674e-06,
+      "loss": 3.5003,
+      "step": 6255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.494050154257408e-06,
+      "loss": 1.7382,
+      "step": 6256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4910171558130016e-06,
+      "loss": 2.9294,
+      "step": 6257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.487984525444613e-06,
+      "loss": 3.158,
+      "step": 6258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.484952263470348e-06,
+      "loss": 3.9264,
+      "step": 6259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.481920370208274e-06,
+      "loss": 2.8505,
+      "step": 6260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.47888884597643e-06,
+      "loss": 2.2432,
+      "step": 6261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.475857691092801e-06,
+      "loss": 3.3304,
+      "step": 6262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.472826905875343e-06,
+      "loss": 2.2448,
+      "step": 6263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.469796490641974e-06,
+      "loss": 2.4113,
+      "step": 6264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.466766445710568e-06,
+      "loss": 2.0747,
+      "step": 6265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.46373677139897e-06,
+      "loss": 2.2246,
+      "step": 6266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.460707468024965e-06,
+      "loss": 2.7006,
+      "step": 6267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4576785359063225e-06,
+      "loss": 3.3305,
+      "step": 6268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.45464997536076e-06,
+      "loss": 4.538,
+      "step": 6269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4516217867059615e-06,
+      "loss": 2.4903,
+      "step": 6270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.448593970259571e-06,
+      "loss": 2.3798,
+      "step": 6271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.445566526339187e-06,
+      "loss": 1.984,
+      "step": 6272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.44253945526238e-06,
+      "loss": 3.2616,
+      "step": 6273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.439512757346671e-06,
+      "loss": 3.4792,
+      "step": 6274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.43648643290955e-06,
+      "loss": 2.1658,
+      "step": 6275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4334604822684645e-06,
+      "loss": 1.501,
+      "step": 6276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.430434905740825e-06,
+      "loss": 3.8247,
+      "step": 6277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.42740970364399e-06,
+      "loss": 3.4163,
+      "step": 6278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4243848762953e-06,
+      "loss": 1.9708,
+      "step": 6279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.421360424012039e-06,
+      "loss": 2.2662,
+      "step": 6280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.418336347111463e-06,
+      "loss": 3.8136,
+      "step": 6281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.415312645910785e-06,
+      "loss": 2.571,
+      "step": 6282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.412289320727167e-06,
+      "loss": 2.0111,
+      "step": 6283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.409266371877751e-06,
+      "loss": 2.784,
+      "step": 6284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.406243799679625e-06,
+      "loss": 1.6735,
+      "step": 6285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.403221604449847e-06,
+      "loss": 3.6904,
+      "step": 6286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400199786505432e-06,
+      "loss": 2.0423,
+      "step": 6287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.397178346163348e-06,
+      "loss": 2.6258,
+      "step": 6288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.394157283740536e-06,
+      "loss": 2.6773,
+      "step": 6289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.39113659955389e-06,
+      "loss": 2.7196,
+      "step": 6290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3881162939202645e-06,
+      "loss": 2.6946,
+      "step": 6291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.38509636715648e-06,
+      "loss": 2.1863,
+      "step": 6292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.382076819579306e-06,
+      "loss": 3.0375,
+      "step": 6293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3790576515054824e-06,
+      "loss": 1.9916,
+      "step": 6294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.376038863251706e-06,
+      "loss": 2.9908,
+      "step": 6295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.373020455134633e-06,
+      "loss": 3.0463,
+      "step": 6296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.370002427470884e-06,
+      "loss": 3.0527,
+      "step": 6297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.36698478057703e-06,
+      "loss": 2.3824,
+      "step": 6298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.363967514769611e-06,
+      "loss": 2.4727,
+      "step": 6299
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.360950630365126e-06,
+      "loss": 3.2935,
+      "step": 6300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357934127680029e-06,
+      "loss": 3.5003,
+      "step": 6301
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.354918007030743e-06,
+      "loss": 2.676,
+      "step": 6302
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.351902268733636e-06,
+      "loss": 3.3727,
+      "step": 6303
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3488869131050505e-06,
+      "loss": 2.3508,
+      "step": 6304
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.345871940461282e-06,
+      "loss": 2.8702,
+      "step": 6305
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.342857351118588e-06,
+      "loss": 2.7731,
+      "step": 6306
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.339843145393188e-06,
+      "loss": 1.7336,
+      "step": 6307
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.33682932360125e-06,
+      "loss": 2.3714,
+      "step": 6308
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.333815886058916e-06,
+      "loss": 2.0362,
+      "step": 6309
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.33080283308228e-06,
+      "loss": 1.7368,
+      "step": 6310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3277901649873974e-06,
+      "loss": 2.0193,
+      "step": 6311
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.324777882090287e-06,
+      "loss": 2.6703,
+      "step": 6312
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.321765984706915e-06,
+      "loss": 2.4953,
+      "step": 6313
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.318754473153221e-06,
+      "loss": 2.7142,
+      "step": 6314
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.315743347745098e-06,
+      "loss": 2.9062,
+      "step": 6315
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3127326087983974e-06,
+      "loss": 3.4021,
+      "step": 6316
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.309722256628938e-06,
+      "loss": 2.5387,
+      "step": 6317
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.306712291552484e-06,
+      "loss": 3.2359,
+      "step": 6318
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.303702713884767e-06,
+      "loss": 2.5098,
+      "step": 6319
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.300693523941481e-06,
+      "loss": 2.5077,
+      "step": 6320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.297684722038277e-06,
+      "loss": 2.1337,
+      "step": 6321
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2946763084907625e-06,
+      "loss": 2.5004,
+      "step": 6322
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.291668283614505e-06,
+      "loss": 1.3623,
+      "step": 6323
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2886606477250345e-06,
+      "loss": 3.072,
+      "step": 6324
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2856534011378365e-06,
+      "loss": 2.941,
+      "step": 6325
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.282646544168358e-06,
+      "loss": 2.7496,
+      "step": 6326
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.279640077132008e-06,
+      "loss": 3.6036,
+      "step": 6327
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.276634000344144e-06,
+      "loss": 3.8973,
+      "step": 6328
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273628314120092e-06,
+      "loss": 2.9764,
+      "step": 6329
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.270623018775135e-06,
+      "loss": 2.5591,
+      "step": 6330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.267618114624515e-06,
+      "loss": 2.9694,
+      "step": 6331
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.264613601983435e-06,
+      "loss": 2.9586,
+      "step": 6332
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.261609481167048e-06,
+      "loss": 2.9515,
+      "step": 6333
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.258605752490477e-06,
+      "loss": 3.0091,
+      "step": 6334
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.255602416268799e-06,
+      "loss": 1.8344,
+      "step": 6335
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2525994728170495e-06,
+      "loss": 2.2271,
+      "step": 6336
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.249596922450221e-06,
+      "loss": 1.8917,
+      "step": 6337
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.246594765483274e-06,
+      "loss": 2.4467,
+      "step": 6338
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.243593002231112e-06,
+      "loss": 3.2615,
+      "step": 6339
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2405916330086106e-06,
+      "loss": 1.8368,
+      "step": 6340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.237590658130599e-06,
+      "loss": 1.4418,
+      "step": 6341
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.234590077911865e-06,
+      "loss": 2.4588,
+      "step": 6342
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.231589892667158e-06,
+      "loss": 4.3495,
+      "step": 6343
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2285901027111806e-06,
+      "loss": 2.1242,
+      "step": 6344
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.225590708358596e-06,
+      "loss": 1.9363,
+      "step": 6345
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.22259170992403e-06,
+      "loss": 3.0527,
+      "step": 6346
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.219593107722062e-06,
+      "loss": 2.8049,
+      "step": 6347
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.216594902067233e-06,
+      "loss": 3.8871,
+      "step": 6348
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.213597093274038e-06,
+      "loss": 2.0273,
+      "step": 6349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.210599681656933e-06,
+      "loss": 2.7159,
+      "step": 6350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.207602667530336e-06,
+      "loss": 2.8949,
+      "step": 6351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204606051208617e-06,
+      "loss": 3.5751,
+      "step": 6352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.201609833006113e-06,
+      "loss": 2.361,
+      "step": 6353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.198614013237105e-06,
+      "loss": 2.6038,
+      "step": 6354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1956185922158445e-06,
+      "loss": 3.7074,
+      "step": 6355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.192623570256535e-06,
+      "loss": 1.6709,
+      "step": 6356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.189628947673344e-06,
+      "loss": 1.9842,
+      "step": 6357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.186634724780394e-06,
+      "loss": 2.3626,
+      "step": 6358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18364090189176e-06,
+      "loss": 3.543,
+      "step": 6359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.180647479321484e-06,
+      "loss": 2.7046,
+      "step": 6360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.177654457383561e-06,
+      "loss": 3.0636,
+      "step": 6361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.174661836391945e-06,
+      "loss": 2.474,
+      "step": 6362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.171669616660553e-06,
+      "loss": 2.9044,
+      "step": 6363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.168677798503246e-06,
+      "loss": 3.5622,
+      "step": 6364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.165686382233856e-06,
+      "loss": 2.8983,
+      "step": 6365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.16269536816617e-06,
+      "loss": 2.9728,
+      "step": 6366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.159704756613929e-06,
+      "loss": 1.9853,
+      "step": 6367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.156714547890838e-06,
+      "loss": 2.7481,
+      "step": 6368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.153724742310551e-06,
+      "loss": 3.5452,
+      "step": 6369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1507353401866896e-06,
+      "loss": 2.6486,
+      "step": 6370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1477463418328255e-06,
+      "loss": 2.3078,
+      "step": 6371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.14475774756249e-06,
+      "loss": 1.3914,
+      "step": 6372
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.141769557689179e-06,
+      "loss": 2.4875,
+      "step": 6373
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13878177252633e-06,
+      "loss": 2.5611,
+      "step": 6374
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.135794392387353e-06,
+      "loss": 3.0754,
+      "step": 6375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13280741758561e-06,
+      "loss": 2.3121,
+      "step": 6376
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12982084843442e-06,
+      "loss": 1.5958,
+      "step": 6377
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.126834685247065e-06,
+      "loss": 1.6276,
+      "step": 6378
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.123848928336771e-06,
+      "loss": 2.281,
+      "step": 6379
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.120863578016736e-06,
+      "loss": 1.9524,
+      "step": 6380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.117878634600107e-06,
+      "loss": 2.1011,
+      "step": 6381
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.114894098399994e-06,
+      "loss": 2.9425,
+      "step": 6382
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.11190996972946e-06,
+      "loss": 1.7231,
+      "step": 6383
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.108926248901521e-06,
+      "loss": 0.9707,
+      "step": 6384
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1059429362291615e-06,
+      "loss": 3.3876,
+      "step": 6385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.102960032025313e-06,
+      "loss": 3.2938,
+      "step": 6386
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.099977536602872e-06,
+      "loss": 3.7074,
+      "step": 6387
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0969954502746916e-06,
+      "loss": 3.2573,
+      "step": 6388
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.094013773353569e-06,
+      "loss": 3.1284,
+      "step": 6389
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.091032506152274e-06,
+      "loss": 2.7415,
+      "step": 6390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0880516489835265e-06,
+      "loss": 2.5494,
+      "step": 6391
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0850712021600044e-06,
+      "loss": 2.582,
+      "step": 6392
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.082091165994345e-06,
+      "loss": 2.5194,
+      "step": 6393
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.07911154079914e-06,
+      "loss": 3.4082,
+      "step": 6394
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.076132326886934e-06,
+      "loss": 2.0972,
+      "step": 6395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.073153524570236e-06,
+      "loss": 1.7895,
+      "step": 6396
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0701751341615086e-06,
+      "loss": 2.7598,
+      "step": 6397
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.067197155973172e-06,
+      "loss": 2.8397,
+      "step": 6398
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.064219590317604e-06,
+      "loss": 2.5665,
+      "step": 6399
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.061242437507131e-06,
+      "loss": 1.1103,
+      "step": 6400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.058265697854044e-06,
+      "loss": 2.0842,
+      "step": 6401
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0552893716705925e-06,
+      "loss": 2.6814,
+      "step": 6402
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.052313459268979e-06,
+      "loss": 3.7522,
+      "step": 6403
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.049337960961362e-06,
+      "loss": 3.595,
+      "step": 6404
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0463628770598574e-06,
+      "loss": 1.0791,
+      "step": 6405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.043388207876535e-06,
+      "loss": 2.7485,
+      "step": 6406
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.040413953723427e-06,
+      "loss": 2.1156,
+      "step": 6407
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.037440114912521e-06,
+      "loss": 1.995,
+      "step": 6408
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.034466691755757e-06,
+      "loss": 3.0749,
+      "step": 6409
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0314936845650296e-06,
+      "loss": 2.8516,
+      "step": 6410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.028521093652195e-06,
+      "loss": 3.1105,
+      "step": 6411
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0255489193290675e-06,
+      "loss": 3.1841,
+      "step": 6412
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0225771619074105e-06,
+      "loss": 3.4351,
+      "step": 6413
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.019605821698953e-06,
+      "loss": 2.2138,
+      "step": 6414
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.016634899015369e-06,
+      "loss": 1.4446,
+      "step": 6415
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.013664394168297e-06,
+      "loss": 3.5408,
+      "step": 6416
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.010694307469327e-06,
+      "loss": 3.655,
+      "step": 6417
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.00772463923001e-06,
+      "loss": 3.878,
+      "step": 6418
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.004755389761854e-06,
+      "loss": 2.7319,
+      "step": 6419
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.00178655937631e-06,
+      "loss": 2.6571,
+      "step": 6420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.998818148384801e-06,
+      "loss": 4.0191,
+      "step": 6421
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.995850157098698e-06,
+      "loss": 2.8449,
+      "step": 6422
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.99288258582933e-06,
+      "loss": 3.0186,
+      "step": 6423
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.989915434887985e-06,
+      "loss": 2.7328,
+      "step": 6424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.986948704585895e-06,
+      "loss": 1.7585,
+      "step": 6425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983982395234261e-06,
+      "loss": 1.781,
+      "step": 6426
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.981016507144235e-06,
+      "loss": 2.8444,
+      "step": 6427
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9780510406269245e-06,
+      "loss": 2.4674,
+      "step": 6428
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.975085995993396e-06,
+      "loss": 2.0339,
+      "step": 6429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.972121373554665e-06,
+      "loss": 3.2297,
+      "step": 6430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.969157173621706e-06,
+      "loss": 1.4809,
+      "step": 6431
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.966193396505452e-06,
+      "loss": 2.8438,
+      "step": 6432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96323004251679e-06,
+      "loss": 2.2921,
+      "step": 6433
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.960267111966565e-06,
+      "loss": 2.1011,
+      "step": 6434
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.957304605165567e-06,
+      "loss": 2.615,
+      "step": 6435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.954342522424553e-06,
+      "loss": 3.9689,
+      "step": 6436
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9513808640542325e-06,
+      "loss": 3.8001,
+      "step": 6437
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.948419630365269e-06,
+      "loss": 2.876,
+      "step": 6438
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.945458821668284e-06,
+      "loss": 3.8489,
+      "step": 6439
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.942498438273849e-06,
+      "loss": 2.6205,
+      "step": 6440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9395384804924974e-06,
+      "loss": 1.8894,
+      "step": 6441
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.936578948634713e-06,
+      "loss": 2.8999,
+      "step": 6442
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.933619843010937e-06,
+      "loss": 2.9267,
+      "step": 6443
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.930661163931572e-06,
+      "loss": 2.2815,
+      "step": 6444
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.927702911706961e-06,
+      "loss": 3.3489,
+      "step": 6445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.924745086647413e-06,
+      "loss": 1.1687,
+      "step": 6446
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.921787689063192e-06,
+      "loss": 1.7719,
+      "step": 6447
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.918830719264514e-06,
+      "loss": 2.6035,
+      "step": 6448
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.915874177561558e-06,
+      "loss": 3.2507,
+      "step": 6449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.912918064264441e-06,
+      "loss": 3.5639,
+      "step": 6450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.909962379683249e-06,
+      "loss": 1.4625,
+      "step": 6451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9070071241280235e-06,
+      "loss": 1.7982,
+      "step": 6452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.904052297908754e-06,
+      "loss": 3.642,
+      "step": 6453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.901097901335388e-06,
+      "loss": 3.046,
+      "step": 6454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.898143934717831e-06,
+      "loss": 2.2506,
+      "step": 6455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.895190398365935e-06,
+      "loss": 3.3261,
+      "step": 6456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.892237292589516e-06,
+      "loss": 3.7703,
+      "step": 6457
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.889284617698339e-06,
+      "loss": 2.891,
+      "step": 6458
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.886332374002128e-06,
+      "loss": 3.3161,
+      "step": 6459
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8833805618105635e-06,
+      "loss": 1.0069,
+      "step": 6460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.880429181433268e-06,
+      "loss": 2.8475,
+      "step": 6461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.877478233179832e-06,
+      "loss": 1.8674,
+      "step": 6462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.874527717359796e-06,
+      "loss": 3.3606,
+      "step": 6463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.871577634282655e-06,
+      "loss": 2.0907,
+      "step": 6464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.868627984257862e-06,
+      "loss": 3.3314,
+      "step": 6465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.865678767594817e-06,
+      "loss": 3.8452,
+      "step": 6466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.862729984602882e-06,
+      "loss": 3.1423,
+      "step": 6467
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8597816355913685e-06,
+      "loss": 2.7072,
+      "step": 6468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.856833720869547e-06,
+      "loss": 2.7522,
+      "step": 6469
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.853886240746643e-06,
+      "loss": 2.2752,
+      "step": 6470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.850939195531826e-06,
+      "loss": 2.2855,
+      "step": 6471
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.84799258553423e-06,
+      "loss": 1.4506,
+      "step": 6472
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845046411062942e-06,
+      "loss": 2.5702,
+      "step": 6473
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.842100672427002e-06,
+      "loss": 2.9351,
+      "step": 6474
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.839155369935407e-06,
+      "loss": 2.7689,
+      "step": 6475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.836210503897099e-06,
+      "loss": 2.766,
+      "step": 6476
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.833266074620987e-06,
+      "loss": 1.2764,
+      "step": 6477
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.830322082415922e-06,
+      "loss": 2.6632,
+      "step": 6478
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.827378527590722e-06,
+      "loss": 2.8316,
+      "step": 6479
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82443541045415e-06,
+      "loss": 3.1094,
+      "step": 6480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82149273131492e-06,
+      "loss": 3.3662,
+      "step": 6481
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.818550490481711e-06,
+      "loss": 2.0496,
+      "step": 6482
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.815608688263148e-06,
+      "loss": 3.633,
+      "step": 6483
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.812667324967813e-06,
+      "loss": 2.1272,
+      "step": 6484
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.809726400904242e-06,
+      "loss": 3.8991,
+      "step": 6485
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.806785916380924e-06,
+      "loss": 2.2755,
+      "step": 6486
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.803845871706302e-06,
+      "loss": 1.8322,
+      "step": 6487
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.800906267188773e-06,
+      "loss": 2.9481,
+      "step": 6488
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.797967103136688e-06,
+      "loss": 3.7306,
+      "step": 6489
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.795028379858355e-06,
+      "loss": 2.9776,
+      "step": 6490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7920900976620265e-06,
+      "loss": 1.9963,
+      "step": 6491
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.789152256855917e-06,
+      "loss": 3.0643,
+      "step": 6492
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.786214857748192e-06,
+      "loss": 3.3386,
+      "step": 6493
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.783277900646971e-06,
+      "loss": 2.5936,
+      "step": 6494
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.780341385860333e-06,
+      "loss": 3.2782,
+      "step": 6495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.777405313696294e-06,
+      "loss": 2.6486,
+      "step": 6496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.774469684462841e-06,
+      "loss": 3.5114,
+      "step": 6497
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.771534498467908e-06,
+      "loss": 1.339,
+      "step": 6498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.76859975601938e-06,
+      "loss": 2.7074,
+      "step": 6499
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.765665457425102e-06,
+      "loss": 1.7395,
+      "step": 6500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.762731602992862e-06,
+      "loss": 1.3832,
+      "step": 6501
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.759798193030413e-06,
+      "loss": 1.6434,
+      "step": 6502
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.756865227845452e-06,
+      "loss": 2.1658,
+      "step": 6503
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.753932707745635e-06,
+      "loss": 2.5338,
+      "step": 6504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.751000633038573e-06,
+      "loss": 1.9417,
+      "step": 6505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.748069004031822e-06,
+      "loss": 2.6922,
+      "step": 6506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7451378210329e-06,
+      "loss": 3.2521,
+      "step": 6507
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.742207084349274e-06,
+      "loss": 2.6598,
+      "step": 6508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.739276794288362e-06,
+      "loss": 2.8891,
+      "step": 6509
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.736346951157544e-06,
+      "loss": 3.4293,
+      "step": 6510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.733417555264139e-06,
+      "loss": 2.6078,
+      "step": 6511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73048860691543e-06,
+      "loss": 1.5769,
+      "step": 6512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.727560106418651e-06,
+      "loss": 2.7587,
+      "step": 6513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.724632054080989e-06,
+      "loss": 2.9235,
+      "step": 6514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.721704450209581e-06,
+      "loss": 2.0424,
+      "step": 6515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.718777295111524e-06,
+      "loss": 3.1183,
+      "step": 6516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.715850589093855e-06,
+      "loss": 3.1754,
+      "step": 6517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.712924332463575e-06,
+      "loss": 3.3604,
+      "step": 6518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.709998525527637e-06,
+      "loss": 2.5596,
+      "step": 6519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.707073168592943e-06,
+      "loss": 2.2886,
+      "step": 6520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.704148261966349e-06,
+      "loss": 3.2046,
+      "step": 6521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.701223805954664e-06,
+      "loss": 3.3861,
+      "step": 6522
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.698299800864651e-06,
+      "loss": 3.3258,
+      "step": 6523
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.695376247003025e-06,
+      "loss": 3.1971,
+      "step": 6524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.692453144676451e-06,
+      "loss": 2.8312,
+      "step": 6525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.689530494191555e-06,
+      "loss": 2.8792,
+      "step": 6526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.686608295854901e-06,
+      "loss": 2.9639,
+      "step": 6527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.683686549973018e-06,
+      "loss": 3.4659,
+      "step": 6528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.680765256852381e-06,
+      "loss": 1.425,
+      "step": 6529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.677844416799424e-06,
+      "loss": 1.904,
+      "step": 6530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.674924030120533e-06,
+      "loss": 3.051,
+      "step": 6531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.672004097122033e-06,
+      "loss": 2.7347,
+      "step": 6532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.669084618110215e-06,
+      "loss": 1.3522,
+      "step": 6533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.666165593391322e-06,
+      "loss": 1.698,
+      "step": 6534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.663247023271543e-06,
+      "loss": 3.3506,
+      "step": 6535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6603289080570274e-06,
+      "loss": 3.4015,
+      "step": 6536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.657411248053866e-06,
+      "loss": 3.356,
+      "step": 6537
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.654494043568109e-06,
+      "loss": 3.3606,
+      "step": 6538
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651577294905759e-06,
+      "loss": 3.5895,
+      "step": 6539
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.648661002372769e-06,
+      "loss": 1.6587,
+      "step": 6540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.645745166275045e-06,
+      "loss": 2.2552,
+      "step": 6541
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6428297869184444e-06,
+      "loss": 2.4257,
+      "step": 6542
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.639914864608777e-06,
+      "loss": 3.0961,
+      "step": 6543
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.637000399651804e-06,
+      "loss": 2.9367,
+      "step": 6544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.63408639235324e-06,
+      "loss": 2.6262,
+      "step": 6545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.631172843018754e-06,
+      "loss": 2.0734,
+      "step": 6546
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6282597519539575e-06,
+      "loss": 2.4069,
+      "step": 6547
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.625347119464422e-06,
+      "loss": 2.7203,
+      "step": 6548
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622434945855671e-06,
+      "loss": 1.398,
+      "step": 6549
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.619523231433177e-06,
+      "loss": 4.0236,
+      "step": 6550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.616611976502369e-06,
+      "loss": 1.9952,
+      "step": 6551
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.613701181368618e-06,
+      "loss": 2.7236,
+      "step": 6552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610790846337255e-06,
+      "loss": 3.3723,
+      "step": 6553
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6078809717135615e-06,
+      "loss": 3.0549,
+      "step": 6554
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.604971557802769e-06,
+      "loss": 1.3749,
+      "step": 6555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602062604910064e-06,
+      "loss": 2.5668,
+      "step": 6556
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.599154113340578e-06,
+      "loss": 2.3244,
+      "step": 6557
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.596246083399402e-06,
+      "loss": 2.194,
+      "step": 6558
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5933385153915735e-06,
+      "loss": 2.6789,
+      "step": 6559
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.590431409622081e-06,
+      "loss": 3.2257,
+      "step": 6560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.587524766395873e-06,
+      "loss": 2.4058,
+      "step": 6561
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5846185860178344e-06,
+      "loss": 2.4132,
+      "step": 6562
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.581712868792814e-06,
+      "loss": 3.2513,
+      "step": 6563
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5788076150256075e-06,
+      "loss": 2.7097,
+      "step": 6564
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.575902825020962e-06,
+      "loss": 1.8805,
+      "step": 6565
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.572998499083583e-06,
+      "loss": 2.8743,
+      "step": 6566
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.570094637518111e-06,
+      "loss": 4.1193,
+      "step": 6567
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.567191240629151e-06,
+      "loss": 2.8106,
+      "step": 6568
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.564288308721255e-06,
+      "loss": 2.3586,
+      "step": 6569
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.56138584209893e-06,
+      "loss": 2.6254,
+      "step": 6570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.558483841066636e-06,
+      "loss": 2.7991,
+      "step": 6571
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.555582305928766e-06,
+      "loss": 3.111,
+      "step": 6572
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5526812369896855e-06,
+      "loss": 2.6364,
+      "step": 6573
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.549780634553704e-06,
+      "loss": 1.8974,
+      "step": 6574
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.546880498925079e-06,
+      "loss": 3.5091,
+      "step": 6575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5439808304080225e-06,
+      "loss": 2.6168,
+      "step": 6576
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.541081629306695e-06,
+      "loss": 2.293,
+      "step": 6577
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.538182895925212e-06,
+      "loss": 2.8213,
+      "step": 6578
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535284630567634e-06,
+      "loss": 2.4634,
+      "step": 6579
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5323868335379775e-06,
+      "loss": 3.2375,
+      "step": 6580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.529489505140206e-06,
+      "loss": 3.1619,
+      "step": 6581
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.526592645678241e-06,
+      "loss": 1.7849,
+      "step": 6582
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.523696255455944e-06,
+      "loss": 2.6815,
+      "step": 6583
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.520800334777132e-06,
+      "loss": 3.0111,
+      "step": 6584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.517904883945577e-06,
+      "loss": 3.0015,
+      "step": 6585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.515009903264998e-06,
+      "loss": 1.8031,
+      "step": 6586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.512115393039071e-06,
+      "loss": 3.0416,
+      "step": 6587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.509221353571404e-06,
+      "loss": 1.9862,
+      "step": 6588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.506327785165574e-06,
+      "loss": 3.1019,
+      "step": 6589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.503434688125104e-06,
+      "loss": 3.6633,
+      "step": 6590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.500542062753467e-06,
+      "loss": 1.811,
+      "step": 6591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.497649909354084e-06,
+      "loss": 1.4909,
+      "step": 6592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.49475822823033e-06,
+      "loss": 2.9549,
+      "step": 6593
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.491867019685528e-06,
+      "loss": 2.9954,
+      "step": 6594
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.488976284022953e-06,
+      "loss": 2.4042,
+      "step": 6595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.486086021545829e-06,
+      "loss": 2.0777,
+      "step": 6596
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.483196232557338e-06,
+      "loss": 2.1892,
+      "step": 6597
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4803069173605915e-06,
+      "loss": 3.0205,
+      "step": 6598
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.477418076258675e-06,
+      "loss": 2.8226,
+      "step": 6599
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4745297095546125e-06,
+      "loss": 3.5808,
+      "step": 6600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4716418175513784e-06,
+      "loss": 2.7277,
+      "step": 6601
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.468754400551905e-06,
+      "loss": 1.3967,
+      "step": 6602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4658674588590625e-06,
+      "loss": 2.4803,
+      "step": 6603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4629809927756794e-06,
+      "loss": 3.4007,
+      "step": 6604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.460095002604533e-06,
+      "loss": 0.6771,
+      "step": 6605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457209488648351e-06,
+      "loss": 1.5416,
+      "step": 6606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.454324451209814e-06,
+      "loss": 2.3153,
+      "step": 6607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.451439890591539e-06,
+      "loss": 2.7843,
+      "step": 6608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.448555807096111e-06,
+      "loss": 3.9707,
+      "step": 6609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445672201026054e-06,
+      "loss": 1.6673,
+      "step": 6610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.442789072683846e-06,
+      "loss": 3.5503,
+      "step": 6611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.439906422371914e-06,
+      "loss": 2.336,
+      "step": 6612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.437024250392634e-06,
+      "loss": 3.4414,
+      "step": 6613
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.434142557048332e-06,
+      "loss": 2.1906,
+      "step": 6614
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.431261342641287e-06,
+      "loss": 2.731,
+      "step": 6615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.42838060747372e-06,
+      "loss": 1.6676,
+      "step": 6616
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4255003518478165e-06,
+      "loss": 2.2938,
+      "step": 6617
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.422620576065689e-06,
+      "loss": 2.9161,
+      "step": 6618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.41974128042942e-06,
+      "loss": 1.9537,
+      "step": 6619
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.416862465241033e-06,
+      "loss": 3.9716,
+      "step": 6620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4139841308025036e-06,
+      "loss": 2.749,
+      "step": 6621
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.411106277415756e-06,
+      "loss": 1.6541,
+      "step": 6622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.40822890538266e-06,
+      "loss": 2.4157,
+      "step": 6623
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.405352015005039e-06,
+      "loss": 1.0192,
+      "step": 6624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.40247560658467e-06,
+      "loss": 3.2533,
+      "step": 6625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.399599680423269e-06,
+      "loss": 1.3481,
+      "step": 6626
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.396724236822513e-06,
+      "loss": 3.153,
+      "step": 6627
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3938492760840176e-06,
+      "loss": 1.748,
+      "step": 6628
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3909747985093565e-06,
+      "loss": 2.3231,
+      "step": 6629
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3881008044000495e-06,
+      "loss": 1.7819,
+      "step": 6630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.385227294057561e-06,
+      "loss": 2.8909,
+      "step": 6631
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.382354267783316e-06,
+      "loss": 2.2442,
+      "step": 6632
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.379481725878674e-06,
+      "loss": 3.171,
+      "step": 6633
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.376609668644954e-06,
+      "loss": 3.2243,
+      "step": 6634
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.373738096383423e-06,
+      "loss": 2.0722,
+      "step": 6635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.370867009395294e-06,
+      "loss": 2.272,
+      "step": 6636
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.367996407981732e-06,
+      "loss": 2.0901,
+      "step": 6637
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.365126292443852e-06,
+      "loss": 2.3539,
+      "step": 6638
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.362256663082708e-06,
+      "loss": 1.8502,
+      "step": 6639
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.359387520199317e-06,
+      "loss": 2.3386,
+      "step": 6640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3565188640946366e-06,
+      "loss": 1.7893,
+      "step": 6641
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.353650695069577e-06,
+      "loss": 3.8377,
+      "step": 6642
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3507830134249985e-06,
+      "loss": 1.9133,
+      "step": 6643
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3479158194617e-06,
+      "loss": 3.0055,
+      "step": 6644
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3450491134804416e-06,
+      "loss": 2.6512,
+      "step": 6645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.342182895781926e-06,
+      "loss": 1.3715,
+      "step": 6646
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3393171666668086e-06,
+      "loss": 3.5386,
+      "step": 6647
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.336451926435688e-06,
+      "loss": 1.9648,
+      "step": 6648
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.333587175389117e-06,
+      "loss": 1.536,
+      "step": 6649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.330722913827594e-06,
+      "loss": 2.9199,
+      "step": 6650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.327859142051565e-06,
+      "loss": 1.9764,
+      "step": 6651
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32499586036143e-06,
+      "loss": 2.797,
+      "step": 6652
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.322133069057535e-06,
+      "loss": 2.5971,
+      "step": 6653
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3192707684401665e-06,
+      "loss": 3.083,
+      "step": 6654
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3164089588095705e-06,
+      "loss": 2.5143,
+      "step": 6655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.313547640465937e-06,
+      "loss": 2.1399,
+      "step": 6656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3106868137094055e-06,
+      "loss": 2.5374,
+      "step": 6657
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.307826478840068e-06,
+      "loss": 2.4108,
+      "step": 6658
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.304966636157952e-06,
+      "loss": 1.7826,
+      "step": 6659
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.302107285963045e-06,
+      "loss": 3.1874,
+      "step": 6660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.29924842855528e-06,
+      "loss": 2.5362,
+      "step": 6661
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2963900642345385e-06,
+      "loss": 3.346,
+      "step": 6662
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.29353219330065e-06,
+      "loss": 3.1228,
+      "step": 6663
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2906748160533895e-06,
+      "loss": 1.2207,
+      "step": 6664
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.287817932792485e-06,
+      "loss": 2.6139,
+      "step": 6665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.284961543817608e-06,
+      "loss": 2.438,
+      "step": 6666
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.282105649428384e-06,
+      "loss": 3.1592,
+      "step": 6667
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.279250249924384e-06,
+      "loss": 2.4453,
+      "step": 6668
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.276395345605119e-06,
+      "loss": 3.5937,
+      "step": 6669
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.273540936770059e-06,
+      "loss": 2.6583,
+      "step": 6670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.270687023718618e-06,
+      "loss": 2.1576,
+      "step": 6671
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.26783360675016e-06,
+      "loss": 2.225,
+      "step": 6672
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2649806861639975e-06,
+      "loss": 3.3824,
+      "step": 6673
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.26212826225938e-06,
+      "loss": 1.3375,
+      "step": 6674
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.259276335335522e-06,
+      "loss": 2.3891,
+      "step": 6675
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2564249056915704e-06,
+      "loss": 2.2249,
+      "step": 6676
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.253573973626634e-06,
+      "loss": 2.2558,
+      "step": 6677
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2507235394397595e-06,
+      "loss": 2.4689,
+      "step": 6678
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2478736034299415e-06,
+      "loss": 1.6975,
+      "step": 6679
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.245024165896126e-06,
+      "loss": 3.1369,
+      "step": 6680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.242175227137209e-06,
+      "loss": 2.532,
+      "step": 6681
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239326787452028e-06,
+      "loss": 3.2104,
+      "step": 6682
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2364788471393714e-06,
+      "loss": 2.0656,
+      "step": 6683
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2336314064979766e-06,
+      "loss": 1.8938,
+      "step": 6684
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2307844658265236e-06,
+      "loss": 3.107,
+      "step": 6685
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2279380254236475e-06,
+      "loss": 3.2596,
+      "step": 6686
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.225092085587925e-06,
+      "loss": 4.0426,
+      "step": 6687
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.222246646617886e-06,
+      "loss": 1.9435,
+      "step": 6688
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.219401708811993e-06,
+      "loss": 2.9042,
+      "step": 6689
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.216557272468675e-06,
+      "loss": 2.6419,
+      "step": 6690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.213713337886299e-06,
+      "loss": 2.6229,
+      "step": 6691
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.210869905363178e-06,
+      "loss": 2.3139,
+      "step": 6692
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.208026975197583e-06,
+      "loss": 2.909,
+      "step": 6693
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.205184547687714e-06,
+      "loss": 2.8006,
+      "step": 6694
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.202342623131731e-06,
+      "loss": 3.4178,
+      "step": 6695
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.199501201827741e-06,
+      "loss": 3.6366,
+      "step": 6696
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.196660284073795e-06,
+      "loss": 2.9675,
+      "step": 6697
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.193819870167893e-06,
+      "loss": 1.6556,
+      "step": 6698
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.19097996040798e-06,
+      "loss": 2.9607,
+      "step": 6699
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.18814055509195e-06,
+      "loss": 2.3531,
+      "step": 6700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.185301654517643e-06,
+      "loss": 3.2289,
+      "step": 6701
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1824632589828465e-06,
+      "loss": 2.1038,
+      "step": 6702
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1796253687852945e-06,
+      "loss": 3.2309,
+      "step": 6703
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1767879842226745e-06,
+      "loss": 1.8893,
+      "step": 6704
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.173951105592605e-06,
+      "loss": 2.0253,
+      "step": 6705
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171114733192667e-06,
+      "loss": 1.8059,
+      "step": 6706
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.168278867320382e-06,
+      "loss": 2.8005,
+      "step": 6707
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.165443508273218e-06,
+      "loss": 2.295,
+      "step": 6708
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.162608656348597e-06,
+      "loss": 1.8567,
+      "step": 6709
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1597743118438725e-06,
+      "loss": 2.9654,
+      "step": 6710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.156940475056358e-06,
+      "loss": 2.976,
+      "step": 6711
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.154107146283311e-06,
+      "loss": 3.9099,
+      "step": 6712
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.151274325821934e-06,
+      "loss": 3.175,
+      "step": 6713
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1484420139693805e-06,
+      "loss": 1.8684,
+      "step": 6714
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.145610211022738e-06,
+      "loss": 3.0065,
+      "step": 6715
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1427789172790565e-06,
+      "loss": 2.3962,
+      "step": 6716
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.139948133035322e-06,
+      "loss": 3.1823,
+      "step": 6717
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.137117858588472e-06,
+      "loss": 3.5342,
+      "step": 6718
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.13428809423539e-06,
+      "loss": 2.0782,
+      "step": 6719
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.131458840272905e-06,
+      "loss": 1.025,
+      "step": 6720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.128630096997791e-06,
+      "loss": 2.3129,
+      "step": 6721
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.125801864706771e-06,
+      "loss": 2.4384,
+      "step": 6722
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.122974143696513e-06,
+      "loss": 2.9917,
+      "step": 6723
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.120146934263638e-06,
+      "loss": 2.5894,
+      "step": 6724
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.117320236704697e-06,
+      "loss": 1.4353,
+      "step": 6725
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.114494051316201e-06,
+      "loss": 2.179,
+      "step": 6726
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1116683783946055e-06,
+      "loss": 1.2126,
+      "step": 6727
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.10884321823631e-06,
+      "loss": 2.7474,
+      "step": 6728
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.106018571137663e-06,
+      "loss": 3.4845,
+      "step": 6729
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.103194437394952e-06,
+      "loss": 2.1695,
+      "step": 6730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.100370817304417e-06,
+      "loss": 2.9737,
+      "step": 6731
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.097547711162243e-06,
+      "loss": 2.2695,
+      "step": 6732
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0947251192645604e-06,
+      "loss": 2.5542,
+      "step": 6733
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.091903041907449e-06,
+      "loss": 2.784,
+      "step": 6734
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.089081479386928e-06,
+      "loss": 2.2796,
+      "step": 6735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.086260431998967e-06,
+      "loss": 3.199,
+      "step": 6736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.083439900039482e-06,
+      "loss": 2.9618,
+      "step": 6737
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.080619883804333e-06,
+      "loss": 2.8748,
+      "step": 6738
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.077800383589331e-06,
+      "loss": 3.5224,
+      "step": 6739
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.074981399690219e-06,
+      "loss": 3.1196,
+      "step": 6740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072162932402701e-06,
+      "loss": 3.5213,
+      "step": 6741
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.069344982022421e-06,
+      "loss": 3.1819,
+      "step": 6742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.066527548844968e-06,
+      "loss": 3.3085,
+      "step": 6743
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0637106331658815e-06,
+      "loss": 1.8126,
+      "step": 6744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.060894235280637e-06,
+      "loss": 1.8159,
+      "step": 6745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.058078355484663e-06,
+      "loss": 2.9404,
+      "step": 6746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.055262994073334e-06,
+      "loss": 2.5094,
+      "step": 6747
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0524481513419675e-06,
+      "loss": 2.7969,
+      "step": 6748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.049633827585832e-06,
+      "loss": 2.2771,
+      "step": 6749
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.046820023100129e-06,
+      "loss": 2.6777,
+      "step": 6750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0440067381800165e-06,
+      "loss": 3.7625,
+      "step": 6751
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.041193973120595e-06,
+      "loss": 3.2763,
+      "step": 6752
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.038381728216912e-06,
+      "loss": 1.4193,
+      "step": 6753
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.035570003763958e-06,
+      "loss": 1.9363,
+      "step": 6754
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.03275880005667e-06,
+      "loss": 2.4167,
+      "step": 6755
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.02994811738993e-06,
+      "loss": 3.5218,
+      "step": 6756
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.027137956058565e-06,
+      "loss": 1.8078,
+      "step": 6757
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.02432831635735e-06,
+      "loss": 3.5942,
+      "step": 6758
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.021519198581002e-06,
+      "loss": 3.8029,
+      "step": 6759
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.018710603024187e-06,
+      "loss": 2.7853,
+      "step": 6760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.015902529981508e-06,
+      "loss": 2.0579,
+      "step": 6761
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.013094979747522e-06,
+      "loss": 2.4427,
+      "step": 6762
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.010287952616726e-06,
+      "loss": 0.8732,
+      "step": 6763
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.007481448883567e-06,
+      "loss": 2.1883,
+      "step": 6764
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.004675468842436e-06,
+      "loss": 2.3656,
+      "step": 6765
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0018700127876606e-06,
+      "loss": 2.7386,
+      "step": 6766
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9990650810135246e-06,
+      "loss": 2.5855,
+      "step": 6767
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.99626067381425e-06,
+      "loss": 3.3154,
+      "step": 6768
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.993456791484009e-06,
+      "loss": 2.2881,
+      "step": 6769
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.990653434316915e-06,
+      "loss": 2.0737,
+      "step": 6770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9878506026070265e-06,
+      "loss": 3.7133,
+      "step": 6771
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.985048296648346e-06,
+      "loss": 3.268,
+      "step": 6772
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.982246516734826e-06,
+      "loss": 2.0343,
+      "step": 6773
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.979445263160357e-06,
+      "loss": 2.7451,
+      "step": 6774
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.976644536218783e-06,
+      "loss": 2.1007,
+      "step": 6775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.973844336203879e-06,
+      "loss": 1.6351,
+      "step": 6776
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.971044663409378e-06,
+      "loss": 3.0861,
+      "step": 6777
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.96824551812895e-06,
+      "loss": 2.4484,
+      "step": 6778
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.965446900656215e-06,
+      "loss": 3.1249,
+      "step": 6779
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9626488112847384e-06,
+      "loss": 2.9683,
+      "step": 6780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.959851250308019e-06,
+      "loss": 1.6851,
+      "step": 6781
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.957054218019511e-06,
+      "loss": 4.1309,
+      "step": 6782
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9542577147126095e-06,
+      "loss": 3.4246,
+      "step": 6783
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.951461740680655e-06,
+      "loss": 1.8686,
+      "step": 6784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.948666296216938e-06,
+      "loss": 1.957,
+      "step": 6785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.945871381614679e-06,
+      "loss": 1.4356,
+      "step": 6786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.943076997167055e-06,
+      "loss": 1.7553,
+      "step": 6787
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9402831431671834e-06,
+      "loss": 3.3272,
+      "step": 6788
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9374898199081245e-06,
+      "loss": 2.5825,
+      "step": 6789
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.934697027682894e-06,
+      "loss": 3.2299,
+      "step": 6790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.931904766784434e-06,
+      "loss": 3.4316,
+      "step": 6791
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.929113037505642e-06,
+      "loss": 2.4971,
+      "step": 6792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.926321840139357e-06,
+      "loss": 3.3822,
+      "step": 6793
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.923531174978363e-06,
+      "loss": 2.0636,
+      "step": 6794
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9207410423153925e-06,
+      "loss": 2.8518,
+      "step": 6795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.91795144244311e-06,
+      "loss": 2.2417,
+      "step": 6796
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9151623756541334e-06,
+      "loss": 2.0387,
+      "step": 6797
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.912373842241025e-06,
+      "loss": 1.7528,
+      "step": 6798
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.909585842496287e-06,
+      "loss": 3.0004,
+      "step": 6799
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9067983767123736e-06,
+      "loss": 3.2364,
+      "step": 6800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.904011445181668e-06,
+      "loss": 3.79,
+      "step": 6801
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.901225048196511e-06,
+      "loss": 2.4318,
+      "step": 6802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.898439186049182e-06,
+      "loss": 2.5639,
+      "step": 6803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.895653859031906e-06,
+      "loss": 3.3303,
+      "step": 6804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8928690674368495e-06,
+      "loss": 2.1737,
+      "step": 6805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890084811556125e-06,
+      "loss": 2.7328,
+      "step": 6806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8873010916817874e-06,
+      "loss": 4.0897,
+      "step": 6807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.884517908105837e-06,
+      "loss": 2.2932,
+      "step": 6808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881735261120216e-06,
+      "loss": 3.6337,
+      "step": 6809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.878953151016816e-06,
+      "loss": 3.5345,
+      "step": 6810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.876171578087459e-06,
+      "loss": 3.1354,
+      "step": 6811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.873390542623922e-06,
+      "loss": 0.9248,
+      "step": 6812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.870610044917925e-06,
+      "loss": 4.1941,
+      "step": 6813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.867830085261128e-06,
+      "loss": 1.5152,
+      "step": 6814
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8650506639451385e-06,
+      "loss": 2.4316,
+      "step": 6815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8622717812615e-06,
+      "loss": 2.5348,
+      "step": 6816
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8594934375017065e-06,
+      "loss": 2.4958,
+      "step": 6817
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856715632957193e-06,
+      "loss": 2.3781,
+      "step": 6818
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853938367919338e-06,
+      "loss": 2.34,
+      "step": 6819
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851161642679466e-06,
+      "loss": 2.4119,
+      "step": 6820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.848385457528845e-06,
+      "loss": 2.7667,
+      "step": 6821
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.845609812758676e-06,
+      "loss": 2.6266,
+      "step": 6822
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.842834708660115e-06,
+      "loss": 2.6203,
+      "step": 6823
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.840060145524254e-06,
+      "loss": 2.4126,
+      "step": 6824
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.837286123642141e-06,
+      "loss": 2.5181,
+      "step": 6825
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.834512643304755e-06,
+      "loss": 3.58,
+      "step": 6826
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.831739704803016e-06,
+      "loss": 2.1104,
+      "step": 6827
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.828967308427795e-06,
+      "loss": 1.837,
+      "step": 6828
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8261954544699045e-06,
+      "loss": 2.0569,
+      "step": 6829
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.823424143220097e-06,
+      "loss": 3.2863,
+      "step": 6830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.820653374969077e-06,
+      "loss": 0.7481,
+      "step": 6831
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.817883150007474e-06,
+      "loss": 2.1256,
+      "step": 6832
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8151134686258794e-06,
+      "loss": 2.6696,
+      "step": 6833
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8123443311148175e-06,
+      "loss": 2.3649,
+      "step": 6834
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.809575737764759e-06,
+      "loss": 1.8497,
+      "step": 6835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.806807688866119e-06,
+      "loss": 2.1315,
+      "step": 6836
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.804040184709248e-06,
+      "loss": 2.1141,
+      "step": 6837
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.801273225584445e-06,
+      "loss": 2.3519,
+      "step": 6838
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.798506811781953e-06,
+      "loss": 3.0092,
+      "step": 6839
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.795740943591955e-06,
+      "loss": 2.9088,
+      "step": 6840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7929756213045796e-06,
+      "loss": 2.1319,
+      "step": 6841
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.790210845209894e-06,
+      "loss": 2.5146,
+      "step": 6842
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7874466155979116e-06,
+      "loss": 2.7777,
+      "step": 6843
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.784682932758588e-06,
+      "loss": 2.8449,
+      "step": 6844
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.781919796981818e-06,
+      "loss": 1.1696,
+      "step": 6845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.779157208557448e-06,
+      "loss": 3.4238,
+      "step": 6846
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.776395167775252e-06,
+      "loss": 2.5199,
+      "step": 6847
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.77363367492496e-06,
+      "loss": 3.0161,
+      "step": 6848
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.770872730296239e-06,
+      "loss": 3.0509,
+      "step": 6849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7681123341787e-06,
+      "loss": 1.2497,
+      "step": 6850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7653524868618975e-06,
+      "loss": 1.6861,
+      "step": 6851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7625931886353215e-06,
+      "loss": 3.3268,
+      "step": 6852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.759834439788412e-06,
+      "loss": 2.5261,
+      "step": 6853
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.75707624061055e-06,
+      "loss": 2.7039,
+      "step": 6854
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.754318591391057e-06,
+      "loss": 1.4417,
+      "step": 6855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7515614924192026e-06,
+      "loss": 3.3762,
+      "step": 6856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.748804943984183e-06,
+      "loss": 3.061,
+      "step": 6857
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.74604894637515e-06,
+      "loss": 2.5958,
+      "step": 6858
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.743293499881203e-06,
+      "loss": 3.0649,
+      "step": 6859
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.740538604791371e-06,
+      "loss": 3.5034,
+      "step": 6860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.737784261394632e-06,
+      "loss": 2.9536,
+      "step": 6861
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7350304699798976e-06,
+      "loss": 2.0072,
+      "step": 6862
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732277230836032e-06,
+      "loss": 2.6765,
+      "step": 6863
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.729524544251837e-06,
+      "loss": 2.2759,
+      "step": 6864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.726772410516055e-06,
+      "loss": 1.7481,
+      "step": 6865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.724020829917378e-06,
+      "loss": 1.8793,
+      "step": 6866
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.721269802744425e-06,
+      "loss": 2.3069,
+      "step": 6867
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.718519329285771e-06,
+      "loss": 2.6664,
+      "step": 6868
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.715769409829926e-06,
+      "loss": 2.6578,
+      "step": 6869
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.713020044665348e-06,
+      "loss": 1.4241,
+      "step": 6870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.71027123408043e-06,
+      "loss": 3.2142,
+      "step": 6871
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.707522978363508e-06,
+      "loss": 2.7367,
+      "step": 6872
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.704775277802861e-06,
+      "loss": 3.1269,
+      "step": 6873
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.702028132686711e-06,
+      "loss": 2.4648,
+      "step": 6874
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.699281543303222e-06,
+      "loss": 3.0376,
+      "step": 6875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.696535509940499e-06,
+      "loss": 2.1773,
+      "step": 6876
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.693790032886585e-06,
+      "loss": 3.0354,
+      "step": 6877
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.69104511242947e-06,
+      "loss": 2.4974,
+      "step": 6878
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.688300748857083e-06,
+      "loss": 1.8425,
+      "step": 6879
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.685556942457296e-06,
+      "loss": 4.2605,
+      "step": 6880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.68281369351792e-06,
+      "loss": 3.6148,
+      "step": 6881
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.680071002326714e-06,
+      "loss": 3.9488,
+      "step": 6882
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.677328869171365e-06,
+      "loss": 2.7998,
+      "step": 6883
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.674587294339513e-06,
+      "loss": 2.0613,
+      "step": 6884
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.67184627811874e-06,
+      "loss": 1.7031,
+      "step": 6885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.669105820796564e-06,
+      "loss": 2.2748,
+      "step": 6886
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6663659226604474e-06,
+      "loss": 2.9802,
+      "step": 6887
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663626583997789e-06,
+      "loss": 1.6969,
+      "step": 6888
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660887805095935e-06,
+      "loss": 3.459,
+      "step": 6889
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.65814958624217e-06,
+      "loss": 2.1757,
+      "step": 6890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.655411927723722e-06,
+      "loss": 2.8641,
+      "step": 6891
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.652674829827762e-06,
+      "loss": 3.1819,
+      "step": 6892
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.649938292841386e-06,
+      "loss": 2.6591,
+      "step": 6893
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.647202317051656e-06,
+      "loss": 2.313,
+      "step": 6894
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6444669027455615e-06,
+      "loss": 1.9674,
+      "step": 6895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.641732050210032e-06,
+      "loss": 3.8918,
+      "step": 6896
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.638997759731946e-06,
+      "loss": 1.9987,
+      "step": 6897
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.63626403159811e-06,
+      "loss": 1.7028,
+      "step": 6898
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6335308660952824e-06,
+      "loss": 2.8579,
+      "step": 6899
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.630798263510162e-06,
+      "loss": 0.9583,
+      "step": 6900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6280662241293835e-06,
+      "loss": 2.8572,
+      "step": 6901
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.625334748239531e-06,
+      "loss": 3.61,
+      "step": 6902
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.622603836127113e-06,
+      "loss": 2.7043,
+      "step": 6903
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.619873488078597e-06,
+      "loss": 2.4788,
+      "step": 6904
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.617143704380382e-06,
+      "loss": 3.251,
+      "step": 6905
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.614414485318809e-06,
+      "loss": 1.8865,
+      "step": 6906
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.611685831180166e-06,
+      "loss": 2.8285,
+      "step": 6907
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.608957742250667e-06,
+      "loss": 2.7341,
+      "step": 6908
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.60623021881648e-06,
+      "loss": 2.9583,
+      "step": 6909
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.60350326116371e-06,
+      "loss": 3.0099,
+      "step": 6910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600776869578403e-06,
+      "loss": 2.7977,
+      "step": 6911
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.598051044346542e-06,
+      "loss": 1.485,
+      "step": 6912
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595325785754056e-06,
+      "loss": 2.9786,
+      "step": 6913
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.592601094086811e-06,
+      "loss": 3.91,
+      "step": 6914
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.589876969630616e-06,
+      "loss": 3.0048,
+      "step": 6915
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.587153412671217e-06,
+      "loss": 3.97,
+      "step": 6916
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.584430423494306e-06,
+      "loss": 2.083,
+      "step": 6917
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.581708002385506e-06,
+      "loss": 1.4248,
+      "step": 6918
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.57898614963039e-06,
+      "loss": 2.5266,
+      "step": 6919
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.576264865514467e-06,
+      "loss": 3.4411,
+      "step": 6920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.573544150323187e-06,
+      "loss": 2.5019,
+      "step": 6921
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.570824004341945e-06,
+      "loss": 2.9202,
+      "step": 6922
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568104427856063e-06,
+      "loss": 2.8088,
+      "step": 6923
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.565385421150817e-06,
+      "loss": 2.8675,
+      "step": 6924
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.562666984511416e-06,
+      "loss": 2.4949,
+      "step": 6925
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5599491182230134e-06,
+      "loss": 2.6902,
+      "step": 6926
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.557231822570704e-06,
+      "loss": 2.3744,
+      "step": 6927
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.554515097839511e-06,
+      "loss": 1.7372,
+      "step": 6928
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.551798944314412e-06,
+      "loss": 2.2707,
+      "step": 6929
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.549083362280318e-06,
+      "loss": 2.3903,
+      "step": 6930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.54636835202208e-06,
+      "loss": 2.4782,
+      "step": 6931
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543653913824496e-06,
+      "loss": 3.479,
+      "step": 6932
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.540940047972289e-06,
+      "loss": 2.029,
+      "step": 6933
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5382267547501335e-06,
+      "loss": 2.4534,
+      "step": 6934
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.535514034442644e-06,
+      "loss": 2.4522,
+      "step": 6935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.53280188733437e-06,
+      "loss": 3.0116,
+      "step": 6936
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530090313709808e-06,
+      "loss": 2.9567,
+      "step": 6937
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.527379313853381e-06,
+      "loss": 3.1107,
+      "step": 6938
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.524668888049466e-06,
+      "loss": 3.211,
+      "step": 6939
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.521959036582372e-06,
+      "loss": 2.8004,
+      "step": 6940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.519249759736351e-06,
+      "loss": 3.3773,
+      "step": 6941
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.516541057795593e-06,
+      "loss": 2.467,
+      "step": 6942
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.513832931044232e-06,
+      "loss": 4.0538,
+      "step": 6943
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.511125379766332e-06,
+      "loss": 1.831,
+      "step": 6944
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.508418404245903e-06,
+      "loss": 3.135,
+      "step": 6945
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5057120047668975e-06,
+      "loss": 2.7117,
+      "step": 6946
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.503006181613203e-06,
+      "loss": 2.9751,
+      "step": 6947
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.500300935068647e-06,
+      "loss": 3.1982,
+      "step": 6948
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497596265416999e-06,
+      "loss": 2.5001,
+      "step": 6949
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.494892172941965e-06,
+      "loss": 1.6854,
+      "step": 6950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.492188657927191e-06,
+      "loss": 2.7989,
+      "step": 6951
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.489485720656266e-06,
+      "loss": 2.1724,
+      "step": 6952
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.486783361412717e-06,
+      "loss": 3.8488,
+      "step": 6953
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4840815804800005e-06,
+      "loss": 2.3942,
+      "step": 6954
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.481380378141528e-06,
+      "loss": 3.412,
+      "step": 6955
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.478679754680639e-06,
+      "loss": 2.4906,
+      "step": 6956
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4759797103806205e-06,
+      "loss": 2.8426,
+      "step": 6957
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.473280245524696e-06,
+      "loss": 3.1436,
+      "step": 6958
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470581360396019e-06,
+      "loss": 2.8732,
+      "step": 6959
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.467883055277696e-06,
+      "loss": 3.6651,
+      "step": 6960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.465185330452763e-06,
+      "loss": 2.3552,
+      "step": 6961
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.462488186204201e-06,
+      "loss": 3.5985,
+      "step": 6962
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.45979162281493e-06,
+      "loss": 1.9494,
+      "step": 6963
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.457095640567804e-06,
+      "loss": 4.293,
+      "step": 6964
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454400239745619e-06,
+      "loss": 3.5805,
+      "step": 6965
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.451705420631111e-06,
+      "loss": 1.5002,
+      "step": 6966
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.449011183506955e-06,
+      "loss": 3.2741,
+      "step": 6967
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4463175286557654e-06,
+      "loss": 1.9736,
+      "step": 6968
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.44362445636009e-06,
+      "loss": 1.7963,
+      "step": 6969
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.440931966902419e-06,
+      "loss": 2.9266,
+      "step": 6970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.438240060565185e-06,
+      "loss": 2.1437,
+      "step": 6971
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435548737630756e-06,
+      "loss": 2.628,
+      "step": 6972
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4328579983814425e-06,
+      "loss": 2.1391,
+      "step": 6973
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.430167843099483e-06,
+      "loss": 2.5371,
+      "step": 6974
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.427478272067066e-06,
+      "loss": 2.9405,
+      "step": 6975
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4247892855663164e-06,
+      "loss": 2.2706,
+      "step": 6976
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.422100883879296e-06,
+      "loss": 4.0757,
+      "step": 6977
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.419413067288006e-06,
+      "loss": 3.3473,
+      "step": 6978
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.416725836074384e-06,
+      "loss": 2.5291,
+      "step": 6979
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.414039190520308e-06,
+      "loss": 2.7618,
+      "step": 6980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.411353130907595e-06,
+      "loss": 2.1941,
+      "step": 6981
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.408667657518002e-06,
+      "loss": 2.4296,
+      "step": 6982
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.405982770633221e-06,
+      "loss": 2.7178,
+      "step": 6983
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403298470534885e-06,
+      "loss": 2.1581,
+      "step": 6984
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.400614757504565e-06,
+      "loss": 2.2505,
+      "step": 6985
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.397931631823767e-06,
+      "loss": 2.6598,
+      "step": 6986
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.395249093773943e-06,
+      "loss": 2.9783,
+      "step": 6987
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.39256714363648e-06,
+      "loss": 3.5545,
+      "step": 6988
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.389885781692695e-06,
+      "loss": 1.614,
+      "step": 6989
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3872050082238535e-06,
+      "loss": 3.5661,
+      "step": 6990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.384524823511157e-06,
+      "loss": 3.2136,
+      "step": 6991
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3818452278357445e-06,
+      "loss": 2.5415,
+      "step": 6992
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.379166221478697e-06,
+      "loss": 2.3499,
+      "step": 6993
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.376487804721022e-06,
+      "loss": 3.5274,
+      "step": 6994
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.373809977843676e-06,
+      "loss": 2.4399,
+      "step": 6995
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371132741127553e-06,
+      "loss": 2.4787,
+      "step": 6996
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36845609485348e-06,
+      "loss": 2.2845,
+      "step": 6997
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3657800393022255e-06,
+      "loss": 2.5239,
+      "step": 6998
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363104574754495e-06,
+      "loss": 3.2764,
+      "step": 6999
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.360429701490935e-06,
+      "loss": 2.9048,
+      "step": 7000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.357755419792123e-06,
+      "loss": 1.9035,
+      "step": 7001
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355081729938583e-06,
+      "loss": 1.8889,
+      "step": 7002
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.35240863221077e-06,
+      "loss": 2.4325,
+      "step": 7003
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.349736126889084e-06,
+      "loss": 2.0001,
+      "step": 7004
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.34706421425385e-06,
+      "loss": 2.8788,
+      "step": 7005
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.344392894585344e-06,
+      "loss": 2.5413,
+      "step": 7006
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.341722168163776e-06,
+      "loss": 2.3232,
+      "step": 7007
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.339052035269291e-06,
+      "loss": 1.9385,
+      "step": 7008
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.336382496181978e-06,
+      "loss": 3.0485,
+      "step": 7009
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3337135511818514e-06,
+      "loss": 2.5085,
+      "step": 7010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.331045200548876e-06,
+      "loss": 3.8002,
+      "step": 7011
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.328377444562948e-06,
+      "loss": 2.9296,
+      "step": 7012
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.325710283503903e-06,
+      "loss": 1.7177,
+      "step": 7013
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.323043717651518e-06,
+      "loss": 1.4118,
+      "step": 7014
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.320377747285497e-06,
+      "loss": 2.9885,
+      "step": 7015
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31771237268549e-06,
+      "loss": 2.6297,
+      "step": 7016
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315047594131081e-06,
+      "loss": 1.6514,
+      "step": 7017
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.312383411901796e-06,
+      "loss": 2.9442,
+      "step": 7018
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.309719826277094e-06,
+      "loss": 3.0819,
+      "step": 7019
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.307056837536373e-06,
+      "loss": 1.5225,
+      "step": 7020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304394445958967e-06,
+      "loss": 2.4168,
+      "step": 7021
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.301732651824151e-06,
+      "loss": 2.8562,
+      "step": 7022
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.299071455411132e-06,
+      "loss": 2.3053,
+      "step": 7023
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.296410856999062e-06,
+      "loss": 2.4646,
+      "step": 7024
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2937508568670194e-06,
+      "loss": 1.678,
+      "step": 7025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.291091455294028e-06,
+      "loss": 1.8227,
+      "step": 7026
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2884326525590466e-06,
+      "loss": 2.9735,
+      "step": 7027
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2857744489409725e-06,
+      "loss": 1.8376,
+      "step": 7028
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2831168447186425e-06,
+      "loss": 2.7087,
+      "step": 7029
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.280459840170818e-06,
+      "loss": 3.3846,
+      "step": 7030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277803435576211e-06,
+      "loss": 2.7642,
+      "step": 7031
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2751476312134655e-06,
+      "loss": 2.8608,
+      "step": 7032
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272492427361165e-06,
+      "loss": 2.5888,
+      "step": 7033
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2698378242978254e-06,
+      "loss": 2.6371,
+      "step": 7034
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.267183822301903e-06,
+      "loss": 3.422,
+      "step": 7035
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.264530421651792e-06,
+      "loss": 2.8855,
+      "step": 7036
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.261877622625821e-06,
+      "loss": 1.9769,
+      "step": 7037
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.259225425502256e-06,
+      "loss": 1.7912,
+      "step": 7038
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.256573830559303e-06,
+      "loss": 2.3863,
+      "step": 7039
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2539228380750955e-06,
+      "loss": 2.2259,
+      "step": 7040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.251272448327713e-06,
+      "loss": 3.4441,
+      "step": 7041
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.248622661595172e-06,
+      "loss": 3.7429,
+      "step": 7042
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.24597347815542e-06,
+      "loss": 2.1681,
+      "step": 7043
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.243324898286349e-06,
+      "loss": 3.5266,
+      "step": 7044
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.240676922265774e-06,
+      "loss": 2.9666,
+      "step": 7045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.238029550371459e-06,
+      "loss": 3.1118,
+      "step": 7046
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2353827828811046e-06,
+      "loss": 2.6864,
+      "step": 7047
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2327366200723404e-06,
+      "loss": 1.5112,
+      "step": 7048
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230091062222742e-06,
+      "loss": 3.2303,
+      "step": 7049
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2274461096098085e-06,
+      "loss": 2.9119,
+      "step": 7050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.224801762510987e-06,
+      "loss": 2.5618,
+      "step": 7051
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.222158021203657e-06,
+      "loss": 2.9626,
+      "step": 7052
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2195148859651344e-06,
+      "loss": 1.6153,
+      "step": 7053
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2168723570726735e-06,
+      "loss": 2.4659,
+      "step": 7054
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.21423043480346e-06,
+      "loss": 2.6778,
+      "step": 7055
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.211589119434622e-06,
+      "loss": 2.8963,
+      "step": 7056
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208948411243222e-06,
+      "loss": 2.9665,
+      "step": 7057
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.206308310506255e-06,
+      "loss": 2.1966,
+      "step": 7058
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203668817500662e-06,
+      "loss": 2.0426,
+      "step": 7059
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.201029932503303e-06,
+      "loss": 2.4801,
+      "step": 7060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.198391655790991e-06,
+      "loss": 1.8657,
+      "step": 7061
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1957539876404685e-06,
+      "loss": 2.4254,
+      "step": 7062
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1931169283284134e-06,
+      "loss": 2.6218,
+      "step": 7063
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.190480478131443e-06,
+      "loss": 2.3125,
+      "step": 7064
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.18784463732611e-06,
+      "loss": 1.7903,
+      "step": 7065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.185209406188898e-06,
+      "loss": 1.3272,
+      "step": 7066
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.182574784996231e-06,
+      "loss": 1.9492,
+      "step": 7067
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.179940774024469e-06,
+      "loss": 3.5589,
+      "step": 7068
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.177307373549908e-06,
+      "loss": 1.6046,
+      "step": 7069
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.17467458384878e-06,
+      "loss": 4.2294,
+      "step": 7070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1720424051972515e-06,
+      "loss": 1.6356,
+      "step": 7071
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.169410837871427e-06,
+      "loss": 3.9687,
+      "step": 7072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.166779882147345e-06,
+      "loss": 3.0746,
+      "step": 7073
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.16414953830098e-06,
+      "loss": 1.3847,
+      "step": 7074
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1615198066082475e-06,
+      "loss": 3.0004,
+      "step": 7075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.158890687344986e-06,
+      "loss": 3.1588,
+      "step": 7076
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.156262180786983e-06,
+      "loss": 2.0977,
+      "step": 7077
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.153634287209955e-06,
+      "loss": 2.4046,
+      "step": 7078
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.151007006889556e-06,
+      "loss": 2.895,
+      "step": 7079
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.14838034010138e-06,
+      "loss": 3.4388,
+      "step": 7080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1457542871209445e-06,
+      "loss": 1.8073,
+      "step": 7081
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.143128848223714e-06,
+      "loss": 2.7626,
+      "step": 7082
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.140504023685085e-06,
+      "loss": 3.7633,
+      "step": 7083
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.137879813780388e-06,
+      "loss": 1.9648,
+      "step": 7084
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.135256218784896e-06,
+      "loss": 2.6177,
+      "step": 7085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.132633238973803e-06,
+      "loss": 2.6609,
+      "step": 7086
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.130010874622253e-06,
+      "loss": 2.2907,
+      "step": 7087
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.127389126005319e-06,
+      "loss": 2.5706,
+      "step": 7088
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.124767993398009e-06,
+      "loss": 3.4934,
+      "step": 7089
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 1.8234,
+      "step": 7090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1195275773119805e-06,
+      "loss": 3.6845,
+      "step": 7091
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.116908294382956e-06,
+      "loss": 1.6615,
+      "step": 7092
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1142896285629455e-06,
+      "loss": 2.9005,
+      "step": 7093
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.111671580126637e-06,
+      "loss": 3.3401,
+      "step": 7094
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1090541493486555e-06,
+      "loss": 2.8195,
+      "step": 7095
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.10643733650355e-06,
+      "loss": 1.8787,
+      "step": 7096
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.103821141865815e-06,
+      "loss": 2.6723,
+      "step": 7097
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.101205565709876e-06,
+      "loss": 2.3237,
+      "step": 7098
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.098590608310097e-06,
+      "loss": 2.1779,
+      "step": 7099
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.095976269940777e-06,
+      "loss": 2.8172,
+      "step": 7100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093362550876141e-06,
+      "loss": 1.2491,
+      "step": 7101
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.09074945139036e-06,
+      "loss": 3.5242,
+      "step": 7102
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.088136971757537e-06,
+      "loss": 2.8013,
+      "step": 7103
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.085525112251706e-06,
+      "loss": 2.9884,
+      "step": 7104
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.082913873146842e-06,
+      "loss": 2.7654,
+      "step": 7105
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.080303254716851e-06,
+      "loss": 3.1604,
+      "step": 7106
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.077693257235573e-06,
+      "loss": 2.5082,
+      "step": 7107
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0750838809767875e-06,
+      "loss": 2.3316,
+      "step": 7108
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.072475126214204e-06,
+      "loss": 2.445,
+      "step": 7109
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.069866993221473e-06,
+      "loss": 1.7275,
+      "step": 7110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.067259482272168e-06,
+      "loss": 2.8238,
+      "step": 7111
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0646525936398086e-06,
+      "loss": 2.9105,
+      "step": 7112
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.062046327597845e-06,
+      "loss": 2.5989,
+      "step": 7113
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0594406844196636e-06,
+      "loss": 3.0453,
+      "step": 7114
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056835664378585e-06,
+      "loss": 1.8365,
+      "step": 7115
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.054231267747862e-06,
+      "loss": 3.0346,
+      "step": 7116
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.05162749480068e-06,
+      "loss": 2.0918,
+      "step": 7117
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.049024345810169e-06,
+      "loss": 1.5979,
+      "step": 7118
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.046421821049385e-06,
+      "loss": 2.9555,
+      "step": 7119
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.043819920791322e-06,
+      "loss": 1.7057,
+      "step": 7120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.041218645308905e-06,
+      "loss": 3.4621,
+      "step": 7121
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.038617994874994e-06,
+      "loss": 3.2578,
+      "step": 7122
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.036017969762389e-06,
+      "loss": 2.4575,
+      "step": 7123
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033418570243819e-06,
+      "loss": 3.8332,
+      "step": 7124
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.03081979659195e-06,
+      "loss": 2.2609,
+      "step": 7125
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.028221649079379e-06,
+      "loss": 2.9814,
+      "step": 7126
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.025624127978642e-06,
+      "loss": 2.9159,
+      "step": 7127
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0230272335622065e-06,
+      "loss": 1.9877,
+      "step": 7128
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.020430966102474e-06,
+      "loss": 2.9818,
+      "step": 7129
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.017835325871781e-06,
+      "loss": 3.7269,
+      "step": 7130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.015240313142401e-06,
+      "loss": 2.8053,
+      "step": 7131
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.012645928186533e-06,
+      "loss": 2.6283,
+      "step": 7132
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.010052171276319e-06,
+      "loss": 2.0626,
+      "step": 7133
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.007459042683832e-06,
+      "loss": 3.8805,
+      "step": 7134
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.004866542681079e-06,
+      "loss": 1.6333,
+      "step": 7135
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.002274671540006e-06,
+      "loss": 2.4248,
+      "step": 7136
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9996834295324796e-06,
+      "loss": 2.2763,
+      "step": 7137
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.997092816930313e-06,
+      "loss": 2.1814,
+      "step": 7138
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994502834005249e-06,
+      "loss": 3.3586,
+      "step": 7139
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.991913481028965e-06,
+      "loss": 2.164,
+      "step": 7140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9893247582730736e-06,
+      "loss": 2.3996,
+      "step": 7141
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9867366660091176e-06,
+      "loss": 4.1248,
+      "step": 7142
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984149204508578e-06,
+      "loss": 2.7406,
+      "step": 7143
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.981562374042867e-06,
+      "loss": 1.9802,
+      "step": 7144
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978976174883329e-06,
+      "loss": 2.5716,
+      "step": 7145
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.976390607301249e-06,
+      "loss": 3.6894,
+      "step": 7146
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9738056715678355e-06,
+      "loss": 3.2026,
+      "step": 7147
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.971221367954239e-06,
+      "loss": 3.3339,
+      "step": 7148
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.96863769673154e-06,
+      "loss": 1.4568,
+      "step": 7149
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.966054658170754e-06,
+      "loss": 1.6586,
+      "step": 7150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.963472252542833e-06,
+      "loss": 2.2945,
+      "step": 7151
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.960890480118653e-06,
+      "loss": 2.5476,
+      "step": 7152
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9583093411690345e-06,
+      "loss": 2.1687,
+      "step": 7153
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.955728835964724e-06,
+      "loss": 3.4183,
+      "step": 7154
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.953148964776408e-06,
+      "loss": 3.7864,
+      "step": 7155
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.950569727874704e-06,
+      "loss": 3.1732,
+      "step": 7156
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.947991125530155e-06,
+      "loss": 2.6534,
+      "step": 7157
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.945413158013249e-06,
+      "loss": 2.2049,
+      "step": 7158
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942835825594403e-06,
+      "loss": 1.7093,
+      "step": 7159
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.940259128543967e-06,
+      "loss": 0.9353,
+      "step": 7160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9376830671322245e-06,
+      "loss": 3.0851,
+      "step": 7161
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.935107641629391e-06,
+      "loss": 2.051,
+      "step": 7162
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932532852305618e-06,
+      "loss": 2.0058,
+      "step": 7163
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9299586994309905e-06,
+      "loss": 2.4947,
+      "step": 7164
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.927385183275522e-06,
+      "loss": 3.6417,
+      "step": 7165
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924812304109167e-06,
+      "loss": 2.183,
+      "step": 7166
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.922240062201803e-06,
+      "loss": 2.4734,
+      "step": 7167
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.919668457823248e-06,
+      "loss": 3.2155,
+      "step": 7168
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.917097491243252e-06,
+      "loss": 1.4808,
+      "step": 7169
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.914527162731498e-06,
+      "loss": 3.0425,
+      "step": 7170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911957472557605e-06,
+      "loss": 2.1124,
+      "step": 7171
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.909388420991113e-06,
+      "loss": 1.8593,
+      "step": 7172
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.906820008301509e-06,
+      "loss": 3.6547,
+      "step": 7173
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.904252234758208e-06,
+      "loss": 2.9454,
+      "step": 7174
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.901685100630554e-06,
+      "loss": 2.9041,
+      "step": 7175
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.899118606187832e-06,
+      "loss": 2.6237,
+      "step": 7176
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8965527516992514e-06,
+      "loss": 1.9392,
+      "step": 7177
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893987537433961e-06,
+      "loss": 2.0632,
+      "step": 7178
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.891422963661039e-06,
+      "loss": 1.6714,
+      "step": 7179
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.888859030649498e-06,
+      "loss": 2.8348,
+      "step": 7180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.886295738668281e-06,
+      "loss": 3.3784,
+      "step": 7181
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.88373308798627e-06,
+      "loss": 1.9677,
+      "step": 7182
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8811710788722674e-06,
+      "loss": 3.4916,
+      "step": 7183
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.878609711595022e-06,
+      "loss": 3.243,
+      "step": 7184
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.876048986423207e-06,
+      "loss": 2.7955,
+      "step": 7185
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.87348890362543e-06,
+      "loss": 2.5595,
+      "step": 7186
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8709294634702374e-06,
+      "loss": 2.2688,
+      "step": 7187
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8683706662260945e-06,
+      "loss": 3.7698,
+      "step": 7188
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8658125121614106e-06,
+      "loss": 3.9698,
+      "step": 7189
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.863255001544526e-06,
+      "loss": 2.7531,
+      "step": 7190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.860698134643709e-06,
+      "loss": 1.4639,
+      "step": 7191
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.858141911727168e-06,
+      "loss": 1.5676,
+      "step": 7192
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855586333063031e-06,
+      "loss": 2.3051,
+      "step": 7193
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.853031398919372e-06,
+      "loss": 3.7348,
+      "step": 7194
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8504771095641905e-06,
+      "loss": 3.8985,
+      "step": 7195
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.847923465265418e-06,
+      "loss": 2.1712,
+      "step": 7196
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.845370466290922e-06,
+      "loss": 3.0159,
+      "step": 7197
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.842818112908498e-06,
+      "loss": 2.545,
+      "step": 7198
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.84026640538588e-06,
+      "loss": 1.749,
+      "step": 7199
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.837715343990727e-06,
+      "loss": 3.6814,
+      "step": 7200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.835164928990635e-06,
+      "loss": 2.5944,
+      "step": 7201
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832615160653131e-06,
+      "loss": 1.9177,
+      "step": 7202
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.830066039245673e-06,
+      "loss": 2.6359,
+      "step": 7203
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8275175650356485e-06,
+      "loss": 2.8337,
+      "step": 7204
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.824969738290386e-06,
+      "loss": 2.4286,
+      "step": 7205
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.822422559277139e-06,
+      "loss": 3.285,
+      "step": 7206
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.819876028263097e-06,
+      "loss": 3.2584,
+      "step": 7207
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817330145515374e-06,
+      "loss": 3.5088,
+      "step": 7208
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.814784911301024e-06,
+      "loss": 3.2051,
+      "step": 7209
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.81224032588703e-06,
+      "loss": 2.2826,
+      "step": 7210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8096963895403084e-06,
+      "loss": 2.3544,
+      "step": 7211
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807153102527704e-06,
+      "loss": 0.931,
+      "step": 7212
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8046104651159987e-06,
+      "loss": 2.5233,
+      "step": 7213
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802068477571901e-06,
+      "loss": 2.9892,
+      "step": 7214
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7995271401620548e-06,
+      "loss": 3.0343,
+      "step": 7215
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7969864531530344e-06,
+      "loss": 2.612,
+      "step": 7216
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7944464168113494e-06,
+      "loss": 2.0808,
+      "step": 7217
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79190703140343e-06,
+      "loss": 3.5801,
+      "step": 7218
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7893682971956512e-06,
+      "loss": 2.672,
+      "step": 7219
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7868302144543146e-06,
+      "loss": 3.3322,
+      "step": 7220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7842927834456512e-06,
+      "loss": 2.5854,
+      "step": 7221
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7817560044358302e-06,
+      "loss": 2.2852,
+      "step": 7222
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7792198776909406e-06,
+      "loss": 3.1153,
+      "step": 7223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7766844034770155e-06,
+      "loss": 3.3745,
+      "step": 7224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7741495820600128e-06,
+      "loss": 2.3584,
+      "step": 7225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7716154137058236e-06,
+      "loss": 3.2069,
+      "step": 7226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7690818986802744e-06,
+      "loss": 2.2085,
+      "step": 7227
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.766549037249112e-06,
+      "loss": 3.3323,
+      "step": 7228
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7640168296780243e-06,
+      "loss": 2.6706,
+      "step": 7229
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7614852762326303e-06,
+      "loss": 2.0091,
+      "step": 7230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7589543771784765e-06,
+      "loss": 1.762,
+      "step": 7231
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7564241327810436e-06,
+      "loss": 2.3637,
+      "step": 7232
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7538945433057407e-06,
+      "loss": 2.8071,
+      "step": 7233
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7513656090179116e-06,
+      "loss": 3.0567,
+      "step": 7234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7488373301828296e-06,
+      "loss": 3.4885,
+      "step": 7235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7463097070656995e-06,
+      "loss": 2.1168,
+      "step": 7236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7437827399316594e-06,
+      "loss": 2.4397,
+      "step": 7237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.741256429045771e-06,
+      "loss": 3.0659,
+      "step": 7238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.738730774673035e-06,
+      "loss": 3.2049,
+      "step": 7239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.736205777078381e-06,
+      "loss": 3.0926,
+      "step": 7240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7336814365266706e-06,
+      "loss": 3.368,
+      "step": 7241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7311577532826946e-06,
+      "loss": 1.8826,
+      "step": 7242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.728634727611179e-06,
+      "loss": 3.2332,
+      "step": 7243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.72611235977677e-06,
+      "loss": 2.7973,
+      "step": 7244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7235906500440576e-06,
+      "loss": 2.8904,
+      "step": 7245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.721069598677556e-06,
+      "loss": 2.747,
+      "step": 7246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7185492059417115e-06,
+      "loss": 1.9967,
+      "step": 7247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7160294721009026e-06,
+      "loss": 3.7832,
+      "step": 7248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7135103974194376e-06,
+      "loss": 2.3745,
+      "step": 7249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7109919821615546e-06,
+      "loss": 4.1047,
+      "step": 7250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.708474226591425e-06,
+      "loss": 3.1313,
+      "step": 7251
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.705957130973149e-06,
+      "loss": 3.5496,
+      "step": 7252
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.703440695570761e-06,
+      "loss": 2.336,
+      "step": 7253
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.700924920648219e-06,
+      "loss": 2.0444,
+      "step": 7254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6984098064694174e-06,
+      "loss": 2.7619,
+      "step": 7255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.69589535329818e-06,
+      "loss": 2.9106,
+      "step": 7256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6933815613982626e-06,
+      "loss": 1.2076,
+      "step": 7257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.690868431033352e-06,
+      "loss": 2.841,
+      "step": 7258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6883559624670585e-06,
+      "loss": 2.4813,
+      "step": 7259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.685844155962931e-06,
+      "loss": 2.6987,
+      "step": 7260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.683333011784447e-06,
+      "loss": 3.0796,
+      "step": 7261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.680822530195013e-06,
+      "loss": 2.3774,
+      "step": 7262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.678312711457971e-06,
+      "loss": 3.1777,
+      "step": 7263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.675803555836582e-06,
+      "loss": 3.3259,
+      "step": 7264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.673295063594049e-06,
+      "loss": 2.4902,
+      "step": 7265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6707872349935016e-06,
+      "loss": 2.5824,
+      "step": 7266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.668280070297994e-06,
+      "loss": 2.1178,
+      "step": 7267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6657735697705267e-06,
+      "loss": 2.0587,
+      "step": 7268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6632677336740107e-06,
+      "loss": 2.841,
+      "step": 7269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6607625622713005e-06,
+      "loss": 3.7465,
+      "step": 7270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6582580558251767e-06,
+      "loss": 1.1541,
+      "step": 7271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6557542145983495e-06,
+      "loss": 1.5798,
+      "step": 7272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6532510388534626e-06,
+      "loss": 2.7714,
+      "step": 7273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.650748528853084e-06,
+      "loss": 1.9426,
+      "step": 7274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6482466848597164e-06,
+      "loss": 2.21,
+      "step": 7275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6457455071357918e-06,
+      "loss": 3.6837,
+      "step": 7276
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.643244995943672e-06,
+      "loss": 2.4709,
+      "step": 7277
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6407451515456537e-06,
+      "loss": 2.753,
+      "step": 7278
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.638245974203951e-06,
+      "loss": 1.1976,
+      "step": 7279
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.63574746418072e-06,
+      "loss": 3.0123,
+      "step": 7280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6332496217380432e-06,
+      "loss": 1.9813,
+      "step": 7281
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6307524471379317e-06,
+      "loss": 1.4383,
+      "step": 7282
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.628255940642328e-06,
+      "loss": 3.8106,
+      "step": 7283
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.625760102513103e-06,
+      "loss": 2.9947,
+      "step": 7284
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6232649330120608e-06,
+      "loss": 3.1178,
+      "step": 7285
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6207704324009308e-06,
+      "loss": 3.1854,
+      "step": 7286
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6182766009413763e-06,
+      "loss": 2.3146,
+      "step": 7287
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6157834388949907e-06,
+      "loss": 3.0592,
+      "step": 7288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.613290946523289e-06,
+      "loss": 3.0469,
+      "step": 7289
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.610799124087725e-06,
+      "loss": 1.8784,
+      "step": 7290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6083079718496804e-06,
+      "loss": 3.9708,
+      "step": 7291
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6058174900704646e-06,
+      "loss": 2.6256,
+      "step": 7292
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.60332767901132e-06,
+      "loss": 2.42,
+      "step": 7293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6008385389334122e-06,
+      "loss": 2.2204,
+      "step": 7294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5983500700978425e-06,
+      "loss": 1.7693,
+      "step": 7295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.595862272765638e-06,
+      "loss": 3.0632,
+      "step": 7296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5933751471977597e-06,
+      "loss": 1.9586,
+      "step": 7297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5908886936550967e-06,
+      "loss": 1.401,
+      "step": 7298
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.588402912398462e-06,
+      "loss": 3.0235,
+      "step": 7299
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.585917803688603e-06,
+      "loss": 2.9227,
+      "step": 7300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5834333677861998e-06,
+      "loss": 2.5465,
+      "step": 7301
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.580949604951851e-06,
+      "loss": 3.4414,
+      "step": 7302
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.578466515446101e-06,
+      "loss": 3.0114,
+      "step": 7303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5759840995294136e-06,
+      "loss": 2.6916,
+      "step": 7304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5735023574621765e-06,
+      "loss": 1.9586,
+      "step": 7305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5710212895047146e-06,
+      "loss": 2.2916,
+      "step": 7306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.568540895917283e-06,
+      "loss": 2.3067,
+      "step": 7307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5660611769600604e-06,
+      "loss": 1.8665,
+      "step": 7308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.563582132893164e-06,
+      "loss": 1.9278,
+      "step": 7309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5611037639766267e-06,
+      "loss": 3.2123,
+      "step": 7310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.55862607047042e-06,
+      "loss": 2.271,
+      "step": 7311
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.556149052634443e-06,
+      "loss": 3.7561,
+      "step": 7312
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.553672710728524e-06,
+      "loss": 1.869,
+      "step": 7313
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5511970450124223e-06,
+      "loss": 2.3243,
+      "step": 7314
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.548722055745818e-06,
+      "loss": 2.5219,
+      "step": 7315
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546247743188328e-06,
+      "loss": 3.0732,
+      "step": 7316
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.543774107599497e-06,
+      "loss": 2.2339,
+      "step": 7317
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.541301149238798e-06,
+      "loss": 1.5941,
+      "step": 7318
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5388288683656336e-06,
+      "loss": 2.9544,
+      "step": 7319
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.536357265239333e-06,
+      "loss": 2.7394,
+      "step": 7320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5338863401191557e-06,
+      "loss": 1.9348,
+      "step": 7321
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5314160932642925e-06,
+      "loss": 2.0999,
+      "step": 7322
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5289465249338585e-06,
+      "loss": 3.1864,
+      "step": 7323
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5264776353869046e-06,
+      "loss": 2.7322,
+      "step": 7324
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5240094248824e-06,
+      "loss": 3.3834,
+      "step": 7325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5215418936792487e-06,
+      "loss": 2.9056,
+      "step": 7326
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5190750420362863e-06,
+      "loss": 1.8554,
+      "step": 7327
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5166088702122738e-06,
+      "loss": 2.0209,
+      "step": 7328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.514143378465903e-06,
+      "loss": 3.4301,
+      "step": 7329
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.511678567055786e-06,
+      "loss": 1.3048,
+      "step": 7330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5092144362404757e-06,
+      "loss": 2.1182,
+      "step": 7331
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5067509862784455e-06,
+      "loss": 2.8504,
+      "step": 7332
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5042882174281e-06,
+      "loss": 3.0417,
+      "step": 7333
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5018261299477772e-06,
+      "loss": 3.0112,
+      "step": 7334
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4993647240957307e-06,
+      "loss": 2.7415,
+      "step": 7335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4969040001301513e-06,
+      "loss": 1.726,
+      "step": 7336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.494443958309164e-06,
+      "loss": 3.4678,
+      "step": 7337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.491984598890812e-06,
+      "loss": 2.8434,
+      "step": 7338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4895259221330736e-06,
+      "loss": 2.4505,
+      "step": 7339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.487067928293848e-06,
+      "loss": 2.0157,
+      "step": 7340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.484610617630968e-06,
+      "loss": 2.7426,
+      "step": 7341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4821539904021963e-06,
+      "loss": 2.6673,
+      "step": 7342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.47969804686522e-06,
+      "loss": 1.8618,
+      "step": 7343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4772427872776606e-06,
+      "loss": 1.67,
+      "step": 7344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4747882118970565e-06,
+      "loss": 1.297,
+      "step": 7345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4723343209808834e-06,
+      "loss": 2.8036,
+      "step": 7346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4698811147865454e-06,
+      "loss": 2.5027,
+      "step": 7347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4674285935713715e-06,
+      "loss": 2.478,
+      "step": 7348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4649767575926217e-06,
+      "loss": 3.4807,
+      "step": 7349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4625256071074776e-06,
+      "loss": 2.6515,
+      "step": 7350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.460075142373056e-06,
+      "loss": 2.1133,
+      "step": 7351
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4576253636463996e-06,
+      "loss": 3.5817,
+      "step": 7352
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.455176271184477e-06,
+      "loss": 1.9551,
+      "step": 7353
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4527278652441896e-06,
+      "loss": 1.8979,
+      "step": 7354
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.450280146082361e-06,
+      "loss": 3.2306,
+      "step": 7355
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4478331139557475e-06,
+      "loss": 2.2454,
+      "step": 7356
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4453867691210307e-06,
+      "loss": 2.9253,
+      "step": 7357
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.442941111834822e-06,
+      "loss": 1.8217,
+      "step": 7358
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.440496142353661e-06,
+      "loss": 2.0723,
+      "step": 7359
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4380518609340076e-06,
+      "loss": 1.6473,
+      "step": 7360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.435608267832259e-06,
+      "loss": 2.3544,
+      "step": 7361
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4331653633047356e-06,
+      "loss": 2.1941,
+      "step": 7362
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4307231476076896e-06,
+      "loss": 2.2234,
+      "step": 7363
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.428281620997296e-06,
+      "loss": 2.9094,
+      "step": 7364
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4258407837296635e-06,
+      "loss": 2.5616,
+      "step": 7365
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4234006360608173e-06,
+      "loss": 3.8585,
+      "step": 7366
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4209611782467224e-06,
+      "loss": 2.9634,
+      "step": 7367
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.418522410543266e-06,
+      "loss": 2.4112,
+      "step": 7368
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4160843332062623e-06,
+      "loss": 2.6101,
+      "step": 7369
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.413646946491458e-06,
+      "loss": 3.6007,
+      "step": 7370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.411210250654514e-06,
+      "loss": 2.4371,
+      "step": 7371
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4087742459510396e-06,
+      "loss": 2.2924,
+      "step": 7372
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.406338932636556e-06,
+      "loss": 2.5015,
+      "step": 7373
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.403904310966515e-06,
+      "loss": 2.3765,
+      "step": 7374
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4014703811963024e-06,
+      "loss": 2.699,
+      "step": 7375
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3990371435812185e-06,
+      "loss": 3.1221,
+      "step": 7376
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.396604598376503e-06,
+      "loss": 2.3869,
+      "step": 7377
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3941727458373177e-06,
+      "loss": 2.9307,
+      "step": 7378
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.391741586218752e-06,
+      "loss": 2.433,
+      "step": 7379
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3893111197758276e-06,
+      "loss": 3.3701,
+      "step": 7380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3868813467634833e-06,
+      "loss": 3.4132,
+      "step": 7381
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3844522674365922e-06,
+      "loss": 1.8737,
+      "step": 7382
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.382023882049954e-06,
+      "loss": 3.2999,
+      "step": 7383
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3795961908582965e-06,
+      "loss": 2.4413,
+      "step": 7384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3771691941162755e-06,
+      "loss": 2.3032,
+      "step": 7385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.374742892078465e-06,
+      "loss": 2.4938,
+      "step": 7386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.372317284999377e-06,
+      "loss": 2.4481,
+      "step": 7387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3698923731334453e-06,
+      "loss": 1.9752,
+      "step": 7388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3674681567350332e-06,
+      "loss": 3.3579,
+      "step": 7389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3650446360584276e-06,
+      "loss": 3.2144,
+      "step": 7390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3626218113578467e-06,
+      "loss": 2.0417,
+      "step": 7391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.360199682887433e-06,
+      "loss": 4.0802,
+      "step": 7392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3577782509012557e-06,
+      "loss": 2.5554,
+      "step": 7393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3553575156533125e-06,
+      "loss": 2.6694,
+      "step": 7394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.35293747739753e-06,
+      "loss": 1.4109,
+      "step": 7395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3505181363877536e-06,
+      "loss": 2.3801,
+      "step": 7396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3480994928777633e-06,
+      "loss": 2.5366,
+      "step": 7397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3456815471212634e-06,
+      "loss": 2.3381,
+      "step": 7398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3432642993718856e-06,
+      "loss": 2.1235,
+      "step": 7399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3408477498831917e-06,
+      "loss": 2.5298,
+      "step": 7400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3384318989086596e-06,
+      "loss": 3.7669,
+      "step": 7401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3360167467017036e-06,
+      "loss": 3.1613,
+      "step": 7402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.333602293515662e-06,
+      "loss": 3.2277,
+      "step": 7403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3311885396038002e-06,
+      "loss": 3.5871,
+      "step": 7404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3287754852193143e-06,
+      "loss": 1.5944,
+      "step": 7405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.32636313061531e-06,
+      "loss": 1.7222,
+      "step": 7406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.323951476044843e-06,
+      "loss": 2.7808,
+      "step": 7407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.321540521760883e-06,
+      "loss": 2.4146,
+      "step": 7408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.319130268016325e-06,
+      "loss": 3.6073,
+      "step": 7409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3167207150640003e-06,
+      "loss": 2.8013,
+      "step": 7410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3143118631566496e-06,
+      "loss": 3.2521,
+      "step": 7411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3119037125469553e-06,
+      "loss": 1.4891,
+      "step": 7412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3094962634875216e-06,
+      "loss": 2.3715,
+      "step": 7413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.307089516230878e-06,
+      "loss": 2.3597,
+      "step": 7414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.304683471029485e-06,
+      "loss": 2.2381,
+      "step": 7415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3022781281357184e-06,
+      "loss": 2.239,
+      "step": 7416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.299873487801891e-06,
+      "loss": 2.745,
+      "step": 7417
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.297469550280239e-06,
+      "loss": 1.5081,
+      "step": 7418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2950663158229233e-06,
+      "loss": 2.8685,
+      "step": 7419
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2926637846820366e-06,
+      "loss": 1.6951,
+      "step": 7420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.290261957109585e-06,
+      "loss": 3.8054,
+      "step": 7421
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2878608333575136e-06,
+      "loss": 1.5609,
+      "step": 7422
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.285460413677689e-06,
+      "loss": 2.0422,
+      "step": 7423
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2830606983219038e-06,
+      "loss": 3.2728,
+      "step": 7424
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.280661687541876e-06,
+      "loss": 3.4747,
+      "step": 7425
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.278263381589253e-06,
+      "loss": 3.516,
+      "step": 7426
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2758657807156046e-06,
+      "loss": 1.7453,
+      "step": 7427
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2734688851724273e-06,
+      "loss": 2.8501,
+      "step": 7428
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.271072695211146e-06,
+      "loss": 2.5868,
+      "step": 7429
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.268677211083109e-06,
+      "loss": 3.3467,
+      "step": 7430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2662824330395946e-06,
+      "loss": 2.6805,
+      "step": 7431
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2638883613317974e-06,
+      "loss": 2.6396,
+      "step": 7432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2614949962108478e-06,
+      "loss": 3.1145,
+      "step": 7433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2591023379277986e-06,
+      "loss": 2.5172,
+      "step": 7434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.256710386733629e-06,
+      "loss": 2.0688,
+      "step": 7435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2543191428792466e-06,
+      "loss": 2.1484,
+      "step": 7436
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2519286066154753e-06,
+      "loss": 2.2269,
+      "step": 7437
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.249538778193074e-06,
+      "loss": 2.9425,
+      "step": 7438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.247149657862725e-06,
+      "loss": 2.5189,
+      "step": 7439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2447612458750365e-06,
+      "loss": 2.849,
+      "step": 7440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2423735424805413e-06,
+      "loss": 1.9796,
+      "step": 7441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2399865479296978e-06,
+      "loss": 2.7266,
+      "step": 7442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.237600262472891e-06,
+      "loss": 1.4148,
+      "step": 7443
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2352146863604317e-06,
+      "loss": 3.1103,
+      "step": 7444
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2328298198425556e-06,
+      "loss": 2.5334,
+      "step": 7445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.230445663169427e-06,
+      "loss": 3.2212,
+      "step": 7446
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2280622165911267e-06,
+      "loss": 2.8719,
+      "step": 7447
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2256794803576707e-06,
+      "loss": 1.722,
+      "step": 7448
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2232974547189967e-06,
+      "loss": 2.2041,
+      "step": 7449
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2209161399249677e-06,
+      "loss": 1.7294,
+      "step": 7450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2185355362253755e-06,
+      "loss": 3.0938,
+      "step": 7451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2161556438699303e-06,
+      "loss": 2.2602,
+      "step": 7452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2137764631082715e-06,
+      "loss": 2.6477,
+      "step": 7453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2113979941899666e-06,
+      "loss": 2.8853,
+      "step": 7454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.209020237364505e-06,
+      "loss": 2.6441,
+      "step": 7455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2066431928813068e-06,
+      "loss": 2.8628,
+      "step": 7456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2042668609897042e-06,
+      "loss": 2.8536,
+      "step": 7457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.201891241938969e-06,
+      "loss": 4.2379,
+      "step": 7458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.199516335978291e-06,
+      "loss": 1.5011,
+      "step": 7459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.197142143356787e-06,
+      "loss": 1.6743,
+      "step": 7460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1947686643235e-06,
+      "loss": 2.1781,
+      "step": 7461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.192395899127395e-06,
+      "loss": 2.3213,
+      "step": 7462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.190023848017365e-06,
+      "loss": 0.9492,
+      "step": 7463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1876525112422283e-06,
+      "loss": 2.2481,
+      "step": 7464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1852818890507255e-06,
+      "loss": 2.8635,
+      "step": 7465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1829119816915276e-06,
+      "loss": 2.7552,
+      "step": 7466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.180542789413221e-06,
+      "loss": 2.233,
+      "step": 7467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.178174312464326e-06,
+      "loss": 2.2558,
+      "step": 7468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.175806551093283e-06,
+      "loss": 2.2569,
+      "step": 7469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1734395055484623e-06,
+      "loss": 3.5564,
+      "step": 7470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1710731760781575e-06,
+      "loss": 2.7325,
+      "step": 7471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1687075629305787e-06,
+      "loss": 1.3997,
+      "step": 7472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.166342666353872e-06,
+      "loss": 3.1591,
+      "step": 7473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.163978486596103e-06,
+      "loss": 2.933,
+      "step": 7474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1616150239052647e-06,
+      "loss": 4.0126,
+      "step": 7475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1592522785292714e-06,
+      "loss": 2.8161,
+      "step": 7476
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.156890250715966e-06,
+      "loss": 3.2281,
+      "step": 7477
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1545289407131128e-06,
+      "loss": 3.3593,
+      "step": 7478
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.152168348768403e-06,
+      "loss": 2.3844,
+      "step": 7479
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1498084751294523e-06,
+      "loss": 2.444,
+      "step": 7480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.147449320043803e-06,
+      "loss": 2.1335,
+      "step": 7481
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.145090883758912e-06,
+      "loss": 2.4827,
+      "step": 7482
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.142733166522174e-06,
+      "loss": 2.3749,
+      "step": 7483
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1403761685809007e-06,
+      "loss": 3.3815,
+      "step": 7484
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1380198901823313e-06,
+      "loss": 2.4851,
+      "step": 7485
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.135664331573627e-06,
+      "loss": 3.2461,
+      "step": 7486
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1333094930018793e-06,
+      "loss": 2.4596,
+      "step": 7487
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.130955374714094e-06,
+      "loss": 3.0388,
+      "step": 7488
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1286019769572095e-06,
+      "loss": 3.6486,
+      "step": 7489
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.126249299978086e-06,
+      "loss": 2.0254,
+      "step": 7490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1238973440235098e-06,
+      "loss": 1.8345,
+      "step": 7491
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1215461093401913e-06,
+      "loss": 1.4195,
+      "step": 7492
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.11919559617476e-06,
+      "loss": 1.6311,
+      "step": 7493
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.116845804773775e-06,
+      "loss": 3.5075,
+      "step": 7494
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1144967353837196e-06,
+      "loss": 3.0112,
+      "step": 7495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1121483882509996e-06,
+      "loss": 4.0138,
+      "step": 7496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1098007636219464e-06,
+      "loss": 3.3949,
+      "step": 7497
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.107453861742815e-06,
+      "loss": 2.4908,
+      "step": 7498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.105107682859785e-06,
+      "loss": 3.3843,
+      "step": 7499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1027622272189572e-06,
+      "loss": 1.8913,
+      "step": 7500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1004174950663613e-06,
+      "loss": 2.9257,
+      "step": 7501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0980734866479525e-06,
+      "loss": 2.9871,
+      "step": 7502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.095730202209597e-06,
+      "loss": 2.9674,
+      "step": 7503
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0933876419971008e-06,
+      "loss": 2.2398,
+      "step": 7504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0910458062561865e-06,
+      "loss": 2.8802,
+      "step": 7505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0887046952325016e-06,
+      "loss": 2.1538,
+      "step": 7506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.086364309171621e-06,
+      "loss": 3.0646,
+      "step": 7507
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0840246483190338e-06,
+      "loss": 3.6306,
+      "step": 7508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0816857129201627e-06,
+      "loss": 3.1625,
+      "step": 7509
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0793475032203513e-06,
+      "loss": 2.5118,
+      "step": 7510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0770100194648677e-06,
+      "loss": 3.1244,
+      "step": 7511
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.074673261898903e-06,
+      "loss": 2.4583,
+      "step": 7512
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.07233723076757e-06,
+      "loss": 2.6655,
+      "step": 7513
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.070001926315911e-06,
+      "loss": 3.2621,
+      "step": 7514
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0676673487888854e-06,
+      "loss": 2.1171,
+      "step": 7515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.065333498431381e-06,
+      "loss": 2.0923,
+      "step": 7516
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.063000375488212e-06,
+      "loss": 2.1809,
+      "step": 7517
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.060667980204104e-06,
+      "loss": 3.1428,
+      "step": 7518
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.058336312823719e-06,
+      "loss": 3.3608,
+      "step": 7519
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0560053735916372e-06,
+      "loss": 2.2821,
+      "step": 7520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0536751627523643e-06,
+      "loss": 2.3292,
+      "step": 7521
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0513456805503305e-06,
+      "loss": 2.6113,
+      "step": 7522
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.049016927229882e-06,
+      "loss": 2.7664,
+      "step": 7523
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0466889030352976e-06,
+      "loss": 3.2505,
+      "step": 7524
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0443616082107753e-06,
+      "loss": 1.9409,
+      "step": 7525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0420350430004388e-06,
+      "loss": 4.0792,
+      "step": 7526
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.039709207648336e-06,
+      "loss": 1.4818,
+      "step": 7527
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.037384102398431e-06,
+      "loss": 3.7633,
+      "step": 7528
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.035059727494619e-06,
+      "loss": 2.1112,
+      "step": 7529
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.032736083180716e-06,
+      "loss": 2.3539,
+      "step": 7530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0304131697004624e-06,
+      "loss": 2.3519,
+      "step": 7531
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0280909872975194e-06,
+      "loss": 1.8532,
+      "step": 7532
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.025769536215475e-06,
+      "loss": 2.6485,
+      "step": 7533
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.023448816697838e-06,
+      "loss": 2.7479,
+      "step": 7534
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0211288289880404e-06,
+      "loss": 2.2649,
+      "step": 7535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0188095733294388e-06,
+      "loss": 2.2541,
+      "step": 7536
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0164910499653155e-06,
+      "loss": 2.7631,
+      "step": 7537
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.014173259138867e-06,
+      "loss": 3.3526,
+      "step": 7538
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.011856201093221e-06,
+      "loss": 2.0925,
+      "step": 7539
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.009539876071427e-06,
+      "loss": 3.1332,
+      "step": 7540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0072242843164555e-06,
+      "loss": 2.4492,
+      "step": 7541
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0049094260712053e-06,
+      "loss": 2.6983,
+      "step": 7542
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.002595301578488e-06,
+      "loss": 3.2417,
+      "step": 7543
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0002819110810475e-06,
+      "loss": 2.6633,
+      "step": 7544
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.997969254821548e-06,
+      "loss": 2.1628,
+      "step": 7545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9956573330425766e-06,
+      "loss": 3.4052,
+      "step": 7546
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.993346145986642e-06,
+      "loss": 1.7422,
+      "step": 7547
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9910356938961782e-06,
+      "loss": 1.9119,
+      "step": 7548
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9887259770135415e-06,
+      "loss": 3.2643,
+      "step": 7549
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9864169955810085e-06,
+      "loss": 1.7191,
+      "step": 7550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9841087498407817e-06,
+      "loss": 2.3338,
+      "step": 7551
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.981801240034985e-06,
+      "loss": 2.3452,
+      "step": 7552
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.979494466405669e-06,
+      "loss": 2.7783,
+      "step": 7553
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.977188429194797e-06,
+      "loss": 1.4364,
+      "step": 7554
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.974883128644266e-06,
+      "loss": 2.8697,
+      "step": 7555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9725785649958895e-06,
+      "loss": 2.6995,
+      "step": 7556
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9702747384914055e-06,
+      "loss": 2.5372,
+      "step": 7557
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9679716493724795e-06,
+      "loss": 2.4739,
+      "step": 7558
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.965669297880688e-06,
+      "loss": 2.0023,
+      "step": 7559
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9633676842575386e-06,
+      "loss": 2.7513,
+      "step": 7560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9610668087444617e-06,
+      "loss": 2.6992,
+      "step": 7561
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.958766671582809e-06,
+      "loss": 3.5884,
+      "step": 7562
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9564672730138542e-06,
+      "loss": 2.8427,
+      "step": 7563
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9541686132787907e-06,
+      "loss": 3.009,
+      "step": 7564
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.951870692618739e-06,
+      "loss": 2.5342,
+      "step": 7565
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.949573511274739e-06,
+      "loss": 2.413,
+      "step": 7566
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9472770694877574e-06,
+      "loss": 1.4994,
+      "step": 7567
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.944981367498677e-06,
+      "loss": 2.8722,
+      "step": 7568
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9426864055483085e-06,
+      "loss": 2.7396,
+      "step": 7569
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.940392183877382e-06,
+      "loss": 1.9129,
+      "step": 7570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9380987027265506e-06,
+      "loss": 2.2499,
+      "step": 7571
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.93580596233639e-06,
+      "loss": 1.8488,
+      "step": 7572
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.933513962947401e-06,
+      "loss": 3.7706,
+      "step": 7573
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9312227047999974e-06,
+      "loss": 1.514,
+      "step": 7574
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 3.4741,
+      "step": 7575
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9266424131912495e-06,
+      "loss": 4.4112,
+      "step": 7576
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9243533802103553e-06,
+      "loss": 3.0964,
+      "step": 7577
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9220650894319557e-06,
+      "loss": 2.6782,
+      "step": 7578
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.919777541096075e-06,
+      "loss": 2.5348,
+      "step": 7579
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9174907354426696e-06,
+      "loss": 2.5296,
+      "step": 7580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9152046727116157e-06,
+      "loss": 2.8818,
+      "step": 7581
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.91291935314271e-06,
+      "loss": 2.5222,
+      "step": 7582
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9106347769756728e-06,
+      "loss": 2.7236,
+      "step": 7583
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9083509444501433e-06,
+      "loss": 1.7185,
+      "step": 7584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9060678558056876e-06,
+      "loss": 3.8873,
+      "step": 7585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9037855112817893e-06,
+      "loss": 2.2338,
+      "step": 7586
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9015039111178566e-06,
+      "loss": 2.9353,
+      "step": 7587
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.899223055553221e-06,
+      "loss": 2.9232,
+      "step": 7588
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.896942944827129e-06,
+      "loss": 2.7033,
+      "step": 7589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8946635791787546e-06,
+      "loss": 2.2784,
+      "step": 7590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.892384958847194e-06,
+      "loss": 2.7604,
+      "step": 7591
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.890107084071465e-06,
+      "loss": 2.1814,
+      "step": 7592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8878299550905064e-06,
+      "loss": 2.6611,
+      "step": 7593
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.885553572143175e-06,
+      "loss": 3.4847,
+      "step": 7594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.883277935468254e-06,
+      "loss": 3.6088,
+      "step": 7595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.881003045304448e-06,
+      "loss": 2.4796,
+      "step": 7596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.878728901890383e-06,
+      "loss": 2.4187,
+      "step": 7597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8764555054646083e-06,
+      "loss": 1.7983,
+      "step": 7598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.874182856265586e-06,
+      "loss": 1.557,
+      "step": 7599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8719109545317102e-06,
+      "loss": 3.7687,
+      "step": 7600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8696398005012927e-06,
+      "loss": 2.462,
+      "step": 7601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.867369394412567e-06,
+      "loss": 1.9007,
+      "step": 7602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.865099736503688e-06,
+      "loss": 2.1873,
+      "step": 7603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8628308270127335e-06,
+      "loss": 2.6048,
+      "step": 7604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8605626661776995e-06,
+      "loss": 2.541,
+      "step": 7605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8582952542365083e-06,
+      "loss": 2.8648,
+      "step": 7606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.856028591426998e-06,
+      "loss": 2.8289,
+      "step": 7607
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.853762677986932e-06,
+      "loss": 2.9137,
+      "step": 7608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8514975141539993e-06,
+      "loss": 2.6874,
+      "step": 7609
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.849233100165795e-06,
+      "loss": 2.7019,
+      "step": 7610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.846969436259852e-06,
+      "loss": 2.414,
+      "step": 7611
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.844706522673616e-06,
+      "loss": 2.2383,
+      "step": 7612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.842444359644457e-06,
+      "loss": 3.2406,
+      "step": 7613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8401829474096696e-06,
+      "loss": 4.4871,
+      "step": 7614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.837922286206457e-06,
+      "loss": 2.6859,
+      "step": 7615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.835662376271957e-06,
+      "loss": 0.9358,
+      "step": 7616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8334032178432234e-06,
+      "loss": 2.9538,
+      "step": 7617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8311448111572304e-06,
+      "loss": 3.4741,
+      "step": 7618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8288871564508745e-06,
+      "loss": 4.3793,
+      "step": 7619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8266302539609747e-06,
+      "loss": 3.2597,
+      "step": 7620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.824374103924269e-06,
+      "loss": 1.9486,
+      "step": 7621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.822118706577416e-06,
+      "loss": 3.7694,
+      "step": 7622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8198640621569973e-06,
+      "loss": 3.3818,
+      "step": 7623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8176101708995174e-06,
+      "loss": 2.4877,
+      "step": 7624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8153570330413925e-06,
+      "loss": 2.5157,
+      "step": 7625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8131046488189705e-06,
+      "loss": 2.6959,
+      "step": 7626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8108530184685143e-06,
+      "loss": 3.1623,
+      "step": 7627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.808602142226212e-06,
+      "loss": 2.1029,
+      "step": 7628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8063520203281714e-06,
+      "loss": 2.3059,
+      "step": 7629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8041026530104144e-06,
+      "loss": 3.9718,
+      "step": 7630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8018540405088914e-06,
+      "loss": 2.1364,
+      "step": 7631
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7996061830594714e-06,
+      "loss": 2.4339,
+      "step": 7632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7973590808979455e-06,
+      "loss": 2.8686,
+      "step": 7633
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.795112734260026e-06,
+      "loss": 3.188,
+      "step": 7634
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7928671433813392e-06,
+      "loss": 2.1035,
+      "step": 7635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7906223084974405e-06,
+      "loss": 1.4667,
+      "step": 7636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7883782298438002e-06,
+      "loss": 3.4465,
+      "step": 7637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.786134907655814e-06,
+      "loss": 1.6441,
+      "step": 7638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.783892342168796e-06,
+      "loss": 3.0435,
+      "step": 7639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.78165053361798e-06,
+      "loss": 1.169,
+      "step": 7640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.779409482238522e-06,
+      "loss": 2.7796,
+      "step": 7641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.777169188265496e-06,
+      "loss": 2.5983,
+      "step": 7642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.774929651933902e-06,
+      "loss": 1.9015,
+      "step": 7643
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.772690873478656e-06,
+      "loss": 2.0622,
+      "step": 7644
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.770452853134593e-06,
+      "loss": 2.3165,
+      "step": 7645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7682155911364718e-06,
+      "loss": 2.8883,
+      "step": 7646
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.765979087718971e-06,
+      "loss": 0.8428,
+      "step": 7647
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7637433431166903e-06,
+      "loss": 2.321,
+      "step": 7648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.761508357564151e-06,
+      "loss": 2.5352,
+      "step": 7649
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.759274131295787e-06,
+      "loss": 1.1232,
+      "step": 7650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.757040664545961e-06,
+      "loss": 1.7848,
+      "step": 7651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.754807957548955e-06,
+      "loss": 2.6693,
+      "step": 7652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7525760105389676e-06,
+      "loss": 3.3072,
+      "step": 7653
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7503448237501208e-06,
+      "loss": 2.326,
+      "step": 7654
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7481143974164548e-06,
+      "loss": 2.543,
+      "step": 7655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.745884731771931e-06,
+      "loss": 3.5795,
+      "step": 7656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.743655827050432e-06,
+      "loss": 2.4757,
+      "step": 7657
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.741427683485759e-06,
+      "loss": 2.6049,
+      "step": 7658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7392003013116362e-06,
+      "loss": 3.1426,
+      "step": 7659
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.736973680761702e-06,
+      "loss": 1.6958,
+      "step": 7660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.734747822069519e-06,
+      "loss": 2.6988,
+      "step": 7661
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7325227254685704e-06,
+      "loss": 3.2417,
+      "step": 7662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7302983911922576e-06,
+      "loss": 3.5072,
+      "step": 7663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.728074819473908e-06,
+      "loss": 2.8203,
+      "step": 7664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7258520105467566e-06,
+      "loss": 2.9286,
+      "step": 7665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.723629964643969e-06,
+      "loss": 2.4208,
+      "step": 7666
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7214086819986272e-06,
+      "loss": 2.3608,
+      "step": 7667
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7191881628437335e-06,
+      "loss": 3.0877,
+      "step": 7668
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7169684074122105e-06,
+      "loss": 3.358,
+      "step": 7669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.714749415936904e-06,
+      "loss": 2.5205,
+      "step": 7670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7125311886505665e-06,
+      "loss": 1.879,
+      "step": 7671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7103137257858867e-06,
+      "loss": 2.5905,
+      "step": 7672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7080970275754647e-06,
+      "loss": 2.2747,
+      "step": 7673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.705881094251821e-06,
+      "loss": 1.8184,
+      "step": 7674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7036659260473973e-06,
+      "loss": 1.548,
+      "step": 7675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7014515231945557e-06,
+      "loss": 2.2294,
+      "step": 7676
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6992378859255754e-06,
+      "loss": 3.5802,
+      "step": 7677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6970250144726563e-06,
+      "loss": 2.9092,
+      "step": 7678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.69481290906792e-06,
+      "loss": 3.1349,
+      "step": 7679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.692601569943407e-06,
+      "loss": 1.5793,
+      "step": 7680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6903909973310725e-06,
+      "loss": 3.4887,
+      "step": 7681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6881811914627976e-06,
+      "loss": 3.5076,
+      "step": 7682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6859721525703797e-06,
+      "loss": 2.6006,
+      "step": 7683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.683763880885538e-06,
+      "loss": 1.6688,
+      "step": 7684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6815563766399122e-06,
+      "loss": 2.3412,
+      "step": 7685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.679349640065053e-06,
+      "loss": 2.451,
+      "step": 7686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6771436713924417e-06,
+      "loss": 3.3525,
+      "step": 7687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.674938470853472e-06,
+      "loss": 2.4066,
+      "step": 7688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.672734038679461e-06,
+      "loss": 2.3395,
+      "step": 7689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.670530375101641e-06,
+      "loss": 1.5057,
+      "step": 7690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.668327480351168e-06,
+      "loss": 3.3604,
+      "step": 7691
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6661253546591158e-06,
+      "loss": 4.2666,
+      "step": 7692
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6639239982564757e-06,
+      "loss": 2.3512,
+      "step": 7693
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.661723411374161e-06,
+      "loss": 3.4349,
+      "step": 7694
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6595235942430044e-06,
+      "loss": 1.6605,
+      "step": 7695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6573245470937527e-06,
+      "loss": 2.0178,
+      "step": 7696
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6551262701570768e-06,
+      "loss": 2.2797,
+      "step": 7697
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.652928763663567e-06,
+      "loss": 3.1295,
+      "step": 7698
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6507320278437296e-06,
+      "loss": 1.4842,
+      "step": 7699
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.648536062927999e-06,
+      "loss": 3.594,
+      "step": 7700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6463408691467107e-06,
+      "loss": 2.804,
+      "step": 7701
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.644146446730138e-06,
+      "loss": 1.9143,
+      "step": 7702
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.641952795908462e-06,
+      "loss": 2.3704,
+      "step": 7703
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.639759916911788e-06,
+      "loss": 2.0466,
+      "step": 7704
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.637567809970143e-06,
+      "loss": 3.5547,
+      "step": 7705
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6353764753134603e-06,
+      "loss": 3.739,
+      "step": 7706
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6331859131716064e-06,
+      "loss": 2.9602,
+      "step": 7707
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6309961237743587e-06,
+      "loss": 1.6128,
+      "step": 7708
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6288071073514176e-06,
+      "loss": 3.7029,
+      "step": 7709
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6266188641324e-06,
+      "loss": 2.5756,
+      "step": 7710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6244313943468426e-06,
+      "loss": 2.1983,
+      "step": 7711
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6222446982242e-06,
+      "loss": 3.9411,
+      "step": 7712
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.620058775993849e-06,
+      "loss": 2.5862,
+      "step": 7713
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6178736278850792e-06,
+      "loss": 3.206,
+      "step": 7714
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6156892541271083e-06,
+      "loss": 2.8102,
+      "step": 7715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.61350565494906e-06,
+      "loss": 1.7096,
+      "step": 7716
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6113228305799867e-06,
+      "loss": 2.5124,
+      "step": 7717
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6091407812488567e-06,
+      "loss": 4.5388,
+      "step": 7718
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6069595071845566e-06,
+      "loss": 2.0434,
+      "step": 7719
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.604779008615895e-06,
+      "loss": 1.3203,
+      "step": 7720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.60259928577159e-06,
+      "loss": 2.8834,
+      "step": 7721
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6004203388802883e-06,
+      "loss": 2.6195,
+      "step": 7722
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5982421681705493e-06,
+      "loss": 2.5046,
+      "step": 7723
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5960647738708553e-06,
+      "loss": 3.7888,
+      "step": 7724
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.593888156209603e-06,
+      "loss": 1.2731,
+      "step": 7725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.59171231541511e-06,
+      "loss": 2.3892,
+      "step": 7726
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5895372517156137e-06,
+      "loss": 3.2965,
+      "step": 7727
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5873629653392653e-06,
+      "loss": 2.0712,
+      "step": 7728
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.585189456514139e-06,
+      "loss": 2.9039,
+      "step": 7729
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.583016725468226e-06,
+      "loss": 2.6907,
+      "step": 7730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.580844772429438e-06,
+      "loss": 3.516,
+      "step": 7731
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.578673597625597e-06,
+      "loss": 2.9588,
+      "step": 7732
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5765032012844526e-06,
+      "loss": 1.9205,
+      "step": 7733
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.574333583633668e-06,
+      "loss": 1.8736,
+      "step": 7734
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.572164744900827e-06,
+      "loss": 2.4706,
+      "step": 7735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.569996685313434e-06,
+      "loss": 4.1914,
+      "step": 7736
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5678294050989017e-06,
+      "loss": 2.9137,
+      "step": 7737
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5656629044845714e-06,
+      "loss": 2.5273,
+      "step": 7738
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5634971836976987e-06,
+      "loss": 3.0941,
+      "step": 7739
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5613322429654573e-06,
+      "loss": 1.8529,
+      "step": 7740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5591680825149423e-06,
+      "loss": 1.9509,
+      "step": 7741
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5570047025731594e-06,
+      "loss": 2.984,
+      "step": 7742
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.554842103367038e-06,
+      "loss": 3.4098,
+      "step": 7743
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5526802851234268e-06,
+      "loss": 3.6792,
+      "step": 7744
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5505192480690865e-06,
+      "loss": 3.336,
+      "step": 7745
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5483589924307086e-06,
+      "loss": 3.1585,
+      "step": 7746
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5461995184348854e-06,
+      "loss": 3.8983,
+      "step": 7747
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5440408263081385e-06,
+      "loss": 2.685,
+      "step": 7748
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5418829162769053e-06,
+      "loss": 2.5041,
+      "step": 7749
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5397257885675396e-06,
+      "loss": 2.6539,
+      "step": 7750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.537569443406317e-06,
+      "loss": 3.2564,
+      "step": 7751
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.535413881019423e-06,
+      "loss": 3.261,
+      "step": 7752
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.533259101632969e-06,
+      "loss": 2.7872,
+      "step": 7753
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5311051054729806e-06,
+      "loss": 2.9045,
+      "step": 7754
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.528951892765402e-06,
+      "loss": 1.5313,
+      "step": 7755
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.526799463736099e-06,
+      "loss": 2.7739,
+      "step": 7756
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.524647818610846e-06,
+      "loss": 3.7506,
+      "step": 7757
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5224969576153413e-06,
+      "loss": 2.8244,
+      "step": 7758
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.520346880975203e-06,
+      "loss": 2.8559,
+      "step": 7759
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5181975889159615e-06,
+      "loss": 2.9405,
+      "step": 7760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.516049081663069e-06,
+      "loss": 2.332,
+      "step": 7761
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.513901359441894e-06,
+      "loss": 4.4483,
+      "step": 7762
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5117544224777222e-06,
+      "loss": 4.0416,
+      "step": 7763
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.509608270995758e-06,
+      "loss": 1.997,
+      "step": 7764
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.507462905221122e-06,
+      "loss": 1.7935,
+      "step": 7765
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.505318325378856e-06,
+      "loss": 2.7986,
+      "step": 7766
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5031745316939106e-06,
+      "loss": 2.6123,
+      "step": 7767
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.501031524391163e-06,
+      "loss": 3.9475,
+      "step": 7768
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4988893036954045e-06,
+      "loss": 2.8903,
+      "step": 7769
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.496747869831345e-06,
+      "loss": 3.514,
+      "step": 7770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.494607223023613e-06,
+      "loss": 2.6154,
+      "step": 7771
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.492467363496747e-06,
+      "loss": 1.9033,
+      "step": 7772
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4903282914752104e-06,
+      "loss": 3.2853,
+      "step": 7773
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.488190007183383e-06,
+      "loss": 2.075,
+      "step": 7774
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.48605251084556e-06,
+      "loss": 3.1573,
+      "step": 7775
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.483915802685959e-06,
+      "loss": 3.3801,
+      "step": 7776
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.481779882928703e-06,
+      "loss": 1.4422,
+      "step": 7777
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.479644751797845e-06,
+      "loss": 2.1737,
+      "step": 7778
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4775104095173495e-06,
+      "loss": 3.5609,
+      "step": 7779
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.475376856311097e-06,
+      "loss": 3.0045,
+      "step": 7780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4732440924028935e-06,
+      "loss": 1.3638,
+      "step": 7781
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.47111211801645e-06,
+      "loss": 2.688,
+      "step": 7782
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4689809333754022e-06,
+      "loss": 2.3763,
+      "step": 7783
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4668505387033025e-06,
+      "loss": 1.6369,
+      "step": 7784
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.464720934223619e-06,
+      "loss": 2.8228,
+      "step": 7785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.462592120159739e-06,
+      "loss": 3.2934,
+      "step": 7786
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4604640967349614e-06,
+      "loss": 2.2644,
+      "step": 7787
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.458336864172508e-06,
+      "loss": 2.5616,
+      "step": 7788
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.456210422695515e-06,
+      "loss": 3.4731,
+      "step": 7789
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4540847725270376e-06,
+      "loss": 1.7814,
+      "step": 7790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4519599138900463e-06,
+      "loss": 2.3965,
+      "step": 7791
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.44983584700743e-06,
+      "loss": 2.7239,
+      "step": 7792
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.44771257210199e-06,
+      "loss": 1.9139,
+      "step": 7793
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.445590089396449e-06,
+      "loss": 2.9698,
+      "step": 7794
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4434683991134476e-06,
+      "loss": 2.6647,
+      "step": 7795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4413475014755396e-06,
+      "loss": 2.6177,
+      "step": 7796
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4392273967051983e-06,
+      "loss": 3.105,
+      "step": 7797
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.437108085024812e-06,
+      "loss": 2.4655,
+      "step": 7798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.434989566656687e-06,
+      "loss": 2.3864,
+      "step": 7799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.432871841823047e-06,
+      "loss": 3.2232,
+      "step": 7800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4307549107460304e-06,
+      "loss": 2.2149,
+      "step": 7801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.428638773647696e-06,
+      "loss": 1.6462,
+      "step": 7802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4265234307500117e-06,
+      "loss": 3.1183,
+      "step": 7803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.42440888227487e-06,
+      "loss": 1.6029,
+      "step": 7804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4222951284440776e-06,
+      "loss": 2.3001,
+      "step": 7805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.420182169479357e-06,
+      "loss": 1.7587,
+      "step": 7806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.41807000560235e-06,
+      "loss": 3.0571,
+      "step": 7807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.415958637034609e-06,
+      "loss": 3.5057,
+      "step": 7808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.413848063997607e-06,
+      "loss": 3.5759,
+      "step": 7809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.411738286712735e-06,
+      "loss": 2.3722,
+      "step": 7810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.409629305401299e-06,
+      "loss": 1.9805,
+      "step": 7811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.407521120284523e-06,
+      "loss": 3.3824,
+      "step": 7812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.405413731583541e-06,
+      "loss": 3.5914,
+      "step": 7813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.403307139519411e-06,
+      "loss": 4.1177,
+      "step": 7814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.401201344313102e-06,
+      "loss": 3.4789,
+      "step": 7815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3990963461855075e-06,
+      "loss": 4.0047,
+      "step": 7816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.396992145357433e-06,
+      "loss": 2.5487,
+      "step": 7817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3948887420495927e-06,
+      "loss": 2.3188,
+      "step": 7818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3927861364826265e-06,
+      "loss": 3.021,
+      "step": 7819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.390684328877089e-06,
+      "loss": 3.0295,
+      "step": 7820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.388583319453449e-06,
+      "loss": 1.6586,
+      "step": 7821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.386483108432096e-06,
+      "loss": 2.962,
+      "step": 7822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3843836960333267e-06,
+      "loss": 3.0446,
+      "step": 7823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3822850824773623e-06,
+      "loss": 3.1243,
+      "step": 7824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3801872679843384e-06,
+      "loss": 2.7499,
+      "step": 7825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3780902527743054e-06,
+      "loss": 2.4453,
+      "step": 7826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.375994037067233e-06,
+      "loss": 2.9265,
+      "step": 7827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3738986210829997e-06,
+      "loss": 2.903,
+      "step": 7828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.371804005041406e-06,
+      "loss": 3.104,
+      "step": 7829
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.36971018916217e-06,
+      "loss": 1.6791,
+      "step": 7830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3676171736649212e-06,
+      "loss": 1.6653,
+      "step": 7831
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3655249587692073e-06,
+      "loss": 2.2527,
+      "step": 7832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3634335446944936e-06,
+      "loss": 2.6482,
+      "step": 7833
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3613429316601577e-06,
+      "loss": 2.3905,
+      "step": 7834
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3592531198854974e-06,
+      "loss": 2.8639,
+      "step": 7835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3571641095897223e-06,
+      "loss": 2.088,
+      "step": 7836
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.355075900991963e-06,
+      "loss": 2.2746,
+      "step": 7837
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.352988494311259e-06,
+      "loss": 2.2063,
+      "step": 7838
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.35090188976657e-06,
+      "loss": 2.6722,
+      "step": 7839
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3488160875767717e-06,
+      "loss": 2.3302,
+      "step": 7840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3467310879606566e-06,
+      "loss": 1.7517,
+      "step": 7841
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3446468911369334e-06,
+      "loss": 3.7505,
+      "step": 7842
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3425634973242182e-06,
+      "loss": 2.2173,
+      "step": 7843
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.340480906741053e-06,
+      "loss": 3.5378,
+      "step": 7844
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3383991196058918e-06,
+      "loss": 2.7324,
+      "step": 7845
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3363181361371047e-06,
+      "loss": 2.6946,
+      "step": 7846
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.334237956552979e-06,
+      "loss": 2.9625,
+      "step": 7847
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.332158581071712e-06,
+      "loss": 2.4642,
+      "step": 7848
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3300800099114186e-06,
+      "loss": 3.008,
+      "step": 7849
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.328002243290138e-06,
+      "loss": 2.5292,
+      "step": 7850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.325925281425816e-06,
+      "loss": 3.0513,
+      "step": 7851
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.323849124536315e-06,
+      "loss": 5.0964,
+      "step": 7852
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.321773772839416e-06,
+      "loss": 2.0774,
+      "step": 7853
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3196992265528118e-06,
+      "loss": 2.0744,
+      "step": 7854
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.317625485894113e-06,
+      "loss": 2.3936,
+      "step": 7855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3155525510808453e-06,
+      "loss": 2.6074,
+      "step": 7856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.31348042233045e-06,
+      "loss": 2.6246,
+      "step": 7857
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.311409099860288e-06,
+      "loss": 3.4722,
+      "step": 7858
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3093385838876236e-06,
+      "loss": 2.9702,
+      "step": 7859
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.307268874629649e-06,
+      "loss": 2.8378,
+      "step": 7860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3051999723034655e-06,
+      "loss": 2.5304,
+      "step": 7861
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3031318771260935e-06,
+      "loss": 1.9172,
+      "step": 7862
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3010645893144666e-06,
+      "loss": 1.5704,
+      "step": 7863
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2989981090854306e-06,
+      "loss": 3.6676,
+      "step": 7864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.296932436655752e-06,
+      "loss": 2.5223,
+      "step": 7865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2948675722421086e-06,
+      "loss": 2.566,
+      "step": 7866
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2928035160610973e-06,
+      "loss": 1.679,
+      "step": 7867
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2907402683292268e-06,
+      "loss": 2.4856,
+      "step": 7868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2886778292629217e-06,
+      "loss": 2.5275,
+      "step": 7869
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2866161990785228e-06,
+      "loss": 1.9356,
+      "step": 7870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2845553779922867e-06,
+      "loss": 1.5861,
+      "step": 7871
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2824953662203832e-06,
+      "loss": 1.9916,
+      "step": 7872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2804361639788997e-06,
+      "loss": 2.2185,
+      "step": 7873
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2783777714838316e-06,
+      "loss": 1.9755,
+      "step": 7874
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2763201889510987e-06,
+      "loss": 3.6163,
+      "step": 7875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2742634165965317e-06,
+      "loss": 2.5079,
+      "step": 7876
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2722074546358763e-06,
+      "loss": 1.9128,
+      "step": 7877
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.270152303284795e-06,
+      "loss": 2.6876,
+      "step": 7878
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2680979627588593e-06,
+      "loss": 3.7039,
+      "step": 7879
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.266044433273562e-06,
+      "loss": 3.1464,
+      "step": 7880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2639917150443104e-06,
+      "loss": 2.3425,
+      "step": 7881
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2619398082864232e-06,
+      "loss": 2.7727,
+      "step": 7882
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2598887132151394e-06,
+      "loss": 2.6648,
+      "step": 7883
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2578384300456014e-06,
+      "loss": 1.9446,
+      "step": 7884
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2557889589928815e-06,
+      "loss": 3.2514,
+      "step": 7885
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.253740300271959e-06,
+      "loss": 3.4962,
+      "step": 7886
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2516924540977268e-06,
+      "loss": 3.6092,
+      "step": 7887
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.249645420684998e-06,
+      "loss": 3.1478,
+      "step": 7888
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2475992002484915e-06,
+      "loss": 3.6164,
+      "step": 7889
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.245553793002849e-06,
+      "loss": 1.8987,
+      "step": 7890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2435091991626233e-06,
+      "loss": 3.251,
+      "step": 7891
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2414654189422845e-06,
+      "loss": 3.4688,
+      "step": 7892
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.239422452556218e-06,
+      "loss": 2.3703,
+      "step": 7893
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2373803002187146e-06,
+      "loss": 1.5767,
+      "step": 7894
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.23533896214399e-06,
+      "loss": 2.3005,
+      "step": 7895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.233298438546172e-06,
+      "loss": 3.6629,
+      "step": 7896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2312587296393005e-06,
+      "loss": 2.5387,
+      "step": 7897
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2292198356373362e-06,
+      "loss": 2.3622,
+      "step": 7898
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.227181756754143e-06,
+      "loss": 2.656,
+      "step": 7899
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2251444932035094e-06,
+      "loss": 2.1555,
+      "step": 7900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2231080451991326e-06,
+      "loss": 2.3148,
+      "step": 7901
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.221072412954629e-06,
+      "loss": 2.7516,
+      "step": 7902
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2190375966835266e-06,
+      "loss": 2.247,
+      "step": 7903
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2170035965992674e-06,
+      "loss": 2.7016,
+      "step": 7904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2149704129152083e-06,
+      "loss": 3.026,
+      "step": 7905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2129380458446226e-06,
+      "loss": 2.2774,
+      "step": 7906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2109064956006944e-06,
+      "loss": 2.3171,
+      "step": 7907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2088757623965263e-06,
+      "loss": 2.1545,
+      "step": 7908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2068458464451292e-06,
+      "loss": 2.6765,
+      "step": 7909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.204816747959434e-06,
+      "loss": 2.0994,
+      "step": 7910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2027884671522835e-06,
+      "loss": 2.0886,
+      "step": 7911
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2007610042364337e-06,
+      "loss": 3.3744,
+      "step": 7912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.198734359424558e-06,
+      "loss": 3.6893,
+      "step": 7913
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1967085329292435e-06,
+      "loss": 2.0001,
+      "step": 7914
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.194683524962986e-06,
+      "loss": 3.7011,
+      "step": 7915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1926593357382e-06,
+      "loss": 1.9364,
+      "step": 7916
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.190635965467215e-06,
+      "loss": 3.2722,
+      "step": 7917
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.188613414362273e-06,
+      "loss": 3.5099,
+      "step": 7918
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1865916826355293e-06,
+      "loss": 1.8498,
+      "step": 7919
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.184570770499056e-06,
+      "loss": 2.8234,
+      "step": 7920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1825506781648365e-06,
+      "loss": 2.5567,
+      "step": 7921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.180531405844768e-06,
+      "loss": 2.981,
+      "step": 7922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1785129537506646e-06,
+      "loss": 2.8697,
+      "step": 7923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.176495322094254e-06,
+      "loss": 3.6514,
+      "step": 7924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1744785110871713e-06,
+      "loss": 1.5031,
+      "step": 7925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1724625209409735e-06,
+      "loss": 2.9233,
+      "step": 7926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1704473518671287e-06,
+      "loss": 3.2906,
+      "step": 7927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1684330040770183e-06,
+      "loss": 3.0441,
+      "step": 7928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1664194777819414e-06,
+      "loss": 3.2402,
+      "step": 7929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1644067731931005e-06,
+      "loss": 1.8672,
+      "step": 7930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.162394890521625e-06,
+      "loss": 2.0345,
+      "step": 7931
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1603838299785486e-06,
+      "loss": 0.9274,
+      "step": 7932
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1583735917748238e-06,
+      "loss": 2.9464,
+      "step": 7933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1563641761213183e-06,
+      "loss": 3.3346,
+      "step": 7934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1543555832288056e-06,
+      "loss": 3.0149,
+      "step": 7935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1523478133079776e-06,
+      "loss": 2.9399,
+      "step": 7936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1503408665694427e-06,
+      "loss": 2.6575,
+      "step": 7937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.148334743223719e-06,
+      "loss": 2.7616,
+      "step": 7938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1463294434812408e-06,
+      "loss": 2.7258,
+      "step": 7939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1443249675523536e-06,
+      "loss": 2.3677,
+      "step": 7940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1423213156473175e-06,
+      "loss": 1.5039,
+      "step": 7941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1403184879763083e-06,
+      "loss": 3.2374,
+      "step": 7942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1383164847494107e-06,
+      "loss": 3.3305,
+      "step": 7943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1363153061766297e-06,
+      "loss": 2.094,
+      "step": 7944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.134314952467873e-06,
+      "loss": 3.0972,
+      "step": 7945
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1323154238329722e-06,
+      "loss": 2.6546,
+      "step": 7946
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1303167204816698e-06,
+      "loss": 2.8807,
+      "step": 7947
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.128318842623618e-06,
+      "loss": 2.5061,
+      "step": 7948
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1263217904683884e-06,
+      "loss": 3.0447,
+      "step": 7949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.124325564225458e-06,
+      "loss": 2.4812,
+      "step": 7950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1223301641042237e-06,
+      "loss": 3.2425,
+      "step": 7951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1203355903139934e-06,
+      "loss": 1.9302,
+      "step": 7952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1183418430639903e-06,
+      "loss": 3.3174,
+      "step": 7953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.116348922563346e-06,
+      "loss": 3.0722,
+      "step": 7954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1143568290211115e-06,
+      "loss": 3.7936,
+      "step": 7955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.112365562646248e-06,
+      "loss": 2.4824,
+      "step": 7956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.110375123647629e-06,
+      "loss": 1.8538,
+      "step": 7957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.108385512234041e-06,
+      "loss": 3.4551,
+      "step": 7958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.106396728614192e-06,
+      "loss": 2.9923,
+      "step": 7959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1044087729966856e-06,
+      "loss": 1.3737,
+      "step": 7960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.102421645590056e-06,
+      "loss": 3.1114,
+      "step": 7961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.100435346602742e-06,
+      "loss": 2.9701,
+      "step": 7962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.098449876243096e-06,
+      "loss": 2.5815,
+      "step": 7963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0964652347193894e-06,
+      "loss": 2.822,
+      "step": 7964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0944814222397948e-06,
+      "loss": 3.5982,
+      "step": 7965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0924984390124093e-06,
+      "loss": 2.3533,
+      "step": 7966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0905162852452366e-06,
+      "loss": 3.6948,
+      "step": 7967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.088534961146197e-06,
+      "loss": 4.0704,
+      "step": 7968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.086554466923125e-06,
+      "loss": 1.8209,
+      "step": 7969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0845748027837585e-06,
+      "loss": 2.777,
+      "step": 7970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0825959689357588e-06,
+      "loss": 2.0276,
+      "step": 7971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0806179655866964e-06,
+      "loss": 2.9548,
+      "step": 7972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0786407929440543e-06,
+      "loss": 2.3473,
+      "step": 7973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.076664451215229e-06,
+      "loss": 2.2485,
+      "step": 7974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.074688940607529e-06,
+      "loss": 2.1324,
+      "step": 7975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.072714261328177e-06,
+      "loss": 2.8179,
+      "step": 7976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.070740413584307e-06,
+      "loss": 2.1903,
+      "step": 7977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.068767397582967e-06,
+      "loss": 1.7974,
+      "step": 7978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0667952135311178e-06,
+      "loss": 3.6504,
+      "step": 7979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.064823861635633e-06,
+      "loss": 2.7774,
+      "step": 7980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0628533421032937e-06,
+      "loss": 2.3681,
+      "step": 7981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.060883655140803e-06,
+      "loss": 2.7877,
+      "step": 7982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.058914800954769e-06,
+      "loss": 2.3947,
+      "step": 7983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0569467797517173e-06,
+      "loss": 2.6665,
+      "step": 7984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0549795917380867e-06,
+      "loss": 2.6413,
+      "step": 7985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0530132371202195e-06,
+      "loss": 2.7047,
+      "step": 7986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0510477161043805e-06,
+      "loss": 2.3563,
+      "step": 7987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0490830288967443e-06,
+      "loss": 2.1548,
+      "step": 7988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047119175703397e-06,
+      "loss": 1.9674,
+      "step": 7989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0451561567303378e-06,
+      "loss": 3.0108,
+      "step": 7990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0431939721834783e-06,
+      "loss": 2.2928,
+      "step": 7991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.041232622268642e-06,
+      "loss": 3.1196,
+      "step": 7992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0392721071915656e-06,
+      "loss": 2.9694,
+      "step": 7993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.037312427157898e-06,
+      "loss": 2.4404,
+      "step": 7994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0353535823732053e-06,
+      "loss": 3.0221,
+      "step": 7995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.033395573042952e-06,
+      "loss": 2.8418,
+      "step": 7996
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.031438399372531e-06,
+      "loss": 2.574,
+      "step": 7997
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.029482061567237e-06,
+      "loss": 2.5065,
+      "step": 7998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0275265598322822e-06,
+      "loss": 2.9461,
+      "step": 7999
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.025571894372794e-06,
+      "loss": 2.5155,
+      "step": 8000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0236180653937998e-06,
+      "loss": 2.8793,
+      "step": 8001
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0216650731002517e-06,
+      "loss": 3.9854,
+      "step": 8002
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0197129176970087e-06,
+      "loss": 2.8678,
+      "step": 8003
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.017761599388842e-06,
+      "loss": 3.0071,
+      "step": 8004
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0158111183804407e-06,
+      "loss": 3.2814,
+      "step": 8005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.013861474876395e-06,
+      "loss": 3.567,
+      "step": 8006
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.011912669081215e-06,
+      "loss": 1.4808,
+      "step": 8007
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0099647011993217e-06,
+      "loss": 2.591,
+      "step": 8008
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0080175714350492e-06,
+      "loss": 2.3865,
+      "step": 8009
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0060712799926407e-06,
+      "loss": 2.4896,
+      "step": 8010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.004125827076255e-06,
+      "loss": 2.7875,
+      "step": 8011
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.00218121288996e-06,
+      "loss": 2.8449,
+      "step": 8012
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0002374376377354e-06,
+      "loss": 3.8823,
+      "step": 8013
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.998294501523477e-06,
+      "loss": 2.541,
+      "step": 8014
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9963524047509898e-06,
+      "loss": 3.1639,
+      "step": 8015
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.994411147523987e-06,
+      "loss": 2.0929,
+      "step": 8016
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9924707300461e-06,
+      "loss": 2.1834,
+      "step": 8017
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.990531152520869e-06,
+      "loss": 2.7233,
+      "step": 8018
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9885924151517477e-06,
+      "loss": 3.0939,
+      "step": 8019
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9866545181421016e-06,
+      "loss": 1.4101,
+      "step": 8020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9847174616952035e-06,
+      "loss": 1.8496,
+      "step": 8021
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9827812460142435e-06,
+      "loss": 3.1259,
+      "step": 8022
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9808458713023215e-06,
+      "loss": 1.0962,
+      "step": 8023
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.97891133776245e-06,
+      "loss": 2.8307,
+      "step": 8024
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.976977645597552e-06,
+      "loss": 2.3913,
+      "step": 8025
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9750447950104635e-06,
+      "loss": 3.9433,
+      "step": 8026
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.97311278620393e-06,
+      "loss": 1.4356,
+      "step": 8027
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.971181619380611e-06,
+      "loss": 2.2452,
+      "step": 8028
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9692512947430788e-06,
+      "loss": 0.8737,
+      "step": 8029
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.967321812493813e-06,
+      "loss": 2.4614,
+      "step": 8030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9653931728352106e-06,
+      "loss": 1.9363,
+      "step": 8031
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.963465375969572e-06,
+      "loss": 2.2076,
+      "step": 8032
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9615384220991173e-06,
+      "loss": 2.7848,
+      "step": 8033
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9596123114259747e-06,
+      "loss": 4.1038,
+      "step": 8034
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9576870441521834e-06,
+      "loss": 1.6025,
+      "step": 8035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.955762620479699e-06,
+      "loss": 3.5914,
+      "step": 8036
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.953839040610377e-06,
+      "loss": 2.9298,
+      "step": 8037
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9519163047459978e-06,
+      "loss": 3.1513,
+      "step": 8038
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9499944130882455e-06,
+      "loss": 2.8747,
+      "step": 8039
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9480733658387175e-06,
+      "loss": 2.7298,
+      "step": 8040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9461531631989262e-06,
+      "loss": 3.3559,
+      "step": 8041
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.944233805370287e-06,
+      "loss": 2.3805,
+      "step": 8042
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.942315292554133e-06,
+      "loss": 1.9138,
+      "step": 8043
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.940397624951709e-06,
+      "loss": 3.8423,
+      "step": 8044
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9384808027641666e-06,
+      "loss": 2.8168,
+      "step": 8045
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.936564826192574e-06,
+      "loss": 2.6204,
+      "step": 8046
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.934649695437908e-06,
+      "loss": 3.1497,
+      "step": 8047
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9327354107010566e-06,
+      "loss": 2.2905,
+      "step": 8048
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9308219721828192e-06,
+      "loss": 2.379,
+      "step": 8049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9289093800839067e-06,
+      "loss": 2.5656,
+      "step": 8050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9269976346049433e-06,
+      "loss": 3.0755,
+      "step": 8051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9250867359464575e-06,
+      "loss": 2.6542,
+      "step": 8052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9231766843088974e-06,
+      "loss": 3.3208,
+      "step": 8053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9212674798926156e-06,
+      "loss": 2.8148,
+      "step": 8054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9193591228978815e-06,
+      "loss": 3.7644,
+      "step": 8055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9174516135248745e-06,
+      "loss": 4.3765,
+      "step": 8056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.915544951973679e-06,
+      "loss": 1.3679,
+      "step": 8057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9136391384442964e-06,
+      "loss": 2.6915,
+      "step": 8058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911734173136638e-06,
+      "loss": 2.5292,
+      "step": 8059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 2.868,
+      "step": 8060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9079267879856946e-06,
+      "loss": 1.5872,
+      "step": 8061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9060243685417857e-06,
+      "loss": 2.3298,
+      "step": 8062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9041227981183553e-06,
+      "loss": 2.268,
+      "step": 8063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.902222076914869e-06,
+      "loss": 2.8533,
+      "step": 8064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9003222051307046e-06,
+      "loss": 1.8141,
+      "step": 8065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8984231829651512e-06,
+      "loss": 2.3222,
+      "step": 8066
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8965250106174026e-06,
+      "loss": 3.3727,
+      "step": 8067
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.894627688286571e-06,
+      "loss": 2.9565,
+      "step": 8068
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8927312161716771e-06,
+      "loss": 2.2387,
+      "step": 8069
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8908355944716516e-06,
+      "loss": 1.2954,
+      "step": 8070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.888940823385339e-06,
+      "loss": 2.0842,
+      "step": 8071
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8870469031114868e-06,
+      "loss": 2.8272,
+      "step": 8072
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8851538338487608e-06,
+      "loss": 2.7717,
+      "step": 8073
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8832616157957352e-06,
+      "loss": 3.3511,
+      "step": 8074
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8813702491508956e-06,
+      "loss": 2.6865,
+      "step": 8075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8794797341126403e-06,
+      "loss": 3.3263,
+      "step": 8076
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8775900708792693e-06,
+      "loss": 2.8555,
+      "step": 8077
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.875701259649002e-06,
+      "loss": 3.1804,
+      "step": 8078
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8738133006199677e-06,
+      "loss": 1.5323,
+      "step": 8079
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8719261939902023e-06,
+      "loss": 3.6793,
+      "step": 8080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8700399399576563e-06,
+      "loss": 3.4576,
+      "step": 8081
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8681545387201883e-06,
+      "loss": 2.5803,
+      "step": 8082
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8662699904755677e-06,
+      "loss": 2.3814,
+      "step": 8083
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8643862954214754e-06,
+      "loss": 2.8185,
+      "step": 8084
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.862503453755502e-06,
+      "loss": 2.4519,
+      "step": 8085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.860621465675152e-06,
+      "loss": 2.3563,
+      "step": 8086
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.858740331377832e-06,
+      "loss": 2.451,
+      "step": 8087
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8568600510608659e-06,
+      "loss": 2.1302,
+      "step": 8088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8549806249214863e-06,
+      "loss": 2.985,
+      "step": 8089
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8531020531568377e-06,
+      "loss": 2.9366,
+      "step": 8090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8512243359639727e-06,
+      "loss": 1.6638,
+      "step": 8091
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8493474735398575e-06,
+      "loss": 3.068,
+      "step": 8092
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8474714660813619e-06,
+      "loss": 2.2549,
+      "step": 8093
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8455963137852728e-06,
+      "loss": 1.1678,
+      "step": 8094
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8437220168482839e-06,
+      "loss": 2.9642,
+      "step": 8095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8418485754670013e-06,
+      "loss": 3.1204,
+      "step": 8096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8399759898379398e-06,
+      "loss": 3.049,
+      "step": 8097
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.838104260157525e-06,
+      "loss": 2.6238,
+      "step": 8098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.836233386622094e-06,
+      "loss": 2.0709,
+      "step": 8099
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8343633694278895e-06,
+      "loss": 2.5962,
+      "step": 8100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8324942087710706e-06,
+      "loss": 1.9852,
+      "step": 8101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.830625904847706e-06,
+      "loss": 2.0575,
+      "step": 8102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8287584578537653e-06,
+      "loss": 2.7576,
+      "step": 8103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8268918679851388e-06,
+      "loss": 2.7483,
+      "step": 8104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.825026135437622e-06,
+      "loss": 3.1348,
+      "step": 8105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8231612604069238e-06,
+      "loss": 1.517,
+      "step": 8106
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8212972430886612e-06,
+      "loss": 1.4288,
+      "step": 8107
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8194340836783565e-06,
+      "loss": 1.6851,
+      "step": 8108
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.817571782371449e-06,
+      "loss": 2.8612,
+      "step": 8109
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8157103393632869e-06,
+      "loss": 2.3718,
+      "step": 8110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.813849754849125e-06,
+      "loss": 1.6793,
+      "step": 8111
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8119900290241331e-06,
+      "loss": 1.9479,
+      "step": 8112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8101311620833828e-06,
+      "loss": 3.175,
+      "step": 8113
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8082731542218635e-06,
+      "loss": 1.9729,
+      "step": 8114
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8064160056344714e-06,
+      "loss": 2.0505,
+      "step": 8115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8045597165160134e-06,
+      "loss": 3.2589,
+      "step": 8116
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8027042870612033e-06,
+      "loss": 1.3505,
+      "step": 8117
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.80084971746467e-06,
+      "loss": 3.1937,
+      "step": 8118
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7989960079209479e-06,
+      "loss": 2.37,
+      "step": 8119
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7971431586244814e-06,
+      "loss": 2.8258,
+      "step": 8120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7952911697696273e-06,
+      "loss": 1.9632,
+      "step": 8121
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7934400415506537e-06,
+      "loss": 2.4451,
+      "step": 8122
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7915897741617283e-06,
+      "loss": 3.3607,
+      "step": 8123
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7897403677969405e-06,
+      "loss": 3.622,
+      "step": 8124
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7878918226502816e-06,
+      "loss": 2.7617,
+      "step": 8125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.786044138915658e-06,
+      "loss": 3.32,
+      "step": 8126
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7841973167868843e-06,
+      "loss": 3.2417,
+      "step": 8127
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7823513564576788e-06,
+      "loss": 1.7952,
+      "step": 8128
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.780506258121677e-06,
+      "loss": 1.9259,
+      "step": 8129
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7786620219724205e-06,
+      "loss": 2.9438,
+      "step": 8130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7768186482033611e-06,
+      "loss": 1.8692,
+      "step": 8131
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.774976137007861e-06,
+      "loss": 2.4191,
+      "step": 8132
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7731344885791911e-06,
+      "loss": 1.9182,
+      "step": 8133
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7712937031105304e-06,
+      "loss": 2.8333,
+      "step": 8134
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7694537807949707e-06,
+      "loss": 2.4433,
+      "step": 8135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7676147218255092e-06,
+      "loss": 2.508,
+      "step": 8136
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7657765263950588e-06,
+      "loss": 2.089,
+      "step": 8137
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7639391946964312e-06,
+      "loss": 2.5916,
+      "step": 8138
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.762102726922359e-06,
+      "loss": 2.869,
+      "step": 8139
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7602671232654755e-06,
+      "loss": 3.6037,
+      "step": 8140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7584323839183305e-06,
+      "loss": 3.6132,
+      "step": 8141
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7565985090733806e-06,
+      "loss": 1.547,
+      "step": 8142
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7547654989229846e-06,
+      "loss": 2.7795,
+      "step": 8143
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7529333536594217e-06,
+      "loss": 2.6313,
+      "step": 8144
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.751102073474873e-06,
+      "loss": 3.194,
+      "step": 8145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7492716585614334e-06,
+      "loss": 2.0489,
+      "step": 8146
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7474421091111061e-06,
+      "loss": 3.6025,
+      "step": 8147
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7456134253157976e-06,
+      "loss": 3.119,
+      "step": 8148
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.743785607367332e-06,
+      "loss": 2.018,
+      "step": 8149
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7419586554574364e-06,
+      "loss": 2.4793,
+      "step": 8150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7401325697777515e-06,
+      "loss": 3.4422,
+      "step": 8151
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7383073505198255e-06,
+      "loss": 1.4974,
+      "step": 8152
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.736482997875114e-06,
+      "loss": 2.0937,
+      "step": 8153
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7346595120349852e-06,
+      "loss": 1.499,
+      "step": 8154
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7328368931907114e-06,
+      "loss": 1.3084,
+      "step": 8155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7310151415334798e-06,
+      "loss": 2.1714,
+      "step": 8156
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7291942572543806e-06,
+      "loss": 1.3482,
+      "step": 8157
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7273742405444217e-06,
+      "loss": 1.9597,
+      "step": 8158
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7255550915945073e-06,
+      "loss": 2.1677,
+      "step": 8159
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.723736810595461e-06,
+      "loss": 3.368,
+      "step": 8160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7219193977380121e-06,
+      "loss": 3.034,
+      "step": 8161
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7201028532127994e-06,
+      "loss": 1.6427,
+      "step": 8162
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7182871772103705e-06,
+      "loss": 3.1183,
+      "step": 8163
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7164723699211782e-06,
+      "loss": 2.9638,
+      "step": 8164
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7146584315355886e-06,
+      "loss": 3.1485,
+      "step": 8165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7128453622438768e-06,
+      "loss": 3.5535,
+      "step": 8166
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7110331622362242e-06,
+      "loss": 1.5392,
+      "step": 8167
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709221831702723e-06,
+      "loss": 3.4068,
+      "step": 8168
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7074113708333727e-06,
+      "loss": 1.5436,
+      "step": 8169
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7056017798180824e-06,
+      "loss": 2.0534,
+      "step": 8170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7037930588466701e-06,
+      "loss": 3.3339,
+      "step": 8171
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7019852081088616e-06,
+      "loss": 2.5063,
+      "step": 8172
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7001782277942958e-06,
+      "loss": 2.8856,
+      "step": 8173
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6983721180925106e-06,
+      "loss": 3.1139,
+      "step": 8174
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.69656687919296e-06,
+      "loss": 2.4502,
+      "step": 8175
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6947625112850074e-06,
+      "loss": 3.4323,
+      "step": 8176
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6929590145579223e-06,
+      "loss": 3.5392,
+      "step": 8177
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.691156389200883e-06,
+      "loss": 1.5675,
+      "step": 8178
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6893546354029754e-06,
+      "loss": 3.7149,
+      "step": 8179
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.687553753353195e-06,
+      "loss": 1.3009,
+      "step": 8180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6857537432404458e-06,
+      "loss": 2.592,
+      "step": 8181
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6839546052535416e-06,
+      "loss": 1.6508,
+      "step": 8182
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6821563395812057e-06,
+      "loss": 2.3249,
+      "step": 8183
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.680358946412064e-06,
+      "loss": 1.9226,
+      "step": 8184
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6785624259346556e-06,
+      "loss": 2.8572,
+      "step": 8185
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6767667783374276e-06,
+      "loss": 3.7856,
+      "step": 8186
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6749720038087359e-06,
+      "loss": 1.6131,
+      "step": 8187
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6731781025368422e-06,
+      "loss": 2.2938,
+      "step": 8188
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6713850747099202e-06,
+      "loss": 2.1705,
+      "step": 8189
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.669592920516049e-06,
+      "loss": 3.3441,
+      "step": 8190
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6678016401432184e-06,
+      "loss": 3.0914,
+      "step": 8191
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6660112337793256e-06,
+      "loss": 2.4511,
+      "step": 8192
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6642217016121765e-06,
+      "loss": 3.0438,
+      "step": 8193
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.662433043829481e-06,
+      "loss": 1.757,
+      "step": 8194
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.660645260618864e-06,
+      "loss": 3.2127,
+      "step": 8195
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6588583521678536e-06,
+      "loss": 2.0846,
+      "step": 8196
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6570723186638904e-06,
+      "loss": 2.6994,
+      "step": 8197
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6552871602943233e-06,
+      "loss": 2.0692,
+      "step": 8198
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6535028772464013e-06,
+      "loss": 2.8934,
+      "step": 8199
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6517194697072903e-06,
+      "loss": 2.5177,
+      "step": 8200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6499369378640606e-06,
+      "loss": 2.8172,
+      "step": 8201
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6481552819036927e-06,
+      "loss": 2.7974,
+      "step": 8202
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6463745020130738e-06,
+      "loss": 2.6395,
+      "step": 8203
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.644594598378999e-06,
+      "loss": 1.5923,
+      "step": 8204
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6428155711881722e-06,
+      "loss": 2.8406,
+      "step": 8205
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6410374206272061e-06,
+      "loss": 2.0909,
+      "step": 8206
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6392601468826187e-06,
+      "loss": 2.7531,
+      "step": 8207
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6374837501408403e-06,
+      "loss": 1.9783,
+      "step": 8208
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6357082305882032e-06,
+      "loss": 1.4467,
+      "step": 8209
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.633933588410952e-06,
+      "loss": 2.2752,
+      "step": 8210
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6321598237952408e-06,
+      "loss": 2.5157,
+      "step": 8211
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6303869369271264e-06,
+      "loss": 2.8043,
+      "step": 8212
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6286149279925778e-06,
+      "loss": 2.612,
+      "step": 8213
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.626843797177473e-06,
+      "loss": 2.4612,
+      "step": 8214
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6250735446675914e-06,
+      "loss": 1.9124,
+      "step": 8215
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6233041706486253e-06,
+      "loss": 1.8834,
+      "step": 8216
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.621535675306174e-06,
+      "loss": 1.5348,
+      "step": 8217
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6197680588257435e-06,
+      "loss": 1.5461,
+      "step": 8218
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6180013213927538e-06,
+      "loss": 1.6345,
+      "step": 8219
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6162354631925203e-06,
+      "loss": 2.0528,
+      "step": 8220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6144704844102776e-06,
+      "loss": 2.3263,
+      "step": 8221
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6127063852311608e-06,
+      "loss": 3.1915,
+      "step": 8222
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6109431658402163e-06,
+      "loss": 0.6285,
+      "step": 8223
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.609180826422404e-06,
+      "loss": 2.0623,
+      "step": 8224
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.607419367162577e-06,
+      "loss": 3.5927,
+      "step": 8225
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6056587882455078e-06,
+      "loss": 2.2426,
+      "step": 8226
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6038990898558715e-06,
+      "loss": 3.1612,
+      "step": 8227
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.602140272178253e-06,
+      "loss": 3.2732,
+      "step": 8228
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.600382335397147e-06,
+      "loss": 2.7218,
+      "step": 8229
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5986252796969482e-06,
+      "loss": 3.2992,
+      "step": 8230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5968691052619657e-06,
+      "loss": 4.532,
+      "step": 8231
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5951138122764132e-06,
+      "loss": 2.7497,
+      "step": 8232
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5933594009244147e-06,
+      "loss": 2.9489,
+      "step": 8233
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5916058713899995e-06,
+      "loss": 4.1251,
+      "step": 8234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.589853223857103e-06,
+      "loss": 2.4487,
+      "step": 8235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.58810145850957e-06,
+      "loss": 2.9459,
+      "step": 8236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5863505755311525e-06,
+      "loss": 2.4823,
+      "step": 8237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5846005751055116e-06,
+      "loss": 2.1471,
+      "step": 8238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5828514574162124e-06,
+      "loss": 2.2041,
+      "step": 8239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5811032226467304e-06,
+      "loss": 3.4986,
+      "step": 8240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5793558709804469e-06,
+      "loss": 2.3575,
+      "step": 8241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5776094026006516e-06,
+      "loss": 2.5902,
+      "step": 8242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5758638176905406e-06,
+      "loss": 2.08,
+      "step": 8243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5741191164332192e-06,
+      "loss": 1.8998,
+      "step": 8244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5723752990116948e-06,
+      "loss": 3.1683,
+      "step": 8245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5706323656088873e-06,
+      "loss": 2.1806,
+      "step": 8246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5688903164076242e-06,
+      "loss": 4.4032,
+      "step": 8247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5671491515906355e-06,
+      "loss": 2.2035,
+      "step": 8248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5654088713405646e-06,
+      "loss": 2.8106,
+      "step": 8249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5636694758399563e-06,
+      "loss": 1.9432,
+      "step": 8250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5619309652712644e-06,
+      "loss": 1.8471,
+      "step": 8251
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5601933398168523e-06,
+      "loss": 3.6414,
+      "step": 8252
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5584565996589884e-06,
+      "loss": 2.8905,
+      "step": 8253
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5567207449798517e-06,
+      "loss": 2.1385,
+      "step": 8254
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5549857759615195e-06,
+      "loss": 3.3219,
+      "step": 8255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5532516927859853e-06,
+      "loss": 1.78,
+      "step": 8256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.551518495635146e-06,
+      "loss": 2.9892,
+      "step": 8257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5497861846908024e-06,
+      "loss": 2.6445,
+      "step": 8258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5480547601346751e-06,
+      "loss": 0.9368,
+      "step": 8259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5463242221483742e-06,
+      "loss": 3.0724,
+      "step": 8260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.544594570913428e-06,
+      "loss": 2.7386,
+      "step": 8261
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5428658066112689e-06,
+      "loss": 2.9153,
+      "step": 8262
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5411379294232355e-06,
+      "loss": 3.5867,
+      "step": 8263
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5394109395305757e-06,
+      "loss": 3.2128,
+      "step": 8264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5376848371144404e-06,
+      "loss": 2.731,
+      "step": 8265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5359596223558903e-06,
+      "loss": 4.1911,
+      "step": 8266
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.534235295435894e-06,
+      "loss": 2.6075,
+      "step": 8267
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5325118565353237e-06,
+      "loss": 1.8419,
+      "step": 8268
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5307893058349632e-06,
+      "loss": 3.1219,
+      "step": 8269
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5290676435154949e-06,
+      "loss": 3.6573,
+      "step": 8270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5273468697575167e-06,
+      "loss": 1.5664,
+      "step": 8271
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5256269847415283e-06,
+      "loss": 3.6389,
+      "step": 8272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.523907988647939e-06,
+      "loss": 2.5275,
+      "step": 8273
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5221898816570623e-06,
+      "loss": 2.1161,
+      "step": 8274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.520472663949122e-06,
+      "loss": 2.6554,
+      "step": 8275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5187563357042423e-06,
+      "loss": 2.5356,
+      "step": 8276
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.517040897102462e-06,
+      "loss": 2.8261,
+      "step": 8277
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.51532634832372e-06,
+      "loss": 3.2006,
+      "step": 8278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5136126895478653e-06,
+      "loss": 2.8473,
+      "step": 8279
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.511899920954656e-06,
+      "loss": 1.0449,
+      "step": 8280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5101880427237481e-06,
+      "loss": 3.1642,
+      "step": 8281
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5084770550347115e-06,
+      "loss": 1.8418,
+      "step": 8282
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5067669580670218e-06,
+      "loss": 3.0099,
+      "step": 8283
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5050577520000608e-06,
+      "loss": 1.5359,
+      "step": 8284
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5033494370131162e-06,
+      "loss": 2.0983,
+      "step": 8285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5016420132853804e-06,
+      "loss": 2.8447,
+      "step": 8286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4999354809959542e-06,
+      "loss": 5.4531,
+      "step": 8287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.498229840323847e-06,
+      "loss": 2.9607,
+      "step": 8288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4965250914479712e-06,
+      "loss": 2.7284,
+      "step": 8289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4948212345471492e-06,
+      "loss": 2.4174,
+      "step": 8290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4931182698001046e-06,
+      "loss": 2.2758,
+      "step": 8291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4914161973854714e-06,
+      "loss": 2.4418,
+      "step": 8292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4897150174817875e-06,
+      "loss": 1.0665,
+      "step": 8293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.488014730267502e-06,
+      "loss": 2.4063,
+      "step": 8294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4863153359209693e-06,
+      "loss": 2.1508,
+      "step": 8295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4846168346204425e-06,
+      "loss": 2.7511,
+      "step": 8296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4829192265440883e-06,
+      "loss": 3.3568,
+      "step": 8297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4812225118699775e-06,
+      "loss": 3.0817,
+      "step": 8298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4795266907760897e-06,
+      "loss": 1.7256,
+      "step": 8299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4778317634403082e-06,
+      "loss": 1.2669,
+      "step": 8300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.476137730040421e-06,
+      "loss": 3.3044,
+      "step": 8301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4744445907541249e-06,
+      "loss": 2.0449,
+      "step": 8302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4727523457590232e-06,
+      "loss": 3.2517,
+      "step": 8303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4710609952326239e-06,
+      "loss": 2.0026,
+      "step": 8304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.469370539352345e-06,
+      "loss": 3.6316,
+      "step": 8305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4676809782955014e-06,
+      "loss": 1.4955,
+      "step": 8306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.465992312239325e-06,
+      "loss": 3.2458,
+      "step": 8307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.464304541360946e-06,
+      "loss": 2.802,
+      "step": 8308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4626176658374058e-06,
+      "loss": 2.644,
+      "step": 8309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.460931685845649e-06,
+      "loss": 3.9912,
+      "step": 8310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4592466015625274e-06,
+      "loss": 3.8311,
+      "step": 8311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.457562413164799e-06,
+      "loss": 3.0372,
+      "step": 8312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4558791208291256e-06,
+      "loss": 1.9718,
+      "step": 8313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4541967247320787e-06,
+      "loss": 3.8783,
+      "step": 8314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4525152250501362e-06,
+      "loss": 3.145,
+      "step": 8315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4508346219596725e-06,
+      "loss": 2.5295,
+      "step": 8316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4491549156369799e-06,
+      "loss": 3.6667,
+      "step": 8317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4474761062582509e-06,
+      "loss": 2.9043,
+      "step": 8318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4457981939995847e-06,
+      "loss": 2.148,
+      "step": 8319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4441211790369892e-06,
+      "loss": 3.1884,
+      "step": 8320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4424450615463714e-06,
+      "loss": 2.2733,
+      "step": 8321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4407698417035488e-06,
+      "loss": 3.3767,
+      "step": 8322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.439095519684245e-06,
+      "loss": 2.9331,
+      "step": 8323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4374220956640895e-06,
+      "loss": 2.4527,
+      "step": 8324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4357495698186186e-06,
+      "loss": 1.8569,
+      "step": 8325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4340779423232665e-06,
+      "loss": 2.8003,
+      "step": 8326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4324072133533807e-06,
+      "loss": 2.0256,
+      "step": 8327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4307373830842174e-06,
+      "loss": 3.8205,
+      "step": 8328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4290684516909314e-06,
+      "loss": 2.7632,
+      "step": 8329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.427400419348588e-06,
+      "loss": 2.8998,
+      "step": 8330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4257332862321516e-06,
+      "loss": 3.271,
+      "step": 8331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.424067052516499e-06,
+      "loss": 3.7598,
+      "step": 8332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4224017183764104e-06,
+      "loss": 2.0238,
+      "step": 8333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.420737283986572e-06,
+      "loss": 1.6211,
+      "step": 8334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4190737495215746e-06,
+      "loss": 2.8142,
+      "step": 8335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4174111151559188e-06,
+      "loss": 1.8299,
+      "step": 8336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4157493810640011e-06,
+      "loss": 3.2389,
+      "step": 8337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4140885474201315e-06,
+      "loss": 2.9733,
+      "step": 8338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4124286143985267e-06,
+      "loss": 3.8079,
+      "step": 8339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4107695821733026e-06,
+      "loss": 2.0106,
+      "step": 8340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4091114509184889e-06,
+      "loss": 2.137,
+      "step": 8341
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.407454220808009e-06,
+      "loss": 2.3007,
+      "step": 8342
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4057978920157033e-06,
+      "loss": 1.2602,
+      "step": 8343
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4041424647153112e-06,
+      "loss": 3.0816,
+      "step": 8344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.402487939080479e-06,
+      "loss": 2.4775,
+      "step": 8345
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4008343152847615e-06,
+      "loss": 2.5114,
+      "step": 8346
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3991815935016128e-06,
+      "loss": 1.6907,
+      "step": 8347
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3975297739043992e-06,
+      "loss": 2.6204,
+      "step": 8348
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.395878856666385e-06,
+      "loss": 2.9397,
+      "step": 8349
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3942288419607476e-06,
+      "loss": 2.3076,
+      "step": 8350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3925797299605649e-06,
+      "loss": 2.2771,
+      "step": 8351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3909315208388185e-06,
+      "loss": 4.092,
+      "step": 8352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3892842147683994e-06,
+      "loss": 3.3923,
+      "step": 8353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3876378119221024e-06,
+      "loss": 2.0897,
+      "step": 8354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3859923124726283e-06,
+      "loss": 1.5609,
+      "step": 8355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3843477165925846e-06,
+      "loss": 3.2617,
+      "step": 8356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3827040244544754e-06,
+      "loss": 1.9716,
+      "step": 8357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3810612362307208e-06,
+      "loss": 2.3103,
+      "step": 8358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3794193520936405e-06,
+      "loss": 2.9295,
+      "step": 8359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3777783722154603e-06,
+      "loss": 3.3168,
+      "step": 8360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3761382967683146e-06,
+      "loss": 2.7618,
+      "step": 8361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3744991259242324e-06,
+      "loss": 2.1048,
+      "step": 8362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.372860859855163e-06,
+      "loss": 1.5145,
+      "step": 8363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3712234987329486e-06,
+      "loss": 2.0567,
+      "step": 8364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.369587042729341e-06,
+      "loss": 1.2452,
+      "step": 8365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.367951492016001e-06,
+      "loss": 2.5736,
+      "step": 8366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.366316846764485e-06,
+      "loss": 1.7175,
+      "step": 8367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3646831071462606e-06,
+      "loss": 3.0937,
+      "step": 8368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.363050273332701e-06,
+      "loss": 1.8883,
+      "step": 8369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3614183454950824e-06,
+      "loss": 2.3284,
+      "step": 8370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.359787323804589e-06,
+      "loss": 1.8697,
+      "step": 8371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3581572084323014e-06,
+      "loss": 2.024,
+      "step": 8372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3565279995492165e-06,
+      "loss": 2.0959,
+      "step": 8373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3548996973262273e-06,
+      "loss": 1.167,
+      "step": 8374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3532723019341376e-06,
+      "loss": 3.1955,
+      "step": 8375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3516458135436539e-06,
+      "loss": 1.5523,
+      "step": 8376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3500202323253853e-06,
+      "loss": 2.9168,
+      "step": 8377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3483955584498476e-06,
+      "loss": 3.0966,
+      "step": 8378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3467717920874624e-06,
+      "loss": 3.5798,
+      "step": 8379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3451489334085555e-06,
+      "loss": 1.1596,
+      "step": 8380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3435269825833575e-06,
+      "loss": 2.7676,
+      "step": 8381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3419059397820033e-06,
+      "loss": 1.9668,
+      "step": 8382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3402858051745315e-06,
+      "loss": 2.3406,
+      "step": 8383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3386665789308885e-06,
+      "loss": 4.5357,
+      "step": 8384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3370482612209224e-06,
+      "loss": 3.4751,
+      "step": 8385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.33543085221439e-06,
+      "loss": 2.8304,
+      "step": 8386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3338143520809465e-06,
+      "loss": 1.9978,
+      "step": 8387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3321987609901553e-06,
+      "loss": 3.7089,
+      "step": 8388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3305840791114854e-06,
+      "loss": 3.0656,
+      "step": 8389
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3289703066143112e-06,
+      "loss": 2.0024,
+      "step": 8390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3273574436679093e-06,
+      "loss": 2.3943,
+      "step": 8391
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.325745490441458e-06,
+      "loss": 2.6323,
+      "step": 8392
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3241344471040473e-06,
+      "loss": 1.8262,
+      "step": 8393
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3225243138246669e-06,
+      "loss": 1.3015,
+      "step": 8394
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3209150907722124e-06,
+      "loss": 1.5151,
+      "step": 8395
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3193067781154835e-06,
+      "loss": 2.7405,
+      "step": 8396
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.317699376023186e-06,
+      "loss": 2.3133,
+      "step": 8397
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3160928846639275e-06,
+      "loss": 2.7388,
+      "step": 8398
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3144873042062222e-06,
+      "loss": 2.6863,
+      "step": 8399
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3128826348184886e-06,
+      "loss": 3.4206,
+      "step": 8400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3112788766690477e-06,
+      "loss": 2.0883,
+      "step": 8401
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3096760299261303e-06,
+      "loss": 3.1615,
+      "step": 8402
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3080740947578618e-06,
+      "loss": 3.2384,
+      "step": 8403
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3064730713322793e-06,
+      "loss": 2.9464,
+      "step": 8404
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3048729598173248e-06,
+      "loss": 2.4537,
+      "step": 8405
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.303273760380841e-06,
+      "loss": 3.4052,
+      "step": 8406
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.301675473190579e-06,
+      "loss": 2.5703,
+      "step": 8407
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3000780984141881e-06,
+      "loss": 1.9777,
+      "step": 8408
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2984816362192265e-06,
+      "loss": 2.755,
+      "step": 8409
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.296886086773157e-06,
+      "loss": 2.8411,
+      "step": 8410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2952914502433433e-06,
+      "loss": 3.4795,
+      "step": 8411
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2936977267970597e-06,
+      "loss": 3.2223,
+      "step": 8412
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2921049166014755e-06,
+      "loss": 2.676,
+      "step": 8413
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2905130198236704e-06,
+      "loss": 1.2916,
+      "step": 8414
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2889220366306276e-06,
+      "loss": 3.23,
+      "step": 8415
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2873319671892337e-06,
+      "loss": 2.8842,
+      "step": 8416
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2857428116662808e-06,
+      "loss": 3.5774,
+      "step": 8417
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2841545702284618e-06,
+      "loss": 2.9374,
+      "step": 8418
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2825672430423763e-06,
+      "loss": 1.6096,
+      "step": 8419
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2809808302745298e-06,
+      "loss": 2.2491,
+      "step": 8420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2793953320913267e-06,
+      "loss": 1.987,
+      "step": 8421
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.277810748659082e-06,
+      "loss": 3.0685,
+      "step": 8422
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.276227080144007e-06,
+      "loss": 2.3221,
+      "step": 8423
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2746443267122233e-06,
+      "loss": 2.8907,
+      "step": 8424
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2730624885297537e-06,
+      "loss": 2.4169,
+      "step": 8425
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2714815657625257e-06,
+      "loss": 2.6213,
+      "step": 8426
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.269901558576374e-06,
+      "loss": 3.214,
+      "step": 8427
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2683224671370286e-06,
+      "loss": 2.1784,
+      "step": 8428
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2667442916101313e-06,
+      "loss": 2.1616,
+      "step": 8429
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2651670321612264e-06,
+      "loss": 1.8967,
+      "step": 8430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2635906889557591e-06,
+      "loss": 2.3397,
+      "step": 8431
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.262015262159082e-06,
+      "loss": 1.9475,
+      "step": 8432
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2604407519364504e-06,
+      "loss": 3.4371,
+      "step": 8433
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.258867158453021e-06,
+      "loss": 1.3991,
+      "step": 8434
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2572944818738587e-06,
+      "loss": 1.6016,
+      "step": 8435
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.255722722363929e-06,
+      "loss": 2.8924,
+      "step": 8436
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2541518800881047e-06,
+      "loss": 2.5667,
+      "step": 8437
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.252581955211155e-06,
+      "loss": 3.5237,
+      "step": 8438
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2510129478977594e-06,
+      "loss": 2.3369,
+      "step": 8439
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.249444858312502e-06,
+      "loss": 3.0302,
+      "step": 8440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.247877686619865e-06,
+      "loss": 1.8371,
+      "step": 8441
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2463114329842406e-06,
+      "loss": 2.4208,
+      "step": 8442
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2447460975699189e-06,
+      "loss": 3.0162,
+      "step": 8443
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2431816805410968e-06,
+      "loss": 2.8059,
+      "step": 8444
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2416181820618745e-06,
+      "loss": 2.8432,
+      "step": 8445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2400556022962562e-06,
+      "loss": 3.1843,
+      "step": 8446
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.238493941408151e-06,
+      "loss": 3.3507,
+      "step": 8447
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2369331995613664e-06,
+      "loss": 3.3633,
+      "step": 8448
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2353733769196174e-06,
+      "loss": 2.065,
+      "step": 8449
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.233814473646524e-06,
+      "loss": 1.8497,
+      "step": 8450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2322564899056077e-06,
+      "loss": 2.7764,
+      "step": 8451
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2306994258602922e-06,
+      "loss": 3.0728,
+      "step": 8452
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2291432816739079e-06,
+      "loss": 3.422,
+      "step": 8453
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2275880575096865e-06,
+      "loss": 2.0174,
+      "step": 8454
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.226033753530763e-06,
+      "loss": 2.6952,
+      "step": 8455
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2244803699001785e-06,
+      "loss": 1.2946,
+      "step": 8456
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2229279067808752e-06,
+      "loss": 2.0252,
+      "step": 8457
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2213763643357002e-06,
+      "loss": 3.0716,
+      "step": 8458
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2198257427273997e-06,
+      "loss": 2.7607,
+      "step": 8459
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.218276042118629e-06,
+      "loss": 1.6944,
+      "step": 8460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.216727262671944e-06,
+      "loss": 1.8613,
+      "step": 8461
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2151794045498045e-06,
+      "loss": 2.9373,
+      "step": 8462
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.213632467914576e-06,
+      "loss": 1.433,
+      "step": 8463
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2120864529285203e-06,
+      "loss": 2.6467,
+      "step": 8464
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2105413597538107e-06,
+      "loss": 4.6732,
+      "step": 8465
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2089971885525187e-06,
+      "loss": 3.4425,
+      "step": 8466
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2074539394866215e-06,
+      "loss": 1.8256,
+      "step": 8467
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2059116127179993e-06,
+      "loss": 3.4016,
+      "step": 8468
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.204370208408433e-06,
+      "loss": 1.7047,
+      "step": 8469
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.202829726719611e-06,
+      "loss": 2.4047,
+      "step": 8470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2012901678131216e-06,
+      "loss": 3.0728,
+      "step": 8471
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.199751531850457e-06,
+      "loss": 2.5309,
+      "step": 8472
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1982138189930159e-06,
+      "loss": 3.6765,
+      "step": 8473
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1966770294020934e-06,
+      "loss": 2.2723,
+      "step": 8474
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.195141163238892e-06,
+      "loss": 1.549,
+      "step": 8475
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1936062206645183e-06,
+      "loss": 2.0999,
+      "step": 8476
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.19207220183998e-06,
+      "loss": 2.5124,
+      "step": 8477
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1905391069261918e-06,
+      "loss": 1.3578,
+      "step": 8478
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.189006936083963e-06,
+      "loss": 2.7857,
+      "step": 8479
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1874756894740137e-06,
+      "loss": 2.5311,
+      "step": 8480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1859453672569632e-06,
+      "loss": 0.8999,
+      "step": 8481
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1844159695933377e-06,
+      "loss": 1.5916,
+      "step": 8482
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1828874966435632e-06,
+      "loss": 1.8715,
+      "step": 8483
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1813599485679684e-06,
+      "loss": 2.2487,
+      "step": 8484
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1798333255267857e-06,
+      "loss": 1.8658,
+      "step": 8485
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1783076276801508e-06,
+      "loss": 2.929,
+      "step": 8486
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1767828551881045e-06,
+      "loss": 3.0846,
+      "step": 8487
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1752590082105863e-06,
+      "loss": 4.0228,
+      "step": 8488
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.173736086907441e-06,
+      "loss": 3.1815,
+      "step": 8489
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1722140914384162e-06,
+      "loss": 2.3626,
+      "step": 8490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1706930219631619e-06,
+      "loss": 3.1371,
+      "step": 8491
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1691728786412315e-06,
+      "loss": 2.896,
+      "step": 8492
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1676536616320833e-06,
+      "loss": 2.3097,
+      "step": 8493
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.166135371095073e-06,
+      "loss": 2.5737,
+      "step": 8494
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1646180071894608e-06,
+      "loss": 3.1047,
+      "step": 8495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1631015700744153e-06,
+      "loss": 3.7457,
+      "step": 8496
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1615860599090012e-06,
+      "loss": 3.6581,
+      "step": 8497
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1600714768521903e-06,
+      "loss": 1.8904,
+      "step": 8498
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1585578210628524e-06,
+      "loss": 2.5946,
+      "step": 8499
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1570450926997657e-06,
+      "loss": 1.6034,
+      "step": 8500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1555332919216057e-06,
+      "loss": 3.094,
+      "step": 8501
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1540224188869565e-06,
+      "loss": 2.4471,
+      "step": 8502
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1525124737542992e-06,
+      "loss": 2.4299,
+      "step": 8503
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1510034566820205e-06,
+      "loss": 3.368,
+      "step": 8504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1494953678284105e-06,
+      "loss": 2.5391,
+      "step": 8505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.14798820735166e-06,
+      "loss": 3.3413,
+      "step": 8506
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1464819754098633e-06,
+      "loss": 2.9798,
+      "step": 8507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.144976672161019e-06,
+      "loss": 2.6766,
+      "step": 8508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.143472297763022e-06,
+      "loss": 2.8495,
+      "step": 8509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1419688523736761e-06,
+      "loss": 3.149,
+      "step": 8510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1404663361506863e-06,
+      "loss": 3.191,
+      "step": 8511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1389647492516598e-06,
+      "loss": 3.4789,
+      "step": 8512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1374640918341074e-06,
+      "loss": 2.2409,
+      "step": 8513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1359643640554384e-06,
+      "loss": 2.5823,
+      "step": 8514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1344655660729676e-06,
+      "loss": 3.7683,
+      "step": 8515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.132967698043913e-06,
+      "loss": 3.3413,
+      "step": 8516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1314707601253938e-06,
+      "loss": 1.4925,
+      "step": 8517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1299747524744309e-06,
+      "loss": 2.7042,
+      "step": 8518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.128479675247952e-06,
+      "loss": 3.6031,
+      "step": 8519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1269855286027798e-06,
+      "loss": 3.2227,
+      "step": 8520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1254923126956441e-06,
+      "loss": 3.3868,
+      "step": 8521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1240000276831765e-06,
+      "loss": 2.4987,
+      "step": 8522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1225086737219114e-06,
+      "loss": 3.9429,
+      "step": 8523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1210182509682854e-06,
+      "loss": 3.2053,
+      "step": 8524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1195287595786352e-06,
+      "loss": 1.9624,
+      "step": 8525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.118040199709204e-06,
+      "loss": 1.6678,
+      "step": 8526
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1165525715161317e-06,
+      "loss": 2.6383,
+      "step": 8527
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1150658751554667e-06,
+      "loss": 1.3307,
+      "step": 8528
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1135801107831568e-06,
+      "loss": 3.056,
+      "step": 8529
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1120952785550477e-06,
+      "loss": 3.2743,
+      "step": 8530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1106113786268935e-06,
+      "loss": 2.6567,
+      "step": 8531
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1091284111543499e-06,
+      "loss": 2.492,
+      "step": 8532
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1076463762929712e-06,
+      "loss": 3.7485,
+      "step": 8533
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1061652741982198e-06,
+      "loss": 1.9343,
+      "step": 8534
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1046851050254504e-06,
+      "loss": 3.4704,
+      "step": 8535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1032058689299297e-06,
+      "loss": 2.4736,
+      "step": 8536
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1017275660668224e-06,
+      "loss": 3.3187,
+      "step": 8537
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.100250196591195e-06,
+      "loss": 2.4738,
+      "step": 8538
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0987737606580174e-06,
+      "loss": 2.771,
+      "step": 8539
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0972982584221592e-06,
+      "loss": 2.3996,
+      "step": 8540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0958236900383957e-06,
+      "loss": 3.0395,
+      "step": 8541
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0943500556614017e-06,
+      "loss": 2.4066,
+      "step": 8542
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.092877355445755e-06,
+      "loss": 2.9614,
+      "step": 8543
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0914055895459353e-06,
+      "loss": 3.4313,
+      "step": 8544
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0899347581163222e-06,
+      "loss": 2.6724,
+      "step": 8545
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0884648613111992e-06,
+      "loss": 3.2336,
+      "step": 8546
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.086995899284753e-06,
+      "loss": 3.0473,
+      "step": 8547
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.08552787219107e-06,
+      "loss": 1.186,
+      "step": 8548
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.084060780184143e-06,
+      "loss": 2.6841,
+      "step": 8549
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0825946234178575e-06,
+      "loss": 3.1557,
+      "step": 8550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0811294020460083e-06,
+      "loss": 2.3236,
+      "step": 8551
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0796651162222916e-06,
+      "loss": 2.4591,
+      "step": 8552
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0782017661003041e-06,
+      "loss": 2.5696,
+      "step": 8553
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0767393518335445e-06,
+      "loss": 2.1554,
+      "step": 8554
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0752778735754121e-06,
+      "loss": 3.8273,
+      "step": 8555
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.07381733147921e-06,
+      "loss": 2.7783,
+      "step": 8556
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.072357725698141e-06,
+      "loss": 2.0405,
+      "step": 8557
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0708990563853127e-06,
+      "loss": 2.8,
+      "step": 8558
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0694413236937329e-06,
+      "loss": 3.205,
+      "step": 8559
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.067984527776309e-06,
+      "loss": 1.5896,
+      "step": 8560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0665286687858544e-06,
+      "loss": 4.4653,
+      "step": 8561
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0650737468750806e-06,
+      "loss": 2.4509,
+      "step": 8562
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0636197621966016e-06,
+      "loss": 3.461,
+      "step": 8563
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.062166714902938e-06,
+      "loss": 2.2728,
+      "step": 8564
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0607146051465011e-06,
+      "loss": 2.435,
+      "step": 8565
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0592634330796137e-06,
+      "loss": 2.2065,
+      "step": 8566
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.057813198854497e-06,
+      "loss": 2.0603,
+      "step": 8567
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0563639026232742e-06,
+      "loss": 2.8694,
+      "step": 8568
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.054915544537971e-06,
+      "loss": 3.5772,
+      "step": 8569
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0534681247505107e-06,
+      "loss": 1.5391,
+      "step": 8570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0520216434127207e-06,
+      "loss": 3.4262,
+      "step": 8571
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0505761006763315e-06,
+      "loss": 3.0901,
+      "step": 8572
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0491314966929743e-06,
+      "loss": 2.77,
+      "step": 8573
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0476878316141814e-06,
+      "loss": 3.3466,
+      "step": 8574
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0462451055913847e-06,
+      "loss": 2.7204,
+      "step": 8575
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.044803318775922e-06,
+      "loss": 3.6547,
+      "step": 8576
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0433624713190281e-06,
+      "loss": 2.6222,
+      "step": 8577
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.041922563371842e-06,
+      "loss": 3.0174,
+      "step": 8578
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.040483595085403e-06,
+      "loss": 2.4486,
+      "step": 8579
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0390455666106547e-06,
+      "loss": 2.5327,
+      "step": 8580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0376084780984353e-06,
+      "loss": 1.8838,
+      "step": 8581
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0361723296994908e-06,
+      "loss": 2.412,
+      "step": 8582
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0347371215644675e-06,
+      "loss": 1.3839,
+      "step": 8583
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0333028538439093e-06,
+      "loss": 2.8757,
+      "step": 8584
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0318695266882696e-06,
+      "loss": 1.6557,
+      "step": 8585
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0304371402478918e-06,
+      "loss": 2.157,
+      "step": 8586
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0290056946730288e-06,
+      "loss": 2.9023,
+      "step": 8587
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.027575190113832e-06,
+      "loss": 2.8352,
+      "step": 8588
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0261456267203561e-06,
+      "loss": 4.1856,
+      "step": 8589
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.024717004642557e-06,
+      "loss": 3.7518,
+      "step": 8590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0232893240302865e-06,
+      "loss": 2.9622,
+      "step": 8591
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.021862585033304e-06,
+      "loss": 3.075,
+      "step": 8592
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0204367878012666e-06,
+      "loss": 2.302,
+      "step": 8593
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0190119324837356e-06,
+      "loss": 2.8479,
+      "step": 8594
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0175880192301713e-06,
+      "loss": 2.8901,
+      "step": 8595
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0161650481899344e-06,
+      "loss": 1.8997,
+      "step": 8596
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0147430195122888e-06,
+      "loss": 2.5679,
+      "step": 8597
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0133219333463983e-06,
+      "loss": 2.885,
+      "step": 8598
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0119017898413286e-06,
+      "loss": 3.4054,
+      "step": 8599
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.010482589146048e-06,
+      "loss": 3.1906,
+      "step": 8600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0090643314094207e-06,
+      "loss": 2.8731,
+      "step": 8601
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0076470167802156e-06,
+      "loss": 3.4368,
+      "step": 8602
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0062306454071047e-06,
+      "loss": 2.4573,
+      "step": 8603
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0048152174386584e-06,
+      "loss": 3.6851,
+      "step": 8604
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0034007330233487e-06,
+      "loss": 1.9936,
+      "step": 8605
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0019871923095459e-06,
+      "loss": 3.1097,
+      "step": 8606
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0005745954455248e-06,
+      "loss": 1.6493,
+      "step": 8607
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.991629425794624e-07,
+      "loss": 2.474,
+      "step": 8608
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.977522338594325e-07,
+      "loss": 3.2412,
+      "step": 8609
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.963424694334122e-07,
+      "loss": 2.8526,
+      "step": 8610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.9493364944928e-07,
+      "loss": 2.2056,
+      "step": 8611
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.935257740548143e-07,
+      "loss": 1.5151,
+      "step": 8612
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.921188433976947e-07,
+      "loss": 3.6622,
+      "step": 8613
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.90712857625501e-07,
+      "loss": 2.4325,
+      "step": 8614
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.893078168857173e-07,
+      "loss": 2.0997,
+      "step": 8615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.879037213257214e-07,
+      "loss": 2.6248,
+      "step": 8616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.865005710927989e-07,
+      "loss": 3.2358,
+      "step": 8617
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.85098366334134e-07,
+      "loss": 1.6115,
+      "step": 8618
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.836971071968104e-07,
+      "loss": 2.3939,
+      "step": 8619
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.822967938278172e-07,
+      "loss": 2.6439,
+      "step": 8620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.808974263740345e-07,
+      "loss": 2.9142,
+      "step": 8621
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.79499004982254e-07,
+      "loss": 3.1018,
+      "step": 8622
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.781015297991626e-07,
+      "loss": 2.0578,
+      "step": 8623
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.767050009713476e-07,
+      "loss": 2.5485,
+      "step": 8624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.753094186453028e-07,
+      "loss": 2.7707,
+      "step": 8625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.739147829674124e-07,
+      "loss": 2.5366,
+      "step": 8626
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.725210940839703e-07,
+      "loss": 2.9189,
+      "step": 8627
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.711283521411674e-07,
+      "loss": 4.0509,
+      "step": 8628
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.697365572850958e-07,
+      "loss": 2.0114,
+      "step": 8629
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.683457096617487e-07,
+      "loss": 1.6072,
+      "step": 8630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.669558094170194e-07,
+      "loss": 2.2758,
+      "step": 8631
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.655668566967026e-07,
+      "loss": 2.4606,
+      "step": 8632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.641788516464922e-07,
+      "loss": 3.4971,
+      "step": 8633
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.627917944119835e-07,
+      "loss": 2.5723,
+      "step": 8634
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.614056851386743e-07,
+      "loss": 1.6684,
+      "step": 8635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.600205239719584e-07,
+      "loss": 2.1719,
+      "step": 8636
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.58636311057133e-07,
+      "loss": 3.6076,
+      "step": 8637
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.57253046539396e-07,
+      "loss": 2.5843,
+      "step": 8638
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.558707305638458e-07,
+      "loss": 1.4689,
+      "step": 8639
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.544893632754816e-07,
+      "loss": 2.7937,
+      "step": 8640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.531089448192022e-07,
+      "loss": 3.6117,
+      "step": 8641
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.517294753398066e-07,
+      "loss": 1.9088,
+      "step": 8642
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.503509549819934e-07,
+      "loss": 2.8804,
+      "step": 8643
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.489733838903648e-07,
+      "loss": 3.1922,
+      "step": 8644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.475967622094207e-07,
+      "loss": 3.1252,
+      "step": 8645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.462210900835622e-07,
+      "loss": 3.3514,
+      "step": 8646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.448463676570929e-07,
+      "loss": 3.6259,
+      "step": 8647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.434725950742119e-07,
+      "loss": 2.6047,
+      "step": 8648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.420997724790238e-07,
+      "loss": 1.2758,
+      "step": 8649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.407279000155311e-07,
+      "loss": 1.184,
+      "step": 8650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.393569778276379e-07,
+      "loss": 2.7677,
+      "step": 8651
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.379870060591434e-07,
+      "loss": 2.6277,
+      "step": 8652
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.366179848537549e-07,
+      "loss": 2.4127,
+      "step": 8653
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.352499143550742e-07,
+      "loss": 0.7525,
+      "step": 8654
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.338827947066076e-07,
+      "loss": 2.4326,
+      "step": 8655
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.325166260517593e-07,
+      "loss": 2.509,
+      "step": 8656
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.311514085338325e-07,
+      "loss": 1.7223,
+      "step": 8657
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.297871422960336e-07,
+      "loss": 2.8852,
+      "step": 8658
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.284238274814672e-07,
+      "loss": 3.9729,
+      "step": 8659
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.270614642331377e-07,
+      "loss": 3.3639,
+      "step": 8660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.257000526939541e-07,
+      "loss": 2.9387,
+      "step": 8661
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.243395930067189e-07,
+      "loss": 1.7894,
+      "step": 8662
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.22980085314138e-07,
+      "loss": 2.1923,
+      "step": 8663
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.216215297588182e-07,
+      "loss": 3.1858,
+      "step": 8664
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.202639264832669e-07,
+      "loss": 2.1835,
+      "step": 8665
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.189072756298889e-07,
+      "loss": 2.5991,
+      "step": 8666
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.175515773409904e-07,
+      "loss": 2.1211,
+      "step": 8667
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.161968317587788e-07,
+      "loss": 3.6449,
+      "step": 8668
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.148430390253615e-07,
+      "loss": 2.0798,
+      "step": 8669
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.134901992827427e-07,
+      "loss": 2.171,
+      "step": 8670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.121383126728322e-07,
+      "loss": 3.1534,
+      "step": 8671
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.107873793374322e-07,
+      "loss": 2.4267,
+      "step": 8672
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.094373994182525e-07,
+      "loss": 2.7441,
+      "step": 8673
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.08088373056899e-07,
+      "loss": 3.8413,
+      "step": 8674
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.067403003948783e-07,
+      "loss": 2.2472,
+      "step": 8675
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.053931815735995e-07,
+      "loss": 1.9288,
+      "step": 8676
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.04047016734364e-07,
+      "loss": 3.3183,
+      "step": 8677
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.027018060183801e-07,
+      "loss": 2.4023,
+      "step": 8678
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.013575495667559e-07,
+      "loss": 2.7754,
+      "step": 8679
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.000142475204965e-07,
+      "loss": 2.3445,
+      "step": 8680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.986719000205079e-07,
+      "loss": 2.6914,
+      "step": 8681
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.973305072075966e-07,
+      "loss": 2.3891,
+      "step": 8682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.959900692224676e-07,
+      "loss": 3.7606,
+      "step": 8683
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.946505862057286e-07,
+      "loss": 3.5041,
+      "step": 8684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.933120582978827e-07,
+      "loss": 2.5736,
+      "step": 8685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.919744856393386e-07,
+      "loss": 3.0678,
+      "step": 8686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.906378683703986e-07,
+      "loss": 3.3086,
+      "step": 8687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.893022066312674e-07,
+      "loss": 2.9916,
+      "step": 8688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.879675005620503e-07,
+      "loss": 2.9072,
+      "step": 8689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.866337503027523e-07,
+      "loss": 2.2156,
+      "step": 8690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.853009559932802e-07,
+      "loss": 2.1656,
+      "step": 8691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.839691177734322e-07,
+      "loss": 1.6796,
+      "step": 8692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.826382357829155e-07,
+      "loss": 2.9451,
+      "step": 8693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.813083101613329e-07,
+      "loss": 3.5347,
+      "step": 8694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.799793410481871e-07,
+      "loss": 3.0722,
+      "step": 8695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.786513285828835e-07,
+      "loss": 1.9988,
+      "step": 8696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.773242729047193e-07,
+      "loss": 3.4471,
+      "step": 8697
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.759981741529e-07,
+      "loss": 2.2524,
+      "step": 8698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.746730324665264e-07,
+      "loss": 1.9929,
+      "step": 8699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.733488479845997e-07,
+      "loss": 2.1158,
+      "step": 8700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.720256208460187e-07,
+      "loss": 1.3667,
+      "step": 8701
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.70703351189589e-07,
+      "loss": 1.1496,
+      "step": 8702
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.693820391540064e-07,
+      "loss": 3.0881,
+      "step": 8703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.680616848778711e-07,
+      "loss": 2.4401,
+      "step": 8704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.667422884996823e-07,
+      "loss": 2.688,
+      "step": 8705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.654238501578393e-07,
+      "loss": 1.8895,
+      "step": 8706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.641063699906404e-07,
+      "loss": 1.5499,
+      "step": 8707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.627898481362817e-07,
+      "loss": 3.1101,
+      "step": 8708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.614742847328594e-07,
+      "loss": 0.8852,
+      "step": 8709
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.60159679918372e-07,
+      "loss": 2.498,
+      "step": 8710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.588460338307159e-07,
+      "loss": 3.9207,
+      "step": 8711
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.575333466076863e-07,
+      "loss": 2.1638,
+      "step": 8712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.562216183869754e-07,
+      "loss": 2.5889,
+      "step": 8713
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.549108493061786e-07,
+      "loss": 2.5635,
+      "step": 8714
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.536010395027905e-07,
+      "loss": 1.829,
+      "step": 8715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.522921891142034e-07,
+      "loss": 2.7295,
+      "step": 8716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.509842982777105e-07,
+      "loss": 3.4226,
+      "step": 8717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.496773671305025e-07,
+      "loss": 1.6965,
+      "step": 8718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.483713958096706e-07,
+      "loss": 2.883,
+      "step": 8719
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.470663844522053e-07,
+      "loss": 1.9573,
+      "step": 8720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.457623331949961e-07,
+      "loss": 2.8613,
+      "step": 8721
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.444592421748332e-07,
+      "loss": 4.3515,
+      "step": 8722
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.431571115284032e-07,
+      "loss": 3.218,
+      "step": 8723
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.418559413922933e-07,
+      "loss": 2.6512,
+      "step": 8724
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.405557319029911e-07,
+      "loss": 2.7393,
+      "step": 8725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.392564831968819e-07,
+      "loss": 2.2576,
+      "step": 8726
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.379581954102534e-07,
+      "loss": 3.6981,
+      "step": 8727
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.366608686792854e-07,
+      "loss": 1.3818,
+      "step": 8728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.353645031400648e-07,
+      "loss": 3.2603,
+      "step": 8729
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.340690989285727e-07,
+      "loss": 2.9706,
+      "step": 8730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.327746561806926e-07,
+      "loss": 3.0199,
+      "step": 8731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.31481175032206e-07,
+      "loss": 2.1918,
+      "step": 8732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.301886556187899e-07,
+      "loss": 3.5231,
+      "step": 8733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.288970980760258e-07,
+      "loss": 2.5239,
+      "step": 8734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.276065025393909e-07,
+      "loss": 1.9257,
+      "step": 8735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.263168691442624e-07,
+      "loss": 2.2238,
+      "step": 8736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.250281980259212e-07,
+      "loss": 2.0299,
+      "step": 8737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.237404893195377e-07,
+      "loss": 1.6949,
+      "step": 8738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.224537431601886e-07,
+      "loss": 2.3286,
+      "step": 8739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.211679596828481e-07,
+      "loss": 2.2165,
+      "step": 8740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.198831390223871e-07,
+      "loss": 2.7633,
+      "step": 8741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.185992813135813e-07,
+      "loss": 1.9869,
+      "step": 8742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.173163866910972e-07,
+      "loss": 2.8115,
+      "step": 8743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.160344552895061e-07,
+      "loss": 1.7582,
+      "step": 8744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.147534872432761e-07,
+      "loss": 2.6177,
+      "step": 8745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.134734826867751e-07,
+      "loss": 1.601,
+      "step": 8746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.121944417542715e-07,
+      "loss": 2.6187,
+      "step": 8747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.109163645799267e-07,
+      "loss": 2.0181,
+      "step": 8748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.096392512978091e-07,
+      "loss": 2.5104,
+      "step": 8749
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.083631020418792e-07,
+      "loss": 2.7911,
+      "step": 8750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.070879169459999e-07,
+      "loss": 2.6788,
+      "step": 8751
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.058136961439333e-07,
+      "loss": 1.8435,
+      "step": 8752
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.045404397693379e-07,
+      "loss": 2.6925,
+      "step": 8753
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.032681479557725e-07,
+      "loss": 3.4125,
+      "step": 8754
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.019968208366958e-07,
+      "loss": 2.5353,
+      "step": 8755
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.007264585454632e-07,
+      "loss": 2.4064,
+      "step": 8756
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.994570612153319e-07,
+      "loss": 1.9292,
+      "step": 8757
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.981886289794516e-07,
+      "loss": 2.1531,
+      "step": 8758
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.969211619708773e-07,
+      "loss": 2.7375,
+      "step": 8759
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.956546603225601e-07,
+      "loss": 3.9562,
+      "step": 8760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.943891241673507e-07,
+      "loss": 1.6192,
+      "step": 8761
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.931245536379973e-07,
+      "loss": 1.249,
+      "step": 8762
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.918609488671503e-07,
+      "loss": 1.9209,
+      "step": 8763
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.905983099873504e-07,
+      "loss": 2.5604,
+      "step": 8764
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.893366371310463e-07,
+      "loss": 2.9526,
+      "step": 8765
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.880759304305818e-07,
+      "loss": 3.1699,
+      "step": 8766
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.86816190018197e-07,
+      "loss": 2.7135,
+      "step": 8767
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.855574160260371e-07,
+      "loss": 0.7424,
+      "step": 8768
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.842996085861365e-07,
+      "loss": 1.2529,
+      "step": 8769
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.830427678304353e-07,
+      "loss": 1.9697,
+      "step": 8770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.817868938907691e-07,
+      "loss": 2.8919,
+      "step": 8771
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.805319868988759e-07,
+      "loss": 3.5352,
+      "step": 8772
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.792780469863903e-07,
+      "loss": 1.7454,
+      "step": 8773
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.780250742848416e-07,
+      "loss": 2.6351,
+      "step": 8774
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.767730689256614e-07,
+      "loss": 1.858,
+      "step": 8775
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.755220310401812e-07,
+      "loss": 1.7016,
+      "step": 8776
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.742719607596261e-07,
+      "loss": 2.5688,
+      "step": 8777
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.730228582151278e-07,
+      "loss": 2.091,
+      "step": 8778
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.717747235377048e-07,
+      "loss": 3.5794,
+      "step": 8779
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.705275568582848e-07,
+      "loss": 2.716,
+      "step": 8780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.692813583076885e-07,
+      "loss": 2.4026,
+      "step": 8781
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.680361280166359e-07,
+      "loss": 3.8205,
+      "step": 8782
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.667918661157492e-07,
+      "loss": 3.3922,
+      "step": 8783
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.655485727355416e-07,
+      "loss": 1.6635,
+      "step": 8784
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.643062480064301e-07,
+      "loss": 2.4855,
+      "step": 8785
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.63064892058728e-07,
+      "loss": 2.9149,
+      "step": 8786
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.618245050226503e-07,
+      "loss": 1.922,
+      "step": 8787
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.60585087028305e-07,
+      "loss": 2.5037,
+      "step": 8788
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.593466382057035e-07,
+      "loss": 3.1252,
+      "step": 8789
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.581091586847522e-07,
+      "loss": 2.0061,
+      "step": 8790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.568726485952571e-07,
+      "loss": 2.0612,
+      "step": 8791
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.556371080669222e-07,
+      "loss": 1.6441,
+      "step": 8792
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.544025372293528e-07,
+      "loss": 2.1045,
+      "step": 8793
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.531689362120442e-07,
+      "loss": 3.7662,
+      "step": 8794
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.519363051443996e-07,
+      "loss": 1.8079,
+      "step": 8795
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.507046441557142e-07,
+      "loss": 1.9206,
+      "step": 8796
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.494739533751849e-07,
+      "loss": 3.9516,
+      "step": 8797
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.482442329319062e-07,
+      "loss": 1.7167,
+      "step": 8798
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.470154829548671e-07,
+      "loss": 1.8263,
+      "step": 8799
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.457877035729588e-07,
+      "loss": 2.6115,
+      "step": 8800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.445608949149708e-07,
+      "loss": 3.8468,
+      "step": 8801
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.433350571095898e-07,
+      "loss": 1.7617,
+      "step": 8802
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.421101902854e-07,
+      "loss": 2.7326,
+      "step": 8803
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.408862945708839e-07,
+      "loss": 1.5868,
+      "step": 8804
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3966337009442e-07,
+      "loss": 1.8074,
+      "step": 8805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.384414169842924e-07,
+      "loss": 2.0819,
+      "step": 8806
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.372204353686751e-07,
+      "loss": 1.5306,
+      "step": 8807
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.360004253756459e-07,
+      "loss": 2.3702,
+      "step": 8808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.347813871331755e-07,
+      "loss": 2.7732,
+      "step": 8809
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.335633207691362e-07,
+      "loss": 2.8241,
+      "step": 8810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.32346226411298e-07,
+      "loss": 1.2454,
+      "step": 8811
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.311301041873276e-07,
+      "loss": 1.9424,
+      "step": 8812
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.299149542247929e-07,
+      "loss": 2.9317,
+      "step": 8813
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.287007766511534e-07,
+      "loss": 2.9038,
+      "step": 8814
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.274875715937746e-07,
+      "loss": 2.6692,
+      "step": 8815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.262753391799127e-07,
+      "loss": 4.6812,
+      "step": 8816
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.250640795367281e-07,
+      "loss": 1.9419,
+      "step": 8817
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.238537927912747e-07,
+      "loss": 2.5538,
+      "step": 8818
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.22644479070509e-07,
+      "loss": 3.1916,
+      "step": 8819
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.21436138501278e-07,
+      "loss": 2.5971,
+      "step": 8820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.202287712103329e-07,
+      "loss": 2.5286,
+      "step": 8821
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.190223773243199e-07,
+      "loss": 2.023,
+      "step": 8822
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.178169569697868e-07,
+      "loss": 2.3636,
+      "step": 8823
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.166125102731735e-07,
+      "loss": 1.9582,
+      "step": 8824
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.154090373608236e-07,
+      "loss": 2.9317,
+      "step": 8825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.142065383589736e-07,
+      "loss": 2.6726,
+      "step": 8826
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.130050133937617e-07,
+      "loss": 2.5164,
+      "step": 8827
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.118044625912213e-07,
+      "loss": 2.9917,
+      "step": 8828
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.106048860772863e-07,
+      "loss": 2.8909,
+      "step": 8829
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.094062839777838e-07,
+      "loss": 3.3769,
+      "step": 8830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.082086564184432e-07,
+      "loss": 2.7279,
+      "step": 8831
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.070120035248906e-07,
+      "loss": 2.6957,
+      "step": 8832
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.05816325422648e-07,
+      "loss": 2.8606,
+      "step": 8833
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.046216222371393e-07,
+      "loss": 2.9394,
+      "step": 8834
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.03427894093679e-07,
+      "loss": 2.2668,
+      "step": 8835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.022351411174866e-07,
+      "loss": 3.0242,
+      "step": 8836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.010433634336755e-07,
+      "loss": 3.0247,
+      "step": 8837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.99852561167258e-07,
+      "loss": 2.0786,
+      "step": 8838
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.98662734443144e-07,
+      "loss": 2.0901,
+      "step": 8839
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.974738833861383e-07,
+      "loss": 2.1984,
+      "step": 8840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.962860081209489e-07,
+      "loss": 2.3733,
+      "step": 8841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.950991087721781e-07,
+      "loss": 2.0898,
+      "step": 8842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.939131854643244e-07,
+      "loss": 1.0768,
+      "step": 8843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.927282383217893e-07,
+      "loss": 2.33,
+      "step": 8844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.915442674688633e-07,
+      "loss": 2.3633,
+      "step": 8845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.903612730297416e-07,
+      "loss": 2.8968,
+      "step": 8846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.891792551285148e-07,
+      "loss": 2.4063,
+      "step": 8847
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.879982138891717e-07,
+      "loss": 1.5046,
+      "step": 8848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.868181494355996e-07,
+      "loss": 2.0459,
+      "step": 8849
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.856390618915775e-07,
+      "loss": 1.988,
+      "step": 8850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.844609513807898e-07,
+      "loss": 2.2678,
+      "step": 8851
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.83283818026812e-07,
+      "loss": 2.7073,
+      "step": 8852
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.82107661953122e-07,
+      "loss": 1.9354,
+      "step": 8853
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.809324832830944e-07,
+      "loss": 3.0339,
+      "step": 8854
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.797582821399973e-07,
+      "loss": 3.177,
+      "step": 8855
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.785850586469989e-07,
+      "loss": 2.7153,
+      "step": 8856
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.774128129271651e-07,
+      "loss": 2.6033,
+      "step": 8857
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.76241545103461e-07,
+      "loss": 2.4646,
+      "step": 8858
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.75071255298746e-07,
+      "loss": 2.1911,
+      "step": 8859
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.739019436357774e-07,
+      "loss": 3.017,
+      "step": 8860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.727336102372106e-07,
+      "loss": 1.9031,
+      "step": 8861
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.715662552255998e-07,
+      "loss": 2.3567,
+      "step": 8862
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.703998787233946e-07,
+      "loss": 2.3868,
+      "step": 8863
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.692344808529427e-07,
+      "loss": 3.5065,
+      "step": 8864
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.680700617364877e-07,
+      "loss": 3.0638,
+      "step": 8865
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.669066214961717e-07,
+      "loss": 1.9255,
+      "step": 8866
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.657441602540348e-07,
+      "loss": 3.2299,
+      "step": 8867
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.645826781320141e-07,
+      "loss": 1.9354,
+      "step": 8868
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.634221752519454e-07,
+      "loss": 2.8215,
+      "step": 8869
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.622626517355557e-07,
+      "loss": 2.7081,
+      "step": 8870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.611041077044766e-07,
+      "loss": 2.6404,
+      "step": 8871
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.599465432802332e-07,
+      "loss": 2.8173,
+      "step": 8872
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.587899585842494e-07,
+      "loss": 2.6263,
+      "step": 8873
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.57634353737846e-07,
+      "loss": 2.9406,
+      "step": 8874
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.564797288622371e-07,
+      "loss": 1.2063,
+      "step": 8875
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.553260840785414e-07,
+      "loss": 2.2733,
+      "step": 8876
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.541734195077687e-07,
+      "loss": 3.0811,
+      "step": 8877
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.530217352708301e-07,
+      "loss": 2.7993,
+      "step": 8878
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.518710314885302e-07,
+      "loss": 2.6008,
+      "step": 8879
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.507213082815745e-07,
+      "loss": 2.1888,
+      "step": 8880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.495725657705609e-07,
+      "loss": 2.5344,
+      "step": 8881
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.484248040759888e-07,
+      "loss": 2.9539,
+      "step": 8882
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.47278023318253e-07,
+      "loss": 2.3492,
+      "step": 8883
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.461322236176438e-07,
+      "loss": 3.4275,
+      "step": 8884
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.449874050943549e-07,
+      "loss": 1.8395,
+      "step": 8885
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.438435678684662e-07,
+      "loss": 3.1623,
+      "step": 8886
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.427007120599649e-07,
+      "loss": 2.2289,
+      "step": 8887
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.415588377887305e-07,
+      "loss": 2.5336,
+      "step": 8888
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.404179451745396e-07,
+      "loss": 3.4369,
+      "step": 8889
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.392780343370686e-07,
+      "loss": 1.0843,
+      "step": 8890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.381391053958863e-07,
+      "loss": 3.5888,
+      "step": 8891
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.370011584704617e-07,
+      "loss": 4.1655,
+      "step": 8892
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.358641936801613e-07,
+      "loss": 3.2375,
+      "step": 8893
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.347282111442455e-07,
+      "loss": 2.7996,
+      "step": 8894
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.335932109818754e-07,
+      "loss": 1.3614,
+      "step": 8895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.324591933121072e-07,
+      "loss": 2.3491,
+      "step": 8896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.313261582538932e-07,
+      "loss": 2.4596,
+      "step": 8897
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.30194105926083e-07,
+      "loss": 3.094,
+      "step": 8898
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.290630364474248e-07,
+      "loss": 2.0856,
+      "step": 8899
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.279329499365649e-07,
+      "loss": 2.035,
+      "step": 8900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.268038465120386e-07,
+      "loss": 1.094,
+      "step": 8901
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.256757262922863e-07,
+      "loss": 1.841,
+      "step": 8902
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.245485893956438e-07,
+      "loss": 3.0465,
+      "step": 8903
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.234224359403407e-07,
+      "loss": 3.358,
+      "step": 8904
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.222972660445082e-07,
+      "loss": 2.3158,
+      "step": 8905
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.211730798261673e-07,
+      "loss": 1.4361,
+      "step": 8906
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.200498774032415e-07,
+      "loss": 3.6156,
+      "step": 8907
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.1892765889355e-07,
+      "loss": 1.9671,
+      "step": 8908
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.178064244148085e-07,
+      "loss": 2.0036,
+      "step": 8909
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.166861740846297e-07,
+      "loss": 2.968,
+      "step": 8910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.155669080205206e-07,
+      "loss": 1.3549,
+      "step": 8911
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.144486263398886e-07,
+      "loss": 2.1461,
+      "step": 8912
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.133313291600362e-07,
+      "loss": 3.1782,
+      "step": 8913
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.122150165981633e-07,
+      "loss": 2.5623,
+      "step": 8914
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.11099688771366e-07,
+      "loss": 1.8506,
+      "step": 8915
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.099853457966342e-07,
+      "loss": 1.4922,
+      "step": 8916
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.088719877908589e-07,
+      "loss": 2.8818,
+      "step": 8917
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.077596148708275e-07,
+      "loss": 2.5754,
+      "step": 8918
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.066482271532215e-07,
+      "loss": 3.3953,
+      "step": 8919
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.055378247546217e-07,
+      "loss": 1.4542,
+      "step": 8920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.044284077915019e-07,
+      "loss": 2.9758,
+      "step": 8921
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.033199763802356e-07,
+      "loss": 2.3203,
+      "step": 8922
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.022125306370929e-07,
+      "loss": 2.4257,
+      "step": 8923
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.01106070678239e-07,
+      "loss": 2.5216,
+      "step": 8924
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.000005966197387e-07,
+      "loss": 2.4834,
+      "step": 8925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.988961085775468e-07,
+      "loss": 1.8947,
+      "step": 8926
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.97792606667521e-07,
+      "loss": 2.3958,
+      "step": 8927
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.966900910054141e-07,
+      "loss": 2.7225,
+      "step": 8928
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.955885617068758e-07,
+      "loss": 2.2069,
+      "step": 8929
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.94488018887448e-07,
+      "loss": 2.7326,
+      "step": 8930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.933884626625764e-07,
+      "loss": 3.4714,
+      "step": 8931
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.922898931475973e-07,
+      "loss": 2.7849,
+      "step": 8932
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.911923104577455e-07,
+      "loss": 2.9453,
+      "step": 8933
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.90095714708152e-07,
+      "loss": 1.5551,
+      "step": 8934
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.890001060138484e-07,
+      "loss": 1.2179,
+      "step": 8935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.879054844897536e-07,
+      "loss": 2.7303,
+      "step": 8936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.868118502506903e-07,
+      "loss": 2.7032,
+      "step": 8937
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.857192034113757e-07,
+      "loss": 3.7228,
+      "step": 8938
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.846275440864246e-07,
+      "loss": 2.6554,
+      "step": 8939
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.835368723903456e-07,
+      "loss": 1.5915,
+      "step": 8940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.824471884375471e-07,
+      "loss": 2.8805,
+      "step": 8941
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.813584923423287e-07,
+      "loss": 3.4961,
+      "step": 8942
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.802707842188915e-07,
+      "loss": 3.4961,
+      "step": 8943
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.791840641813295e-07,
+      "loss": 3.0169,
+      "step": 8944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.780983323436374e-07,
+      "loss": 2.757,
+      "step": 8945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.770135888197004e-07,
+      "loss": 2.57,
+      "step": 8946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.759298337233054e-07,
+      "loss": 2.7576,
+      "step": 8947
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.748470671681328e-07,
+      "loss": 1.2608,
+      "step": 8948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.737652892677592e-07,
+      "loss": 2.1721,
+      "step": 8949
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.726845001356573e-07,
+      "loss": 2.9284,
+      "step": 8950
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.716046998851999e-07,
+      "loss": 2.0776,
+      "step": 8951
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.705258886296494e-07,
+      "loss": 3.3054,
+      "step": 8952
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6944806648217e-07,
+      "loss": 1.9136,
+      "step": 8953
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.683712335558189e-07,
+      "loss": 2.8946,
+      "step": 8954
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.672953899635524e-07,
+      "loss": 2.6703,
+      "step": 8955
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.662205358182226e-07,
+      "loss": 1.779,
+      "step": 8956
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.651466712325737e-07,
+      "loss": 2.3145,
+      "step": 8957
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.640737963192511e-07,
+      "loss": 3.0264,
+      "step": 8958
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.630019111907925e-07,
+      "loss": 3.5398,
+      "step": 8959
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.619310159596358e-07,
+      "loss": 3.0643,
+      "step": 8960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.608611107381135e-07,
+      "loss": 2.8431,
+      "step": 8961
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5979219563845e-07,
+      "loss": 2.3649,
+      "step": 8962
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.587242707727725e-07,
+      "loss": 3.0132,
+      "step": 8963
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.576573362531001e-07,
+      "loss": 2.5897,
+      "step": 8964
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.565913921913513e-07,
+      "loss": 2.866,
+      "step": 8965
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.555264386993353e-07,
+      "loss": 2.9015,
+      "step": 8966
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.54462475888764e-07,
+      "loss": 2.7972,
+      "step": 8967
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.533995038712403e-07,
+      "loss": 3.5999,
+      "step": 8968
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.523375227582662e-07,
+      "loss": 2.2685,
+      "step": 8969
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.51276532661238e-07,
+      "loss": 1.2513,
+      "step": 8970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5021653369145e-07,
+      "loss": 1.7128,
+      "step": 8971
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.491575259600879e-07,
+      "loss": 1.9325,
+      "step": 8972
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.480995095782382e-07,
+      "loss": 2.6774,
+      "step": 8973
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.470424846568833e-07,
+      "loss": 1.9384,
+      "step": 8974
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.459864513068991e-07,
+      "loss": 3.6498,
+      "step": 8975
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.449314096390601e-07,
+      "loss": 4.1159,
+      "step": 8976
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.438773597640323e-07,
+      "loss": 2.9467,
+      "step": 8977
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.428243017923817e-07,
+      "loss": 2.653,
+      "step": 8978
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.417722358345712e-07,
+      "loss": 1.6739,
+      "step": 8979
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.407211620009545e-07,
+      "loss": 2.599,
+      "step": 8980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.396710804017868e-07,
+      "loss": 1.596,
+      "step": 8981
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.386219911472168e-07,
+      "loss": 2.3797,
+      "step": 8982
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.375738943472874e-07,
+      "loss": 2.0665,
+      "step": 8983
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.365267901119398e-07,
+      "loss": 1.9208,
+      "step": 8984
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.354806785510113e-07,
+      "loss": 2.6437,
+      "step": 8985
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.344355597742345e-07,
+      "loss": 2.8242,
+      "step": 8986
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.333914338912349e-07,
+      "loss": 2.2729,
+      "step": 8987
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.323483010115382e-07,
+      "loss": 2.0808,
+      "step": 8988
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.313061612445636e-07,
+      "loss": 2.4396,
+      "step": 8989
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.30265014699628e-07,
+      "loss": 2.4991,
+      "step": 8990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.292248614859419e-07,
+      "loss": 2.1408,
+      "step": 8991
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.281857017126124e-07,
+      "loss": 3.8052,
+      "step": 8992
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.271475354886424e-07,
+      "loss": 3.0438,
+      "step": 8993
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.261103629229314e-07,
+      "loss": 2.9988,
+      "step": 8994
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.250741841242735e-07,
+      "loss": 3.4832,
+      "step": 8995
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.240389992013606e-07,
+      "loss": 2.6404,
+      "step": 8996
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.230048082627748e-07,
+      "loss": 2.6684,
+      "step": 8997
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.219716114170026e-07,
+      "loss": 2.527,
+      "step": 8998
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.209394087724174e-07,
+      "loss": 3.204,
+      "step": 8999
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.199082004372958e-07,
+      "loss": 3.1911,
+      "step": 9000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.188779865198057e-07,
+      "loss": 1.595,
+      "step": 9001
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.178487671280108e-07,
+      "loss": 3.3,
+      "step": 9002
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.168205423698724e-07,
+      "loss": 2.4188,
+      "step": 9003
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.157933123532466e-07,
+      "loss": 4.4781,
+      "step": 9004
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.147670771858848e-07,
+      "loss": 3.2998,
+      "step": 9005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.137418369754343e-07,
+      "loss": 1.8452,
+      "step": 9006
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.127175918294392e-07,
+      "loss": 3.3715,
+      "step": 9007
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.116943418553355e-07,
+      "loss": 3.4695,
+      "step": 9008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.106720871604586e-07,
+      "loss": 3.2702,
+      "step": 9009
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.096508278520385e-07,
+      "loss": 1.0246,
+      "step": 9010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.086305640372002e-07,
+      "loss": 2.1487,
+      "step": 9011
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.076112958229673e-07,
+      "loss": 3.3375,
+      "step": 9012
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.06593023316252e-07,
+      "loss": 2.8176,
+      "step": 9013
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.055757466238676e-07,
+      "loss": 3.0588,
+      "step": 9014
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.045594658525232e-07,
+      "loss": 1.9317,
+      "step": 9015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.035441811088204e-07,
+      "loss": 2.2815,
+      "step": 9016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.025298924992594e-07,
+      "loss": 2.8662,
+      "step": 9017
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.01516600130233e-07,
+      "loss": 3.3361,
+      "step": 9018
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.005043041080327e-07,
+      "loss": 2.4797,
+      "step": 9019
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.994930045388414e-07,
+      "loss": 2.3502,
+      "step": 9020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.984827015287397e-07,
+      "loss": 3.5397,
+      "step": 9021
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.974733951837074e-07,
+      "loss": 1.7033,
+      "step": 9022
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.964650856096121e-07,
+      "loss": 2.7122,
+      "step": 9023
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.954577729122212e-07,
+      "loss": 2.4849,
+      "step": 9024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.944514571971981e-07,
+      "loss": 3.848,
+      "step": 9025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.934461385701006e-07,
+      "loss": 3.1392,
+      "step": 9026
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.924418171363843e-07,
+      "loss": 1.1595,
+      "step": 9027
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.914384930013927e-07,
+      "loss": 2.5394,
+      "step": 9028
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.904361662703739e-07,
+      "loss": 2.048,
+      "step": 9029
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 2.8486,
+      "step": 9030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.884345054407014e-07,
+      "loss": 4.0605,
+      "step": 9031
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.874351715520154e-07,
+      "loss": 2.3857,
+      "step": 9032
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.864368354872306e-07,
+      "loss": 1.2492,
+      "step": 9033
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.854394973510667e-07,
+      "loss": 2.3515,
+      "step": 9034
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.844431572481412e-07,
+      "loss": 2.8057,
+      "step": 9035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.834478152829658e-07,
+      "loss": 1.5306,
+      "step": 9036
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.824534715599461e-07,
+      "loss": 2.1475,
+      "step": 9037
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.814601261833851e-07,
+      "loss": 2.625,
+      "step": 9038
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.804677792574808e-07,
+      "loss": 2.0359,
+      "step": 9039
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.794764308863242e-07,
+      "loss": 2.9324,
+      "step": 9040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.784860811739045e-07,
+      "loss": 2.9273,
+      "step": 9041
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.774967302241051e-07,
+      "loss": 2.7349,
+      "step": 9042
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7650837814070203e-07,
+      "loss": 2.1176,
+      "step": 9043
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.755210250273701e-07,
+      "loss": 3.0307,
+      "step": 9044
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.745346709876786e-07,
+      "loss": 2.5955,
+      "step": 9045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.735493161250915e-07,
+      "loss": 2.1549,
+      "step": 9046
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7256496054296827e-07,
+      "loss": 2.3248,
+      "step": 9047
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.715816043445609e-07,
+      "loss": 2.1787,
+      "step": 9048
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7059924763302013e-07,
+      "loss": 2.9009,
+      "step": 9049
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.696178905113913e-07,
+      "loss": 2.6706,
+      "step": 9050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.686375330826143e-07,
+      "loss": 3.3886,
+      "step": 9051
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.676581754495235e-07,
+      "loss": 2.5797,
+      "step": 9052
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6667981771484903e-07,
+      "loss": 2.4208,
+      "step": 9053
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6570245998121654e-07,
+      "loss": 2.4226,
+      "step": 9054
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6472610235114513e-07,
+      "loss": 3.3596,
+      "step": 9055
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6375074492705173e-07,
+      "loss": 2.5388,
+      "step": 9056
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.627763878112479e-07,
+      "loss": 3.5322,
+      "step": 9057
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.618030311059352e-07,
+      "loss": 1.5586,
+      "step": 9058
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6083067491321633e-07,
+      "loss": 2.9979,
+      "step": 9059
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5985931933508757e-07,
+      "loss": 3.1893,
+      "step": 9060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5888896447343843e-07,
+      "loss": 2.9597,
+      "step": 9061
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.579196104300554e-07,
+      "loss": 3.1844,
+      "step": 9062
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.569512573066204e-07,
+      "loss": 2.5757,
+      "step": 9063
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.559839052047066e-07,
+      "loss": 1.8808,
+      "step": 9064
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.550175542257862e-07,
+      "loss": 2.8171,
+      "step": 9065
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.540522044712248e-07,
+      "loss": 2.1895,
+      "step": 9066
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5308785604228245e-07,
+      "loss": 3.3133,
+      "step": 9067
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.521245090401172e-07,
+      "loss": 1.7234,
+      "step": 9068
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.511621635657759e-07,
+      "loss": 1.1927,
+      "step": 9069
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.502008197202068e-07,
+      "loss": 3.4992,
+      "step": 9070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.492404776042503e-07,
+      "loss": 2.9384,
+      "step": 9071
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.482811373186402e-07,
+      "loss": 3.2133,
+      "step": 9072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.473227989640083e-07,
+      "loss": 1.848,
+      "step": 9073
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.463654626408798e-07,
+      "loss": 1.8768,
+      "step": 9074
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.454091284496731e-07,
+      "loss": 1.954,
+      "step": 9075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4445379649070587e-07,
+      "loss": 2.0098,
+      "step": 9076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.434994668641868e-07,
+      "loss": 2.821,
+      "step": 9077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.425461396702213e-07,
+      "loss": 3.2787,
+      "step": 9078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.415938150088073e-07,
+      "loss": 2.1761,
+      "step": 9079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.406424929798403e-07,
+      "loss": 2.0116,
+      "step": 9080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3969217368310834e-07,
+      "loss": 2.8563,
+      "step": 9081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3874285721829835e-07,
+      "loss": 4.2598,
+      "step": 9082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.377945436849873e-07,
+      "loss": 1.9631,
+      "step": 9083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.368472331826479e-07,
+      "loss": 3.5848,
+      "step": 9084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3590092581065055e-07,
+      "loss": 2.4318,
+      "step": 9085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3495562166825714e-07,
+      "loss": 2.4557,
+      "step": 9086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3401132085462705e-07,
+      "loss": 2.8324,
+      "step": 9087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3306802346881116e-07,
+      "loss": 4.1121,
+      "step": 9088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.321257296097592e-07,
+      "loss": 1.5513,
+      "step": 9089
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3118443937631094e-07,
+      "loss": 2.7931,
+      "step": 9090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3024415286720635e-07,
+      "loss": 3.3243,
+      "step": 9091
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2930487018107425e-07,
+      "loss": 2.3183,
+      "step": 9092
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.283665914164437e-07,
+      "loss": 1.5249,
+      "step": 9093
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.274293166717336e-07,
+      "loss": 2.7621,
+      "step": 9094
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.26493046045261e-07,
+      "loss": 3.0802,
+      "step": 9095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2555777963523506e-07,
+      "loss": 2.6411,
+      "step": 9096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2462351753976174e-07,
+      "loss": 2.2868,
+      "step": 9097
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2369025985684264e-07,
+      "loss": 3.6394,
+      "step": 9098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.227580066843684e-07,
+      "loss": 3.0473,
+      "step": 9099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.218267581201296e-07,
+      "loss": 3.3179,
+      "step": 9100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2089651426181044e-07,
+      "loss": 1.9312,
+      "step": 9101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.199672752069894e-07,
+      "loss": 2.6901,
+      "step": 9102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1903904105313976e-07,
+      "loss": 1.8816,
+      "step": 9103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1811181189762684e-07,
+      "loss": 2.0733,
+      "step": 9104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.17185587837714e-07,
+      "loss": 3.5403,
+      "step": 9105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.16260368970558e-07,
+      "loss": 3.1495,
+      "step": 9106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1533615539320893e-07,
+      "loss": 1.5874,
+      "step": 9107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1441294720261373e-07,
+      "loss": 2.5602,
+      "step": 9108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1349074449561267e-07,
+      "loss": 3.1816,
+      "step": 9109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.125695473689406e-07,
+      "loss": 3.3047,
+      "step": 9110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1164935591922583e-07,
+      "loss": 2.1375,
+      "step": 9111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.107301702429922e-07,
+      "loss": 2.2965,
+      "step": 9112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.098119904366604e-07,
+      "loss": 2.8937,
+      "step": 9113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0889481659654006e-07,
+      "loss": 2.9084,
+      "step": 9114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0797864881883977e-07,
+      "loss": 2.6822,
+      "step": 9115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.070634871996615e-07,
+      "loss": 2.828,
+      "step": 9116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.061493318350007e-07,
+      "loss": 2.3113,
+      "step": 9117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0523618282074964e-07,
+      "loss": 3.2706,
+      "step": 9118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.043240402526916e-07,
+      "loss": 2.8007,
+      "step": 9119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.034129042265067e-07,
+      "loss": 2.5203,
+      "step": 9120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0250277483776854e-07,
+      "loss": 1.92,
+      "step": 9121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0159365218194633e-07,
+      "loss": 1.8058,
+      "step": 9122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0068553635440264e-07,
+      "loss": 1.8421,
+      "step": 9123
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9977842745039464e-07,
+      "loss": 2.8706,
+      "step": 9124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.988723255650728e-07,
+      "loss": 2.9058,
+      "step": 9125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9796723079348455e-07,
+      "loss": 1.7108,
+      "step": 9126
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9706314323056936e-07,
+      "loss": 2.6288,
+      "step": 9127
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.961600629711615e-07,
+      "loss": 1.9449,
+      "step": 9128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9525799010999064e-07,
+      "loss": 3.3974,
+      "step": 9129
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.943569247416801e-07,
+      "loss": 2.1358,
+      "step": 9130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9345686696074526e-07,
+      "loss": 2.9311,
+      "step": 9131
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.925578168616007e-07,
+      "loss": 3.054,
+      "step": 9132
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.91659774538552e-07,
+      "loss": 2.5622,
+      "step": 9133
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.907627400857994e-07,
+      "loss": 1.9172,
+      "step": 9134
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8986671359743767e-07,
+      "loss": 1.9575,
+      "step": 9135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.889716951674549e-07,
+      "loss": 3.1217,
+      "step": 9136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.88077684889735e-07,
+      "loss": 2.7793,
+      "step": 9137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.87184682858055e-07,
+      "loss": 3.8083,
+      "step": 9138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.862926891660901e-07,
+      "loss": 3.2201,
+      "step": 9139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8540170390740097e-07,
+      "loss": 2.6654,
+      "step": 9140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.845117271754517e-07,
+      "loss": 2.8398,
+      "step": 9141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.836227590635944e-07,
+      "loss": 2.1222,
+      "step": 9142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8273479966507876e-07,
+      "loss": 2.8701,
+      "step": 9143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8184784907304704e-07,
+      "loss": 2.2729,
+      "step": 9144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8096190738053815e-07,
+      "loss": 3.9804,
+      "step": 9145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.800769746804811e-07,
+      "loss": 2.8652,
+      "step": 9146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.791930510657016e-07,
+      "loss": 2.5032,
+      "step": 9147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.783101366289199e-07,
+      "loss": 2.3507,
+      "step": 9148
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7742823146274977e-07,
+      "loss": 1.4373,
+      "step": 9149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7654733565969826e-07,
+      "loss": 2.1358,
+      "step": 9150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7566744931216704e-07,
+      "loss": 3.313,
+      "step": 9151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.747885725124523e-07,
+      "loss": 2.9633,
+      "step": 9152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7391070535274465e-07,
+      "loss": 2.7094,
+      "step": 9153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.730338479251283e-07,
+      "loss": 1.8295,
+      "step": 9154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.721580003215808e-07,
+      "loss": 3.0503,
+      "step": 9155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.712831626339752e-07,
+      "loss": 3.2211,
+      "step": 9156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7040933495407606e-07,
+      "loss": 3.1949,
+      "step": 9157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.695365173735466e-07,
+      "loss": 3.0278,
+      "step": 9158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.686647099839402e-07,
+      "loss": 2.042,
+      "step": 9159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.67793912876705e-07,
+      "loss": 3.6915,
+      "step": 9160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.669241261431833e-07,
+      "loss": 3.5715,
+      "step": 9161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6605534987461335e-07,
+      "loss": 2.6184,
+      "step": 9162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.651875841621244e-07,
+      "loss": 2.8,
+      "step": 9163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.643208290967415e-07,
+      "loss": 1.3784,
+      "step": 9164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6345508476938296e-07,
+      "loss": 4.5114,
+      "step": 9165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6259035127086173e-07,
+      "loss": 2.5794,
+      "step": 9166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6172662869188303e-07,
+      "loss": 1.5102,
+      "step": 9167
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.608639171230488e-07,
+      "loss": 2.8933,
+      "step": 9168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6000221665485445e-07,
+      "loss": 3.376,
+      "step": 9169
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.591415273776855e-07,
+      "loss": 2.7182,
+      "step": 9170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.582818493818252e-07,
+      "loss": 2.2741,
+      "step": 9171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5742318275745147e-07,
+      "loss": 2.5068,
+      "step": 9172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5656552759463227e-07,
+      "loss": 0.8941,
+      "step": 9173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.557088839833345e-07,
+      "loss": 1.9166,
+      "step": 9174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.548532520134129e-07,
+      "loss": 2.2303,
+      "step": 9175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5399863177462024e-07,
+      "loss": 2.2542,
+      "step": 9176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5314502335660363e-07,
+      "loss": 2.2677,
+      "step": 9177
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.522924268489003e-07,
+      "loss": 1.8176,
+      "step": 9178
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5144084234094544e-07,
+      "loss": 1.8886,
+      "step": 9179
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5059026992206645e-07,
+      "loss": 2.2123,
+      "step": 9180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.497407096814842e-07,
+      "loss": 2.9841,
+      "step": 9181
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4889216170831187e-07,
+      "loss": 1.4604,
+      "step": 9182
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4804462609156043e-07,
+      "loss": 2.3212,
+      "step": 9183
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4719810292013214e-07,
+      "loss": 2.2223,
+      "step": 9184
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4635259228282256e-07,
+      "loss": 2.4465,
+      "step": 9185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.45508094268322e-07,
+      "loss": 2.9056,
+      "step": 9186
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.446646089652139e-07,
+      "loss": 3.545,
+      "step": 9187
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.438221364619776e-07,
+      "loss": 2.3592,
+      "step": 9188
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.429806768469823e-07,
+      "loss": 3.6694,
+      "step": 9189
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.421402302084953e-07,
+      "loss": 3.7049,
+      "step": 9190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4130079663467485e-07,
+      "loss": 2.8309,
+      "step": 9191
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.404623762135728e-07,
+      "loss": 3.7325,
+      "step": 9192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3962496903313544e-07,
+      "loss": 3.1591,
+      "step": 9193
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.387885751812048e-07,
+      "loss": 2.6473,
+      "step": 9194
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.379531947455128e-07,
+      "loss": 2.2893,
+      "step": 9195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.371188278136883e-07,
+      "loss": 2.6722,
+      "step": 9196
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3628547447325133e-07,
+      "loss": 3.3729,
+      "step": 9197
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3545313481161743e-07,
+      "loss": 2.4977,
+      "step": 9198
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3462180891609573e-07,
+      "loss": 2.9458,
+      "step": 9199
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3379149687388866e-07,
+      "loss": 1.9686,
+      "step": 9200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.329621987720899e-07,
+      "loss": 2.5028,
+      "step": 9201
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.32133914697691e-07,
+      "loss": 2.7802,
+      "step": 9202
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3130664473757344e-07,
+      "loss": 3.1984,
+      "step": 9203
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3048038897851576e-07,
+      "loss": 2.5069,
+      "step": 9204
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2965514750718964e-07,
+      "loss": 3.677,
+      "step": 9205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2883092041015476e-07,
+      "loss": 1.88,
+      "step": 9206
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.280077077738708e-07,
+      "loss": 2.1507,
+      "step": 9207
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.271855096846899e-07,
+      "loss": 2.037,
+      "step": 9208
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2636432622885627e-07,
+      "loss": 1.7144,
+      "step": 9209
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.255441574925089e-07,
+      "loss": 2.1453,
+      "step": 9210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.247250035616778e-07,
+      "loss": 2.8933,
+      "step": 9211
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2390686452228983e-07,
+      "loss": 2.4864,
+      "step": 9212
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2308974046016296e-07,
+      "loss": 2.5966,
+      "step": 9213
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.222736314610098e-07,
+      "loss": 1.6138,
+      "step": 9214
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2145853761043844e-07,
+      "loss": 3.9337,
+      "step": 9215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2064445899394723e-07,
+      "loss": 3.2392,
+      "step": 9216
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.198313956969279e-07,
+      "loss": 2.934,
+      "step": 9217
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.190193478046677e-07,
+      "loss": 2.3924,
+      "step": 9218
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1820831540234856e-07,
+      "loss": 2.4614,
+      "step": 9219
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1739829857504235e-07,
+      "loss": 2.4061,
+      "step": 9220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.165892974077156e-07,
+      "loss": 1.9347,
+      "step": 9221
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.157813119852304e-07,
+      "loss": 2.2637,
+      "step": 9222
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1497434239233904e-07,
+      "loss": 1.8006,
+      "step": 9223
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1416838871368925e-07,
+      "loss": 1.9416,
+      "step": 9224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.133634510338235e-07,
+      "loss": 2.1709,
+      "step": 9225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1255952943717305e-07,
+      "loss": 3.5793,
+      "step": 9226
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1175662400806716e-07,
+      "loss": 2.1434,
+      "step": 9227
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1095473483072733e-07,
+      "loss": 2.2764,
+      "step": 9228
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1015386198926856e-07,
+      "loss": 3.116,
+      "step": 9229
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.093540055676958e-07,
+      "loss": 2.7151,
+      "step": 9230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0855516564991307e-07,
+      "loss": 2.211,
+      "step": 9231
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0775734231971443e-07,
+      "loss": 1.8341,
+      "step": 9232
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.069605356607874e-07,
+      "loss": 1.328,
+      "step": 9233
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0616474575671297e-07,
+      "loss": 2.0909,
+      "step": 9234
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.053699726909676e-07,
+      "loss": 1.7497,
+      "step": 9235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.045762165469168e-07,
+      "loss": 3.1919,
+      "step": 9236
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.03783477407823e-07,
+      "loss": 2.0546,
+      "step": 9237
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.029917553568407e-07,
+      "loss": 3.0961,
+      "step": 9238
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0220105047701787e-07,
+      "loss": 2.8179,
+      "step": 9239
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0141136285129825e-07,
+      "loss": 3.1506,
+      "step": 9240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.006226925625122e-07,
+      "loss": 1.2437,
+      "step": 9241
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9983503969338913e-07,
+      "loss": 2.5616,
+      "step": 9242
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.990484043265507e-07,
+      "loss": 2.5771,
+      "step": 9243
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.982627865445109e-07,
+      "loss": 1.6886,
+      "step": 9244
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.974781864296783e-07,
+      "loss": 2.3108,
+      "step": 9245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.966946040643548e-07,
+      "loss": 3.1147,
+      "step": 9246
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9591203953073024e-07,
+      "loss": 1.8157,
+      "step": 9247
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.951304929108956e-07,
+      "loss": 2.5522,
+      "step": 9248
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.943499642868297e-07,
+      "loss": 3.5922,
+      "step": 9249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.935704537404083e-07,
+      "loss": 3.3622,
+      "step": 9250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9279196135339806e-07,
+      "loss": 2.1985,
+      "step": 9251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9201448720745706e-07,
+      "loss": 2.7846,
+      "step": 9252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.912380313841412e-07,
+      "loss": 2.8497,
+      "step": 9253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.904625939648953e-07,
+      "loss": 3.2117,
+      "step": 9254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8968817503105984e-07,
+      "loss": 1.9102,
+      "step": 9255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8891477466386987e-07,
+      "loss": 2.5939,
+      "step": 9256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.881423929444471e-07,
+      "loss": 1.3771,
+      "step": 9257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.873710299538146e-07,
+      "loss": 2.5187,
+      "step": 9258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.86600685772882e-07,
+      "loss": 1.7366,
+      "step": 9259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8583136048245697e-07,
+      "loss": 3.1924,
+      "step": 9260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8506305416323823e-07,
+      "loss": 3.4032,
+      "step": 9261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.842957668958157e-07,
+      "loss": 2.8855,
+      "step": 9262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8352949876067516e-07,
+      "loss": 1.6932,
+      "step": 9263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.827642498381955e-07,
+      "loss": 2.9595,
+      "step": 9264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.820000202086459e-07,
+      "loss": 1.8736,
+      "step": 9265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8123680995219226e-07,
+      "loss": 3.5081,
+      "step": 9266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.804746191488905e-07,
+      "loss": 1.5072,
+      "step": 9267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7971344787869114e-07,
+      "loss": 2.1947,
+      "step": 9268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7895329622143917e-07,
+      "loss": 1.6497,
+      "step": 9269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.781941642568686e-07,
+      "loss": 2.8581,
+      "step": 9270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.774360520646113e-07,
+      "loss": 3.1428,
+      "step": 9271
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.76678959724187e-07,
+      "loss": 3.3308,
+      "step": 9272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.759228873150122e-07,
+      "loss": 2.3502,
+      "step": 9273
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7516783491639666e-07,
+      "loss": 2.2898,
+      "step": 9274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.744138026075405e-07,
+      "loss": 2.8999,
+      "step": 9275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7366079046753925e-07,
+      "loss": 2.3144,
+      "step": 9276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.729087985753798e-07,
+      "loss": 1.3581,
+      "step": 9277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.721578270099412e-07,
+      "loss": 2.4045,
+      "step": 9278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.714078758499994e-07,
+      "loss": 3.2851,
+      "step": 9279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.706589451742181e-07,
+      "loss": 2.9651,
+      "step": 9280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6991103506116e-07,
+      "loss": 2.092,
+      "step": 9281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6916414558927575e-07,
+      "loss": 1.6706,
+      "step": 9282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.684182768369081e-07,
+      "loss": 3.7512,
+      "step": 9283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6767342888229907e-07,
+      "loss": 3.2831,
+      "step": 9284
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.669296018035772e-07,
+      "loss": 2.8517,
+      "step": 9285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6618679567877006e-07,
+      "loss": 2.7253,
+      "step": 9286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6544501058578976e-07,
+      "loss": 2.4888,
+      "step": 9287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.647042466024485e-07,
+      "loss": 3.9761,
+      "step": 9288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6396450380644976e-07,
+      "loss": 1.4087,
+      "step": 9289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.632257822753881e-07,
+      "loss": 1.9207,
+      "step": 9290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.624880820867526e-07,
+      "loss": 3.4537,
+      "step": 9291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.617514033179236e-07,
+      "loss": 1.9805,
+      "step": 9292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6101574604617485e-07,
+      "loss": 2.4098,
+      "step": 9293
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6028111034867466e-07,
+      "loss": 3.4322,
+      "step": 9294
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5954749630248355e-07,
+      "loss": 3.1509,
+      "step": 9295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.588149039845533e-07,
+      "loss": 3.0528,
+      "step": 9296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5808333347172807e-07,
+      "loss": 1.9182,
+      "step": 9297
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5735278484074865e-07,
+      "loss": 3.2413,
+      "step": 9298
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.566232581682449e-07,
+      "loss": 2.7492,
+      "step": 9299
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5589475353073987e-07,
+      "loss": 3.452,
+      "step": 9300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.551672710046527e-07,
+      "loss": 3.1976,
+      "step": 9301
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.54440810666291e-07,
+      "loss": 3.2649,
+      "step": 9302
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.537153725918573e-07,
+      "loss": 2.5215,
+      "step": 9303
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5299095685744734e-07,
+      "loss": 3.0487,
+      "step": 9304
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5226756353904925e-07,
+      "loss": 3.7416,
+      "step": 9305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.515451927125423e-07,
+      "loss": 2.2519,
+      "step": 9306
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5082384445370254e-07,
+      "loss": 2.7138,
+      "step": 9307
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5010351883819283e-07,
+      "loss": 2.1398,
+      "step": 9308
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.493842159415738e-07,
+      "loss": 3.6505,
+      "step": 9309
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.486659358392951e-07,
+      "loss": 2.0561,
+      "step": 9310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4794867860670425e-07,
+      "loss": 2.9533,
+      "step": 9311
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.472324443190355e-07,
+      "loss": 3.1947,
+      "step": 9312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4651723305141986e-07,
+      "loss": 2.8308,
+      "step": 9313
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.458030448788784e-07,
+      "loss": 2.0547,
+      "step": 9314
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.450898798763268e-07,
+      "loss": 2.5557,
+      "step": 9315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4437773811857304e-07,
+      "loss": 3.0242,
+      "step": 9316
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4366661968031836e-07,
+      "loss": 2.9082,
+      "step": 9317
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.429565246361532e-07,
+      "loss": 3.0785,
+      "step": 9318
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.422474530605656e-07,
+      "loss": 2.3195,
+      "step": 9319
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4153940502793185e-07,
+      "loss": 2.7138,
+      "step": 9320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4083238061252565e-07,
+      "loss": 3.1644,
+      "step": 9321
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4012637988850894e-07,
+      "loss": 2.1232,
+      "step": 9322
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.39421402929938e-07,
+      "loss": 3.1424,
+      "step": 9323
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.387174498107614e-07,
+      "loss": 3.38,
+      "step": 9324
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.380145206048201e-07,
+      "loss": 2.7083,
+      "step": 9325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.373126153858496e-07,
+      "loss": 2.1024,
+      "step": 9326
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.366117342274754e-07,
+      "loss": 2.4012,
+      "step": 9327
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.359118772032176e-07,
+      "loss": 3.0724,
+      "step": 9328
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.352130443864864e-07,
+      "loss": 3.0638,
+      "step": 9329
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3451523585058756e-07,
+      "loss": 1.3128,
+      "step": 9330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3381845166871698e-07,
+      "loss": 3.3665,
+      "step": 9331
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3312269191396619e-07,
+      "loss": 1.9603,
+      "step": 9332
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.324279566593135e-07,
+      "loss": 2.2478,
+      "step": 9333
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3173424597763615e-07,
+      "loss": 2.423,
+      "step": 9334
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3104155994170042e-07,
+      "loss": 2.7508,
+      "step": 9335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.30349898624167e-07,
+      "loss": 3.1744,
+      "step": 9336
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2965926209758572e-07,
+      "loss": 1.6866,
+      "step": 9337
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.289696504344019e-07,
+      "loss": 2.5867,
+      "step": 9338
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2828106370695325e-07,
+      "loss": 3.7536,
+      "step": 9339
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2759350198746978e-07,
+      "loss": 2.0975,
+      "step": 9340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2690696534807267e-07,
+      "loss": 2.5739,
+      "step": 9341
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2622145386077653e-07,
+      "loss": 2.7025,
+      "step": 9342
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2553696759748834e-07,
+      "loss": 3.7709,
+      "step": 9343
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2485350663000727e-07,
+      "loss": 2.1758,
+      "step": 9344
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.24171071030026e-07,
+      "loss": 1.9857,
+      "step": 9345
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2348966086912725e-07,
+      "loss": 2.7497,
+      "step": 9346
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2280927621879056e-07,
+      "loss": 3.2909,
+      "step": 9347
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2212991715038324e-07,
+      "loss": 2.1754,
+      "step": 9348
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2145158373516606e-07,
+      "loss": 2.5836,
+      "step": 9349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2077427604429435e-07,
+      "loss": 3.0645,
+      "step": 9350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.200979941488146e-07,
+      "loss": 2.0993,
+      "step": 9351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1942273811966563e-07,
+      "loss": 2.8092,
+      "step": 9352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1874850802767634e-07,
+      "loss": 2.5493,
+      "step": 9353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1807530394357457e-07,
+      "loss": 2.7227,
+      "step": 9354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1740312593797274e-07,
+      "loss": 2.7799,
+      "step": 9355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1673197408138115e-07,
+      "loss": 1.8899,
+      "step": 9356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.160618484442012e-07,
+      "loss": 3.4951,
+      "step": 9357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1539274909672337e-07,
+      "loss": 2.5073,
+      "step": 9358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1472467610913482e-07,
+      "loss": 3.6132,
+      "step": 9359
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1405762955151178e-07,
+      "loss": 3.3057,
+      "step": 9360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1339160949382597e-07,
+      "loss": 2.2126,
+      "step": 9361
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1272661600594046e-07,
+      "loss": 2.6577,
+      "step": 9362
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1206264915760833e-07,
+      "loss": 1.8947,
+      "step": 9363
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1139970901847607e-07,
+      "loss": 2.9434,
+      "step": 9364
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1073779565808471e-07,
+      "loss": 1.9628,
+      "step": 9365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1007690914586542e-07,
+      "loss": 3.7183,
+      "step": 9366
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0941704955114162e-07,
+      "loss": 2.0107,
+      "step": 9367
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0875821694313014e-07,
+      "loss": 2.5247,
+      "step": 9368
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0810041139093906e-07,
+      "loss": 1.7135,
+      "step": 9369
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0744363296356872e-07,
+      "loss": 1.4765,
+      "step": 9370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0678788172991294e-07,
+      "loss": 1.9273,
+      "step": 9371
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0613315775875665e-07,
+      "loss": 2.6513,
+      "step": 9372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0547946111877714e-07,
+      "loss": 3.1139,
+      "step": 9373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0482679187854405e-07,
+      "loss": 2.1207,
+      "step": 9374
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0417515010652032e-07,
+      "loss": 3.0551,
+      "step": 9375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0352453587105914e-07,
+      "loss": 3.3285,
+      "step": 9376
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0287494924040806e-07,
+      "loss": 2.3575,
+      "step": 9377
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0222639028270486e-07,
+      "loss": 2.8627,
+      "step": 9378
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0157885906597952e-07,
+      "loss": 2.5092,
+      "step": 9379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.009323556581566e-07,
+      "loss": 2.107,
+      "step": 9380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0028688012705077e-07,
+      "loss": 2.3901,
+      "step": 9381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9964243254036897e-07,
+      "loss": 1.4308,
+      "step": 9382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9899901296571268e-07,
+      "loss": 2.472,
+      "step": 9383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9835662147057012e-07,
+      "loss": 3.6702,
+      "step": 9384
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.977152581223274e-07,
+      "loss": 2.4193,
+      "step": 9385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.970749229882607e-07,
+      "loss": 2.2115,
+      "step": 9386
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.964356161355374e-07,
+      "loss": 2.6008,
+      "step": 9387
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9579733763121943e-07,
+      "loss": 2.7299,
+      "step": 9388
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9516008754225658e-07,
+      "loss": 2.9365,
+      "step": 9389
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9452386593549534e-07,
+      "loss": 2.638,
+      "step": 9390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9388867287767232e-07,
+      "loss": 3.433,
+      "step": 9391
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.932545084354154e-07,
+      "loss": 3.0212,
+      "step": 9392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9262137267524684e-07,
+      "loss": 2.7811,
+      "step": 9393
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.91989265663578e-07,
+      "loss": 3.2098,
+      "step": 9394
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9135818746671587e-07,
+      "loss": 1.3345,
+      "step": 9395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9072813815085523e-07,
+      "loss": 2.132,
+      "step": 9396
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9009911778208766e-07,
+      "loss": 2.5862,
+      "step": 9397
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8947112642639376e-07,
+      "loss": 2.6437,
+      "step": 9398
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8884416414964524e-07,
+      "loss": 3.0196,
+      "step": 9399
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8821823101760949e-07,
+      "loss": 1.9107,
+      "step": 9400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8759332709594403e-07,
+      "loss": 2.089,
+      "step": 9401
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8696945245019638e-07,
+      "loss": 2.3497,
+      "step": 9402
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8634660714581198e-07,
+      "loss": 2.8668,
+      "step": 9403
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.857247912481197e-07,
+      "loss": 4.3094,
+      "step": 9404
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8510400482234848e-07,
+      "loss": 2.9763,
+      "step": 9405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8448424793361398e-07,
+      "loss": 2.1975,
+      "step": 9406
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.838655206469264e-07,
+      "loss": 3.0486,
+      "step": 9407
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8324782302718835e-07,
+      "loss": 2.7546,
+      "step": 9408
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.826311551391924e-07,
+      "loss": 2.8595,
+      "step": 9409
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8201551704762453e-07,
+      "loss": 2.4907,
+      "step": 9410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8140090881706207e-07,
+      "loss": 2.1462,
+      "step": 9411
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8078733051197561e-07,
+      "loss": 2.8164,
+      "step": 9412
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.801747821967259e-07,
+      "loss": 2.911,
+      "step": 9413
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.79563263935566e-07,
+      "loss": 2.8983,
+      "step": 9414
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7895277579264015e-07,
+      "loss": 2.8448,
+      "step": 9415
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7834331783198933e-07,
+      "loss": 2.3331,
+      "step": 9416
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7773489011753908e-07,
+      "loss": 2.465,
+      "step": 9417
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7712749271311392e-07,
+      "loss": 1.9676,
+      "step": 9418
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7652112568242398e-07,
+      "loss": 3.1816,
+      "step": 9419
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7591578908907724e-07,
+      "loss": 2.7279,
+      "step": 9420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.753114829965674e-07,
+      "loss": 1.8735,
+      "step": 9421
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.74708207468286e-07,
+      "loss": 1.7298,
+      "step": 9422
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7410596256751343e-07,
+      "loss": 2.9581,
+      "step": 9423
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.735047483574215e-07,
+      "loss": 1.5167,
+      "step": 9424
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7290456490107522e-07,
+      "loss": 3.2526,
+      "step": 9425
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7230541226143205e-07,
+      "loss": 1.8274,
+      "step": 9426
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7170729050133838e-07,
+      "loss": 4.2976,
+      "step": 9427
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7111019968353625e-07,
+      "loss": 2.0868,
+      "step": 9428
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7051413987065668e-07,
+      "loss": 2.5465,
+      "step": 9429
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.699191111252241e-07,
+      "loss": 2.159,
+      "step": 9430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6932511350965298e-07,
+      "loss": 2.8922,
+      "step": 9431
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.687321470862524e-07,
+      "loss": 2.38,
+      "step": 9432
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6814021191722042e-07,
+      "loss": 2.4311,
+      "step": 9433
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6754930806465063e-07,
+      "loss": 2.3673,
+      "step": 9434
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6695943559052463e-07,
+      "loss": 1.5182,
+      "step": 9435
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6637059455671623e-07,
+      "loss": 2.6785,
+      "step": 9436
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6578278502499378e-07,
+      "loss": 2.0576,
+      "step": 9437
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6519600705701465e-07,
+      "loss": 2.1468,
+      "step": 9438
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6461026071433074e-07,
+      "loss": 3.1042,
+      "step": 9439
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6402554605838173e-07,
+      "loss": 3.6533,
+      "step": 9440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6344186315050303e-07,
+      "loss": 2.9177,
+      "step": 9441
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6285921205192012e-07,
+      "loss": 3.1867,
+      "step": 9442
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6227759282375079e-07,
+      "loss": 2.8527,
+      "step": 9443
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6169700552700284e-07,
+      "loss": 1.9997,
+      "step": 9444
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6111745022257873e-07,
+      "loss": 2.7168,
+      "step": 9445
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6053892697126984e-07,
+      "loss": 1.9645,
+      "step": 9446
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.599614358337609e-07,
+      "loss": 3.58,
+      "step": 9447
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5938497687062905e-07,
+      "loss": 2.6785,
+      "step": 9448
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5880955014234144e-07,
+      "loss": 3.152,
+      "step": 9449
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5823515570925763e-07,
+      "loss": 3.15,
+      "step": 9450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5766179363162936e-07,
+      "loss": 4.052,
+      "step": 9451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5708946396959856e-07,
+      "loss": 2.4789,
+      "step": 9452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5651816678320054e-07,
+      "loss": 2.8845,
+      "step": 9453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5594790213236288e-07,
+      "loss": 2.4674,
+      "step": 9454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5537867007690111e-07,
+      "loss": 1.6245,
+      "step": 9455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5481047067652744e-07,
+      "loss": 3.4739,
+      "step": 9456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5424330399084196e-07,
+      "loss": 3.0494,
+      "step": 9457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5367717007933826e-07,
+      "loss": 3.5945,
+      "step": 9458
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5311206900140095e-07,
+      "loss": 2.9433,
+      "step": 9459
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5254800081630828e-07,
+      "loss": 2.9124,
+      "step": 9460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5198496558322506e-07,
+      "loss": 2.1006,
+      "step": 9461
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.514229633612141e-07,
+      "loss": 2.776,
+      "step": 9462
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5086199420922486e-07,
+      "loss": 2.8506,
+      "step": 9463
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5030205818610255e-07,
+      "loss": 1.0943,
+      "step": 9464
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4974315535058016e-07,
+      "loss": 3.6743,
+      "step": 9465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4918528576128298e-07,
+      "loss": 3.2038,
+      "step": 9466
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4862844947673093e-07,
+      "loss": 2.0754,
+      "step": 9467
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4807264655533282e-07,
+      "loss": 3.0717,
+      "step": 9468
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.475178770553909e-07,
+      "loss": 2.0704,
+      "step": 9469
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.469641410350964e-07,
+      "loss": 2.5099,
+      "step": 9470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4641143855253283e-07,
+      "loss": 2.681,
+      "step": 9471
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4585976966567826e-07,
+      "loss": 2.1104,
+      "step": 9472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.453091344323998e-07,
+      "loss": 3.0346,
+      "step": 9473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4475953291045674e-07,
+      "loss": 2.028,
+      "step": 9474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4421096515749855e-07,
+      "loss": 3.4526,
+      "step": 9475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4366343123106697e-07,
+      "loss": 1.2686,
+      "step": 9476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4311693118859716e-07,
+      "loss": 2.5652,
+      "step": 9477
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4257146508741436e-07,
+      "loss": 2.8776,
+      "step": 9478
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4202703298473508e-07,
+      "loss": 2.1333,
+      "step": 9479
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4148363493766803e-07,
+      "loss": 2.6742,
+      "step": 9480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4094127100321208e-07,
+      "loss": 2.6321,
+      "step": 9481
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.403999412382606e-07,
+      "loss": 1.9399,
+      "step": 9482
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3985964569959486e-07,
+      "loss": 2.0653,
+      "step": 9483
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3932038444389063e-07,
+      "loss": 3.0181,
+      "step": 9484
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3878215752771264e-07,
+      "loss": 2.3026,
+      "step": 9485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3824496500751904e-07,
+      "loss": 1.3659,
+      "step": 9486
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3770880693965816e-07,
+      "loss": 2.2899,
+      "step": 9487
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3717368338037163e-07,
+      "loss": 1.7959,
+      "step": 9488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.366395943857912e-07,
+      "loss": 3.6986,
+      "step": 9489
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.361065400119399e-07,
+      "loss": 4.2251,
+      "step": 9490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3557452031473296e-07,
+      "loss": 1.9039,
+      "step": 9491
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3504353534997682e-07,
+      "loss": 2.9838,
+      "step": 9492
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.345135851733681e-07,
+      "loss": 2.2421,
+      "step": 9493
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3398466984049786e-07,
+      "loss": 2.1001,
+      "step": 9494
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3345678940684615e-07,
+      "loss": 1.8287,
+      "step": 9495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3292994392778535e-07,
+      "loss": 2.8911,
+      "step": 9496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3240413345857795e-07,
+      "loss": 3.6003,
+      "step": 9497
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.318793580543809e-07,
+      "loss": 2.273,
+      "step": 9498
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.313556177702402e-07,
+      "loss": 1.6259,
+      "step": 9499
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.30832912661093e-07,
+      "loss": 2.855,
+      "step": 9500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3031124278176875e-07,
+      "loss": 2.388,
+      "step": 9501
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.297906081869893e-07,
+      "loss": 3.0634,
+      "step": 9502
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2927100893136534e-07,
+      "loss": 3.7964,
+      "step": 9503
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.287524450694011e-07,
+      "loss": 3.5635,
+      "step": 9504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2823491665549193e-07,
+      "loss": 2.5638,
+      "step": 9505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.277184237439233e-07,
+      "loss": 2.5777,
+      "step": 9506
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2720296638887297e-07,
+      "loss": 2.3623,
+      "step": 9507
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2668854464441104e-07,
+      "loss": 2.7091,
+      "step": 9508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.261751585644977e-07,
+      "loss": 2.4094,
+      "step": 9509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2566280820298427e-07,
+      "loss": 3.4311,
+      "step": 9510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2515149361361335e-07,
+      "loss": 2.1411,
+      "step": 9511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.246412148500198e-07,
+      "loss": 1.6846,
+      "step": 9512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.241319719657308e-07,
+      "loss": 1.8513,
+      "step": 9513
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2362376501416251e-07,
+      "loss": 3.0324,
+      "step": 9514
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.231165940486234e-07,
+      "loss": 3.3306,
+      "step": 9515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2261045912231318e-07,
+      "loss": 1.1361,
+      "step": 9516
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2210536028832377e-07,
+      "loss": 3.454,
+      "step": 9517
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2160129759963723e-07,
+      "loss": 1.4676,
+      "step": 9518
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2109827110912687e-07,
+      "loss": 3.5726,
+      "step": 9519
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2059628086956044e-07,
+      "loss": 2.8501,
+      "step": 9520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2009532693359027e-07,
+      "loss": 2.2362,
+      "step": 9521
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1959540935376769e-07,
+      "loss": 2.6915,
+      "step": 9522
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.190965281825296e-07,
+      "loss": 1.6847,
+      "step": 9523
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1859868347220749e-07,
+      "loss": 3.1352,
+      "step": 9524
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1810187527502182e-07,
+      "loss": 2.7412,
+      "step": 9525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1760610364308645e-07,
+      "loss": 3.4835,
+      "step": 9526
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1711136862840533e-07,
+      "loss": 2.6771,
+      "step": 9527
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1661767028287363e-07,
+      "loss": 2.5776,
+      "step": 9528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.161250086582777e-07,
+      "loss": 2.2951,
+      "step": 9529
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1563338380629618e-07,
+      "loss": 1.0748,
+      "step": 9530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1514279577849786e-07,
+      "loss": 2.4801,
+      "step": 9531
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1465324462634375e-07,
+      "loss": 3.0019,
+      "step": 9532
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1416473040118504e-07,
+      "loss": 2.2634,
+      "step": 9533
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1367725315426403e-07,
+      "loss": 2.4473,
+      "step": 9534
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1319081293671541e-07,
+      "loss": 2.5822,
+      "step": 9535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1270540979956501e-07,
+      "loss": 4.212,
+      "step": 9536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1222104379372767e-07,
+      "loss": 2.9317,
+      "step": 9537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1173771497001273e-07,
+      "loss": 2.7592,
+      "step": 9538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1125542337911854e-07,
+      "loss": 3.0418,
+      "step": 9539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1077416907163573e-07,
+      "loss": 1.9634,
+      "step": 9540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1029395209804505e-07,
+      "loss": 3.0624,
+      "step": 9541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0981477250871952e-07,
+      "loss": 2.6248,
+      "step": 9542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0933663035392228e-07,
+      "loss": 2.4957,
+      "step": 9543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0885952568380764e-07,
+      "loss": 3.8626,
+      "step": 9544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0838345854842447e-07,
+      "loss": 1.4927,
+      "step": 9545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0790842899770726e-07,
+      "loss": 1.761,
+      "step": 9546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0743443708148504e-07,
+      "loss": 2.5157,
+      "step": 9547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0696148284947694e-07,
+      "loss": 2.4113,
+      "step": 9548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0648956635129548e-07,
+      "loss": 3.319,
+      "step": 9549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0601868763643997e-07,
+      "loss": 3.5267,
+      "step": 9550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0554884675430532e-07,
+      "loss": 3.4984,
+      "step": 9551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0508004375417546e-07,
+      "loss": 3.553,
+      "step": 9552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0461227868522439e-07,
+      "loss": 2.3616,
+      "step": 9553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0414555159652062e-07,
+      "loss": 2.0706,
+      "step": 9554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0367986253701945e-07,
+      "loss": 1.8412,
+      "step": 9555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.032152115555718e-07,
+      "loss": 2.3323,
+      "step": 9556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0275159870091422e-07,
+      "loss": 2.2293,
+      "step": 9557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0228902402168118e-07,
+      "loss": 2.0096,
+      "step": 9558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.018274875663916e-07,
+      "loss": 1.8495,
+      "step": 9559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0136698938346012e-07,
+      "loss": 2.8692,
+      "step": 9560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0090752952119032e-07,
+      "loss": 4.1573,
+      "step": 9561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0044910802777807e-07,
+      "loss": 2.2455,
+      "step": 9562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999172495130827e-08,
+      "loss": 2.2221,
+      "step": 9563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.953538033975918e-08,
+      "loss": 2.9356,
+      "step": 9564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.90800742410003e-08,
+      "loss": 2.8319,
+      "step": 9565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.862580670278899e-08,
+      "loss": 2.6262,
+      "step": 9566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.81725777727771e-08,
+      "loss": 1.9264,
+      "step": 9567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.772038749850665e-08,
+      "loss": 3.2303,
+      "step": 9568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.726923592740855e-08,
+      "loss": 2.5973,
+      "step": 9569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.68191231068083e-08,
+      "loss": 3.1101,
+      "step": 9570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.637004908392033e-08,
+      "loss": 1.9393,
+      "step": 9571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.59220139058492e-08,
+      "loss": 3.3788,
+      "step": 9572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.547501761959288e-08,
+      "loss": 2.7769,
+      "step": 9573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.50290602720394e-08,
+      "loss": 3.1993,
+      "step": 9574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.45841419099669e-08,
+      "loss": 2.531,
+      "step": 9575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.414026258004583e-08,
+      "loss": 2.1658,
+      "step": 9576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.369742232883561e-08,
+      "loss": 3.0338,
+      "step": 9577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.32556212027902e-08,
+      "loss": 4.1441,
+      "step": 9578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.281485924825029e-08,
+      "loss": 1.9128,
+      "step": 9579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.237513651145224e-08,
+      "loss": 3.3357,
+      "step": 9580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.193645303851917e-08,
+      "loss": 2.4434,
+      "step": 9581
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.149880887546759e-08,
+      "loss": 3.2585,
+      "step": 9582
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.10622040682041e-08,
+      "loss": 2.2584,
+      "step": 9583
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.062663866252541e-08,
+      "loss": 3.6181,
+      "step": 9584
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.019211270412275e-08,
+      "loss": 3.2195,
+      "step": 9585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.975862623857523e-08,
+      "loss": 2.8089,
+      "step": 9586
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.932617931135202e-08,
+      "loss": 2.1622,
+      "step": 9587
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.889477196781571e-08,
+      "loss": 2.6645,
+      "step": 9588
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.8464404253219e-08,
+      "loss": 2.3646,
+      "step": 9589
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.80350762127058e-08,
+      "loss": 1.5145,
+      "step": 9590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.760678789131005e-08,
+      "loss": 3.4088,
+      "step": 9591
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.717953933395695e-08,
+      "loss": 2.5086,
+      "step": 9592
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.675333058546397e-08,
+      "loss": 3.9527,
+      "step": 9593
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.63281616905376e-08,
+      "loss": 3.1892,
+      "step": 9594
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.590403269377656e-08,
+      "loss": 2.9403,
+      "step": 9595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.548094363966974e-08,
+      "loss": 3.1637,
+      "step": 9596
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.505889457259719e-08,
+      "loss": 3.1658,
+      "step": 9597
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.463788553683017e-08,
+      "loss": 1.8922,
+      "step": 9598
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.421791657652889e-08,
+      "loss": 1.4121,
+      "step": 9599
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.379898773574924e-08,
+      "loss": 2.8919,
+      "step": 9600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.338109905843273e-08,
+      "loss": 2.4899,
+      "step": 9601
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.296425058841429e-08,
+      "loss": 1.9776,
+      "step": 9602
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.254844236941895e-08,
+      "loss": 2.6246,
+      "step": 9603
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.213367444506515e-08,
+      "loss": 2.4922,
+      "step": 9604
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.171994685885698e-08,
+      "loss": 2.8308,
+      "step": 9605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.130725965419528e-08,
+      "loss": 2.8787,
+      "step": 9606
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.089561287436876e-08,
+      "loss": 2.7097,
+      "step": 9607
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.04850065625551e-08,
+      "loss": 1.8501,
+      "step": 9608
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.00754407618276e-08,
+      "loss": 2.7194,
+      "step": 9609
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.966691551514527e-08,
+      "loss": 2.3187,
+      "step": 9610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.92594308653627e-08,
+      "loss": 2.1474,
+      "step": 9611
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.885298685522235e-08,
+      "loss": 1.3217,
+      "step": 9612
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.844758352735793e-08,
+      "loss": 2.7668,
+      "step": 9613
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.804322092429429e-08,
+      "loss": 3.7336,
+      "step": 9614
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.763989908844749e-08,
+      "loss": 3.3854,
+      "step": 9615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.723761806212371e-08,
+      "loss": 6.0065,
+      "step": 9616
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.683637788752252e-08,
+      "loss": 2.6929,
+      "step": 9617
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.643617860672914e-08,
+      "loss": 3.3948,
+      "step": 9618
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.603702026172443e-08,
+      "loss": 2.6908,
+      "step": 9619
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.563890289437825e-08,
+      "loss": 2.2887,
+      "step": 9620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.52418265464494e-08,
+      "loss": 2.9306,
+      "step": 9621
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.484579125959124e-08,
+      "loss": 2.9737,
+      "step": 9622
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.445079707534608e-08,
+      "loss": 2.7442,
+      "step": 9623
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.405684403514635e-08,
+      "loss": 3.1123,
+      "step": 9624
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.366393218031564e-08,
+      "loss": 1.7612,
+      "step": 9625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.327206155206879e-08,
+      "loss": 2.2259,
+      "step": 9626
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.288123219151178e-08,
+      "loss": 3.6612,
+      "step": 9627
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.24914441396396e-08,
+      "loss": 1.5919,
+      "step": 9628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.210269743734067e-08,
+      "loss": 2.6026,
+      "step": 9629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.171499212539124e-08,
+      "loss": 3.5018,
+      "step": 9630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.132832824446101e-08,
+      "loss": 2.7988,
+      "step": 9631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.094270583510976e-08,
+      "loss": 1.7981,
+      "step": 9632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.055812493778513e-08,
+      "loss": 1.4856,
+      "step": 9633
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.01745855928293e-08,
+      "loss": 2.1332,
+      "step": 9634
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.979208784047454e-08,
+      "loss": 2.7776,
+      "step": 9635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.941063172084094e-08,
+      "loss": 2.866,
+      "step": 9636
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.903021727394321e-08,
+      "loss": 1.902,
+      "step": 9637
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.865084453968495e-08,
+      "loss": 3.2928,
+      "step": 9638
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.827251355786102e-08,
+      "loss": 2.2138,
+      "step": 9639
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.78952243681541e-08,
+      "loss": 2.6443,
+      "step": 9640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.751897701014365e-08,
+      "loss": 1.7933,
+      "step": 9641
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.714377152329366e-08,
+      "loss": 1.3081,
+      "step": 9642
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.676960794696153e-08,
+      "loss": 1.7434,
+      "step": 9643
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.639648632039697e-08,
+      "loss": 1.846,
+      "step": 9644
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.602440668273758e-08,
+      "loss": 3.5534,
+      "step": 9645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.565336907301212e-08,
+      "loss": 2.7736,
+      "step": 9646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.52833735301428e-08,
+      "loss": 2.5022,
+      "step": 9647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.491442009293858e-08,
+      "loss": 3.0523,
+      "step": 9648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.454650880010182e-08,
+      "loss": 2.3121,
+      "step": 9649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.417963969022389e-08,
+      "loss": 2.0964,
+      "step": 9650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.381381280178733e-08,
+      "loss": 2.4301,
+      "step": 9651
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.344902817316811e-08,
+      "loss": 1.9773,
+      "step": 9652
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.308528584262896e-08,
+      "loss": 2.208,
+      "step": 9653
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.27225858483238e-08,
+      "loss": 1.7342,
+      "step": 9654
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.236092822829887e-08,
+      "loss": 3.0059,
+      "step": 9655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.200031302049048e-08,
+      "loss": 1.9129,
+      "step": 9656
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.164074026272615e-08,
+      "loss": 2.0695,
+      "step": 9657
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.12822099927235e-08,
+      "loss": 2.4111,
+      "step": 9658
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.092472224808909e-08,
+      "loss": 3.1897,
+      "step": 9659
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.056827706632185e-08,
+      "loss": 2.2911,
+      "step": 9660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.021287448481294e-08,
+      "loss": 1.7556,
+      "step": 9661
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.985851454084257e-08,
+      "loss": 3.3091,
+      "step": 9662
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9505197271578776e-08,
+      "loss": 2.9695,
+      "step": 9663
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.915292271408524e-08,
+      "loss": 2.5575,
+      "step": 9664
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.880169090531351e-08,
+      "loss": 3.6337,
+      "step": 9665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.845150188210524e-08,
+      "loss": 3.2002,
+      "step": 9666
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8102355681194375e-08,
+      "loss": 3.4783,
+      "step": 9667
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7754252339204955e-08,
+      "loss": 3.9159,
+      "step": 9668
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7407191892651095e-08,
+      "loss": 2.9569,
+      "step": 9669
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7061174377937015e-08,
+      "loss": 2.7357,
+      "step": 9670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.671619983136034e-08,
+      "loss": 2.652,
+      "step": 9671
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.637226828910436e-08,
+      "loss": 1.2151,
+      "step": 9672
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6029379787249094e-08,
+      "loss": 2.1394,
+      "step": 9673
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.568753436176022e-08,
+      "loss": 2.2822,
+      "step": 9674
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.534673204849572e-08,
+      "loss": 1.8251,
+      "step": 9675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5006972883204776e-08,
+      "loss": 3.4323,
+      "step": 9676
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4668256901525554e-08,
+      "loss": 2.2383,
+      "step": 9677
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4330584138989615e-08,
+      "loss": 2.0202,
+      "step": 9678
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.399395463101531e-08,
+      "loss": 2.1322,
+      "step": 9679
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.365836841291439e-08,
+      "loss": 1.9243,
+      "step": 9680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.332382551988868e-08,
+      "loss": 1.4714,
+      "step": 9681
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2990325987029025e-08,
+      "loss": 3.6304,
+      "step": 9682
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.265786984931964e-08,
+      "loss": 1.8637,
+      "step": 9683
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.232645714163265e-08,
+      "loss": 1.5194,
+      "step": 9684
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.199608789873134e-08,
+      "loss": 1.7967,
+      "step": 9685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.166676215526911e-08,
+      "loss": 3.2891,
+      "step": 9686
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.133847994579388e-08,
+      "loss": 1.4857,
+      "step": 9687
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1011241304738115e-08,
+      "loss": 1.9955,
+      "step": 9688
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.068504626642767e-08,
+      "loss": 2.7979,
+      "step": 9689
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.035989486508075e-08,
+      "loss": 2.86,
+      "step": 9690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.003578713480228e-08,
+      "loss": 2.6162,
+      "step": 9691
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9712723109590636e-08,
+      "loss": 2.7743,
+      "step": 9692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.939070282333425e-08,
+      "loss": 1.9608,
+      "step": 9693
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.906972630981055e-08,
+      "loss": 2.3134,
+      "step": 9694
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.874979360268928e-08,
+      "loss": 3.5625,
+      "step": 9695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.843090473552914e-08,
+      "loss": 2.3689,
+      "step": 9696
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.811305974178004e-08,
+      "loss": 3.3998,
+      "step": 9697
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.779625865478421e-08,
+      "loss": 2.7708,
+      "step": 9698
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.748050150776951e-08,
+      "loss": 2.3025,
+      "step": 9699
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.716578833386054e-08,
+      "loss": 2.3811,
+      "step": 9700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.685211916606647e-08,
+      "loss": 3.3532,
+      "step": 9701
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.653949403729208e-08,
+      "loss": 2.5579,
+      "step": 9702
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6227912980330026e-08,
+      "loss": 3.9936,
+      "step": 9703
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5917376027861945e-08,
+      "loss": 3.4904,
+      "step": 9704
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.56078832124629e-08,
+      "loss": 2.6454,
+      "step": 9705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.529943456659691e-08,
+      "loss": 3.1862,
+      "step": 9706
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4992030122619216e-08,
+      "loss": 2.0221,
+      "step": 9707
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.468566991277512e-08,
+      "loss": 3.0138,
+      "step": 9708
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.438035396920004e-08,
+      "loss": 2.0804,
+      "step": 9709
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4076082323920576e-08,
+      "loss": 2.7779,
+      "step": 9710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.377285500885231e-08,
+      "loss": 2.8273,
+      "step": 9711
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.347067205580424e-08,
+      "loss": 2.249,
+      "step": 9712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3169533496473237e-08,
+      "loss": 3.8277,
+      "step": 9713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2869439362446254e-08,
+      "loss": 1.6961,
+      "step": 9714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.257038968520366e-08,
+      "loss": 3.0921,
+      "step": 9715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2272384496112597e-08,
+      "loss": 1.3799,
+      "step": 9716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1975423826433605e-08,
+      "loss": 1.2755,
+      "step": 9717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1679507707315106e-08,
+      "loss": 2.1877,
+      "step": 9718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.138463616979893e-08,
+      "loss": 2.8179,
+      "step": 9719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.109080924481479e-08,
+      "loss": 2.2197,
+      "step": 9720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.079802696318469e-08,
+      "loss": 1.8562,
+      "step": 9721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.050628935561851e-08,
+      "loss": 2.4675,
+      "step": 9722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0215596452720665e-08,
+      "loss": 3.6691,
+      "step": 9723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9925948284980086e-08,
+      "loss": 3.0564,
+      "step": 9724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.963734488278248e-08,
+      "loss": 2.6546,
+      "step": 9725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.934978627639918e-08,
+      "loss": 1.7279,
+      "step": 9726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.906327249599495e-08,
+      "loss": 2.0585,
+      "step": 9727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.877780357162353e-08,
+      "loss": 1.6825,
+      "step": 9728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.849337953322874e-08,
+      "loss": 2.2509,
+      "step": 9729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.82100004106456e-08,
+      "loss": 3.423,
+      "step": 9730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.792766623360034e-08,
+      "loss": 2.667,
+      "step": 9731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.764637703170593e-08,
+      "loss": 2.9529,
+      "step": 9732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.736613283447099e-08,
+      "loss": 3.6723,
+      "step": 9733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7086933671290904e-08,
+      "loss": 3.67,
+      "step": 9734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.680877957145112e-08,
+      "loss": 3.7444,
+      "step": 9735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.653167056413054e-08,
+      "loss": 1.9196,
+      "step": 9736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6255606678394786e-08,
+      "loss": 1.9948,
+      "step": 9737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.598058794320402e-08,
+      "loss": 2.4328,
+      "step": 9738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5706614387404084e-08,
+      "loss": 3.1135,
+      "step": 9739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.543368603973529e-08,
+      "loss": 2.7272,
+      "step": 9740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5161802928825874e-08,
+      "loss": 2.4389,
+      "step": 9741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.489096508319412e-08,
+      "loss": 2.5382,
+      "step": 9742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4621172531251744e-08,
+      "loss": 2.1024,
+      "step": 9743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.435242530129723e-08,
+      "loss": 3.5668,
+      "step": 9744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.408472342152136e-08,
+      "loss": 1.4452,
+      "step": 9745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.381806692000389e-08,
+      "loss": 2.025,
+      "step": 9746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3552455824718e-08,
+      "loss": 3.3624,
+      "step": 9747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3287890163523626e-08,
+      "loss": 2.4016,
+      "step": 9748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3024369964171907e-08,
+      "loss": 2.172,
+      "step": 9749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2761895254306285e-08,
+      "loss": 3.5477,
+      "step": 9750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2500466061458066e-08,
+      "loss": 1.8423,
+      "step": 9751
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.224008241304977e-08,
+      "loss": 2.6274,
+      "step": 9752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1980744336396195e-08,
+      "loss": 3.9657,
+      "step": 9753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.172245185869893e-08,
+      "loss": 2.6186,
+      "step": 9754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1465205007052965e-08,
+      "loss": 2.5599,
+      "step": 9755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.120900380844116e-08,
+      "loss": 2.452,
+      "step": 9756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.095384828973758e-08,
+      "loss": 3.9783,
+      "step": 9757
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0699738477708576e-08,
+      "loss": 3.3118,
+      "step": 9758
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.044667439900728e-08,
+      "loss": 2.4244,
+      "step": 9759
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.019465608018024e-08,
+      "loss": 3.4026,
+      "step": 9760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.994368354766186e-08,
+      "loss": 3.2233,
+      "step": 9761
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9693756827778862e-08,
+      "loss": 2.8024,
+      "step": 9762
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9444875946746943e-08,
+      "loss": 3.1048,
+      "step": 9763
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9197040930674102e-08,
+      "loss": 3.4543,
+      "step": 9764
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8950251805553997e-08,
+      "loss": 2.7641,
+      "step": 9765
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8704508597277024e-08,
+      "loss": 2.5727,
+      "step": 9766
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8459811331617016e-08,
+      "loss": 2.955,
+      "step": 9767
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8216160034244544e-08,
+      "loss": 2.8788,
+      "step": 9768
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7973554730716947e-08,
+      "loss": 3.7545,
+      "step": 9769
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.773199544648164e-08,
+      "loss": 2.9209,
+      "step": 9770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.749148220687614e-08,
+      "loss": 3.4602,
+      "step": 9771
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7252015037131373e-08,
+      "loss": 3.0067,
+      "step": 9772
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7013593962365025e-08,
+      "loss": 1.573,
+      "step": 9773
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.677621900758709e-08,
+      "loss": 3.4118,
+      "step": 9774
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6539890197695428e-08,
+      "loss": 2.8326,
+      "step": 9775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6304607557481322e-08,
+      "loss": 1.9419,
+      "step": 9776
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6070371111625026e-08,
+      "loss": 2.9751,
+      "step": 9777
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.583718088469689e-08,
+      "loss": 3.7224,
+      "step": 9778
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5605036901156233e-08,
+      "loss": 3.8945,
+      "step": 9779
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.537393918535358e-08,
+      "loss": 2.0413,
+      "step": 9780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.514388776153287e-08,
+      "loss": 3.7196,
+      "step": 9781
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4914882653822582e-08,
+      "loss": 2.8966,
+      "step": 9782
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.468692388624572e-08,
+      "loss": 2.1337,
+      "step": 9783
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4460011482713153e-08,
+      "loss": 2.1472,
+      "step": 9784
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.423414546702807e-08,
+      "loss": 2.9982,
+      "step": 9785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4009325862881517e-08,
+      "loss": 2.1985,
+      "step": 9786
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.378555269385796e-08,
+      "loss": 1.5218,
+      "step": 9787
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3562825983427517e-08,
+      "loss": 2.2221,
+      "step": 9788
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.334114575495483e-08,
+      "loss": 3.1117,
+      "step": 9789
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.312051203169352e-08,
+      "loss": 1.8407,
+      "step": 9790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2900924836786188e-08,
+      "loss": 3.7781,
+      "step": 9791
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2682384193266625e-08,
+      "loss": 2.8358,
+      "step": 9792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2464890124057614e-08,
+      "loss": 2.7514,
+      "step": 9793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2248442651975345e-08,
+      "loss": 1.8495,
+      "step": 9794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2033041799723877e-08,
+      "loss": 2.5485,
+      "step": 9795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1818687589896248e-08,
+      "loss": 3.1655,
+      "step": 9796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1605380044978918e-08,
+      "loss": 2.2689,
+      "step": 9797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1393119187345103e-08,
+      "loss": 1.9464,
+      "step": 9798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.118190503926143e-08,
+      "loss": 2.601,
+      "step": 9799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0971737622883515e-08,
+      "loss": 4.143,
+      "step": 9800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0762616960255943e-08,
+      "loss": 3.2967,
+      "step": 9801
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0554543073313393e-08,
+      "loss": 1.6454,
+      "step": 9802
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0347515983885068e-08,
+      "loss": 1.4223,
+      "step": 9803
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.01415357136836e-08,
+      "loss": 2.483,
+      "step": 9804
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9936602284318375e-08,
+      "loss": 3.4539,
+      "step": 9805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.973271571728441e-08,
+      "loss": 3.6347,
+      "step": 9806
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9529876033969052e-08,
+      "loss": 2.4734,
+      "step": 9807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.93280832556475e-08,
+      "loss": 2.2935,
+      "step": 9808
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9127337403489486e-08,
+      "loss": 2.7162,
+      "step": 9809
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8927638498551502e-08,
+      "loss": 2.4271,
+      "step": 9810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8728986561780127e-08,
+      "loss": 2.0909,
+      "step": 9811
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8531381614013133e-08,
+      "loss": 2.1117,
+      "step": 9812
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8334823675978387e-08,
+      "loss": 2.4902,
+      "step": 9813
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8139312768294946e-08,
+      "loss": 2.2797,
+      "step": 9814
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7944848911470857e-08,
+      "loss": 2.3217,
+      "step": 9815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7751432125903133e-08,
+      "loss": 1.2899,
+      "step": 9816
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.75590624318811e-08,
+      "loss": 3.0949,
+      "step": 9817
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7367739849584174e-08,
+      "loss": 2.6586,
+      "step": 9818
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7177464399080746e-08,
+      "loss": 2.705,
+      "step": 9819
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.698823610032929e-08,
+      "loss": 3.0219,
+      "step": 9820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.680005497317838e-08,
+      "loss": 3.1568,
+      "step": 9821
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.661292103736889e-08,
+      "loss": 2.2164,
+      "step": 9822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.642683431252956e-08,
+      "loss": 2.3441,
+      "step": 9823
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6241794818180333e-08,
+      "loss": 2.6775,
+      "step": 9824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.605780257373124e-08,
+      "loss": 3.0687,
+      "step": 9825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5874857598481286e-08,
+      "loss": 3.0537,
+      "step": 9826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5692959911620676e-08,
+      "loss": 3.0109,
+      "step": 9827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5512109532229703e-08,
+      "loss": 1.8777,
+      "step": 9828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5332306479279858e-08,
+      "loss": 4.0893,
+      "step": 9829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5153550771630498e-08,
+      "loss": 2.4411,
+      "step": 9830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4975842428032182e-08,
+      "loss": 3.7634,
+      "step": 9831
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4799181467125557e-08,
+      "loss": 2.4629,
+      "step": 9832
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.462356790744357e-08,
+      "loss": 2.5089,
+      "step": 9833
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4449001767404824e-08,
+      "loss": 2.6755,
+      "step": 9834
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4275483065321338e-08,
+      "loss": 3.0239,
+      "step": 9835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4103011819395218e-08,
+      "loss": 2.0139,
+      "step": 9836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3931588047716438e-08,
+      "loss": 2.7061,
+      "step": 9837
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.376121176826728e-08,
+      "loss": 2.9025,
+      "step": 9838
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3591882998920113e-08,
+      "loss": 1.6017,
+      "step": 9839
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3423601757436289e-08,
+      "loss": 1.9978,
+      "step": 9840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.325636806146613e-08,
+      "loss": 2.1527,
+      "step": 9841
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.309018192855449e-08,
+      "loss": 1.5295,
+      "step": 9842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2925043376130764e-08,
+      "loss": 2.3507,
+      "step": 9843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.276095242151998e-08,
+      "loss": 2.3393,
+      "step": 9844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2597909081931702e-08,
+      "loss": 1.3633,
+      "step": 9845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2435913374471142e-08,
+      "loss": 1.942,
+      "step": 9846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.227496531612804e-08,
+      "loss": 2.6932,
+      "step": 9847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2115064923787778e-08,
+      "loss": 1.2962,
+      "step": 9848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1956212214221385e-08,
+      "loss": 4.2957,
+      "step": 9849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.179840720409331e-08,
+      "loss": 2.7201,
+      "step": 9850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1641649909954755e-08,
+      "loss": 3.3578,
+      "step": 9851
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1485940348249235e-08,
+      "loss": 2.6787,
+      "step": 9852
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1331278535311464e-08,
+      "loss": 1.8478,
+      "step": 9853
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1177664487362905e-08,
+      "loss": 2.1139,
+      "step": 9854
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.102509822051845e-08,
+      "loss": 2.3023,
+      "step": 9855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.087357975078085e-08,
+      "loss": 3.3854,
+      "step": 9856
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.072310909404406e-08,
+      "loss": 3.843,
+      "step": 9857
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.057368626609101e-08,
+      "loss": 2.9489,
+      "step": 9858
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.042531128259583e-08,
+      "loss": 2.5523,
+      "step": 9859
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0277984159122734e-08,
+      "loss": 3.5013,
+      "step": 9860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0131704911124917e-08,
+      "loss": 1.9157,
+      "step": 9861
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.986473553946774e-09,
+      "loss": 1.9776,
+      "step": 9862
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.842290102823893e-09,
+      "loss": 2.966,
+      "step": 9863
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.699154572877511e-09,
+      "loss": 3.18,
+      "step": 9864
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.557066979123398e-09,
+      "loss": 2.0025,
+      "step": 9865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.416027336466294e-09,
+      "loss": 2.8399,
+      "step": 9866
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.276035659701032e-09,
+      "loss": 3.1724,
+      "step": 9867
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.137091963510314e-09,
+      "loss": 1.4465,
+      "step": 9868
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.999196262469146e-09,
+      "loss": 3.4499,
+      "step": 9869
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.862348571043733e-09,
+      "loss": 1.9814,
+      "step": 9870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.72654890358704e-09,
+      "loss": 2.2123,
+      "step": 9871
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.591797274344338e-09,
+      "loss": 1.7425,
+      "step": 9872
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.458093697450987e-09,
+      "loss": 1.7006,
+      "step": 9873
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.325438186931323e-09,
+      "loss": 2.0189,
+      "step": 9874
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.193830756699773e-09,
+      "loss": 3.1354,
+      "step": 9875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.063271420563068e-09,
+      "loss": 1.6961,
+      "step": 9876
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.933760192214702e-09,
+      "loss": 3.2497,
+      "step": 9877
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.80529708523936e-09,
+      "loss": 1.4606,
+      "step": 9878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.677882113114043e-09,
+      "loss": 2.6222,
+      "step": 9879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.551515289203615e-09,
+      "loss": 2.8016,
+      "step": 9880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.426196626761917e-09,
+      "loss": 1.6982,
+      "step": 9881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.3019261389351005e-09,
+      "loss": 3.0996,
+      "step": 9882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.178703838759405e-09,
+      "loss": 2.1805,
+      "step": 9883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.056529739158935e-09,
+      "loss": 2.2483,
+      "step": 9884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.935403852950107e-09,
+      "loss": 2.997,
+      "step": 9885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.8153261928383115e-09,
+      "loss": 2.9771,
+      "step": 9886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.696296771419031e-09,
+      "loss": 3.8516,
+      "step": 9887
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.5783156011778315e-09,
+      "loss": 1.3577,
+      "step": 9888
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.46138269449037e-09,
+      "loss": 2.4769,
+      "step": 9889
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.345498063622391e-09,
+      "loss": 1.8673,
+      "step": 9890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.2306617207297245e-09,
+      "loss": 1.5947,
+      "step": 9891
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.116873677858293e-09,
+      "loss": 2.2604,
+      "step": 9892
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.004133946942992e-09,
+      "loss": 2.2753,
+      "step": 9893
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.892442539811028e-09,
+      "loss": 3.0677,
+      "step": 9894
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.781799468177473e-09,
+      "loss": 2.6952,
+      "step": 9895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6722047436497115e-09,
+      "loss": 2.4656,
+      "step": 9896
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.563658377720771e-09,
+      "loss": 1.9755,
+      "step": 9897
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.456160381779319e-09,
+      "loss": 2.9605,
+      "step": 9898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.34971076710078e-09,
+      "loss": 3.4515,
+      "step": 9899
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2443095448506674e-09,
+      "loss": 3.3589,
+      "step": 9900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1399567260856935e-09,
+      "loss": 3.4361,
+      "step": 9901
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.036652321751545e-09,
+      "loss": 1.5832,
+      "step": 9902
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9343963426840006e-09,
+      "loss": 3.6392,
+      "step": 9903
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.833188799610033e-09,
+      "loss": 2.4083,
+      "step": 9904
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.733029703146708e-09,
+      "loss": 1.7656,
+      "step": 9905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.633919063797843e-09,
+      "loss": 1.9133,
+      "step": 9906
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.535856891960677e-09,
+      "loss": 1.7091,
+      "step": 9907
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.438843197922538e-09,
+      "loss": 2.3202,
+      "step": 9908
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.342877991858618e-09,
+      "loss": 1.9692,
+      "step": 9909
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.247961283835311e-09,
+      "loss": 3.2509,
+      "step": 9910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.154093083809096e-09,
+      "loss": 3.6975,
+      "step": 9911
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.061273401627653e-09,
+      "loss": 2.7767,
+      "step": 9912
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.969502247024304e-09,
+      "loss": 2.8935,
+      "step": 9913
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.878779629628016e-09,
+      "loss": 2.3247,
+      "step": 9914
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.789105558954509e-09,
+      "loss": 1.7969,
+      "step": 9915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7004800444095933e-09,
+      "loss": 3.9593,
+      "step": 9916
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6129030952902767e-09,
+      "loss": 2.7862,
+      "step": 9917
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.526374720782544e-09,
+      "loss": 3.4697,
+      "step": 9918
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4408949299624683e-09,
+      "loss": 1.86,
+      "step": 9919
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3564637317984318e-09,
+      "loss": 1.3794,
+      "step": 9920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2730811351444624e-09,
+      "loss": 3.3904,
+      "step": 9921
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.190747148748008e-09,
+      "loss": 3.2625,
+      "step": 9922
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.109461781246603e-09,
+      "loss": 3.5687,
+      "step": 9923
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0292250411645406e-09,
+      "loss": 2.5924,
+      "step": 9924
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9500369369195313e-09,
+      "loss": 1.9579,
+      "step": 9925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8718974768182638e-09,
+      "loss": 2.94,
+      "step": 9926
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7948066690564045e-09,
+      "loss": 2.1399,
+      "step": 9927
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7187645217219283e-09,
+      "loss": 3.3369,
+      "step": 9928
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.643771042789567e-09,
+      "loss": 3.049,
+      "step": 9929
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5698262401263607e-09,
+      "loss": 2.2737,
+      "step": 9930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.496930121488328e-09,
+      "loss": 3.5666,
+      "step": 9931
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4250826945226847e-09,
+      "loss": 0.9782,
+      "step": 9932
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.354283966766735e-09,
+      "loss": 2.3689,
+      "step": 9933
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2845339456445402e-09,
+      "loss": 1.9613,
+      "step": 9934
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.215832638474691e-09,
+      "loss": 3.7064,
+      "step": 9935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.148180052462534e-09,
+      "loss": 2.7932,
+      "step": 9936
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0815761947057254e-09,
+      "loss": 2.9779,
+      "step": 9937
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0160210721886788e-09,
+      "loss": 2.5771,
+      "step": 9938
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9515146917892248e-09,
+      "loss": 2.7357,
+      "step": 9939
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.888057060274173e-09,
+      "loss": 2.4995,
+      "step": 9940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.825648184298201e-09,
+      "loss": 2.1411,
+      "step": 9941
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7642880704094033e-09,
+      "loss": 4.3829,
+      "step": 9942
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7039767250437433e-09,
+      "loss": 2.1555,
+      "step": 9943
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6447141545272717e-09,
+      "loss": 3.484,
+      "step": 9944
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5865003650761268e-09,
+      "loss": 3.0765,
+      "step": 9945
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5293353627965358e-09,
+      "loss": 2.2664,
+      "step": 9946
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4732191536870332e-09,
+      "loss": 1.8362,
+      "step": 9947
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4181517436306913e-09,
+      "loss": 3.8808,
+      "step": 9948
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3641331384062207e-09,
+      "loss": 2.2644,
+      "step": 9949
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3111633436779792e-09,
+      "loss": 2.9343,
+      "step": 9950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2592423650048535e-09,
+      "loss": 2.7784,
+      "step": 9951
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2083702078302673e-09,
+      "loss": 1.2186,
+      "step": 9952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.158546877493283e-09,
+      "loss": 2.531,
+      "step": 9953
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1097723792174998e-09,
+      "loss": 2.7405,
+      "step": 9954
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.062046718121046e-09,
+      "loss": 2.3194,
+      "step": 9955
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0153698992088069e-09,
+      "loss": 1.7553,
+      "step": 9956
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.69741927376866e-10,
+      "loss": 3.1184,
+      "step": 9957
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.251628074136154e-10,
+      "loss": 1.5738,
+      "step": 9958
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.81632543993094e-10,
+      "loss": 3.3712,
+      "step": 9959
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.391511416816489e-10,
+      "loss": 3.7246,
+      "step": 9960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.977186049357156e-10,
+      "loss": 3.2879,
+      "step": 9961
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.573349381018169e-10,
+      "loss": 2.5339,
+      "step": 9962
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.18000145415454e-10,
+      "loss": 2.0978,
+      "step": 9963
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.797142310022154e-10,
+      "loss": 1.5378,
+      "step": 9964
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.424771988788881e-10,
+      "loss": 2.9987,
+      "step": 9965
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.062890529512366e-10,
+      "loss": 2.8763,
+      "step": 9966
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.711497970162238e-10,
+      "loss": 2.7802,
+      "step": 9967
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.370594347575697e-10,
+      "loss": 3.3359,
+      "step": 9968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.040179697524128e-10,
+      "loss": 2.8348,
+      "step": 9969
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.720254054679796e-10,
+      "loss": 2.3973,
+      "step": 9970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4108174525825386e-10,
+      "loss": 2.4526,
+      "step": 9971
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1118699236841753e-10,
+      "loss": 1.3045,
+      "step": 9972
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8234114993707105e-10,
+      "loss": 1.7761,
+      "step": 9973
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5454422098735176e-10,
+      "loss": 2.8059,
+      "step": 9974
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.277962084369257e-10,
+      "loss": 3.1439,
+      "step": 9975
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0209711509132657e-10,
+      "loss": 2.2898,
+      "step": 9976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7744694364506554e-10,
+      "loss": 2.6435,
+      "step": 9977
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.538456966838521e-10,
+      "loss": 2.0429,
+      "step": 9978
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3129337668459372e-10,
+      "loss": 2.3367,
+      "step": 9979
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0978998601206558e-10,
+      "loss": 3.0911,
+      "step": 9980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8933552692224078e-10,
+      "loss": 1.8637,
+      "step": 9981
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6993000156118044e-10,
+      "loss": 2.8884,
+      "step": 9982
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5157341196281317e-10,
+      "loss": 3.6979,
+      "step": 9983
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.342657600544861e-10,
+      "loss": 2.0679,
+      "step": 9984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1800704765030367e-10,
+      "loss": 3.4271,
+      "step": 9985
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.027972764566787e-10,
+      "loss": 3.2793,
+      "step": 9986
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.86364480678914e-11,
+      "loss": 1.3911,
+      "step": 9987
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.552456397053042e-11,
+      "loss": 3.4475,
+      "step": 9988
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.34616255401621e-11,
+      "loss": 2.4209,
+      "step": 9989
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.244763404133046e-11,
+      "loss": 3.4604,
+      "step": 9990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2482590628667444e-11,
+      "loss": 2.068,
+      "step": 9991
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3566496349113355e-11,
+      "loss": 3.7631,
+      "step": 9992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.569935213636576e-11,
+      "loss": 3.5455,
+      "step": 9993
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8881158817540824e-11,
+      "loss": 2.4202,
+      "step": 9994
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.311191710651194e-11,
+      "loss": 2.0411,
+      "step": 9995
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.39162760835066e-12,
+      "loss": 2.3086,
+      "step": 9996
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.720290818216455e-12,
+      "loss": 2.3724,
+      "step": 9997
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0979071224669357e-12,
+      "loss": 2.5021,
+      "step": 9998
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.244767942169659e-13,
+      "loss": 2.8417,
+      "step": 9999
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 3.6138,
+      "step": 10000
+    },
+    {
+      "epoch": 1.0,
+      "step": 10000,
+      "total_flos": 832965072764928.0,
+      "train_loss": 2.7026718805015086,
+      "train_runtime": 1306.547,
+      "train_samples_per_second": 15.308,
+      "train_steps_per_second": 7.654
+    }
+  ],
+  "max_steps": 10000,
+  "num_train_epochs": 1,
+  "total_flos": 832965072764928.0,
+  "trial_name": null,
+  "trial_params": null
+}