diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,26507 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0661077844311375,
+  "eval_steps": 418,
+  "global_step": 4400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 1.7834,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 1.705169439315796,
+      "eval_runtime": 209.5946,
+      "eval_samples_per_second": 80.15,
+      "eval_steps_per_second": 10.019,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3333333333333336e-07,
+      "loss": 1.7675,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.666666666666667e-07,
+      "loss": 1.7085,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 1.7556,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.333333333333335e-07,
+      "loss": 1.6492,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 1.5909,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.604,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.333333333333334e-07,
+      "loss": 1.6458,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.066666666666667e-06,
+      "loss": 1.621,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.5373,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 1.5868,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4666666666666669e-06,
+      "loss": 1.6981,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.6067,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7333333333333336e-06,
+      "loss": 1.5937,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8666666666666669e-06,
+      "loss": 1.5503,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.5161,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.133333333333334e-06,
+      "loss": 1.5115,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.266666666666667e-06,
+      "loss": 1.5653,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.5911,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5333333333333338e-06,
+      "loss": 1.5638,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.5087,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 1.501,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9333333333333338e-06,
+      "loss": 1.4502,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.066666666666667e-06,
+      "loss": 1.4783,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.4357,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.4822,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4666666666666672e-06,
+      "loss": 1.4888,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 1.4632,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7333333333333337e-06,
+      "loss": 1.443,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.866666666666667e-06,
+      "loss": 1.5011,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.4159,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.133333333333333e-06,
+      "loss": 1.4779,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.266666666666668e-06,
+      "loss": 1.4188,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4e-06,
+      "loss": 1.4443,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.533333333333334e-06,
+      "loss": 1.4362,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 1.4092,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.4526,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.933333333333334e-06,
+      "loss": 1.4514,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0666666666666676e-06,
+      "loss": 1.4167,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2e-06,
+      "loss": 1.4679,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.4219,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.466666666666667e-06,
+      "loss": 1.3988,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.3799,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.733333333333334e-06,
+      "loss": 1.444,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8666666666666675e-06,
+      "loss": 1.425,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-06,
+      "loss": 1.3184,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.133333333333334e-06,
+      "loss": 1.3679,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.266666666666668e-06,
+      "loss": 1.3959,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.4015,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.533333333333334e-06,
+      "loss": 1.3711,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.3107,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 1.5036,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.9333333333333344e-06,
+      "loss": 1.4919,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.066666666666667e-06,
+      "loss": 1.3959,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 1.4575,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 1.3783,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 1.3554,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 1.2484,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.733333333333334e-06,
+      "loss": 1.3534,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.866666666666667e-06,
+      "loss": 1.359,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.2728,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.133333333333334e-06,
+      "loss": 1.3594,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.266666666666667e-06,
+      "loss": 1.3447,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.3821,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 1.3598,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 1.3309,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.8e-06,
+      "loss": 1.3311,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.933333333333333e-06,
+      "loss": 1.3158,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 1.3868,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 1.3843,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 1.4091,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.466666666666667e-06,
+      "loss": 1.3161,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.311,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.733333333333334e-06,
+      "loss": 1.3776,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.866666666666668e-06,
+      "loss": 1.2554,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1e-05,
+      "loss": 1.3485,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0133333333333335e-05,
+      "loss": 1.3391,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0266666666666668e-05,
+      "loss": 1.2648,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.04e-05,
+      "loss": 1.3058,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0533333333333333e-05,
+      "loss": 1.2558,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 1.3259,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 1.3801,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0933333333333334e-05,
+      "loss": 1.3407,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1066666666666669e-05,
+      "loss": 1.2973,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 1.2884,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1333333333333334e-05,
+      "loss": 1.2705,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1466666666666668e-05,
+      "loss": 1.3055,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.16e-05,
+      "loss": 1.1777,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1733333333333335e-05,
+      "loss": 1.2491,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.186666666666667e-05,
+      "loss": 1.2686,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2e-05,
+      "loss": 1.2838,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2133333333333335e-05,
+      "loss": 1.3587,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2266666666666667e-05,
+      "loss": 1.3023,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2400000000000002e-05,
+      "loss": 1.4266,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2533333333333336e-05,
+      "loss": 1.318,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2666666666666667e-05,
+      "loss": 1.3466,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 1.236,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2933333333333334e-05,
+      "loss": 1.2114,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3066666666666668e-05,
+      "loss": 1.28,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3200000000000002e-05,
+      "loss": 1.2843,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.2859,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3466666666666668e-05,
+      "loss": 1.2856,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 1.3029,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3733333333333335e-05,
+      "loss": 1.2598,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3866666666666669e-05,
+      "loss": 1.2493,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4e-05,
+      "loss": 1.3551,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4133333333333334e-05,
+      "loss": 1.3001,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4266666666666668e-05,
+      "loss": 1.2932,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 1.2735,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4533333333333335e-05,
+      "loss": 1.2659,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 1.1751,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.48e-05,
+      "loss": 1.2406,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4933333333333335e-05,
+      "loss": 1.2634,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5066666666666668e-05,
+      "loss": 1.3114,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 1.2798,
+      "step": 115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5333333333333334e-05,
+      "loss": 1.198,
+      "step": 116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.546666666666667e-05,
+      "loss": 1.2142,
+      "step": 117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5600000000000003e-05,
+      "loss": 1.3102,
+      "step": 118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5733333333333334e-05,
+      "loss": 1.2144,
+      "step": 119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.586666666666667e-05,
+      "loss": 1.24,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.2744,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6133333333333334e-05,
+      "loss": 1.2307,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6266666666666668e-05,
+      "loss": 1.4341,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.64e-05,
+      "loss": 1.2288,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6533333333333333e-05,
+      "loss": 1.236,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.1723,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 1.2339,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6933333333333336e-05,
+      "loss": 1.2388,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.706666666666667e-05,
+      "loss": 1.2666,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.72e-05,
+      "loss": 1.1829,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 1.1917,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7466666666666667e-05,
+      "loss": 1.2933,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.76e-05,
+      "loss": 1.2942,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7733333333333335e-05,
+      "loss": 1.2721,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7866666666666666e-05,
+      "loss": 1.2391,
+      "step": 135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8e-05,
+      "loss": 1.3181,
+      "step": 136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8133333333333335e-05,
+      "loss": 1.2428,
+      "step": 137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.826666666666667e-05,
+      "loss": 1.1372,
+      "step": 138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 1.2083,
+      "step": 139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8533333333333334e-05,
+      "loss": 1.2264,
+      "step": 140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 1.3273,
+      "step": 141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.88e-05,
+      "loss": 1.2065,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8933333333333334e-05,
+      "loss": 1.2081,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9066666666666668e-05,
+      "loss": 1.2571,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 1.2027,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9333333333333333e-05,
+      "loss": 1.1851,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9466666666666668e-05,
+      "loss": 1.2532,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 1.2525,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9733333333333336e-05,
+      "loss": 1.254,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9866666666666667e-05,
+      "loss": 1.2051,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 1.2878,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999756038058063e-05,
+      "loss": 1.2315,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999512076116126e-05,
+      "loss": 1.2669,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999268114174189e-05,
+      "loss": 1.1973,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999024152232252e-05,
+      "loss": 1.2596,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9987801902903147e-05,
+      "loss": 1.2562,
+      "step": 156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9985362283483777e-05,
+      "loss": 1.1849,
+      "step": 157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9982922664064406e-05,
+      "loss": 1.3004,
+      "step": 158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9980483044645036e-05,
+      "loss": 1.2603,
+      "step": 159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9978043425225666e-05,
+      "loss": 1.2333,
+      "step": 160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9975603805806296e-05,
+      "loss": 1.219,
+      "step": 161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9973164186386925e-05,
+      "loss": 1.2749,
+      "step": 162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9970724566967555e-05,
+      "loss": 1.1883,
+      "step": 163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9968284947548185e-05,
+      "loss": 1.1976,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9965845328128815e-05,
+      "loss": 1.2106,
+      "step": 165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.996340570870944e-05,
+      "loss": 1.2118,
+      "step": 166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9960966089290074e-05,
+      "loss": 1.2363,
+      "step": 167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9958526469870704e-05,
+      "loss": 1.2239,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.995608685045133e-05,
+      "loss": 1.147,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.995364723103196e-05,
+      "loss": 1.2208,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.995120761161259e-05,
+      "loss": 1.2007,
+      "step": 171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.994876799219322e-05,
+      "loss": 1.2096,
+      "step": 172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.994632837277385e-05,
+      "loss": 1.3066,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.994388875335448e-05,
+      "loss": 1.2188,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9941449133935105e-05,
+      "loss": 1.2083,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.993900951451574e-05,
+      "loss": 1.2238,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9936569895096368e-05,
+      "loss": 1.2118,
+      "step": 177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9934130275676995e-05,
+      "loss": 1.3051,
+      "step": 178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9931690656257624e-05,
+      "loss": 1.205,
+      "step": 179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9929251036838254e-05,
+      "loss": 1.1149,
+      "step": 180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9926811417418884e-05,
+      "loss": 1.2111,
+      "step": 181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9924371797999514e-05,
+      "loss": 1.1801,
+      "step": 182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9921932178580143e-05,
+      "loss": 1.2142,
+      "step": 183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9919492559160773e-05,
+      "loss": 1.1803,
+      "step": 184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9917052939741403e-05,
+      "loss": 1.2109,
+      "step": 185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9914613320322033e-05,
+      "loss": 1.2119,
+      "step": 186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9912173700902662e-05,
+      "loss": 1.1718,
+      "step": 187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.990973408148329e-05,
+      "loss": 1.2502,
+      "step": 188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.990729446206392e-05,
+      "loss": 1.2071,
+      "step": 189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.990485484264455e-05,
+      "loss": 1.2962,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9902415223225178e-05,
+      "loss": 1.2433,
+      "step": 191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9899975603805808e-05,
+      "loss": 1.2124,
+      "step": 192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9897535984386437e-05,
+      "loss": 1.1996,
+      "step": 193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9895096364967067e-05,
+      "loss": 1.1896,
+      "step": 194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9892656745547697e-05,
+      "loss": 1.2105,
+      "step": 195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9890217126128327e-05,
+      "loss": 1.1356,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9887777506708953e-05,
+      "loss": 1.1636,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9885337887289583e-05,
+      "loss": 1.1127,
+      "step": 198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9882898267870216e-05,
+      "loss": 1.2217,
+      "step": 199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9880458648450846e-05,
+      "loss": 1.23,
+      "step": 200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9878019029031472e-05,
+      "loss": 1.1982,
+      "step": 201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.98755794096121e-05,
+      "loss": 1.1973,
+      "step": 202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.987313979019273e-05,
+      "loss": 1.1723,
+      "step": 203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.987070017077336e-05,
+      "loss": 1.1517,
+      "step": 204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.986826055135399e-05,
+      "loss": 1.1476,
+      "step": 205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.986582093193462e-05,
+      "loss": 1.1435,
+      "step": 206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9863381312515247e-05,
+      "loss": 1.151,
+      "step": 207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.986094169309588e-05,
+      "loss": 1.1957,
+      "step": 208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.985850207367651e-05,
+      "loss": 1.1992,
+      "step": 209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9856062454257136e-05,
+      "loss": 1.1877,
+      "step": 210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9853622834837766e-05,
+      "loss": 1.1376,
+      "step": 211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9851183215418396e-05,
+      "loss": 1.1861,
+      "step": 212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9848743595999025e-05,
+      "loss": 1.1546,
+      "step": 213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9846303976579655e-05,
+      "loss": 1.2118,
+      "step": 214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9843864357160285e-05,
+      "loss": 1.1835,
+      "step": 215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9841424737740915e-05,
+      "loss": 1.214,
+      "step": 216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9838985118321544e-05,
+      "loss": 1.1848,
+      "step": 217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9836545498902174e-05,
+      "loss": 1.1699,
+      "step": 218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9834105879482804e-05,
+      "loss": 1.2104,
+      "step": 219
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.983166626006343e-05,
+      "loss": 1.1823,
+      "step": 220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.982922664064406e-05,
+      "loss": 1.1107,
+      "step": 221
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9826787021224693e-05,
+      "loss": 1.1997,
+      "step": 222
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.982434740180532e-05,
+      "loss": 1.2453,
+      "step": 223
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.982190778238595e-05,
+      "loss": 1.2009,
+      "step": 224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.981946816296658e-05,
+      "loss": 1.1926,
+      "step": 225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.981702854354721e-05,
+      "loss": 1.1397,
+      "step": 226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.981458892412784e-05,
+      "loss": 1.246,
+      "step": 227
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9812149304708468e-05,
+      "loss": 1.192,
+      "step": 228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9809709685289095e-05,
+      "loss": 1.2343,
+      "step": 229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9807270065869724e-05,
+      "loss": 1.1803,
+      "step": 230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9804830446450357e-05,
+      "loss": 1.2141,
+      "step": 231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9802390827030984e-05,
+      "loss": 1.1721,
+      "step": 232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9799951207611614e-05,
+      "loss": 1.1657,
+      "step": 233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9797511588192243e-05,
+      "loss": 1.1708,
+      "step": 234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9795071968772873e-05,
+      "loss": 1.0995,
+      "step": 235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9792632349353503e-05,
+      "loss": 1.1849,
+      "step": 236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9790192729934133e-05,
+      "loss": 1.2096,
+      "step": 237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.978775311051476e-05,
+      "loss": 1.1689,
+      "step": 238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9785313491095392e-05,
+      "loss": 1.2285,
+      "step": 239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9782873871676022e-05,
+      "loss": 1.1346,
+      "step": 240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.978043425225665e-05,
+      "loss": 1.1342,
+      "step": 241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9777994632837278e-05,
+      "loss": 1.1717,
+      "step": 242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9775555013417908e-05,
+      "loss": 1.1334,
+      "step": 243
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9773115393998537e-05,
+      "loss": 1.1522,
+      "step": 244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9770675774579167e-05,
+      "loss": 1.1735,
+      "step": 245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9768236155159797e-05,
+      "loss": 1.175,
+      "step": 246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9765796535740427e-05,
+      "loss": 1.1758,
+      "step": 247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9763356916321056e-05,
+      "loss": 1.1449,
+      "step": 248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9760917296901686e-05,
+      "loss": 1.1551,
+      "step": 249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9758477677482316e-05,
+      "loss": 1.2255,
+      "step": 250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9756038058062942e-05,
+      "loss": 1.1611,
+      "step": 251
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9753598438643572e-05,
+      "loss": 1.1895,
+      "step": 252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.97511588192242e-05,
+      "loss": 1.2135,
+      "step": 253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.974871919980483e-05,
+      "loss": 1.1551,
+      "step": 254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.974627958038546e-05,
+      "loss": 1.1105,
+      "step": 255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.974383996096609e-05,
+      "loss": 1.1565,
+      "step": 256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.974140034154672e-05,
+      "loss": 1.126,
+      "step": 257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.973896072212735e-05,
+      "loss": 1.1807,
+      "step": 258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.973652110270798e-05,
+      "loss": 1.237,
+      "step": 259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.973408148328861e-05,
+      "loss": 1.16,
+      "step": 260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9731641863869236e-05,
+      "loss": 1.1118,
+      "step": 261
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9729202244449866e-05,
+      "loss": 1.1849,
+      "step": 262
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.97267626250305e-05,
+      "loss": 1.2236,
+      "step": 263
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9724323005611125e-05,
+      "loss": 1.153,
+      "step": 264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9721883386191755e-05,
+      "loss": 1.1505,
+      "step": 265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9719443766772385e-05,
+      "loss": 1.1572,
+      "step": 266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9717004147353015e-05,
+      "loss": 1.1357,
+      "step": 267
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9714564527933644e-05,
+      "loss": 1.1233,
+      "step": 268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9712124908514274e-05,
+      "loss": 1.087,
+      "step": 269
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.97096852890949e-05,
+      "loss": 1.1145,
+      "step": 270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9707245669675534e-05,
+      "loss": 1.205,
+      "step": 271
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9704806050256163e-05,
+      "loss": 1.1617,
+      "step": 272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.970236643083679e-05,
+      "loss": 1.1116,
+      "step": 273
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.969992681141742e-05,
+      "loss": 1.1934,
+      "step": 274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.969748719199805e-05,
+      "loss": 1.1563,
+      "step": 275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.969504757257868e-05,
+      "loss": 1.2357,
+      "step": 276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.969260795315931e-05,
+      "loss": 1.1676,
+      "step": 277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.969016833373994e-05,
+      "loss": 1.1695,
+      "step": 278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9687728714320568e-05,
+      "loss": 1.174,
+      "step": 279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9685289094901198e-05,
+      "loss": 1.2076,
+      "step": 280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9682849475481828e-05,
+      "loss": 1.2055,
+      "step": 281
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9680409856062457e-05,
+      "loss": 1.1462,
+      "step": 282
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9677970236643084e-05,
+      "loss": 1.13,
+      "step": 283
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9675530617223714e-05,
+      "loss": 1.1121,
+      "step": 284
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9673090997804343e-05,
+      "loss": 1.1281,
+      "step": 285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9670651378384973e-05,
+      "loss": 1.174,
+      "step": 286
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9668211758965603e-05,
+      "loss": 1.0995,
+      "step": 287
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9665772139546233e-05,
+      "loss": 1.1655,
+      "step": 288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9663332520126862e-05,
+      "loss": 1.1924,
+      "step": 289
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9660892900707492e-05,
+      "loss": 1.0869,
+      "step": 290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9658453281288122e-05,
+      "loss": 1.1323,
+      "step": 291
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9656013661868748e-05,
+      "loss": 1.1408,
+      "step": 292
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9653574042449378e-05,
+      "loss": 1.1475,
+      "step": 293
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.965113442303001e-05,
+      "loss": 1.1622,
+      "step": 294
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9648694803610637e-05,
+      "loss": 1.1627,
+      "step": 295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9646255184191267e-05,
+      "loss": 1.1685,
+      "step": 296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9643815564771897e-05,
+      "loss": 1.1606,
+      "step": 297
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9641375945352527e-05,
+      "loss": 1.067,
+      "step": 298
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9638936325933156e-05,
+      "loss": 1.2286,
+      "step": 299
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9636496706513786e-05,
+      "loss": 1.1475,
+      "step": 300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9634057087094416e-05,
+      "loss": 1.2003,
+      "step": 301
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9631617467675042e-05,
+      "loss": 1.1763,
+      "step": 302
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9629177848255675e-05,
+      "loss": 1.171,
+      "step": 303
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9626738228836305e-05,
+      "loss": 1.0576,
+      "step": 304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.962429860941693e-05,
+      "loss": 1.1389,
+      "step": 305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.962185898999756e-05,
+      "loss": 1.1867,
+      "step": 306
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.961941937057819e-05,
+      "loss": 1.221,
+      "step": 307
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.961697975115882e-05,
+      "loss": 1.0795,
+      "step": 308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.961454013173945e-05,
+      "loss": 1.164,
+      "step": 309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.961210051232008e-05,
+      "loss": 1.1847,
+      "step": 310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9609660892900707e-05,
+      "loss": 1.0842,
+      "step": 311
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.960722127348134e-05,
+      "loss": 1.12,
+      "step": 312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.960478165406197e-05,
+      "loss": 1.1408,
+      "step": 313
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9602342034642596e-05,
+      "loss": 1.1419,
+      "step": 314
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9599902415223226e-05,
+      "loss": 1.1099,
+      "step": 315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9597462795803855e-05,
+      "loss": 1.1743,
+      "step": 316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.959502317638449e-05,
+      "loss": 1.1894,
+      "step": 317
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9592583556965115e-05,
+      "loss": 1.202,
+      "step": 318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9590143937545744e-05,
+      "loss": 1.1269,
+      "step": 319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9587704318126374e-05,
+      "loss": 1.1651,
+      "step": 320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9585264698707004e-05,
+      "loss": 1.1229,
+      "step": 321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9582825079287634e-05,
+      "loss": 1.0953,
+      "step": 322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9580385459868263e-05,
+      "loss": 1.0795,
+      "step": 323
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.957794584044889e-05,
+      "loss": 1.1625,
+      "step": 324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.957550622102952e-05,
+      "loss": 1.1576,
+      "step": 325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9573066601610153e-05,
+      "loss": 1.0593,
+      "step": 326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.957062698219078e-05,
+      "loss": 1.1111,
+      "step": 327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.956818736277141e-05,
+      "loss": 1.1164,
+      "step": 328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.956574774335204e-05,
+      "loss": 1.0891,
+      "step": 329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9563308123932668e-05,
+      "loss": 1.2232,
+      "step": 330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9560868504513298e-05,
+      "loss": 1.1344,
+      "step": 331
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9558428885093928e-05,
+      "loss": 1.1172,
+      "step": 332
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9555989265674554e-05,
+      "loss": 1.0891,
+      "step": 333
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9553549646255184e-05,
+      "loss": 1.1724,
+      "step": 334
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9551110026835817e-05,
+      "loss": 1.0828,
+      "step": 335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9548670407416447e-05,
+      "loss": 1.1415,
+      "step": 336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9546230787997073e-05,
+      "loss": 1.1311,
+      "step": 337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9543791168577703e-05,
+      "loss": 1.1437,
+      "step": 338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9541351549158333e-05,
+      "loss": 1.1942,
+      "step": 339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9538911929738962e-05,
+      "loss": 1.1284,
+      "step": 340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9536472310319592e-05,
+      "loss": 1.0725,
+      "step": 341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9534032690900222e-05,
+      "loss": 1.0888,
+      "step": 342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.953159307148085e-05,
+      "loss": 1.1848,
+      "step": 343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.952915345206148e-05,
+      "loss": 1.1947,
+      "step": 344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.952671383264211e-05,
+      "loss": 1.102,
+      "step": 345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9524274213222737e-05,
+      "loss": 1.1824,
+      "step": 346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9521834593803367e-05,
+      "loss": 1.1009,
+      "step": 347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9519394974383997e-05,
+      "loss": 1.1678,
+      "step": 348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9516955354964627e-05,
+      "loss": 1.0832,
+      "step": 349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9514515735545256e-05,
+      "loss": 1.104,
+      "step": 350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9512076116125886e-05,
+      "loss": 1.1134,
+      "step": 351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9509636496706516e-05,
+      "loss": 1.1108,
+      "step": 352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9507196877287146e-05,
+      "loss": 1.0886,
+      "step": 353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9504757257867775e-05,
+      "loss": 1.0529,
+      "step": 354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9502317638448405e-05,
+      "loss": 1.1571,
+      "step": 355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.949987801902903e-05,
+      "loss": 1.1496,
+      "step": 356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.949743839960966e-05,
+      "loss": 1.0431,
+      "step": 357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9494998780190294e-05,
+      "loss": 1.1344,
+      "step": 358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.949255916077092e-05,
+      "loss": 1.056,
+      "step": 359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.949011954135155e-05,
+      "loss": 1.1128,
+      "step": 360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.948767992193218e-05,
+      "loss": 1.1154,
+      "step": 361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.948524030251281e-05,
+      "loss": 1.1734,
+      "step": 362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.948280068309344e-05,
+      "loss": 1.1493,
+      "step": 363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.948036106367407e-05,
+      "loss": 1.1627,
+      "step": 364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9477921444254696e-05,
+      "loss": 1.1471,
+      "step": 365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.947548182483533e-05,
+      "loss": 1.1862,
+      "step": 366
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.947304220541596e-05,
+      "loss": 1.1126,
+      "step": 367
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9470602585996585e-05,
+      "loss": 1.1573,
+      "step": 368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9468162966577215e-05,
+      "loss": 1.1531,
+      "step": 369
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9465723347157845e-05,
+      "loss": 1.1038,
+      "step": 370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9463283727738474e-05,
+      "loss": 1.1326,
+      "step": 371
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9460844108319104e-05,
+      "loss": 1.1588,
+      "step": 372
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9458404488899734e-05,
+      "loss": 1.1843,
+      "step": 373
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.945596486948036e-05,
+      "loss": 1.1215,
+      "step": 374
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9453525250060993e-05,
+      "loss": 1.1192,
+      "step": 375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9451085630641623e-05,
+      "loss": 1.1728,
+      "step": 376
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9448646011222253e-05,
+      "loss": 1.1039,
+      "step": 377
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.944620639180288e-05,
+      "loss": 1.1583,
+      "step": 378
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.944376677238351e-05,
+      "loss": 1.1323,
+      "step": 379
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.944132715296414e-05,
+      "loss": 1.0776,
+      "step": 380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.943888753354477e-05,
+      "loss": 1.1168,
+      "step": 381
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9436447914125398e-05,
+      "loss": 1.0915,
+      "step": 382
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9434008294706028e-05,
+      "loss": 1.1626,
+      "step": 383
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9431568675286658e-05,
+      "loss": 1.0919,
+      "step": 384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9429129055867287e-05,
+      "loss": 1.1398,
+      "step": 385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9426689436447917e-05,
+      "loss": 1.1476,
+      "step": 386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9424249817028543e-05,
+      "loss": 1.1229,
+      "step": 387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9421810197609173e-05,
+      "loss": 1.0961,
+      "step": 388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9419370578189803e-05,
+      "loss": 1.1145,
+      "step": 389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9416930958770433e-05,
+      "loss": 1.0843,
+      "step": 390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9414491339351062e-05,
+      "loss": 1.1695,
+      "step": 391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9412051719931692e-05,
+      "loss": 1.2127,
+      "step": 392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9409612100512322e-05,
+      "loss": 1.0992,
+      "step": 393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.940717248109295e-05,
+      "loss": 1.172,
+      "step": 394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.940473286167358e-05,
+      "loss": 1.2447,
+      "step": 395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.940229324225421e-05,
+      "loss": 1.0941,
+      "step": 396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9399853622834837e-05,
+      "loss": 1.1537,
+      "step": 397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.939741400341547e-05,
+      "loss": 1.085,
+      "step": 398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.93949743839961e-05,
+      "loss": 1.1218,
+      "step": 399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9392534764576727e-05,
+      "loss": 1.0764,
+      "step": 400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9390095145157356e-05,
+      "loss": 1.131,
+      "step": 401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9387655525737986e-05,
+      "loss": 1.1528,
+      "step": 402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9385215906318616e-05,
+      "loss": 1.1731,
+      "step": 403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9382776286899246e-05,
+      "loss": 1.0976,
+      "step": 404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9380336667479875e-05,
+      "loss": 1.1691,
+      "step": 405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9377897048060502e-05,
+      "loss": 1.1015,
+      "step": 406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9375457428641135e-05,
+      "loss": 1.0753,
+      "step": 407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9373017809221765e-05,
+      "loss": 1.1999,
+      "step": 408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.937057818980239e-05,
+      "loss": 1.1682,
+      "step": 409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.936813857038302e-05,
+      "loss": 1.1594,
+      "step": 410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.936569895096365e-05,
+      "loss": 1.0447,
+      "step": 411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.936325933154428e-05,
+      "loss": 1.0797,
+      "step": 412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.936081971212491e-05,
+      "loss": 1.1568,
+      "step": 413
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.935838009270554e-05,
+      "loss": 1.0864,
+      "step": 414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.935594047328617e-05,
+      "loss": 1.1283,
+      "step": 415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.93535008538668e-05,
+      "loss": 1.1013,
+      "step": 416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.935106123444743e-05,
+      "loss": 1.2111,
+      "step": 417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.934862161502806e-05,
+      "loss": 1.1012,
+      "step": 418
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 1.121598720550537,
+      "eval_runtime": 207.8122,
+      "eval_samples_per_second": 80.837,
+      "eval_steps_per_second": 10.105,
+      "step": 418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9346181995608685e-05,
+      "loss": 1.1107,
+      "step": 419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9343742376189315e-05,
+      "loss": 1.1065,
+      "step": 420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9341302756769948e-05,
+      "loss": 1.0603,
+      "step": 421
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9338863137350574e-05,
+      "loss": 1.132,
+      "step": 422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9336423517931204e-05,
+      "loss": 1.194,
+      "step": 423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9333983898511834e-05,
+      "loss": 1.038,
+      "step": 424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9331544279092464e-05,
+      "loss": 1.1841,
+      "step": 425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9329104659673093e-05,
+      "loss": 1.2119,
+      "step": 426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9326665040253723e-05,
+      "loss": 1.1323,
+      "step": 427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.932422542083435e-05,
+      "loss": 1.199,
+      "step": 428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.932178580141498e-05,
+      "loss": 1.1349,
+      "step": 429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9319346181995612e-05,
+      "loss": 1.14,
+      "step": 430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.931690656257624e-05,
+      "loss": 1.1639,
+      "step": 431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.931446694315687e-05,
+      "loss": 1.1654,
+      "step": 432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9312027323737498e-05,
+      "loss": 1.137,
+      "step": 433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9309587704318128e-05,
+      "loss": 1.1143,
+      "step": 434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9307148084898758e-05,
+      "loss": 1.1207,
+      "step": 435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9304708465479387e-05,
+      "loss": 1.1611,
+      "step": 436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9302268846060017e-05,
+      "loss": 1.0054,
+      "step": 437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9299829226640643e-05,
+      "loss": 1.0629,
+      "step": 438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9297389607221277e-05,
+      "loss": 1.1448,
+      "step": 439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9294949987801906e-05,
+      "loss": 1.1693,
+      "step": 440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9292510368382533e-05,
+      "loss": 1.0572,
+      "step": 441
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9290070748963162e-05,
+      "loss": 1.1373,
+      "step": 442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9287631129543792e-05,
+      "loss": 1.1935,
+      "step": 443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9285191510124422e-05,
+      "loss": 1.0687,
+      "step": 444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.928275189070505e-05,
+      "loss": 1.0922,
+      "step": 445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.928031227128568e-05,
+      "loss": 1.1649,
+      "step": 446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.927787265186631e-05,
+      "loss": 1.113,
+      "step": 447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.927543303244694e-05,
+      "loss": 1.1389,
+      "step": 448
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.927299341302757e-05,
+      "loss": 1.1519,
+      "step": 449
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9270553793608197e-05,
+      "loss": 1.1186,
+      "step": 450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9268114174188827e-05,
+      "loss": 1.1568,
+      "step": 451
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9265674554769456e-05,
+      "loss": 1.1461,
+      "step": 452
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.926323493535009e-05,
+      "loss": 0.9733,
+      "step": 453
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9260795315930716e-05,
+      "loss": 1.0806,
+      "step": 454
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9258355696511346e-05,
+      "loss": 1.1184,
+      "step": 455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9255916077091975e-05,
+      "loss": 1.1734,
+      "step": 456
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9253476457672605e-05,
+      "loss": 1.0851,
+      "step": 457
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9251036838253235e-05,
+      "loss": 1.0895,
+      "step": 458
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9248597218833865e-05,
+      "loss": 1.0526,
+      "step": 459
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.924615759941449e-05,
+      "loss": 1.0781,
+      "step": 460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.924371797999512e-05,
+      "loss": 1.1509,
+      "step": 461
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9241278360575754e-05,
+      "loss": 1.093,
+      "step": 462
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.923883874115638e-05,
+      "loss": 1.1784,
+      "step": 463
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.923639912173701e-05,
+      "loss": 1.1402,
+      "step": 464
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.923395950231764e-05,
+      "loss": 1.1432,
+      "step": 465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.923151988289827e-05,
+      "loss": 1.1059,
+      "step": 466
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.92290802634789e-05,
+      "loss": 1.118,
+      "step": 467
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.922664064405953e-05,
+      "loss": 1.11,
+      "step": 468
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9224201024640155e-05,
+      "loss": 1.1351,
+      "step": 469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.922176140522079e-05,
+      "loss": 1.1354,
+      "step": 470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9219321785801418e-05,
+      "loss": 1.1048,
+      "step": 471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9216882166382048e-05,
+      "loss": 1.0658,
+      "step": 472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9214442546962674e-05,
+      "loss": 1.1475,
+      "step": 473
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9212002927543304e-05,
+      "loss": 1.2179,
+      "step": 474
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9209563308123934e-05,
+      "loss": 1.081,
+      "step": 475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9207123688704564e-05,
+      "loss": 1.1456,
+      "step": 476
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9204684069285193e-05,
+      "loss": 1.1261,
+      "step": 477
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9202244449865823e-05,
+      "loss": 1.0643,
+      "step": 478
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9199804830446453e-05,
+      "loss": 1.1333,
+      "step": 479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9197365211027083e-05,
+      "loss": 1.0688,
+      "step": 480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9194925591607712e-05,
+      "loss": 1.0861,
+      "step": 481
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.919248597218834e-05,
+      "loss": 1.0927,
+      "step": 482
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.919004635276897e-05,
+      "loss": 1.124,
+      "step": 483
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9187606733349598e-05,
+      "loss": 1.0937,
+      "step": 484
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9185167113930228e-05,
+      "loss": 1.168,
+      "step": 485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9182727494510858e-05,
+      "loss": 1.1025,
+      "step": 486
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9180287875091487e-05,
+      "loss": 1.0983,
+      "step": 487
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9177848255672117e-05,
+      "loss": 1.0469,
+      "step": 488
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9175408636252747e-05,
+      "loss": 1.1186,
+      "step": 489
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9172969016833377e-05,
+      "loss": 1.1261,
+      "step": 490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9170529397414006e-05,
+      "loss": 1.0348,
+      "step": 491
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9168089777994633e-05,
+      "loss": 1.109,
+      "step": 492
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9165650158575266e-05,
+      "loss": 1.1387,
+      "step": 493
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9163210539155896e-05,
+      "loss": 1.0415,
+      "step": 494
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9160770919736522e-05,
+      "loss": 1.1319,
+      "step": 495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.915833130031715e-05,
+      "loss": 1.0754,
+      "step": 496
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.915589168089778e-05,
+      "loss": 1.118,
+      "step": 497
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.915345206147841e-05,
+      "loss": 1.1083,
+      "step": 498
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.915101244205904e-05,
+      "loss": 1.0942,
+      "step": 499
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.914857282263967e-05,
+      "loss": 1.0602,
+      "step": 500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9146133203220297e-05,
+      "loss": 1.0663,
+      "step": 501
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.914369358380093e-05,
+      "loss": 1.0867,
+      "step": 502
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.914125396438156e-05,
+      "loss": 1.0884,
+      "step": 503
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9138814344962186e-05,
+      "loss": 1.1414,
+      "step": 504
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9136374725542816e-05,
+      "loss": 1.0728,
+      "step": 505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9133935106123446e-05,
+      "loss": 1.1435,
+      "step": 506
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9131495486704075e-05,
+      "loss": 1.082,
+      "step": 507
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9129055867284705e-05,
+      "loss": 1.0782,
+      "step": 508
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9126616247865335e-05,
+      "loss": 1.0428,
+      "step": 509
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.912417662844596e-05,
+      "loss": 1.151,
+      "step": 510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9121737009026594e-05,
+      "loss": 1.1337,
+      "step": 511
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9119297389607224e-05,
+      "loss": 1.0654,
+      "step": 512
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9116857770187854e-05,
+      "loss": 1.0166,
+      "step": 513
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.911441815076848e-05,
+      "loss": 1.0807,
+      "step": 514
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.911197853134911e-05,
+      "loss": 1.1172,
+      "step": 515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.910953891192974e-05,
+      "loss": 1.0826,
+      "step": 516
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.910709929251037e-05,
+      "loss": 1.1192,
+      "step": 517
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9104659673091e-05,
+      "loss": 1.0543,
+      "step": 518
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.910222005367163e-05,
+      "loss": 1.1473,
+      "step": 519
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.909978043425226e-05,
+      "loss": 1.1292,
+      "step": 520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.909734081483289e-05,
+      "loss": 1.0673,
+      "step": 521
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9094901195413518e-05,
+      "loss": 1.0832,
+      "step": 522
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9092461575994145e-05,
+      "loss": 1.1038,
+      "step": 523
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9090021956574774e-05,
+      "loss": 1.0732,
+      "step": 524
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9087582337155407e-05,
+      "loss": 1.0827,
+      "step": 525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9085142717736034e-05,
+      "loss": 1.0765,
+      "step": 526
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9082703098316664e-05,
+      "loss": 1.1085,
+      "step": 527
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9080263478897293e-05,
+      "loss": 1.086,
+      "step": 528
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9077823859477923e-05,
+      "loss": 1.1345,
+      "step": 529
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9075384240058553e-05,
+      "loss": 1.0962,
+      "step": 530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9072944620639183e-05,
+      "loss": 1.0894,
+      "step": 531
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9070505001219812e-05,
+      "loss": 1.0347,
+      "step": 532
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.906806538180044e-05,
+      "loss": 1.1364,
+      "step": 533
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9065625762381072e-05,
+      "loss": 1.0791,
+      "step": 534
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.90631861429617e-05,
+      "loss": 1.1096,
+      "step": 535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9060746523542328e-05,
+      "loss": 1.0511,
+      "step": 536
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9058306904122958e-05,
+      "loss": 1.043,
+      "step": 537
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9055867284703587e-05,
+      "loss": 1.0793,
+      "step": 538
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9053427665284217e-05,
+      "loss": 1.0914,
+      "step": 539
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9050988045864847e-05,
+      "loss": 1.1449,
+      "step": 540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9048548426445477e-05,
+      "loss": 1.0606,
+      "step": 541
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9046108807026106e-05,
+      "loss": 1.0569,
+      "step": 542
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9043669187606736e-05,
+      "loss": 1.0596,
+      "step": 543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9041229568187366e-05,
+      "loss": 1.0781,
+      "step": 544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9038789948767992e-05,
+      "loss": 1.084,
+      "step": 545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9036350329348622e-05,
+      "loss": 1.0404,
+      "step": 546
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.903391070992925e-05,
+      "loss": 1.1647,
+      "step": 547
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.903147109050988e-05,
+      "loss": 1.128,
+      "step": 548
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.902903147109051e-05,
+      "loss": 1.0781,
+      "step": 549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.902659185167114e-05,
+      "loss": 1.0831,
+      "step": 550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.902415223225177e-05,
+      "loss": 1.117,
+      "step": 551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.90217126128324e-05,
+      "loss": 1.0601,
+      "step": 552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.901927299341303e-05,
+      "loss": 1.1177,
+      "step": 553
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.901683337399366e-05,
+      "loss": 1.0434,
+      "step": 554
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9014393754574286e-05,
+      "loss": 1.1361,
+      "step": 555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9011954135154916e-05,
+      "loss": 1.045,
+      "step": 556
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.900951451573555e-05,
+      "loss": 1.0672,
+      "step": 557
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9007074896316175e-05,
+      "loss": 1.0952,
+      "step": 558
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9004635276896805e-05,
+      "loss": 1.0118,
+      "step": 559
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9002195657477435e-05,
+      "loss": 1.0616,
+      "step": 560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8999756038058065e-05,
+      "loss": 1.0601,
+      "step": 561
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8997316418638694e-05,
+      "loss": 1.1062,
+      "step": 562
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8994876799219324e-05,
+      "loss": 1.0735,
+      "step": 563
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.899243717979995e-05,
+      "loss": 1.0575,
+      "step": 564
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.898999756038058e-05,
+      "loss": 1.0646,
+      "step": 565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8987557940961213e-05,
+      "loss": 1.0707,
+      "step": 566
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.898511832154184e-05,
+      "loss": 1.0439,
+      "step": 567
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.898267870212247e-05,
+      "loss": 1.079,
+      "step": 568
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.89802390827031e-05,
+      "loss": 1.103,
+      "step": 569
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.897779946328373e-05,
+      "loss": 1.0878,
+      "step": 570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.897535984386436e-05,
+      "loss": 1.1308,
+      "step": 571
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.897292022444499e-05,
+      "loss": 1.0889,
+      "step": 572
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8970480605025618e-05,
+      "loss": 1.088,
+      "step": 573
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8968040985606248e-05,
+      "loss": 1.0666,
+      "step": 574
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8965601366186878e-05,
+      "loss": 1.057,
+      "step": 575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8963161746767507e-05,
+      "loss": 1.0681,
+      "step": 576
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8960722127348134e-05,
+      "loss": 1.0642,
+      "step": 577
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8958282507928764e-05,
+      "loss": 1.0912,
+      "step": 578
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8955842888509393e-05,
+      "loss": 1.005,
+      "step": 579
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8953403269090023e-05,
+      "loss": 1.1446,
+      "step": 580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8950963649670653e-05,
+      "loss": 1.0272,
+      "step": 581
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8948524030251283e-05,
+      "loss": 1.0633,
+      "step": 582
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8946084410831912e-05,
+      "loss": 1.0701,
+      "step": 583
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8943644791412542e-05,
+      "loss": 1.1662,
+      "step": 584
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8941205171993172e-05,
+      "loss": 1.0838,
+      "step": 585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8938765552573798e-05,
+      "loss": 1.0893,
+      "step": 586
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8936325933154428e-05,
+      "loss": 1.0468,
+      "step": 587
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8933886313735058e-05,
+      "loss": 1.0883,
+      "step": 588
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.893144669431569e-05,
+      "loss": 1.0331,
+      "step": 589
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8929007074896317e-05,
+      "loss": 1.0405,
+      "step": 590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8926567455476947e-05,
+      "loss": 1.0618,
+      "step": 591
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8924127836057577e-05,
+      "loss": 1.0508,
+      "step": 592
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8921688216638206e-05,
+      "loss": 1.1264,
+      "step": 593
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8919248597218836e-05,
+      "loss": 1.0283,
+      "step": 594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8916808977799466e-05,
+      "loss": 1.1046,
+      "step": 595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8914369358380092e-05,
+      "loss": 1.0548,
+      "step": 596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8911929738960725e-05,
+      "loss": 1.0488,
+      "step": 597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8909490119541355e-05,
+      "loss": 1.0538,
+      "step": 598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.890705050012198e-05,
+      "loss": 1.0178,
+      "step": 599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.890461088070261e-05,
+      "loss": 1.0757,
+      "step": 600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.890217126128324e-05,
+      "loss": 1.0624,
+      "step": 601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.889973164186387e-05,
+      "loss": 1.0751,
+      "step": 602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.88972920224445e-05,
+      "loss": 1.106,
+      "step": 603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.889485240302513e-05,
+      "loss": 1.0562,
+      "step": 604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8892412783605757e-05,
+      "loss": 1.0511,
+      "step": 605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.888997316418639e-05,
+      "loss": 1.0767,
+      "step": 606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.888753354476702e-05,
+      "loss": 1.0059,
+      "step": 607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.888509392534765e-05,
+      "loss": 1.1089,
+      "step": 608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8882654305928275e-05,
+      "loss": 1.1137,
+      "step": 609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8880214686508905e-05,
+      "loss": 1.0348,
+      "step": 610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8877775067089535e-05,
+      "loss": 1.1074,
+      "step": 611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8875335447670165e-05,
+      "loss": 1.0199,
+      "step": 612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8872895828250794e-05,
+      "loss": 1.1085,
+      "step": 613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8870456208831424e-05,
+      "loss": 1.0587,
+      "step": 614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8868016589412054e-05,
+      "loss": 1.0418,
+      "step": 615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8865576969992684e-05,
+      "loss": 1.1098,
+      "step": 616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8863137350573313e-05,
+      "loss": 1.0672,
+      "step": 617
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.886069773115394e-05,
+      "loss": 1.0093,
+      "step": 618
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.885825811173457e-05,
+      "loss": 1.0801,
+      "step": 619
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8855818492315203e-05,
+      "loss": 1.1067,
+      "step": 620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.885337887289583e-05,
+      "loss": 1.0165,
+      "step": 621
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.885093925347646e-05,
+      "loss": 1.0327,
+      "step": 622
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.884849963405709e-05,
+      "loss": 1.055,
+      "step": 623
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8846060014637718e-05,
+      "loss": 1.0979,
+      "step": 624
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8843620395218348e-05,
+      "loss": 1.1102,
+      "step": 625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8841180775798978e-05,
+      "loss": 1.0495,
+      "step": 626
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8838741156379607e-05,
+      "loss": 1.0738,
+      "step": 627
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8836301536960234e-05,
+      "loss": 1.0558,
+      "step": 628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8833861917540867e-05,
+      "loss": 1.0851,
+      "step": 629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8831422298121497e-05,
+      "loss": 1.1186,
+      "step": 630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8828982678702123e-05,
+      "loss": 1.1491,
+      "step": 631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8826543059282753e-05,
+      "loss": 1.1003,
+      "step": 632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8824103439863383e-05,
+      "loss": 0.994,
+      "step": 633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8821663820444012e-05,
+      "loss": 1.0964,
+      "step": 634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8819224201024642e-05,
+      "loss": 1.0801,
+      "step": 635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8816784581605272e-05,
+      "loss": 1.1159,
+      "step": 636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8814344962185898e-05,
+      "loss": 1.0844,
+      "step": 637
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.881190534276653e-05,
+      "loss": 1.0993,
+      "step": 638
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.880946572334716e-05,
+      "loss": 1.0684,
+      "step": 639
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8807026103927787e-05,
+      "loss": 1.1314,
+      "step": 640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8804586484508417e-05,
+      "loss": 1.0547,
+      "step": 641
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8802146865089047e-05,
+      "loss": 1.2133,
+      "step": 642
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8799707245669677e-05,
+      "loss": 1.0521,
+      "step": 643
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8797267626250306e-05,
+      "loss": 1.128,
+      "step": 644
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8794828006830936e-05,
+      "loss": 1.0169,
+      "step": 645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8792388387411566e-05,
+      "loss": 1.1265,
+      "step": 646
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8789948767992196e-05,
+      "loss": 1.0253,
+      "step": 647
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8787509148572825e-05,
+      "loss": 1.0844,
+      "step": 648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8785069529153455e-05,
+      "loss": 1.0268,
+      "step": 649
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.878262990973408e-05,
+      "loss": 1.0844,
+      "step": 650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.878019029031471e-05,
+      "loss": 1.0059,
+      "step": 651
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8777750670895344e-05,
+      "loss": 1.1092,
+      "step": 652
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.877531105147597e-05,
+      "loss": 1.1884,
+      "step": 653
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.87728714320566e-05,
+      "loss": 1.0944,
+      "step": 654
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.877043181263723e-05,
+      "loss": 1.1617,
+      "step": 655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.876799219321786e-05,
+      "loss": 1.0215,
+      "step": 656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.876555257379849e-05,
+      "loss": 1.0259,
+      "step": 657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.876311295437912e-05,
+      "loss": 1.0887,
+      "step": 658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8760673334959746e-05,
+      "loss": 1.0571,
+      "step": 659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8758233715540376e-05,
+      "loss": 1.054,
+      "step": 660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.875579409612101e-05,
+      "loss": 1.0815,
+      "step": 661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8753354476701635e-05,
+      "loss": 1.0595,
+      "step": 662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8750914857282265e-05,
+      "loss": 1.0864,
+      "step": 663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8748475237862894e-05,
+      "loss": 1.0847,
+      "step": 664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8746035618443524e-05,
+      "loss": 1.0236,
+      "step": 665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8743595999024154e-05,
+      "loss": 1.0952,
+      "step": 666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8741156379604784e-05,
+      "loss": 1.0352,
+      "step": 667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8738716760185413e-05,
+      "loss": 1.1308,
+      "step": 668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8736277140766043e-05,
+      "loss": 1.0357,
+      "step": 669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8733837521346673e-05,
+      "loss": 1.0264,
+      "step": 670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8731397901927303e-05,
+      "loss": 1.043,
+      "step": 671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.872895828250793e-05,
+      "loss": 1.042,
+      "step": 672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.872651866308856e-05,
+      "loss": 1.0191,
+      "step": 673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.872407904366919e-05,
+      "loss": 1.0834,
+      "step": 674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8721639424249818e-05,
+      "loss": 1.1077,
+      "step": 675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8719199804830448e-05,
+      "loss": 1.0748,
+      "step": 676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8716760185411078e-05,
+      "loss": 1.0717,
+      "step": 677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8714320565991708e-05,
+      "loss": 1.0918,
+      "step": 678
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8711880946572337e-05,
+      "loss": 0.9862,
+      "step": 679
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8709441327152967e-05,
+      "loss": 1.0533,
+      "step": 680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8707001707733593e-05,
+      "loss": 1.0824,
+      "step": 681
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8704562088314223e-05,
+      "loss": 1.0795,
+      "step": 682
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8702122468894853e-05,
+      "loss": 1.0021,
+      "step": 683
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8699682849475483e-05,
+      "loss": 1.1045,
+      "step": 684
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8697243230056112e-05,
+      "loss": 1.0653,
+      "step": 685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8694803610636742e-05,
+      "loss": 1.0815,
+      "step": 686
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8692363991217372e-05,
+      "loss": 1.0786,
+      "step": 687
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8689924371798e-05,
+      "loss": 1.0504,
+      "step": 688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.868748475237863e-05,
+      "loss": 1.0282,
+      "step": 689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.868504513295926e-05,
+      "loss": 1.0141,
+      "step": 690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8682605513539887e-05,
+      "loss": 1.0963,
+      "step": 691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8680165894120517e-05,
+      "loss": 1.0837,
+      "step": 692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.867772627470115e-05,
+      "loss": 1.0834,
+      "step": 693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8675286655281777e-05,
+      "loss": 1.0896,
+      "step": 694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8672847035862406e-05,
+      "loss": 1.0635,
+      "step": 695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8670407416443036e-05,
+      "loss": 1.0612,
+      "step": 696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8667967797023666e-05,
+      "loss": 1.0077,
+      "step": 697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8665528177604296e-05,
+      "loss": 1.0488,
+      "step": 698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8663088558184925e-05,
+      "loss": 1.0646,
+      "step": 699
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8660648938765552e-05,
+      "loss": 1.0544,
+      "step": 700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8658209319346185e-05,
+      "loss": 1.0725,
+      "step": 701
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8655769699926815e-05,
+      "loss": 1.0653,
+      "step": 702
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.865333008050744e-05,
+      "loss": 1.1989,
+      "step": 703
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.865089046108807e-05,
+      "loss": 1.0131,
+      "step": 704
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.86484508416687e-05,
+      "loss": 1.0709,
+      "step": 705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.864601122224933e-05,
+      "loss": 1.0624,
+      "step": 706
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.864357160282996e-05,
+      "loss": 1.0315,
+      "step": 707
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.864113198341059e-05,
+      "loss": 1.1129,
+      "step": 708
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.863869236399122e-05,
+      "loss": 1.0429,
+      "step": 709
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.863625274457185e-05,
+      "loss": 1.0596,
+      "step": 710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.863381312515248e-05,
+      "loss": 1.096,
+      "step": 711
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.863137350573311e-05,
+      "loss": 1.0251,
+      "step": 712
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8628933886313735e-05,
+      "loss": 1.0901,
+      "step": 713
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8626494266894365e-05,
+      "loss": 1.0162,
+      "step": 714
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8624054647474995e-05,
+      "loss": 1.0126,
+      "step": 715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8621615028055624e-05,
+      "loss": 1.0002,
+      "step": 716
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8619175408636254e-05,
+      "loss": 1.0152,
+      "step": 717
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8616735789216884e-05,
+      "loss": 1.048,
+      "step": 718
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8614296169797513e-05,
+      "loss": 1.0679,
+      "step": 719
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8611856550378143e-05,
+      "loss": 1.0649,
+      "step": 720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8609416930958773e-05,
+      "loss": 1.0252,
+      "step": 721
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.86069773115394e-05,
+      "loss": 1.1268,
+      "step": 722
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.860453769212003e-05,
+      "loss": 1.0698,
+      "step": 723
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8602098072700662e-05,
+      "loss": 1.0329,
+      "step": 724
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8599658453281292e-05,
+      "loss": 1.1713,
+      "step": 725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.859721883386192e-05,
+      "loss": 1.0974,
+      "step": 726
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8594779214442548e-05,
+      "loss": 1.0258,
+      "step": 727
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8592339595023178e-05,
+      "loss": 1.0478,
+      "step": 728
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8589899975603808e-05,
+      "loss": 1.1165,
+      "step": 729
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8587460356184437e-05,
+      "loss": 1.0797,
+      "step": 730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8585020736765067e-05,
+      "loss": 1.111,
+      "step": 731
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8582581117345693e-05,
+      "loss": 1.0996,
+      "step": 732
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8580141497926327e-05,
+      "loss": 1.0613,
+      "step": 733
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8577701878506956e-05,
+      "loss": 1.1332,
+      "step": 734
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8575262259087583e-05,
+      "loss": 1.0871,
+      "step": 735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8572822639668212e-05,
+      "loss": 1.0115,
+      "step": 736
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8570383020248842e-05,
+      "loss": 1.0812,
+      "step": 737
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8567943400829472e-05,
+      "loss": 1.0795,
+      "step": 738
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.85655037814101e-05,
+      "loss": 1.0735,
+      "step": 739
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.856306416199073e-05,
+      "loss": 0.9623,
+      "step": 740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8560624542571358e-05,
+      "loss": 1.0512,
+      "step": 741
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.855818492315199e-05,
+      "loss": 1.0862,
+      "step": 742
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.855574530373262e-05,
+      "loss": 1.0749,
+      "step": 743
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.855330568431325e-05,
+      "loss": 1.109,
+      "step": 744
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8550866064893877e-05,
+      "loss": 1.0507,
+      "step": 745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8548426445474506e-05,
+      "loss": 1.0708,
+      "step": 746
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.854598682605514e-05,
+      "loss": 1.05,
+      "step": 747
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8543547206635766e-05,
+      "loss": 1.0027,
+      "step": 748
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8541107587216396e-05,
+      "loss": 1.0177,
+      "step": 749
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8538667967797025e-05,
+      "loss": 1.0395,
+      "step": 750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8536228348377655e-05,
+      "loss": 0.9994,
+      "step": 751
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8533788728958285e-05,
+      "loss": 1.0572,
+      "step": 752
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8531349109538915e-05,
+      "loss": 1.0041,
+      "step": 753
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.852890949011954e-05,
+      "loss": 1.1039,
+      "step": 754
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.852646987070017e-05,
+      "loss": 1.0529,
+      "step": 755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8524030251280804e-05,
+      "loss": 1.0253,
+      "step": 756
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.852159063186143e-05,
+      "loss": 1.0928,
+      "step": 757
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.851915101244206e-05,
+      "loss": 1.0359,
+      "step": 758
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.851671139302269e-05,
+      "loss": 1.0171,
+      "step": 759
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.851427177360332e-05,
+      "loss": 1.082,
+      "step": 760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.851183215418395e-05,
+      "loss": 1.0933,
+      "step": 761
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.850939253476458e-05,
+      "loss": 1.0874,
+      "step": 762
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.850695291534521e-05,
+      "loss": 1.0196,
+      "step": 763
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8504513295925835e-05,
+      "loss": 1.0272,
+      "step": 764
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8502073676506468e-05,
+      "loss": 1.0379,
+      "step": 765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8499634057087098e-05,
+      "loss": 1.0536,
+      "step": 766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8497194437667724e-05,
+      "loss": 1.0565,
+      "step": 767
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8494754818248354e-05,
+      "loss": 1.0877,
+      "step": 768
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8492315198828984e-05,
+      "loss": 1.0423,
+      "step": 769
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8489875579409614e-05,
+      "loss": 0.9537,
+      "step": 770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8487435959990243e-05,
+      "loss": 1.0703,
+      "step": 771
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8484996340570873e-05,
+      "loss": 1.0673,
+      "step": 772
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8482556721151503e-05,
+      "loss": 1.0235,
+      "step": 773
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8480117101732132e-05,
+      "loss": 1.0979,
+      "step": 774
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8477677482312762e-05,
+      "loss": 1.0294,
+      "step": 775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.847523786289339e-05,
+      "loss": 1.0501,
+      "step": 776
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.847279824347402e-05,
+      "loss": 1.0968,
+      "step": 777
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8470358624054648e-05,
+      "loss": 1.1092,
+      "step": 778
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8467919004635278e-05,
+      "loss": 1.0737,
+      "step": 779
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8465479385215908e-05,
+      "loss": 1.0379,
+      "step": 780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8463039765796537e-05,
+      "loss": 1.0523,
+      "step": 781
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8460600146377167e-05,
+      "loss": 1.1,
+      "step": 782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8458160526957797e-05,
+      "loss": 1.0407,
+      "step": 783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8455720907538427e-05,
+      "loss": 1.0227,
+      "step": 784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8453281288119056e-05,
+      "loss": 0.9887,
+      "step": 785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8450841668699683e-05,
+      "loss": 1.0912,
+      "step": 786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8448402049280312e-05,
+      "loss": 1.0297,
+      "step": 787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8445962429860946e-05,
+      "loss": 1.0761,
+      "step": 788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8443522810441572e-05,
+      "loss": 1.1252,
+      "step": 789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.84410831910222e-05,
+      "loss": 1.0313,
+      "step": 790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.843864357160283e-05,
+      "loss": 1.0543,
+      "step": 791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.843620395218346e-05,
+      "loss": 1.1355,
+      "step": 792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.843376433276409e-05,
+      "loss": 0.9838,
+      "step": 793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.843132471334472e-05,
+      "loss": 1.1023,
+      "step": 794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8428885093925347e-05,
+      "loss": 1.0389,
+      "step": 795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.842644547450598e-05,
+      "loss": 1.0328,
+      "step": 796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.842400585508661e-05,
+      "loss": 1.2162,
+      "step": 797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8421566235667236e-05,
+      "loss": 1.084,
+      "step": 798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8419126616247866e-05,
+      "loss": 0.9897,
+      "step": 799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8416686996828496e-05,
+      "loss": 1.1175,
+      "step": 800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8414247377409125e-05,
+      "loss": 1.0185,
+      "step": 801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8411807757989755e-05,
+      "loss": 1.0387,
+      "step": 802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8409368138570385e-05,
+      "loss": 1.0585,
+      "step": 803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8406928519151015e-05,
+      "loss": 1.0514,
+      "step": 804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8404488899731644e-05,
+      "loss": 0.9732,
+      "step": 805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8402049280312274e-05,
+      "loss": 1.0507,
+      "step": 806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8399609660892904e-05,
+      "loss": 1.0892,
+      "step": 807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.839717004147353e-05,
+      "loss": 1.0707,
+      "step": 808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.839473042205416e-05,
+      "loss": 1.0472,
+      "step": 809
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.839229080263479e-05,
+      "loss": 1.0141,
+      "step": 810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.838985118321542e-05,
+      "loss": 1.0481,
+      "step": 811
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.838741156379605e-05,
+      "loss": 1.0265,
+      "step": 812
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.838497194437668e-05,
+      "loss": 1.062,
+      "step": 813
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.838253232495731e-05,
+      "loss": 1.0149,
+      "step": 814
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.838009270553794e-05,
+      "loss": 1.0327,
+      "step": 815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8377653086118568e-05,
+      "loss": 1.0078,
+      "step": 816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8375213466699195e-05,
+      "loss": 1.0238,
+      "step": 817
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8372773847279824e-05,
+      "loss": 1.0716,
+      "step": 818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8370334227860454e-05,
+      "loss": 1.0752,
+      "step": 819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8367894608441084e-05,
+      "loss": 1.0825,
+      "step": 820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8365454989021714e-05,
+      "loss": 1.0286,
+      "step": 821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8363015369602343e-05,
+      "loss": 1.0752,
+      "step": 822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8360575750182973e-05,
+      "loss": 1.0563,
+      "step": 823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8358136130763603e-05,
+      "loss": 1.0815,
+      "step": 824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8355696511344233e-05,
+      "loss": 1.0213,
+      "step": 825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8353256891924862e-05,
+      "loss": 1.0288,
+      "step": 826
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.835081727250549e-05,
+      "loss": 1.0902,
+      "step": 827
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8348377653086122e-05,
+      "loss": 0.9559,
+      "step": 828
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.834593803366675e-05,
+      "loss": 1.0118,
+      "step": 829
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8343498414247378e-05,
+      "loss": 1.0748,
+      "step": 830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8341058794828008e-05,
+      "loss": 1.1046,
+      "step": 831
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8338619175408637e-05,
+      "loss": 1.0558,
+      "step": 832
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8336179555989267e-05,
+      "loss": 1.059,
+      "step": 833
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8333739936569897e-05,
+      "loss": 1.067,
+      "step": 834
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8331300317150527e-05,
+      "loss": 0.9523,
+      "step": 835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8328860697731153e-05,
+      "loss": 0.9826,
+      "step": 836
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 1.0543705224990845,
+      "eval_runtime": 207.8815,
+      "eval_samples_per_second": 80.81,
+      "eval_steps_per_second": 10.102,
+      "step": 836
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8326421078311786e-05,
+      "loss": 1.0802,
+      "step": 837
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8323981458892416e-05,
+      "loss": 1.0423,
+      "step": 838
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8321541839473042e-05,
+      "loss": 1.2377,
+      "step": 839
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8319102220053672e-05,
+      "loss": 1.1044,
+      "step": 840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.83166626006343e-05,
+      "loss": 1.0628,
+      "step": 841
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.831422298121493e-05,
+      "loss": 1.1256,
+      "step": 842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.831178336179556e-05,
+      "loss": 1.0241,
+      "step": 843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.830934374237619e-05,
+      "loss": 1.0926,
+      "step": 844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.830690412295682e-05,
+      "loss": 0.9993,
+      "step": 845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.830446450353745e-05,
+      "loss": 1.0483,
+      "step": 846
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.830202488411808e-05,
+      "loss": 1.0425,
+      "step": 847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.829958526469871e-05,
+      "loss": 1.0572,
+      "step": 848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8297145645279336e-05,
+      "loss": 1.1099,
+      "step": 849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8294706025859966e-05,
+      "loss": 0.9964,
+      "step": 850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.82922664064406e-05,
+      "loss": 1.046,
+      "step": 851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8289826787021225e-05,
+      "loss": 1.0139,
+      "step": 852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8287387167601855e-05,
+      "loss": 1.0187,
+      "step": 853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8284947548182485e-05,
+      "loss": 1.0487,
+      "step": 854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8282507928763115e-05,
+      "loss": 1.0402,
+      "step": 855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8280068309343744e-05,
+      "loss": 1.0399,
+      "step": 856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8277628689924374e-05,
+      "loss": 1.0238,
+      "step": 857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8275189070505e-05,
+      "loss": 1.0748,
+      "step": 858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.827274945108563e-05,
+      "loss": 1.0784,
+      "step": 859
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8270309831666263e-05,
+      "loss": 1.0381,
+      "step": 860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8267870212246893e-05,
+      "loss": 1.0642,
+      "step": 861
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.826543059282752e-05,
+      "loss": 0.9604,
+      "step": 862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.826299097340815e-05,
+      "loss": 1.0081,
+      "step": 863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.826055135398878e-05,
+      "loss": 0.9446,
+      "step": 864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.825811173456941e-05,
+      "loss": 1.0669,
+      "step": 865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.825567211515004e-05,
+      "loss": 0.9849,
+      "step": 866
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8253232495730668e-05,
+      "loss": 1.0384,
+      "step": 867
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8250792876311295e-05,
+      "loss": 1.0492,
+      "step": 868
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8248353256891928e-05,
+      "loss": 1.001,
+      "step": 869
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8245913637472557e-05,
+      "loss": 1.0457,
+      "step": 870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8243474018053184e-05,
+      "loss": 0.9735,
+      "step": 871
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8241034398633814e-05,
+      "loss": 1.0187,
+      "step": 872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8238594779214443e-05,
+      "loss": 1.0872,
+      "step": 873
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8236155159795073e-05,
+      "loss": 1.0618,
+      "step": 874
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8233715540375703e-05,
+      "loss": 1.0548,
+      "step": 875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8231275920956333e-05,
+      "loss": 1.0044,
+      "step": 876
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8228836301536962e-05,
+      "loss": 0.9912,
+      "step": 877
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8226396682117592e-05,
+      "loss": 1.0724,
+      "step": 878
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8223957062698222e-05,
+      "loss": 1.0386,
+      "step": 879
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.822151744327885e-05,
+      "loss": 1.045,
+      "step": 880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8219077823859478e-05,
+      "loss": 1.0799,
+      "step": 881
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8216638204440108e-05,
+      "loss": 1.106,
+      "step": 882
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.821419858502074e-05,
+      "loss": 1.0024,
+      "step": 883
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8211758965601367e-05,
+      "loss": 1.0134,
+      "step": 884
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8209319346181997e-05,
+      "loss": 1.0912,
+      "step": 885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8206879726762627e-05,
+      "loss": 0.989,
+      "step": 886
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8204440107343256e-05,
+      "loss": 0.8884,
+      "step": 887
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8202000487923886e-05,
+      "loss": 0.9997,
+      "step": 888
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8199560868504516e-05,
+      "loss": 1.1022,
+      "step": 889
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8197121249085142e-05,
+      "loss": 1.0792,
+      "step": 890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8194681629665772e-05,
+      "loss": 1.0095,
+      "step": 891
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8192242010246405e-05,
+      "loss": 1.0677,
+      "step": 892
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.818980239082703e-05,
+      "loss": 1.0592,
+      "step": 893
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.818736277140766e-05,
+      "loss": 1.0771,
+      "step": 894
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.818492315198829e-05,
+      "loss": 0.9941,
+      "step": 895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.818248353256892e-05,
+      "loss": 1.0235,
+      "step": 896
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.818004391314955e-05,
+      "loss": 1.0525,
+      "step": 897
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.817760429373018e-05,
+      "loss": 1.0277,
+      "step": 898
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.817516467431081e-05,
+      "loss": 1.0751,
+      "step": 899
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.817272505489144e-05,
+      "loss": 1.0538,
+      "step": 900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.817028543547207e-05,
+      "loss": 1.0837,
+      "step": 901
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.81678458160527e-05,
+      "loss": 1.0487,
+      "step": 902
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8165406196633325e-05,
+      "loss": 1.0088,
+      "step": 903
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8162966577213955e-05,
+      "loss": 0.9884,
+      "step": 904
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8160526957794585e-05,
+      "loss": 1.0361,
+      "step": 905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8158087338375215e-05,
+      "loss": 1.0034,
+      "step": 906
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8155647718955844e-05,
+      "loss": 1.1115,
+      "step": 907
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8153208099536474e-05,
+      "loss": 1.0812,
+      "step": 908
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8150768480117104e-05,
+      "loss": 1.0237,
+      "step": 909
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8148328860697734e-05,
+      "loss": 1.067,
+      "step": 910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8145889241278363e-05,
+      "loss": 1.0987,
+      "step": 911
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.814344962185899e-05,
+      "loss": 1.0805,
+      "step": 912
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.814101000243962e-05,
+      "loss": 1.0117,
+      "step": 913
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.813857038302025e-05,
+      "loss": 1.0792,
+      "step": 914
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.813613076360088e-05,
+      "loss": 0.9906,
+      "step": 915
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.813369114418151e-05,
+      "loss": 1.0912,
+      "step": 916
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.813125152476214e-05,
+      "loss": 1.0295,
+      "step": 917
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8128811905342768e-05,
+      "loss": 1.0115,
+      "step": 918
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8126372285923398e-05,
+      "loss": 1.0242,
+      "step": 919
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8123932666504028e-05,
+      "loss": 1.1317,
+      "step": 920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8121493047084657e-05,
+      "loss": 1.023,
+      "step": 921
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8119053427665284e-05,
+      "loss": 1.1007,
+      "step": 922
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8116613808245917e-05,
+      "loss": 0.9901,
+      "step": 923
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8114174188826547e-05,
+      "loss": 1.0359,
+      "step": 924
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8111734569407173e-05,
+      "loss": 1.0925,
+      "step": 925
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8109294949987803e-05,
+      "loss": 1.0149,
+      "step": 926
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8106855330568433e-05,
+      "loss": 1.1541,
+      "step": 927
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8104415711149062e-05,
+      "loss": 1.0538,
+      "step": 928
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8101976091729692e-05,
+      "loss": 1.0555,
+      "step": 929
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8099536472310322e-05,
+      "loss": 1.0927,
+      "step": 930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8097096852890948e-05,
+      "loss": 1.0669,
+      "step": 931
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.809465723347158e-05,
+      "loss": 1.01,
+      "step": 932
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.809221761405221e-05,
+      "loss": 1.113,
+      "step": 933
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8089777994632837e-05,
+      "loss": 1.0429,
+      "step": 934
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8087338375213467e-05,
+      "loss": 1.0401,
+      "step": 935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8084898755794097e-05,
+      "loss": 1.116,
+      "step": 936
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8082459136374727e-05,
+      "loss": 0.9721,
+      "step": 937
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8080019516955356e-05,
+      "loss": 1.0297,
+      "step": 938
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8077579897535986e-05,
+      "loss": 1.0802,
+      "step": 939
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8075140278116616e-05,
+      "loss": 1.0573,
+      "step": 940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8072700658697246e-05,
+      "loss": 1.0788,
+      "step": 941
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8070261039277875e-05,
+      "loss": 1.0372,
+      "step": 942
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8067821419858505e-05,
+      "loss": 1.0641,
+      "step": 943
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.806538180043913e-05,
+      "loss": 0.9897,
+      "step": 944
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.806294218101976e-05,
+      "loss": 1.0793,
+      "step": 945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.806050256160039e-05,
+      "loss": 1.0346,
+      "step": 946
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.805806294218102e-05,
+      "loss": 1.0076,
+      "step": 947
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.805562332276165e-05,
+      "loss": 0.9664,
+      "step": 948
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.805318370334228e-05,
+      "loss": 1.0229,
+      "step": 949
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.805074408392291e-05,
+      "loss": 1.0295,
+      "step": 950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.804830446450354e-05,
+      "loss": 0.9936,
+      "step": 951
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.804586484508417e-05,
+      "loss": 1.0468,
+      "step": 952
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8043425225664796e-05,
+      "loss": 1.0912,
+      "step": 953
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8040985606245425e-05,
+      "loss": 0.9963,
+      "step": 954
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.803854598682606e-05,
+      "loss": 1.0094,
+      "step": 955
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8036106367406685e-05,
+      "loss": 1.0419,
+      "step": 956
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8033666747987315e-05,
+      "loss": 1.0286,
+      "step": 957
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8031227128567944e-05,
+      "loss": 1.0029,
+      "step": 958
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8028787509148574e-05,
+      "loss": 0.9659,
+      "step": 959
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8026347889729204e-05,
+      "loss": 1.0996,
+      "step": 960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8023908270309834e-05,
+      "loss": 1.0324,
+      "step": 961
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8021468650890463e-05,
+      "loss": 1.0008,
+      "step": 962
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.801902903147109e-05,
+      "loss": 1.0223,
+      "step": 963
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8016589412051723e-05,
+      "loss": 1.0595,
+      "step": 964
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8014149792632353e-05,
+      "loss": 1.024,
+      "step": 965
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.801171017321298e-05,
+      "loss": 1.0447,
+      "step": 966
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.800927055379361e-05,
+      "loss": 0.9743,
+      "step": 967
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.800683093437424e-05,
+      "loss": 1.0045,
+      "step": 968
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8004391314954868e-05,
+      "loss": 1.0661,
+      "step": 969
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8001951695535498e-05,
+      "loss": 1.0501,
+      "step": 970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7999512076116128e-05,
+      "loss": 1.0329,
+      "step": 971
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7997072456696758e-05,
+      "loss": 1.0342,
+      "step": 972
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7994632837277387e-05,
+      "loss": 0.9851,
+      "step": 973
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7992193217858017e-05,
+      "loss": 1.0603,
+      "step": 974
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7989753598438643e-05,
+      "loss": 1.039,
+      "step": 975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7987313979019273e-05,
+      "loss": 1.0811,
+      "step": 976
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7984874359599903e-05,
+      "loss": 1.0639,
+      "step": 977
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7982434740180536e-05,
+      "loss": 0.9864,
+      "step": 978
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7979995120761162e-05,
+      "loss": 1.0345,
+      "step": 979
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7977555501341792e-05,
+      "loss": 1.052,
+      "step": 980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7975115881922422e-05,
+      "loss": 1.0024,
+      "step": 981
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.797267626250305e-05,
+      "loss": 0.9473,
+      "step": 982
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.797023664308368e-05,
+      "loss": 1.0451,
+      "step": 983
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.796779702366431e-05,
+      "loss": 0.9733,
+      "step": 984
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7965357404244937e-05,
+      "loss": 0.9793,
+      "step": 985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7962917784825567e-05,
+      "loss": 1.1115,
+      "step": 986
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.79604781654062e-05,
+      "loss": 1.01,
+      "step": 987
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7958038545986827e-05,
+      "loss": 1.0232,
+      "step": 988
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7955598926567456e-05,
+      "loss": 1.0202,
+      "step": 989
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7953159307148086e-05,
+      "loss": 1.0075,
+      "step": 990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7950719687728716e-05,
+      "loss": 0.9709,
+      "step": 991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7948280068309346e-05,
+      "loss": 0.9765,
+      "step": 992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7945840448889975e-05,
+      "loss": 1.0296,
+      "step": 993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7943400829470602e-05,
+      "loss": 1.0862,
+      "step": 994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.794096121005123e-05,
+      "loss": 1.03,
+      "step": 995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7938521590631865e-05,
+      "loss": 1.0395,
+      "step": 996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7936081971212494e-05,
+      "loss": 1.0475,
+      "step": 997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.793364235179312e-05,
+      "loss": 1.0555,
+      "step": 998
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.793120273237375e-05,
+      "loss": 1.0451,
+      "step": 999
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.792876311295438e-05,
+      "loss": 0.9844,
+      "step": 1000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.792632349353501e-05,
+      "loss": 1.0083,
+      "step": 1001
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.792388387411564e-05,
+      "loss": 1.0391,
+      "step": 1002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.792144425469627e-05,
+      "loss": 0.9868,
+      "step": 1003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.79190046352769e-05,
+      "loss": 1.0302,
+      "step": 1004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.791656501585753e-05,
+      "loss": 1.0556,
+      "step": 1005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.791412539643816e-05,
+      "loss": 0.9987,
+      "step": 1006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7911685777018785e-05,
+      "loss": 0.9837,
+      "step": 1007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7909246157599415e-05,
+      "loss": 1.0832,
+      "step": 1008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7906806538180044e-05,
+      "loss": 1.0053,
+      "step": 1009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7904366918760674e-05,
+      "loss": 0.9813,
+      "step": 1010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7901927299341304e-05,
+      "loss": 1.0475,
+      "step": 1011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7899487679921934e-05,
+      "loss": 1.0466,
+      "step": 1012
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7897048060502563e-05,
+      "loss": 1.0047,
+      "step": 1013
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7894608441083193e-05,
+      "loss": 0.983,
+      "step": 1014
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7892168821663823e-05,
+      "loss": 1.053,
+      "step": 1015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7889729202244453e-05,
+      "loss": 0.9403,
+      "step": 1016
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.788728958282508e-05,
+      "loss": 1.0549,
+      "step": 1017
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.788484996340571e-05,
+      "loss": 1.0716,
+      "step": 1018
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7882410343986342e-05,
+      "loss": 1.0435,
+      "step": 1019
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7879970724566968e-05,
+      "loss": 1.038,
+      "step": 1020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7877531105147598e-05,
+      "loss": 1.1034,
+      "step": 1021
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7875091485728228e-05,
+      "loss": 1.0562,
+      "step": 1022
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7872651866308858e-05,
+      "loss": 1.0219,
+      "step": 1023
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7870212246889487e-05,
+      "loss": 1.0142,
+      "step": 1024
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7867772627470117e-05,
+      "loss": 1.0293,
+      "step": 1025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7865333008050743e-05,
+      "loss": 1.0146,
+      "step": 1026
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7862893388631377e-05,
+      "loss": 0.9949,
+      "step": 1027
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7860453769212006e-05,
+      "loss": 0.992,
+      "step": 1028
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7858014149792633e-05,
+      "loss": 1.0503,
+      "step": 1029
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7855574530373262e-05,
+      "loss": 0.956,
+      "step": 1030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7853134910953892e-05,
+      "loss": 1.0935,
+      "step": 1031
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7850695291534522e-05,
+      "loss": 0.9644,
+      "step": 1032
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.784825567211515e-05,
+      "loss": 0.9466,
+      "step": 1033
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.784581605269578e-05,
+      "loss": 1.0247,
+      "step": 1034
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.784337643327641e-05,
+      "loss": 1.0843,
+      "step": 1035
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.784093681385704e-05,
+      "loss": 1.0175,
+      "step": 1036
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.783849719443767e-05,
+      "loss": 0.9933,
+      "step": 1037
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.78360575750183e-05,
+      "loss": 1.03,
+      "step": 1038
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7833617955598927e-05,
+      "loss": 0.9799,
+      "step": 1039
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7831178336179556e-05,
+      "loss": 0.9537,
+      "step": 1040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7828738716760186e-05,
+      "loss": 1.0785,
+      "step": 1041
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7826299097340816e-05,
+      "loss": 1.0322,
+      "step": 1042
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7823859477921446e-05,
+      "loss": 1.069,
+      "step": 1043
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7821419858502075e-05,
+      "loss": 1.0642,
+      "step": 1044
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7818980239082705e-05,
+      "loss": 0.9737,
+      "step": 1045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7816540619663335e-05,
+      "loss": 1.0681,
+      "step": 1046
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7814101000243965e-05,
+      "loss": 1.031,
+      "step": 1047
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.781166138082459e-05,
+      "loss": 1.0265,
+      "step": 1048
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.780922176140522e-05,
+      "loss": 1.009,
+      "step": 1049
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7806782141985854e-05,
+      "loss": 0.9615,
+      "step": 1050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.780434252256648e-05,
+      "loss": 1.0125,
+      "step": 1051
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.780190290314711e-05,
+      "loss": 1.0316,
+      "step": 1052
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.779946328372774e-05,
+      "loss": 1.0556,
+      "step": 1053
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.779702366430837e-05,
+      "loss": 1.0068,
+      "step": 1054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7794584044889e-05,
+      "loss": 1.0018,
+      "step": 1055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.779214442546963e-05,
+      "loss": 1.0767,
+      "step": 1056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.778970480605026e-05,
+      "loss": 1.097,
+      "step": 1057
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7787265186630885e-05,
+      "loss": 0.958,
+      "step": 1058
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7784825567211518e-05,
+      "loss": 1.0029,
+      "step": 1059
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7782385947792148e-05,
+      "loss": 1.0401,
+      "step": 1060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7779946328372774e-05,
+      "loss": 0.9807,
+      "step": 1061
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7777506708953404e-05,
+      "loss": 1.0181,
+      "step": 1062
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7775067089534034e-05,
+      "loss": 1.0674,
+      "step": 1063
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7772627470114663e-05,
+      "loss": 1.0089,
+      "step": 1064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7770187850695293e-05,
+      "loss": 1.1045,
+      "step": 1065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7767748231275923e-05,
+      "loss": 1.0442,
+      "step": 1066
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.776530861185655e-05,
+      "loss": 1.0302,
+      "step": 1067
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7762868992437182e-05,
+      "loss": 1.0431,
+      "step": 1068
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7760429373017812e-05,
+      "loss": 1.1045,
+      "step": 1069
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.775798975359844e-05,
+      "loss": 1.052,
+      "step": 1070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.775555013417907e-05,
+      "loss": 0.9753,
+      "step": 1071
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7753110514759698e-05,
+      "loss": 1.016,
+      "step": 1072
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7750670895340328e-05,
+      "loss": 0.9908,
+      "step": 1073
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7748231275920958e-05,
+      "loss": 1.0082,
+      "step": 1074
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7745791656501587e-05,
+      "loss": 1.0308,
+      "step": 1075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7743352037082217e-05,
+      "loss": 1.0256,
+      "step": 1076
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7740912417662847e-05,
+      "loss": 0.9633,
+      "step": 1077
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7738472798243477e-05,
+      "loss": 1.0483,
+      "step": 1078
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7736033178824106e-05,
+      "loss": 0.9933,
+      "step": 1079
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7733593559404733e-05,
+      "loss": 1.0529,
+      "step": 1080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7731153939985362e-05,
+      "loss": 1.0333,
+      "step": 1081
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7728714320565996e-05,
+      "loss": 0.996,
+      "step": 1082
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7726274701146622e-05,
+      "loss": 1.0438,
+      "step": 1083
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.772383508172725e-05,
+      "loss": 1.0354,
+      "step": 1084
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.772139546230788e-05,
+      "loss": 0.9855,
+      "step": 1085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.771895584288851e-05,
+      "loss": 0.9761,
+      "step": 1086
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.771651622346914e-05,
+      "loss": 1.0197,
+      "step": 1087
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.771407660404977e-05,
+      "loss": 1.1032,
+      "step": 1088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7711636984630397e-05,
+      "loss": 0.9631,
+      "step": 1089
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7709197365211027e-05,
+      "loss": 1.008,
+      "step": 1090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.770675774579166e-05,
+      "loss": 0.9738,
+      "step": 1091
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7704318126372286e-05,
+      "loss": 1.0318,
+      "step": 1092
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7701878506952916e-05,
+      "loss": 1.0921,
+      "step": 1093
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7699438887533546e-05,
+      "loss": 1.0233,
+      "step": 1094
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7696999268114175e-05,
+      "loss": 1.0473,
+      "step": 1095
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7694559648694805e-05,
+      "loss": 1.0569,
+      "step": 1096
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7692120029275435e-05,
+      "loss": 1.0453,
+      "step": 1097
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7689680409856065e-05,
+      "loss": 0.9745,
+      "step": 1098
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7687240790436694e-05,
+      "loss": 1.0726,
+      "step": 1099
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7684801171017324e-05,
+      "loss": 0.965,
+      "step": 1100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7682361551597954e-05,
+      "loss": 1.0466,
+      "step": 1101
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.767992193217858e-05,
+      "loss": 0.9727,
+      "step": 1102
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.767748231275921e-05,
+      "loss": 0.9586,
+      "step": 1103
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.767504269333984e-05,
+      "loss": 1.0402,
+      "step": 1104
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.767260307392047e-05,
+      "loss": 1.0177,
+      "step": 1105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.76701634545011e-05,
+      "loss": 1.054,
+      "step": 1106
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.766772383508173e-05,
+      "loss": 1.0334,
+      "step": 1107
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.766528421566236e-05,
+      "loss": 1.0871,
+      "step": 1108
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.766284459624299e-05,
+      "loss": 1.0533,
+      "step": 1109
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7660404976823618e-05,
+      "loss": 1.026,
+      "step": 1110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7657965357404245e-05,
+      "loss": 1.0716,
+      "step": 1111
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7655525737984874e-05,
+      "loss": 1.0739,
+      "step": 1112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7653086118565504e-05,
+      "loss": 0.9835,
+      "step": 1113
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7650646499146137e-05,
+      "loss": 0.9776,
+      "step": 1114
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7648206879726764e-05,
+      "loss": 0.9951,
+      "step": 1115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7645767260307393e-05,
+      "loss": 0.9961,
+      "step": 1116
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7643327640888023e-05,
+      "loss": 1.0157,
+      "step": 1117
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7640888021468653e-05,
+      "loss": 1.0641,
+      "step": 1118
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7638448402049282e-05,
+      "loss": 1.0037,
+      "step": 1119
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7636008782629912e-05,
+      "loss": 0.9915,
+      "step": 1120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.763356916321054e-05,
+      "loss": 1.0424,
+      "step": 1121
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.763112954379117e-05,
+      "loss": 0.9707,
+      "step": 1122
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.76286899243718e-05,
+      "loss": 1.0116,
+      "step": 1123
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7626250304952428e-05,
+      "loss": 1.0017,
+      "step": 1124
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7623810685533058e-05,
+      "loss": 1.0239,
+      "step": 1125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7621371066113687e-05,
+      "loss": 1.0243,
+      "step": 1126
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7618931446694317e-05,
+      "loss": 0.9955,
+      "step": 1127
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7616491827274947e-05,
+      "loss": 0.9732,
+      "step": 1128
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7614052207855577e-05,
+      "loss": 1.0367,
+      "step": 1129
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7611612588436203e-05,
+      "loss": 1.0748,
+      "step": 1130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7609172969016836e-05,
+      "loss": 1.0891,
+      "step": 1131
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7606733349597466e-05,
+      "loss": 0.9751,
+      "step": 1132
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7604293730178096e-05,
+      "loss": 1.032,
+      "step": 1133
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7601854110758722e-05,
+      "loss": 0.9906,
+      "step": 1134
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.759941449133935e-05,
+      "loss": 0.9418,
+      "step": 1135
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.759697487191998e-05,
+      "loss": 1.0854,
+      "step": 1136
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.759453525250061e-05,
+      "loss": 0.9611,
+      "step": 1137
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.759209563308124e-05,
+      "loss": 1.0155,
+      "step": 1138
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.758965601366187e-05,
+      "loss": 1.0633,
+      "step": 1139
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.75872163942425e-05,
+      "loss": 0.98,
+      "step": 1140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.758477677482313e-05,
+      "loss": 0.958,
+      "step": 1141
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.758233715540376e-05,
+      "loss": 1.0428,
+      "step": 1142
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7579897535984386e-05,
+      "loss": 0.9898,
+      "step": 1143
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7577457916565016e-05,
+      "loss": 1.0016,
+      "step": 1144
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7575018297145646e-05,
+      "loss": 1.0307,
+      "step": 1145
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7572578677726275e-05,
+      "loss": 1.0365,
+      "step": 1146
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7570139058306905e-05,
+      "loss": 0.9419,
+      "step": 1147
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7567699438887535e-05,
+      "loss": 1.0331,
+      "step": 1148
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7565259819468165e-05,
+      "loss": 1.0808,
+      "step": 1149
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7562820200048794e-05,
+      "loss": 1.0306,
+      "step": 1150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7560380580629424e-05,
+      "loss": 1.0131,
+      "step": 1151
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7557940961210054e-05,
+      "loss": 0.9611,
+      "step": 1152
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.755550134179068e-05,
+      "loss": 0.9753,
+      "step": 1153
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7553061722371313e-05,
+      "loss": 1.1147,
+      "step": 1154
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7550622102951943e-05,
+      "loss": 1.0443,
+      "step": 1155
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.754818248353257e-05,
+      "loss": 1.0872,
+      "step": 1156
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.75457428641132e-05,
+      "loss": 1.0138,
+      "step": 1157
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.754330324469383e-05,
+      "loss": 0.9919,
+      "step": 1158
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.754086362527446e-05,
+      "loss": 1.0749,
+      "step": 1159
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.753842400585509e-05,
+      "loss": 1.0741,
+      "step": 1160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7535984386435718e-05,
+      "loss": 0.9741,
+      "step": 1161
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7533544767016345e-05,
+      "loss": 1.0932,
+      "step": 1162
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7531105147596978e-05,
+      "loss": 0.9875,
+      "step": 1163
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7528665528177607e-05,
+      "loss": 0.9966,
+      "step": 1164
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7526225908758234e-05,
+      "loss": 0.9927,
+      "step": 1165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7523786289338864e-05,
+      "loss": 1.0496,
+      "step": 1166
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7521346669919493e-05,
+      "loss": 1.1114,
+      "step": 1167
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7518907050500123e-05,
+      "loss": 1.0501,
+      "step": 1168
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7516467431080753e-05,
+      "loss": 1.0026,
+      "step": 1169
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7514027811661383e-05,
+      "loss": 1.0247,
+      "step": 1170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7511588192242012e-05,
+      "loss": 1.048,
+      "step": 1171
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7509148572822642e-05,
+      "loss": 1.09,
+      "step": 1172
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7506708953403272e-05,
+      "loss": 1.0525,
+      "step": 1173
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.75042693339839e-05,
+      "loss": 1.0679,
+      "step": 1174
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7501829714564528e-05,
+      "loss": 1.0521,
+      "step": 1175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7499390095145158e-05,
+      "loss": 1.0881,
+      "step": 1176
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.749695047572579e-05,
+      "loss": 0.9949,
+      "step": 1177
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7494510856306417e-05,
+      "loss": 0.9461,
+      "step": 1178
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7492071236887047e-05,
+      "loss": 1.0103,
+      "step": 1179
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7489631617467677e-05,
+      "loss": 1.0272,
+      "step": 1180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7487191998048306e-05,
+      "loss": 1.0289,
+      "step": 1181
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7484752378628936e-05,
+      "loss": 1.0417,
+      "step": 1182
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7482312759209566e-05,
+      "loss": 1.0268,
+      "step": 1183
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7479873139790192e-05,
+      "loss": 0.9574,
+      "step": 1184
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7477433520370822e-05,
+      "loss": 1.0169,
+      "step": 1185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7474993900951455e-05,
+      "loss": 1.0023,
+      "step": 1186
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.747255428153208e-05,
+      "loss": 1.0561,
+      "step": 1187
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.747011466211271e-05,
+      "loss": 1.0776,
+      "step": 1188
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.746767504269334e-05,
+      "loss": 0.9836,
+      "step": 1189
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.746523542327397e-05,
+      "loss": 1.0264,
+      "step": 1190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.74627958038546e-05,
+      "loss": 1.013,
+      "step": 1191
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.746035618443523e-05,
+      "loss": 1.05,
+      "step": 1192
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.745791656501586e-05,
+      "loss": 1.1041,
+      "step": 1193
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7455476945596486e-05,
+      "loss": 1.0273,
+      "step": 1194
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.745303732617712e-05,
+      "loss": 1.0088,
+      "step": 1195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.745059770675775e-05,
+      "loss": 1.037,
+      "step": 1196
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7448158087338375e-05,
+      "loss": 1.0118,
+      "step": 1197
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7445718467919005e-05,
+      "loss": 0.989,
+      "step": 1198
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7443278848499635e-05,
+      "loss": 1.0122,
+      "step": 1199
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7440839229080265e-05,
+      "loss": 1.0736,
+      "step": 1200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7438399609660894e-05,
+      "loss": 1.0668,
+      "step": 1201
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7435959990241524e-05,
+      "loss": 1.0327,
+      "step": 1202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7433520370822154e-05,
+      "loss": 0.9586,
+      "step": 1203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7431080751402784e-05,
+      "loss": 1.0646,
+      "step": 1204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7428641131983413e-05,
+      "loss": 1.0227,
+      "step": 1205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.742620151256404e-05,
+      "loss": 1.0188,
+      "step": 1206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.742376189314467e-05,
+      "loss": 1.0252,
+      "step": 1207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.74213222737253e-05,
+      "loss": 1.0248,
+      "step": 1208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.741888265430593e-05,
+      "loss": 1.0531,
+      "step": 1209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.741644303488656e-05,
+      "loss": 1.0458,
+      "step": 1210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.741400341546719e-05,
+      "loss": 0.9696,
+      "step": 1211
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7411563796047818e-05,
+      "loss": 1.0569,
+      "step": 1212
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7409124176628448e-05,
+      "loss": 0.9953,
+      "step": 1213
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7406684557209078e-05,
+      "loss": 1.0692,
+      "step": 1214
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7404244937789707e-05,
+      "loss": 1.0126,
+      "step": 1215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7401805318370334e-05,
+      "loss": 1.0845,
+      "step": 1216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7399365698950964e-05,
+      "loss": 0.9981,
+      "step": 1217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7396926079531597e-05,
+      "loss": 0.9697,
+      "step": 1218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7394486460112223e-05,
+      "loss": 1.0493,
+      "step": 1219
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7392046840692853e-05,
+      "loss": 0.9452,
+      "step": 1220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7389607221273483e-05,
+      "loss": 1.0221,
+      "step": 1221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7387167601854112e-05,
+      "loss": 1.0125,
+      "step": 1222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7384727982434742e-05,
+      "loss": 0.9522,
+      "step": 1223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7382288363015372e-05,
+      "loss": 1.0135,
+      "step": 1224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7379848743595998e-05,
+      "loss": 1.0129,
+      "step": 1225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.737740912417663e-05,
+      "loss": 0.9567,
+      "step": 1226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.737496950475726e-05,
+      "loss": 0.9379,
+      "step": 1227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7372529885337887e-05,
+      "loss": 0.9915,
+      "step": 1228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7370090265918517e-05,
+      "loss": 1.0394,
+      "step": 1229
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7367650646499147e-05,
+      "loss": 1.0612,
+      "step": 1230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7365211027079777e-05,
+      "loss": 0.9903,
+      "step": 1231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7362771407660406e-05,
+      "loss": 0.987,
+      "step": 1232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7360331788241036e-05,
+      "loss": 0.9716,
+      "step": 1233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7357892168821666e-05,
+      "loss": 1.0176,
+      "step": 1234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7355452549402296e-05,
+      "loss": 1.0149,
+      "step": 1235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7353012929982925e-05,
+      "loss": 1.0001,
+      "step": 1236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7350573310563555e-05,
+      "loss": 1.0513,
+      "step": 1237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.734813369114418e-05,
+      "loss": 0.9947,
+      "step": 1238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.734569407172481e-05,
+      "loss": 0.9913,
+      "step": 1239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.734325445230544e-05,
+      "loss": 1.0246,
+      "step": 1240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.734081483288607e-05,
+      "loss": 1.0301,
+      "step": 1241
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.73383752134667e-05,
+      "loss": 0.9842,
+      "step": 1242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.733593559404733e-05,
+      "loss": 1.0133,
+      "step": 1243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.733349597462796e-05,
+      "loss": 1.0025,
+      "step": 1244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.733105635520859e-05,
+      "loss": 1.0734,
+      "step": 1245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.732861673578922e-05,
+      "loss": 1.0264,
+      "step": 1246
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7326177116369846e-05,
+      "loss": 1.0224,
+      "step": 1247
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7323737496950475e-05,
+      "loss": 1.0746,
+      "step": 1248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.732129787753111e-05,
+      "loss": 0.9542,
+      "step": 1249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.731885825811174e-05,
+      "loss": 0.9404,
+      "step": 1250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7316418638692365e-05,
+      "loss": 1.0088,
+      "step": 1251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7313979019272994e-05,
+      "loss": 1.0263,
+      "step": 1252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7311539399853624e-05,
+      "loss": 1.0091,
+      "step": 1253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7309099780434254e-05,
+      "loss": 0.9752,
+      "step": 1254
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 1.0167489051818848,
+      "eval_runtime": 205.5551,
+      "eval_samples_per_second": 81.725,
+      "eval_steps_per_second": 10.216,
+      "step": 1254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7306660161014884e-05,
+      "loss": 1.0544,
+      "step": 1255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7304220541595513e-05,
+      "loss": 0.9941,
+      "step": 1256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.730178092217614e-05,
+      "loss": 0.9207,
+      "step": 1257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7299341302756773e-05,
+      "loss": 1.0439,
+      "step": 1258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7296901683337403e-05,
+      "loss": 0.9603,
+      "step": 1259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.729446206391803e-05,
+      "loss": 0.9378,
+      "step": 1260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.729202244449866e-05,
+      "loss": 0.9759,
+      "step": 1261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.728958282507929e-05,
+      "loss": 1.0251,
+      "step": 1262
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7287143205659918e-05,
+      "loss": 1.0102,
+      "step": 1263
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7284703586240548e-05,
+      "loss": 0.9404,
+      "step": 1264
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7282263966821178e-05,
+      "loss": 0.9977,
+      "step": 1265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7279824347401804e-05,
+      "loss": 0.9765,
+      "step": 1266
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7277384727982437e-05,
+      "loss": 0.9064,
+      "step": 1267
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7274945108563067e-05,
+      "loss": 1.0471,
+      "step": 1268
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7272505489143697e-05,
+      "loss": 0.9511,
+      "step": 1269
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7270065869724323e-05,
+      "loss": 1.0092,
+      "step": 1270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7267626250304953e-05,
+      "loss": 1.0036,
+      "step": 1271
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7265186630885583e-05,
+      "loss": 0.933,
+      "step": 1272
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7262747011466212e-05,
+      "loss": 1.0389,
+      "step": 1273
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7260307392046842e-05,
+      "loss": 0.9477,
+      "step": 1274
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7257867772627472e-05,
+      "loss": 1.0156,
+      "step": 1275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.72554281532081e-05,
+      "loss": 1.0797,
+      "step": 1276
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.725298853378873e-05,
+      "loss": 0.9962,
+      "step": 1277
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.725054891436936e-05,
+      "loss": 1.116,
+      "step": 1278
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7248109294949987e-05,
+      "loss": 1.0433,
+      "step": 1279
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7245669675530617e-05,
+      "loss": 1.0276,
+      "step": 1280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.724323005611125e-05,
+      "loss": 1.0115,
+      "step": 1281
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7240790436691877e-05,
+      "loss": 0.9989,
+      "step": 1282
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7238350817272506e-05,
+      "loss": 1.0029,
+      "step": 1283
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7235911197853136e-05,
+      "loss": 0.9186,
+      "step": 1284
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7233471578433766e-05,
+      "loss": 0.9542,
+      "step": 1285
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7231031959014396e-05,
+      "loss": 0.9825,
+      "step": 1286
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7228592339595025e-05,
+      "loss": 0.995,
+      "step": 1287
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7226152720175655e-05,
+      "loss": 1.0168,
+      "step": 1288
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.722371310075628e-05,
+      "loss": 0.9888,
+      "step": 1289
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7221273481336915e-05,
+      "loss": 1.0245,
+      "step": 1290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7218833861917544e-05,
+      "loss": 1.0318,
+      "step": 1291
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.721639424249817e-05,
+      "loss": 1.0232,
+      "step": 1292
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.72139546230788e-05,
+      "loss": 0.9565,
+      "step": 1293
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.721151500365943e-05,
+      "loss": 1.0067,
+      "step": 1294
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.720907538424006e-05,
+      "loss": 0.9992,
+      "step": 1295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.720663576482069e-05,
+      "loss": 0.9937,
+      "step": 1296
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.720419614540132e-05,
+      "loss": 1.0571,
+      "step": 1297
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.720175652598195e-05,
+      "loss": 0.9921,
+      "step": 1298
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.719931690656258e-05,
+      "loss": 0.9472,
+      "step": 1299
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.719687728714321e-05,
+      "loss": 0.9825,
+      "step": 1300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7194437667723835e-05,
+      "loss": 1.0554,
+      "step": 1301
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7191998048304465e-05,
+      "loss": 0.9698,
+      "step": 1302
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7189558428885094e-05,
+      "loss": 0.9307,
+      "step": 1303
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7187118809465724e-05,
+      "loss": 1.068,
+      "step": 1304
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7184679190046354e-05,
+      "loss": 0.9871,
+      "step": 1305
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7182239570626984e-05,
+      "loss": 0.9872,
+      "step": 1306
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7179799951207613e-05,
+      "loss": 1.1256,
+      "step": 1307
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7177360331788243e-05,
+      "loss": 0.9917,
+      "step": 1308
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7174920712368873e-05,
+      "loss": 1.0447,
+      "step": 1309
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7172481092949503e-05,
+      "loss": 1.0332,
+      "step": 1310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.717004147353013e-05,
+      "loss": 1.088,
+      "step": 1311
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.716760185411076e-05,
+      "loss": 1.043,
+      "step": 1312
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7165162234691392e-05,
+      "loss": 1.0716,
+      "step": 1313
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7162722615272018e-05,
+      "loss": 1.0358,
+      "step": 1314
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7160282995852648e-05,
+      "loss": 1.1014,
+      "step": 1315
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7157843376433278e-05,
+      "loss": 1.0396,
+      "step": 1316
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7155403757013908e-05,
+      "loss": 0.9754,
+      "step": 1317
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7152964137594537e-05,
+      "loss": 1.0685,
+      "step": 1318
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7150524518175167e-05,
+      "loss": 0.9847,
+      "step": 1319
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7148084898755793e-05,
+      "loss": 0.9828,
+      "step": 1320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7145645279336423e-05,
+      "loss": 1.0584,
+      "step": 1321
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7143205659917056e-05,
+      "loss": 0.9303,
+      "step": 1322
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7140766040497683e-05,
+      "loss": 0.9869,
+      "step": 1323
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7138326421078312e-05,
+      "loss": 0.962,
+      "step": 1324
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7135886801658942e-05,
+      "loss": 1.0069,
+      "step": 1325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7133447182239572e-05,
+      "loss": 1.0346,
+      "step": 1326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.71310075628202e-05,
+      "loss": 0.9488,
+      "step": 1327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.712856794340083e-05,
+      "loss": 0.9573,
+      "step": 1328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.712612832398146e-05,
+      "loss": 1.0426,
+      "step": 1329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.712368870456209e-05,
+      "loss": 1.0269,
+      "step": 1330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.712124908514272e-05,
+      "loss": 0.9857,
+      "step": 1331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.711880946572335e-05,
+      "loss": 1.0651,
+      "step": 1332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7116369846303977e-05,
+      "loss": 0.9637,
+      "step": 1333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7113930226884606e-05,
+      "loss": 1.0283,
+      "step": 1334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7111490607465236e-05,
+      "loss": 1.0113,
+      "step": 1335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7109050988045866e-05,
+      "loss": 1.0631,
+      "step": 1336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7106611368626496e-05,
+      "loss": 1.022,
+      "step": 1337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7104171749207125e-05,
+      "loss": 0.9986,
+      "step": 1338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7101732129787755e-05,
+      "loss": 1.0474,
+      "step": 1339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7099292510368385e-05,
+      "loss": 1.002,
+      "step": 1340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7096852890949015e-05,
+      "loss": 0.9587,
+      "step": 1341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.709441327152964e-05,
+      "loss": 0.9933,
+      "step": 1342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.709197365211027e-05,
+      "loss": 0.9865,
+      "step": 1343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.70895340326909e-05,
+      "loss": 0.9584,
+      "step": 1344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.708709441327153e-05,
+      "loss": 1.0066,
+      "step": 1345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.708465479385216e-05,
+      "loss": 1.0605,
+      "step": 1346
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.708221517443279e-05,
+      "loss": 1.0381,
+      "step": 1347
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.707977555501342e-05,
+      "loss": 1.0367,
+      "step": 1348
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.707733593559405e-05,
+      "loss": 1.0182,
+      "step": 1349
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.707489631617468e-05,
+      "loss": 1.0357,
+      "step": 1350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.707245669675531e-05,
+      "loss": 1.0102,
+      "step": 1351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7070017077335935e-05,
+      "loss": 0.9648,
+      "step": 1352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7067577457916568e-05,
+      "loss": 0.9824,
+      "step": 1353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7065137838497198e-05,
+      "loss": 1.0484,
+      "step": 1354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7062698219077824e-05,
+      "loss": 0.9234,
+      "step": 1355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7060258599658454e-05,
+      "loss": 0.9674,
+      "step": 1356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7057818980239084e-05,
+      "loss": 1.0331,
+      "step": 1357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7055379360819713e-05,
+      "loss": 1.1185,
+      "step": 1358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7052939741400343e-05,
+      "loss": 0.9919,
+      "step": 1359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7050500121980973e-05,
+      "loss": 0.9888,
+      "step": 1360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.70480605025616e-05,
+      "loss": 0.9765,
+      "step": 1361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7045620883142232e-05,
+      "loss": 1.0662,
+      "step": 1362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7043181263722862e-05,
+      "loss": 1.0405,
+      "step": 1363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.704074164430349e-05,
+      "loss": 1.0257,
+      "step": 1364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7038302024884118e-05,
+      "loss": 1.0078,
+      "step": 1365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7035862405464748e-05,
+      "loss": 1.0479,
+      "step": 1366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7033422786045378e-05,
+      "loss": 1.0061,
+      "step": 1367
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7030983166626008e-05,
+      "loss": 1.0085,
+      "step": 1368
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7028543547206637e-05,
+      "loss": 0.9531,
+      "step": 1369
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7026103927787267e-05,
+      "loss": 0.9211,
+      "step": 1370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7023664308367897e-05,
+      "loss": 1.0205,
+      "step": 1371
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7021224688948527e-05,
+      "loss": 1.0293,
+      "step": 1372
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7018785069529156e-05,
+      "loss": 1.0329,
+      "step": 1373
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7016345450109783e-05,
+      "loss": 1.0565,
+      "step": 1374
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7013905830690412e-05,
+      "loss": 0.9846,
+      "step": 1375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7011466211271045e-05,
+      "loss": 1.0257,
+      "step": 1376
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7009026591851672e-05,
+      "loss": 0.972,
+      "step": 1377
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.70065869724323e-05,
+      "loss": 0.9848,
+      "step": 1378
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.700414735301293e-05,
+      "loss": 1.0185,
+      "step": 1379
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.700170773359356e-05,
+      "loss": 1.0277,
+      "step": 1380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.699926811417419e-05,
+      "loss": 1.0046,
+      "step": 1381
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.699682849475482e-05,
+      "loss": 0.9428,
+      "step": 1382
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6994388875335447e-05,
+      "loss": 1.0162,
+      "step": 1383
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6991949255916077e-05,
+      "loss": 0.9809,
+      "step": 1384
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.698950963649671e-05,
+      "loss": 1.0304,
+      "step": 1385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.698707001707734e-05,
+      "loss": 1.018,
+      "step": 1386
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6984630397657966e-05,
+      "loss": 1.0498,
+      "step": 1387
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6982190778238596e-05,
+      "loss": 1.033,
+      "step": 1388
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6979751158819225e-05,
+      "loss": 0.9751,
+      "step": 1389
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6977311539399855e-05,
+      "loss": 0.9334,
+      "step": 1390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6974871919980485e-05,
+      "loss": 0.9234,
+      "step": 1391
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6972432300561115e-05,
+      "loss": 1.121,
+      "step": 1392
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.696999268114174e-05,
+      "loss": 1.0316,
+      "step": 1393
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6967553061722374e-05,
+      "loss": 1.0243,
+      "step": 1394
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6965113442303004e-05,
+      "loss": 1.0735,
+      "step": 1395
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.696267382288363e-05,
+      "loss": 1.0211,
+      "step": 1396
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.696023420346426e-05,
+      "loss": 1.0168,
+      "step": 1397
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.695779458404489e-05,
+      "loss": 0.963,
+      "step": 1398
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.695535496462552e-05,
+      "loss": 0.983,
+      "step": 1399
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.695291534520615e-05,
+      "loss": 0.9516,
+      "step": 1400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.695047572578678e-05,
+      "loss": 0.9693,
+      "step": 1401
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.694803610636741e-05,
+      "loss": 1.0619,
+      "step": 1402
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.694559648694804e-05,
+      "loss": 0.9932,
+      "step": 1403
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6943156867528668e-05,
+      "loss": 0.9662,
+      "step": 1404
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6940717248109298e-05,
+      "loss": 1.0562,
+      "step": 1405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6938277628689924e-05,
+      "loss": 0.9401,
+      "step": 1406
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6935838009270554e-05,
+      "loss": 1.0228,
+      "step": 1407
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6933398389851187e-05,
+      "loss": 1.0522,
+      "step": 1408
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6930958770431814e-05,
+      "loss": 1.0253,
+      "step": 1409
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6928519151012443e-05,
+      "loss": 1.1054,
+      "step": 1410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6926079531593073e-05,
+      "loss": 1.0061,
+      "step": 1411
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6923639912173703e-05,
+      "loss": 0.9559,
+      "step": 1412
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6921200292754332e-05,
+      "loss": 0.9604,
+      "step": 1413
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6918760673334962e-05,
+      "loss": 1.0443,
+      "step": 1414
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.691632105391559e-05,
+      "loss": 0.9052,
+      "step": 1415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.691388143449622e-05,
+      "loss": 0.9645,
+      "step": 1416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.691144181507685e-05,
+      "loss": 1.0892,
+      "step": 1417
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6909002195657478e-05,
+      "loss": 0.9846,
+      "step": 1418
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6906562576238108e-05,
+      "loss": 0.9974,
+      "step": 1419
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6904122956818737e-05,
+      "loss": 1.0208,
+      "step": 1420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6901683337399367e-05,
+      "loss": 0.9403,
+      "step": 1421
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6899243717979997e-05,
+      "loss": 0.9326,
+      "step": 1422
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6896804098560627e-05,
+      "loss": 0.967,
+      "step": 1423
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6894364479141256e-05,
+      "loss": 0.987,
+      "step": 1424
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6891924859721886e-05,
+      "loss": 1.0533,
+      "step": 1425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6889485240302516e-05,
+      "loss": 0.9533,
+      "step": 1426
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6887045620883146e-05,
+      "loss": 1.0347,
+      "step": 1427
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6884606001463772e-05,
+      "loss": 0.9794,
+      "step": 1428
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.68821663820444e-05,
+      "loss": 0.9722,
+      "step": 1429
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.687972676262503e-05,
+      "loss": 0.9885,
+      "step": 1430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.687728714320566e-05,
+      "loss": 1.0196,
+      "step": 1431
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.687484752378629e-05,
+      "loss": 1.0177,
+      "step": 1432
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.687240790436692e-05,
+      "loss": 1.0964,
+      "step": 1433
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.686996828494755e-05,
+      "loss": 0.9739,
+      "step": 1434
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.686752866552818e-05,
+      "loss": 1.0117,
+      "step": 1435
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.686508904610881e-05,
+      "loss": 0.9652,
+      "step": 1436
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6862649426689436e-05,
+      "loss": 0.9293,
+      "step": 1437
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6860209807270066e-05,
+      "loss": 1.0234,
+      "step": 1438
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6857770187850696e-05,
+      "loss": 0.9267,
+      "step": 1439
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6855330568431325e-05,
+      "loss": 1.0139,
+      "step": 1440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6852890949011955e-05,
+      "loss": 1.0108,
+      "step": 1441
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6850451329592585e-05,
+      "loss": 0.9696,
+      "step": 1442
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6848011710173215e-05,
+      "loss": 0.9881,
+      "step": 1443
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6845572090753844e-05,
+      "loss": 1.0502,
+      "step": 1444
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6843132471334474e-05,
+      "loss": 1.0788,
+      "step": 1445
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6840692851915104e-05,
+      "loss": 0.9951,
+      "step": 1446
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.683825323249573e-05,
+      "loss": 1.0121,
+      "step": 1447
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.683581361307636e-05,
+      "loss": 0.9573,
+      "step": 1448
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6833373993656993e-05,
+      "loss": 1.0649,
+      "step": 1449
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.683093437423762e-05,
+      "loss": 1.0093,
+      "step": 1450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.682849475481825e-05,
+      "loss": 0.9892,
+      "step": 1451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.682605513539888e-05,
+      "loss": 0.9849,
+      "step": 1452
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.682361551597951e-05,
+      "loss": 1.0232,
+      "step": 1453
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.682117589656014e-05,
+      "loss": 1.1099,
+      "step": 1454
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6818736277140768e-05,
+      "loss": 0.9937,
+      "step": 1455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6816296657721395e-05,
+      "loss": 1.0378,
+      "step": 1456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6813857038302028e-05,
+      "loss": 1.0372,
+      "step": 1457
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6811417418882657e-05,
+      "loss": 1.0209,
+      "step": 1458
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6808977799463284e-05,
+      "loss": 0.9917,
+      "step": 1459
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6806538180043914e-05,
+      "loss": 0.9927,
+      "step": 1460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6804098560624543e-05,
+      "loss": 0.9841,
+      "step": 1461
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6801658941205173e-05,
+      "loss": 0.9957,
+      "step": 1462
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6799219321785803e-05,
+      "loss": 1.0104,
+      "step": 1463
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6796779702366433e-05,
+      "loss": 0.9926,
+      "step": 1464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6794340082947062e-05,
+      "loss": 0.9821,
+      "step": 1465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6791900463527692e-05,
+      "loss": 0.9755,
+      "step": 1466
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6789460844108322e-05,
+      "loss": 0.9798,
+      "step": 1467
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.678702122468895e-05,
+      "loss": 1.0321,
+      "step": 1468
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6784581605269578e-05,
+      "loss": 0.9971,
+      "step": 1469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6782141985850208e-05,
+      "loss": 0.9644,
+      "step": 1470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6779702366430837e-05,
+      "loss": 1.034,
+      "step": 1471
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6777262747011467e-05,
+      "loss": 1.0038,
+      "step": 1472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6774823127592097e-05,
+      "loss": 1.0368,
+      "step": 1473
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6772383508172727e-05,
+      "loss": 0.9817,
+      "step": 1474
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6769943888753356e-05,
+      "loss": 0.9418,
+      "step": 1475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6767504269333986e-05,
+      "loss": 1.0895,
+      "step": 1476
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6765064649914616e-05,
+      "loss": 1.0188,
+      "step": 1477
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6762625030495242e-05,
+      "loss": 0.9318,
+      "step": 1478
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6760185411075872e-05,
+      "loss": 0.9131,
+      "step": 1479
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6757745791656505e-05,
+      "loss": 1.0191,
+      "step": 1480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.675530617223713e-05,
+      "loss": 0.9906,
+      "step": 1481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.675286655281776e-05,
+      "loss": 0.9896,
+      "step": 1482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.675042693339839e-05,
+      "loss": 1.0089,
+      "step": 1483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.674798731397902e-05,
+      "loss": 0.9897,
+      "step": 1484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.674554769455965e-05,
+      "loss": 1.0355,
+      "step": 1485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.674310807514028e-05,
+      "loss": 0.9888,
+      "step": 1486
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.674066845572091e-05,
+      "loss": 0.9733,
+      "step": 1487
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6738228836301536e-05,
+      "loss": 1.0217,
+      "step": 1488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.673578921688217e-05,
+      "loss": 1.01,
+      "step": 1489
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.67333495974628e-05,
+      "loss": 1.0941,
+      "step": 1490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6730909978043425e-05,
+      "loss": 0.9861,
+      "step": 1491
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6728470358624055e-05,
+      "loss": 1.0447,
+      "step": 1492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6726030739204685e-05,
+      "loss": 1.0638,
+      "step": 1493
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6723591119785315e-05,
+      "loss": 1.0137,
+      "step": 1494
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6721151500365944e-05,
+      "loss": 0.9992,
+      "step": 1495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6718711880946574e-05,
+      "loss": 0.9629,
+      "step": 1496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.67162722615272e-05,
+      "loss": 1.0132,
+      "step": 1497
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6713832642107834e-05,
+      "loss": 1.0473,
+      "step": 1498
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6711393022688463e-05,
+      "loss": 1.0473,
+      "step": 1499
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.670895340326909e-05,
+      "loss": 0.9651,
+      "step": 1500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.670651378384972e-05,
+      "loss": 0.9419,
+      "step": 1501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.670407416443035e-05,
+      "loss": 1.0266,
+      "step": 1502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6701634545010982e-05,
+      "loss": 0.9784,
+      "step": 1503
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.669919492559161e-05,
+      "loss": 0.9694,
+      "step": 1504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.669675530617224e-05,
+      "loss": 1.0844,
+      "step": 1505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6694315686752868e-05,
+      "loss": 0.9789,
+      "step": 1506
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6691876067333498e-05,
+      "loss": 0.9984,
+      "step": 1507
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6689436447914128e-05,
+      "loss": 0.9351,
+      "step": 1508
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6686996828494757e-05,
+      "loss": 0.9604,
+      "step": 1509
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6684557209075384e-05,
+      "loss": 1.0099,
+      "step": 1510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6682117589656014e-05,
+      "loss": 1.0313,
+      "step": 1511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6679677970236647e-05,
+      "loss": 1.0108,
+      "step": 1512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6677238350817273e-05,
+      "loss": 1.0334,
+      "step": 1513
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6674798731397903e-05,
+      "loss": 1.0475,
+      "step": 1514
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6672359111978533e-05,
+      "loss": 1.066,
+      "step": 1515
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6669919492559162e-05,
+      "loss": 1.0249,
+      "step": 1516
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6667479873139792e-05,
+      "loss": 0.9523,
+      "step": 1517
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6665040253720422e-05,
+      "loss": 1.0273,
+      "step": 1518
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6662600634301048e-05,
+      "loss": 0.9684,
+      "step": 1519
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6660161014881678e-05,
+      "loss": 1.0408,
+      "step": 1520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.665772139546231e-05,
+      "loss": 1.0449,
+      "step": 1521
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.665528177604294e-05,
+      "loss": 1.0495,
+      "step": 1522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6652842156623567e-05,
+      "loss": 0.9673,
+      "step": 1523
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6650402537204197e-05,
+      "loss": 1.0383,
+      "step": 1524
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6647962917784827e-05,
+      "loss": 1.0057,
+      "step": 1525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6645523298365456e-05,
+      "loss": 0.9068,
+      "step": 1526
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6643083678946086e-05,
+      "loss": 0.9637,
+      "step": 1527
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6640644059526716e-05,
+      "loss": 1.0695,
+      "step": 1528
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6638204440107346e-05,
+      "loss": 0.9994,
+      "step": 1529
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6635764820687975e-05,
+      "loss": 0.917,
+      "step": 1530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6633325201268605e-05,
+      "loss": 0.9926,
+      "step": 1531
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.663088558184923e-05,
+      "loss": 1.0431,
+      "step": 1532
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.662844596242986e-05,
+      "loss": 1.0207,
+      "step": 1533
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.662600634301049e-05,
+      "loss": 0.9756,
+      "step": 1534
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.662356672359112e-05,
+      "loss": 0.9932,
+      "step": 1535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.662112710417175e-05,
+      "loss": 1.0605,
+      "step": 1536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.661868748475238e-05,
+      "loss": 0.9848,
+      "step": 1537
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.661624786533301e-05,
+      "loss": 1.0033,
+      "step": 1538
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.661380824591364e-05,
+      "loss": 1.0371,
+      "step": 1539
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.661136862649427e-05,
+      "loss": 1.0096,
+      "step": 1540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.66089290070749e-05,
+      "loss": 0.9631,
+      "step": 1541
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6606489387655525e-05,
+      "loss": 0.9521,
+      "step": 1542
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6604049768236155e-05,
+      "loss": 0.9858,
+      "step": 1543
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.660161014881679e-05,
+      "loss": 0.9821,
+      "step": 1544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6599170529397415e-05,
+      "loss": 0.9858,
+      "step": 1545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6596730909978044e-05,
+      "loss": 1.0029,
+      "step": 1546
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6594291290558674e-05,
+      "loss": 0.996,
+      "step": 1547
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6591851671139304e-05,
+      "loss": 0.9597,
+      "step": 1548
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6589412051719934e-05,
+      "loss": 0.9788,
+      "step": 1549
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6586972432300563e-05,
+      "loss": 0.9332,
+      "step": 1550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.658453281288119e-05,
+      "loss": 0.9777,
+      "step": 1551
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6582093193461823e-05,
+      "loss": 1.0118,
+      "step": 1552
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6579653574042453e-05,
+      "loss": 0.9674,
+      "step": 1553
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.657721395462308e-05,
+      "loss": 0.9856,
+      "step": 1554
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.657477433520371e-05,
+      "loss": 1.028,
+      "step": 1555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.657233471578434e-05,
+      "loss": 0.9046,
+      "step": 1556
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6569895096364968e-05,
+      "loss": 0.9577,
+      "step": 1557
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6567455476945598e-05,
+      "loss": 1.0324,
+      "step": 1558
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6565015857526228e-05,
+      "loss": 0.9537,
+      "step": 1559
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6562576238106857e-05,
+      "loss": 1.0643,
+      "step": 1560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6560136618687487e-05,
+      "loss": 0.955,
+      "step": 1561
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6557696999268117e-05,
+      "loss": 1.0505,
+      "step": 1562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6555257379848747e-05,
+      "loss": 1.0277,
+      "step": 1563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6552817760429373e-05,
+      "loss": 0.9618,
+      "step": 1564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6550378141010003e-05,
+      "loss": 1.0164,
+      "step": 1565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6547938521590633e-05,
+      "loss": 0.9676,
+      "step": 1566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6545498902171262e-05,
+      "loss": 1.0563,
+      "step": 1567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6543059282751892e-05,
+      "loss": 1.0067,
+      "step": 1568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6540619663332522e-05,
+      "loss": 0.9611,
+      "step": 1569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.653818004391315e-05,
+      "loss": 0.9362,
+      "step": 1570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.653574042449378e-05,
+      "loss": 0.9801,
+      "step": 1571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.653330080507441e-05,
+      "loss": 1.0274,
+      "step": 1572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6530861185655037e-05,
+      "loss": 0.9465,
+      "step": 1573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6528421566235667e-05,
+      "loss": 0.951,
+      "step": 1574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6525981946816297e-05,
+      "loss": 0.9261,
+      "step": 1575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6523542327396927e-05,
+      "loss": 0.9341,
+      "step": 1576
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6521102707977556e-05,
+      "loss": 1.0242,
+      "step": 1577
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6518663088558186e-05,
+      "loss": 1.019,
+      "step": 1578
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6516223469138816e-05,
+      "loss": 0.9795,
+      "step": 1579
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6513783849719446e-05,
+      "loss": 0.9621,
+      "step": 1580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6511344230300075e-05,
+      "loss": 0.9443,
+      "step": 1581
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6508904610880705e-05,
+      "loss": 0.9984,
+      "step": 1582
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.650646499146133e-05,
+      "loss": 1.011,
+      "step": 1583
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6504025372041965e-05,
+      "loss": 1.0212,
+      "step": 1584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6501585752622594e-05,
+      "loss": 0.979,
+      "step": 1585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.649914613320322e-05,
+      "loss": 0.9752,
+      "step": 1586
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.649670651378385e-05,
+      "loss": 0.9254,
+      "step": 1587
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.649426689436448e-05,
+      "loss": 0.9917,
+      "step": 1588
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.649182727494511e-05,
+      "loss": 1.027,
+      "step": 1589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.648938765552574e-05,
+      "loss": 1.012,
+      "step": 1590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.648694803610637e-05,
+      "loss": 1.0144,
+      "step": 1591
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6484508416686996e-05,
+      "loss": 1.0046,
+      "step": 1592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.648206879726763e-05,
+      "loss": 0.9867,
+      "step": 1593
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.647962917784826e-05,
+      "loss": 1.0056,
+      "step": 1594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6477189558428885e-05,
+      "loss": 1.0108,
+      "step": 1595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6474749939009515e-05,
+      "loss": 1.059,
+      "step": 1596
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6472310319590144e-05,
+      "loss": 0.9779,
+      "step": 1597
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6469870700170774e-05,
+      "loss": 0.9827,
+      "step": 1598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6467431080751404e-05,
+      "loss": 0.9983,
+      "step": 1599
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6464991461332034e-05,
+      "loss": 1.0407,
+      "step": 1600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6462551841912663e-05,
+      "loss": 0.9815,
+      "step": 1601
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6460112222493293e-05,
+      "loss": 0.9471,
+      "step": 1602
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6457672603073923e-05,
+      "loss": 0.9123,
+      "step": 1603
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6455232983654553e-05,
+      "loss": 0.9414,
+      "step": 1604
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.645279336423518e-05,
+      "loss": 1.0237,
+      "step": 1605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.645035374481581e-05,
+      "loss": 0.9883,
+      "step": 1606
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6447914125396442e-05,
+      "loss": 1.0253,
+      "step": 1607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6445474505977068e-05,
+      "loss": 0.9854,
+      "step": 1608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6443034886557698e-05,
+      "loss": 0.94,
+      "step": 1609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6440595267138328e-05,
+      "loss": 0.9751,
+      "step": 1610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6438155647718957e-05,
+      "loss": 0.9664,
+      "step": 1611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6435716028299587e-05,
+      "loss": 0.9747,
+      "step": 1612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6433276408880217e-05,
+      "loss": 1.0427,
+      "step": 1613
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6430836789460843e-05,
+      "loss": 0.9928,
+      "step": 1614
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6428397170041473e-05,
+      "loss": 0.9542,
+      "step": 1615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6425957550622106e-05,
+      "loss": 1.0377,
+      "step": 1616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6423517931202733e-05,
+      "loss": 0.9969,
+      "step": 1617
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6421078311783362e-05,
+      "loss": 1.0143,
+      "step": 1618
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6418638692363992e-05,
+      "loss": 0.9711,
+      "step": 1619
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6416199072944622e-05,
+      "loss": 0.9749,
+      "step": 1620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.641375945352525e-05,
+      "loss": 0.9785,
+      "step": 1621
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.641131983410588e-05,
+      "loss": 0.9962,
+      "step": 1622
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.640888021468651e-05,
+      "loss": 1.062,
+      "step": 1623
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6406440595267137e-05,
+      "loss": 0.966,
+      "step": 1624
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.640400097584777e-05,
+      "loss": 0.9838,
+      "step": 1625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.64015613564284e-05,
+      "loss": 0.9861,
+      "step": 1626
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6399121737009027e-05,
+      "loss": 1.0273,
+      "step": 1627
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6396682117589656e-05,
+      "loss": 0.9662,
+      "step": 1628
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6394242498170286e-05,
+      "loss": 0.9564,
+      "step": 1629
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6391802878750916e-05,
+      "loss": 0.9242,
+      "step": 1630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6389363259331546e-05,
+      "loss": 1.0074,
+      "step": 1631
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6386923639912175e-05,
+      "loss": 0.9614,
+      "step": 1632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6384484020492805e-05,
+      "loss": 0.9388,
+      "step": 1633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6382044401073435e-05,
+      "loss": 1.0316,
+      "step": 1634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6379604781654065e-05,
+      "loss": 0.9554,
+      "step": 1635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.637716516223469e-05,
+      "loss": 0.9997,
+      "step": 1636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.637472554281532e-05,
+      "loss": 0.9497,
+      "step": 1637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.637228592339595e-05,
+      "loss": 0.9798,
+      "step": 1638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6369846303976584e-05,
+      "loss": 1.0165,
+      "step": 1639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.636740668455721e-05,
+      "loss": 0.9489,
+      "step": 1640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.636496706513784e-05,
+      "loss": 1.0148,
+      "step": 1641
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.636252744571847e-05,
+      "loss": 1.0301,
+      "step": 1642
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.63600878262991e-05,
+      "loss": 0.9428,
+      "step": 1643
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.635764820687973e-05,
+      "loss": 1.0135,
+      "step": 1644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.635520858746036e-05,
+      "loss": 0.9707,
+      "step": 1645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6352768968040985e-05,
+      "loss": 0.9974,
+      "step": 1646
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6350329348621615e-05,
+      "loss": 0.9979,
+      "step": 1647
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6347889729202248e-05,
+      "loss": 0.9561,
+      "step": 1648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6345450109782874e-05,
+      "loss": 0.967,
+      "step": 1649
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6343010490363504e-05,
+      "loss": 1.0044,
+      "step": 1650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6340570870944134e-05,
+      "loss": 0.9744,
+      "step": 1651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6338131251524763e-05,
+      "loss": 0.9949,
+      "step": 1652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6335691632105393e-05,
+      "loss": 0.9341,
+      "step": 1653
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6333252012686023e-05,
+      "loss": 1.0028,
+      "step": 1654
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.633081239326665e-05,
+      "loss": 0.9703,
+      "step": 1655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6328372773847282e-05,
+      "loss": 0.929,
+      "step": 1656
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6325933154427912e-05,
+      "loss": 0.9891,
+      "step": 1657
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6323493535008542e-05,
+      "loss": 0.9497,
+      "step": 1658
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6321053915589168e-05,
+      "loss": 0.9845,
+      "step": 1659
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6318614296169798e-05,
+      "loss": 0.9834,
+      "step": 1660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6316174676750428e-05,
+      "loss": 0.9966,
+      "step": 1661
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6313735057331058e-05,
+      "loss": 1.0891,
+      "step": 1662
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6311295437911687e-05,
+      "loss": 0.9901,
+      "step": 1663
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6308855818492317e-05,
+      "loss": 0.9919,
+      "step": 1664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6306416199072947e-05,
+      "loss": 0.9456,
+      "step": 1665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6303976579653576e-05,
+      "loss": 0.9949,
+      "step": 1666
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6301536960234206e-05,
+      "loss": 1.0551,
+      "step": 1667
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6299097340814833e-05,
+      "loss": 0.9577,
+      "step": 1668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6296657721395462e-05,
+      "loss": 0.9818,
+      "step": 1669
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6294218101976092e-05,
+      "loss": 0.9937,
+      "step": 1670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6291778482556722e-05,
+      "loss": 1.0228,
+      "step": 1671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.628933886313735e-05,
+      "loss": 0.9237,
+      "step": 1672
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 0.9911256432533264,
+      "eval_runtime": 205.4583,
+      "eval_samples_per_second": 81.764,
+      "eval_steps_per_second": 10.221,
+      "step": 1672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.628689924371798e-05,
+      "loss": 0.9343,
+      "step": 1673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.628445962429861e-05,
+      "loss": 0.9882,
+      "step": 1674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.628202000487924e-05,
+      "loss": 0.9451,
+      "step": 1675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.627958038545987e-05,
+      "loss": 1.0386,
+      "step": 1676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.62771407660405e-05,
+      "loss": 0.9465,
+      "step": 1677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6274701146621127e-05,
+      "loss": 0.9814,
+      "step": 1678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.627226152720176e-05,
+      "loss": 1.0233,
+      "step": 1679
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.626982190778239e-05,
+      "loss": 1.0072,
+      "step": 1680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6267382288363016e-05,
+      "loss": 1.0707,
+      "step": 1681
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6264942668943646e-05,
+      "loss": 1.0158,
+      "step": 1682
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6262503049524275e-05,
+      "loss": 0.9611,
+      "step": 1683
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6260063430104905e-05,
+      "loss": 0.9644,
+      "step": 1684
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6257623810685535e-05,
+      "loss": 0.9407,
+      "step": 1685
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6255184191266165e-05,
+      "loss": 0.9981,
+      "step": 1686
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.625274457184679e-05,
+      "loss": 1.0149,
+      "step": 1687
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6250304952427424e-05,
+      "loss": 0.9148,
+      "step": 1688
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6247865333008054e-05,
+      "loss": 1.042,
+      "step": 1689
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.624542571358868e-05,
+      "loss": 0.9696,
+      "step": 1690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.624298609416931e-05,
+      "loss": 0.9959,
+      "step": 1691
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.624054647474994e-05,
+      "loss": 1.0129,
+      "step": 1692
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.623810685533057e-05,
+      "loss": 1.002,
+      "step": 1693
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.62356672359112e-05,
+      "loss": 0.9765,
+      "step": 1694
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.623322761649183e-05,
+      "loss": 0.9784,
+      "step": 1695
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.623078799707246e-05,
+      "loss": 1.0018,
+      "step": 1696
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.622834837765309e-05,
+      "loss": 0.9905,
+      "step": 1697
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6225908758233718e-05,
+      "loss": 1.0037,
+      "step": 1698
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6223469138814348e-05,
+      "loss": 1.0149,
+      "step": 1699
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6221029519394974e-05,
+      "loss": 0.9472,
+      "step": 1700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6218589899975604e-05,
+      "loss": 0.9952,
+      "step": 1701
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6216150280556234e-05,
+      "loss": 0.9967,
+      "step": 1702
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6213710661136863e-05,
+      "loss": 1.0006,
+      "step": 1703
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6211271041717493e-05,
+      "loss": 0.9889,
+      "step": 1704
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6208831422298123e-05,
+      "loss": 1.0225,
+      "step": 1705
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6206391802878753e-05,
+      "loss": 0.9587,
+      "step": 1706
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6203952183459382e-05,
+      "loss": 1.0544,
+      "step": 1707
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6201512564040012e-05,
+      "loss": 0.9707,
+      "step": 1708
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.619907294462064e-05,
+      "loss": 1.0395,
+      "step": 1709
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6196633325201268e-05,
+      "loss": 0.9311,
+      "step": 1710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.61941937057819e-05,
+      "loss": 0.9873,
+      "step": 1711
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6191754086362528e-05,
+      "loss": 0.971,
+      "step": 1712
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6189314466943158e-05,
+      "loss": 0.9894,
+      "step": 1713
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6186874847523787e-05,
+      "loss": 0.9722,
+      "step": 1714
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6184435228104417e-05,
+      "loss": 1.1108,
+      "step": 1715
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6181995608685047e-05,
+      "loss": 0.9737,
+      "step": 1716
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6179555989265677e-05,
+      "loss": 0.9458,
+      "step": 1717
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6177116369846306e-05,
+      "loss": 0.9608,
+      "step": 1718
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6174676750426933e-05,
+      "loss": 0.9332,
+      "step": 1719
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6172237131007566e-05,
+      "loss": 0.9524,
+      "step": 1720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6169797511588195e-05,
+      "loss": 0.9776,
+      "step": 1721
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6167357892168822e-05,
+      "loss": 0.9491,
+      "step": 1722
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.616491827274945e-05,
+      "loss": 0.9347,
+      "step": 1723
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.616247865333008e-05,
+      "loss": 0.9941,
+      "step": 1724
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.616003903391071e-05,
+      "loss": 0.9549,
+      "step": 1725
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.615759941449134e-05,
+      "loss": 0.9824,
+      "step": 1726
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.615515979507197e-05,
+      "loss": 1.0165,
+      "step": 1727
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.61527201756526e-05,
+      "loss": 1.0196,
+      "step": 1728
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.615028055623323e-05,
+      "loss": 0.962,
+      "step": 1729
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.614784093681386e-05,
+      "loss": 1.0312,
+      "step": 1730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6145401317394486e-05,
+      "loss": 0.9452,
+      "step": 1731
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6142961697975116e-05,
+      "loss": 0.9703,
+      "step": 1732
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6140522078555746e-05,
+      "loss": 1.0418,
+      "step": 1733
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6138082459136375e-05,
+      "loss": 0.9435,
+      "step": 1734
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6135642839717005e-05,
+      "loss": 0.8983,
+      "step": 1735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6133203220297635e-05,
+      "loss": 0.9329,
+      "step": 1736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6130763600878265e-05,
+      "loss": 1.0081,
+      "step": 1737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6128323981458894e-05,
+      "loss": 1.0042,
+      "step": 1738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6125884362039524e-05,
+      "loss": 0.9337,
+      "step": 1739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6123444742620154e-05,
+      "loss": 0.9772,
+      "step": 1740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.612100512320078e-05,
+      "loss": 0.9457,
+      "step": 1741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.611856550378141e-05,
+      "loss": 0.9725,
+      "step": 1742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6116125884362043e-05,
+      "loss": 0.9353,
+      "step": 1743
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.611368626494267e-05,
+      "loss": 0.8915,
+      "step": 1744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.61112466455233e-05,
+      "loss": 1.0125,
+      "step": 1745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.610880702610393e-05,
+      "loss": 1.0046,
+      "step": 1746
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.610636740668456e-05,
+      "loss": 1.0369,
+      "step": 1747
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.610392778726519e-05,
+      "loss": 0.9725,
+      "step": 1748
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6101488167845818e-05,
+      "loss": 0.9687,
+      "step": 1749
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6099048548426445e-05,
+      "loss": 1.0329,
+      "step": 1750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6096608929007074e-05,
+      "loss": 0.9872,
+      "step": 1751
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6094169309587707e-05,
+      "loss": 1.0222,
+      "step": 1752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6091729690168334e-05,
+      "loss": 0.9174,
+      "step": 1753
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6089290070748964e-05,
+      "loss": 0.9731,
+      "step": 1754
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6086850451329593e-05,
+      "loss": 0.9957,
+      "step": 1755
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6084410831910223e-05,
+      "loss": 0.9269,
+      "step": 1756
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6081971212490853e-05,
+      "loss": 0.9403,
+      "step": 1757
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6079531593071482e-05,
+      "loss": 1.0078,
+      "step": 1758
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6077091973652112e-05,
+      "loss": 0.9384,
+      "step": 1759
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6074652354232742e-05,
+      "loss": 0.9902,
+      "step": 1760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6072212734813372e-05,
+      "loss": 0.9596,
+      "step": 1761
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6069773115394e-05,
+      "loss": 1.0005,
+      "step": 1762
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6067333495974628e-05,
+      "loss": 0.9886,
+      "step": 1763
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6064893876555258e-05,
+      "loss": 1.0121,
+      "step": 1764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6062454257135887e-05,
+      "loss": 0.9796,
+      "step": 1765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6060014637716517e-05,
+      "loss": 0.9735,
+      "step": 1766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6057575018297147e-05,
+      "loss": 0.9692,
+      "step": 1767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6055135398877777e-05,
+      "loss": 0.9985,
+      "step": 1768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6052695779458406e-05,
+      "loss": 0.9937,
+      "step": 1769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6050256160039036e-05,
+      "loss": 0.9411,
+      "step": 1770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6047816540619666e-05,
+      "loss": 0.929,
+      "step": 1771
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6045376921200292e-05,
+      "loss": 1.0047,
+      "step": 1772
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6042937301780922e-05,
+      "loss": 1.0315,
+      "step": 1773
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.604049768236155e-05,
+      "loss": 0.9838,
+      "step": 1774
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6038058062942185e-05,
+      "loss": 1.035,
+      "step": 1775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.603561844352281e-05,
+      "loss": 0.9532,
+      "step": 1776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.603317882410344e-05,
+      "loss": 0.9218,
+      "step": 1777
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.603073920468407e-05,
+      "loss": 0.9292,
+      "step": 1778
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.60282995852647e-05,
+      "loss": 0.9136,
+      "step": 1779
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.602585996584533e-05,
+      "loss": 1.0703,
+      "step": 1780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.602342034642596e-05,
+      "loss": 0.9828,
+      "step": 1781
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6020980727006586e-05,
+      "loss": 1.0905,
+      "step": 1782
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.601854110758722e-05,
+      "loss": 0.9837,
+      "step": 1783
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.601610148816785e-05,
+      "loss": 0.9937,
+      "step": 1784
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6013661868748475e-05,
+      "loss": 0.9895,
+      "step": 1785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6011222249329105e-05,
+      "loss": 0.9362,
+      "step": 1786
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6008782629909735e-05,
+      "loss": 0.9338,
+      "step": 1787
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6006343010490365e-05,
+      "loss": 0.967,
+      "step": 1788
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6003903391070994e-05,
+      "loss": 0.8759,
+      "step": 1789
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6001463771651624e-05,
+      "loss": 1.0249,
+      "step": 1790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.599902415223225e-05,
+      "loss": 1.0393,
+      "step": 1791
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5996584532812884e-05,
+      "loss": 0.9371,
+      "step": 1792
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5994144913393513e-05,
+      "loss": 0.9069,
+      "step": 1793
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5991705293974143e-05,
+      "loss": 0.9431,
+      "step": 1794
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.598926567455477e-05,
+      "loss": 0.9587,
+      "step": 1795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.59868260551354e-05,
+      "loss": 0.9109,
+      "step": 1796
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.598438643571603e-05,
+      "loss": 0.9774,
+      "step": 1797
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.598194681629666e-05,
+      "loss": 0.9672,
+      "step": 1798
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.597950719687729e-05,
+      "loss": 0.9348,
+      "step": 1799
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5977067577457918e-05,
+      "loss": 0.9715,
+      "step": 1800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5974627958038548e-05,
+      "loss": 1.0276,
+      "step": 1801
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5972188338619178e-05,
+      "loss": 0.9549,
+      "step": 1802
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5969748719199807e-05,
+      "loss": 0.9903,
+      "step": 1803
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5967309099780434e-05,
+      "loss": 0.9633,
+      "step": 1804
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5964869480361064e-05,
+      "loss": 0.9561,
+      "step": 1805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5962429860941697e-05,
+      "loss": 1.0072,
+      "step": 1806
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5959990241522323e-05,
+      "loss": 0.9563,
+      "step": 1807
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5957550622102953e-05,
+      "loss": 1.0036,
+      "step": 1808
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5955111002683583e-05,
+      "loss": 0.9713,
+      "step": 1809
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5952671383264212e-05,
+      "loss": 1.0212,
+      "step": 1810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5950231763844842e-05,
+      "loss": 0.9835,
+      "step": 1811
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5947792144425472e-05,
+      "loss": 1.0165,
+      "step": 1812
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.59453525250061e-05,
+      "loss": 0.9858,
+      "step": 1813
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5942912905586728e-05,
+      "loss": 1.0478,
+      "step": 1814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.594047328616736e-05,
+      "loss": 1.0751,
+      "step": 1815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.593803366674799e-05,
+      "loss": 0.9889,
+      "step": 1816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5935594047328617e-05,
+      "loss": 0.999,
+      "step": 1817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5933154427909247e-05,
+      "loss": 0.9451,
+      "step": 1818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5930714808489877e-05,
+      "loss": 0.8945,
+      "step": 1819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5928275189070506e-05,
+      "loss": 0.9873,
+      "step": 1820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5925835569651136e-05,
+      "loss": 0.9439,
+      "step": 1821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5923395950231766e-05,
+      "loss": 1.0292,
+      "step": 1822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5920956330812392e-05,
+      "loss": 1.0294,
+      "step": 1823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5918516711393025e-05,
+      "loss": 0.9556,
+      "step": 1824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5916077091973655e-05,
+      "loss": 1.0034,
+      "step": 1825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.591363747255428e-05,
+      "loss": 0.9683,
+      "step": 1826
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.591119785313491e-05,
+      "loss": 1.0867,
+      "step": 1827
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.590875823371554e-05,
+      "loss": 0.9463,
+      "step": 1828
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.590631861429617e-05,
+      "loss": 1.0436,
+      "step": 1829
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.59038789948768e-05,
+      "loss": 0.9129,
+      "step": 1830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.590143937545743e-05,
+      "loss": 0.9691,
+      "step": 1831
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.589899975603806e-05,
+      "loss": 1.0327,
+      "step": 1832
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.589656013661869e-05,
+      "loss": 1.0101,
+      "step": 1833
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.589412051719932e-05,
+      "loss": 0.9803,
+      "step": 1834
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.589168089777995e-05,
+      "loss": 1.0084,
+      "step": 1835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5889241278360575e-05,
+      "loss": 0.9994,
+      "step": 1836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5886801658941205e-05,
+      "loss": 0.9858,
+      "step": 1837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.588436203952184e-05,
+      "loss": 0.9728,
+      "step": 1838
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5881922420102465e-05,
+      "loss": 1.0413,
+      "step": 1839
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5879482800683094e-05,
+      "loss": 1.0167,
+      "step": 1840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5877043181263724e-05,
+      "loss": 1.0486,
+      "step": 1841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5874603561844354e-05,
+      "loss": 0.9528,
+      "step": 1842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5872163942424984e-05,
+      "loss": 1.0002,
+      "step": 1843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5869724323005613e-05,
+      "loss": 0.9198,
+      "step": 1844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.586728470358624e-05,
+      "loss": 0.9889,
+      "step": 1845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.586484508416687e-05,
+      "loss": 1.0011,
+      "step": 1846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5862405464747503e-05,
+      "loss": 0.9795,
+      "step": 1847
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.585996584532813e-05,
+      "loss": 1.0046,
+      "step": 1848
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.585752622590876e-05,
+      "loss": 1.0368,
+      "step": 1849
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.585508660648939e-05,
+      "loss": 0.9786,
+      "step": 1850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5852646987070018e-05,
+      "loss": 1.0035,
+      "step": 1851
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5850207367650648e-05,
+      "loss": 0.955,
+      "step": 1852
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5847767748231278e-05,
+      "loss": 0.9867,
+      "step": 1853
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5845328128811907e-05,
+      "loss": 0.9683,
+      "step": 1854
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5842888509392537e-05,
+      "loss": 0.9441,
+      "step": 1855
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5840448889973167e-05,
+      "loss": 0.9822,
+      "step": 1856
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5838009270553797e-05,
+      "loss": 0.9967,
+      "step": 1857
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5835569651134423e-05,
+      "loss": 0.9533,
+      "step": 1858
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5833130031715053e-05,
+      "loss": 0.9639,
+      "step": 1859
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5830690412295683e-05,
+      "loss": 0.9916,
+      "step": 1860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5828250792876312e-05,
+      "loss": 0.9574,
+      "step": 1861
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5825811173456942e-05,
+      "loss": 0.9752,
+      "step": 1862
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5823371554037572e-05,
+      "loss": 0.9723,
+      "step": 1863
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.58209319346182e-05,
+      "loss": 1.0603,
+      "step": 1864
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.581849231519883e-05,
+      "loss": 0.892,
+      "step": 1865
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.581605269577946e-05,
+      "loss": 0.9202,
+      "step": 1866
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5813613076360087e-05,
+      "loss": 0.9434,
+      "step": 1867
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5811173456940717e-05,
+      "loss": 1.0135,
+      "step": 1868
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5808733837521347e-05,
+      "loss": 1.0189,
+      "step": 1869
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5806294218101977e-05,
+      "loss": 1.0379,
+      "step": 1870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5803854598682606e-05,
+      "loss": 0.9536,
+      "step": 1871
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5801414979263236e-05,
+      "loss": 1.0219,
+      "step": 1872
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5798975359843866e-05,
+      "loss": 0.9293,
+      "step": 1873
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5796535740424496e-05,
+      "loss": 1.0242,
+      "step": 1874
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5794096121005125e-05,
+      "loss": 0.9367,
+      "step": 1875
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5791656501585755e-05,
+      "loss": 0.9914,
+      "step": 1876
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.578921688216638e-05,
+      "loss": 0.9545,
+      "step": 1877
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.578677726274701e-05,
+      "loss": 0.9328,
+      "step": 1878
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5784337643327644e-05,
+      "loss": 0.9427,
+      "step": 1879
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.578189802390827e-05,
+      "loss": 0.9642,
+      "step": 1880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.57794584044889e-05,
+      "loss": 0.9296,
+      "step": 1881
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.577701878506953e-05,
+      "loss": 0.9622,
+      "step": 1882
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.577457916565016e-05,
+      "loss": 0.9736,
+      "step": 1883
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.577213954623079e-05,
+      "loss": 0.9469,
+      "step": 1884
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.576969992681142e-05,
+      "loss": 0.9725,
+      "step": 1885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5767260307392046e-05,
+      "loss": 0.9722,
+      "step": 1886
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.576482068797268e-05,
+      "loss": 0.9258,
+      "step": 1887
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.576238106855331e-05,
+      "loss": 0.9562,
+      "step": 1888
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5759941449133935e-05,
+      "loss": 0.9716,
+      "step": 1889
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5757501829714565e-05,
+      "loss": 1.07,
+      "step": 1890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5755062210295194e-05,
+      "loss": 0.9947,
+      "step": 1891
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5752622590875824e-05,
+      "loss": 0.9944,
+      "step": 1892
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5750182971456454e-05,
+      "loss": 1.0509,
+      "step": 1893
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5747743352037084e-05,
+      "loss": 0.952,
+      "step": 1894
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5745303732617713e-05,
+      "loss": 0.9893,
+      "step": 1895
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5742864113198343e-05,
+      "loss": 0.98,
+      "step": 1896
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5740424493778973e-05,
+      "loss": 0.9698,
+      "step": 1897
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5737984874359603e-05,
+      "loss": 0.9923,
+      "step": 1898
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.573554525494023e-05,
+      "loss": 0.9595,
+      "step": 1899
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.573310563552086e-05,
+      "loss": 0.9615,
+      "step": 1900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.573066601610149e-05,
+      "loss": 0.9648,
+      "step": 1901
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5728226396682118e-05,
+      "loss": 0.9625,
+      "step": 1902
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5725786777262748e-05,
+      "loss": 0.9661,
+      "step": 1903
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5723347157843378e-05,
+      "loss": 0.9563,
+      "step": 1904
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5720907538424007e-05,
+      "loss": 1.0533,
+      "step": 1905
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5718467919004637e-05,
+      "loss": 0.9881,
+      "step": 1906
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5716028299585267e-05,
+      "loss": 0.9602,
+      "step": 1907
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5713588680165893e-05,
+      "loss": 0.9283,
+      "step": 1908
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5711149060746523e-05,
+      "loss": 0.874,
+      "step": 1909
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5708709441327156e-05,
+      "loss": 0.947,
+      "step": 1910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5706269821907786e-05,
+      "loss": 0.9443,
+      "step": 1911
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5703830202488412e-05,
+      "loss": 0.9441,
+      "step": 1912
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5701390583069042e-05,
+      "loss": 0.998,
+      "step": 1913
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5698950963649672e-05,
+      "loss": 0.9427,
+      "step": 1914
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.56965113442303e-05,
+      "loss": 1.0263,
+      "step": 1915
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.569407172481093e-05,
+      "loss": 0.978,
+      "step": 1916
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.569163210539156e-05,
+      "loss": 0.9805,
+      "step": 1917
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5689192485972187e-05,
+      "loss": 1.0621,
+      "step": 1918
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.568675286655282e-05,
+      "loss": 0.9933,
+      "step": 1919
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.568431324713345e-05,
+      "loss": 0.942,
+      "step": 1920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5681873627714077e-05,
+      "loss": 0.8764,
+      "step": 1921
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5679434008294706e-05,
+      "loss": 0.9518,
+      "step": 1922
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5676994388875336e-05,
+      "loss": 1.0079,
+      "step": 1923
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5674554769455966e-05,
+      "loss": 0.9371,
+      "step": 1924
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5672115150036596e-05,
+      "loss": 1.0219,
+      "step": 1925
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5669675530617225e-05,
+      "loss": 0.9179,
+      "step": 1926
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.566723591119785e-05,
+      "loss": 0.9353,
+      "step": 1927
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5664796291778485e-05,
+      "loss": 0.9853,
+      "step": 1928
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5662356672359115e-05,
+      "loss": 0.9902,
+      "step": 1929
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5659917052939744e-05,
+      "loss": 0.9478,
+      "step": 1930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.565747743352037e-05,
+      "loss": 1.002,
+      "step": 1931
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5655037814101e-05,
+      "loss": 0.997,
+      "step": 1932
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5652598194681634e-05,
+      "loss": 1.0093,
+      "step": 1933
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.565015857526226e-05,
+      "loss": 1.0124,
+      "step": 1934
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.564771895584289e-05,
+      "loss": 0.9268,
+      "step": 1935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.564527933642352e-05,
+      "loss": 0.9817,
+      "step": 1936
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.564283971700415e-05,
+      "loss": 0.9754,
+      "step": 1937
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.564040009758478e-05,
+      "loss": 0.9348,
+      "step": 1938
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.563796047816541e-05,
+      "loss": 0.9143,
+      "step": 1939
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5635520858746035e-05,
+      "loss": 1.0233,
+      "step": 1940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5633081239326665e-05,
+      "loss": 0.9658,
+      "step": 1941
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5630641619907298e-05,
+      "loss": 0.9893,
+      "step": 1942
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5628202000487924e-05,
+      "loss": 0.9235,
+      "step": 1943
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5625762381068554e-05,
+      "loss": 1.0176,
+      "step": 1944
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5623322761649184e-05,
+      "loss": 0.9952,
+      "step": 1945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5620883142229813e-05,
+      "loss": 1.0521,
+      "step": 1946
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5618443522810443e-05,
+      "loss": 0.9867,
+      "step": 1947
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5616003903391073e-05,
+      "loss": 1.0136,
+      "step": 1948
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5613564283971703e-05,
+      "loss": 1.0359,
+      "step": 1949
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.561112466455233e-05,
+      "loss": 1.0269,
+      "step": 1950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5608685045132962e-05,
+      "loss": 1.0752,
+      "step": 1951
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5606245425713592e-05,
+      "loss": 0.9277,
+      "step": 1952
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5603805806294218e-05,
+      "loss": 0.9575,
+      "step": 1953
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5601366186874848e-05,
+      "loss": 0.9741,
+      "step": 1954
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5598926567455478e-05,
+      "loss": 0.972,
+      "step": 1955
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5596486948036107e-05,
+      "loss": 0.9444,
+      "step": 1956
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5594047328616737e-05,
+      "loss": 1.0108,
+      "step": 1957
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5591607709197367e-05,
+      "loss": 1.0431,
+      "step": 1958
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5589168089777997e-05,
+      "loss": 0.9687,
+      "step": 1959
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5586728470358626e-05,
+      "loss": 0.9578,
+      "step": 1960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5584288850939256e-05,
+      "loss": 0.9798,
+      "step": 1961
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5581849231519883e-05,
+      "loss": 0.9686,
+      "step": 1962
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5579409612100512e-05,
+      "loss": 1.0219,
+      "step": 1963
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5576969992681142e-05,
+      "loss": 0.9724,
+      "step": 1964
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5574530373261772e-05,
+      "loss": 1.0109,
+      "step": 1965
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.55720907538424e-05,
+      "loss": 0.8818,
+      "step": 1966
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.556965113442303e-05,
+      "loss": 1.0169,
+      "step": 1967
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.556721151500366e-05,
+      "loss": 1.0032,
+      "step": 1968
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.556477189558429e-05,
+      "loss": 0.8983,
+      "step": 1969
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.556233227616492e-05,
+      "loss": 0.9923,
+      "step": 1970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.555989265674555e-05,
+      "loss": 0.9719,
+      "step": 1971
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5557453037326177e-05,
+      "loss": 1.0394,
+      "step": 1972
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5555013417906806e-05,
+      "loss": 0.9816,
+      "step": 1973
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.555257379848744e-05,
+      "loss": 0.9724,
+      "step": 1974
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5550134179068066e-05,
+      "loss": 0.9855,
+      "step": 1975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5547694559648696e-05,
+      "loss": 1.0061,
+      "step": 1976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5545254940229325e-05,
+      "loss": 0.9669,
+      "step": 1977
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5542815320809955e-05,
+      "loss": 0.9524,
+      "step": 1978
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5540375701390585e-05,
+      "loss": 0.9711,
+      "step": 1979
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5537936081971215e-05,
+      "loss": 0.9498,
+      "step": 1980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.553549646255184e-05,
+      "loss": 0.9365,
+      "step": 1981
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5533056843132474e-05,
+      "loss": 1.0496,
+      "step": 1982
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5530617223713104e-05,
+      "loss": 0.9681,
+      "step": 1983
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.552817760429373e-05,
+      "loss": 0.9844,
+      "step": 1984
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.552573798487436e-05,
+      "loss": 0.918,
+      "step": 1985
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.552329836545499e-05,
+      "loss": 1.0312,
+      "step": 1986
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.552085874603562e-05,
+      "loss": 1.0042,
+      "step": 1987
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.551841912661625e-05,
+      "loss": 0.9783,
+      "step": 1988
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.551597950719688e-05,
+      "loss": 0.9688,
+      "step": 1989
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.551353988777751e-05,
+      "loss": 0.9712,
+      "step": 1990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.551110026835814e-05,
+      "loss": 0.9129,
+      "step": 1991
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5508660648938768e-05,
+      "loss": 0.9991,
+      "step": 1992
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5506221029519398e-05,
+      "loss": 0.9646,
+      "step": 1993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5503781410100024e-05,
+      "loss": 1.0041,
+      "step": 1994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5501341790680654e-05,
+      "loss": 0.9636,
+      "step": 1995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5498902171261284e-05,
+      "loss": 0.955,
+      "step": 1996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5496462551841913e-05,
+      "loss": 0.9095,
+      "step": 1997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5494022932422543e-05,
+      "loss": 0.9757,
+      "step": 1998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5491583313003173e-05,
+      "loss": 0.9516,
+      "step": 1999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5489143693583803e-05,
+      "loss": 0.9746,
+      "step": 2000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5486704074164432e-05,
+      "loss": 0.974,
+      "step": 2001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5484264454745062e-05,
+      "loss": 1.0471,
+      "step": 2002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.548182483532569e-05,
+      "loss": 0.9898,
+      "step": 2003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5479385215906318e-05,
+      "loss": 1.002,
+      "step": 2004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5476945596486948e-05,
+      "loss": 1.001,
+      "step": 2005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5474505977067578e-05,
+      "loss": 0.8983,
+      "step": 2006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5472066357648208e-05,
+      "loss": 0.9968,
+      "step": 2007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5469626738228837e-05,
+      "loss": 0.9672,
+      "step": 2008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5467187118809467e-05,
+      "loss": 0.9718,
+      "step": 2009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5464747499390097e-05,
+      "loss": 1.0282,
+      "step": 2010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5462307879970726e-05,
+      "loss": 0.9416,
+      "step": 2011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5459868260551356e-05,
+      "loss": 0.9662,
+      "step": 2012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5457428641131983e-05,
+      "loss": 0.9437,
+      "step": 2013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5454989021712616e-05,
+      "loss": 0.982,
+      "step": 2014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5452549402293245e-05,
+      "loss": 0.9567,
+      "step": 2015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5450109782873872e-05,
+      "loss": 0.8858,
+      "step": 2016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.54476701634545e-05,
+      "loss": 1.0503,
+      "step": 2017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.544523054403513e-05,
+      "loss": 0.9626,
+      "step": 2018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.544279092461576e-05,
+      "loss": 0.9681,
+      "step": 2019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.544035130519639e-05,
+      "loss": 0.9699,
+      "step": 2020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.543791168577702e-05,
+      "loss": 0.958,
+      "step": 2021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5435472066357647e-05,
+      "loss": 0.891,
+      "step": 2022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.543303244693828e-05,
+      "loss": 0.9276,
+      "step": 2023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.543059282751891e-05,
+      "loss": 0.9416,
+      "step": 2024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5428153208099536e-05,
+      "loss": 0.9144,
+      "step": 2025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5425713588680166e-05,
+      "loss": 0.9719,
+      "step": 2026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5423273969260796e-05,
+      "loss": 0.9404,
+      "step": 2027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5420834349841425e-05,
+      "loss": 0.9595,
+      "step": 2028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5418394730422055e-05,
+      "loss": 0.8985,
+      "step": 2029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5415955111002685e-05,
+      "loss": 0.9787,
+      "step": 2030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5413515491583315e-05,
+      "loss": 0.974,
+      "step": 2031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5411075872163944e-05,
+      "loss": 0.9391,
+      "step": 2032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5408636252744574e-05,
+      "loss": 1.0121,
+      "step": 2033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5406196633325204e-05,
+      "loss": 0.9722,
+      "step": 2034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.540375701390583e-05,
+      "loss": 0.9199,
+      "step": 2035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.540131739448646e-05,
+      "loss": 0.997,
+      "step": 2036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5398877775067093e-05,
+      "loss": 0.9638,
+      "step": 2037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.539643815564772e-05,
+      "loss": 0.9298,
+      "step": 2038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.539399853622835e-05,
+      "loss": 0.9416,
+      "step": 2039
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.539155891680898e-05,
+      "loss": 0.986,
+      "step": 2040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.538911929738961e-05,
+      "loss": 0.9694,
+      "step": 2041
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.538667967797024e-05,
+      "loss": 0.964,
+      "step": 2042
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5384240058550868e-05,
+      "loss": 0.9556,
+      "step": 2043
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5381800439131495e-05,
+      "loss": 1.109,
+      "step": 2044
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5379360819712124e-05,
+      "loss": 0.9821,
+      "step": 2045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5376921200292757e-05,
+      "loss": 0.9946,
+      "step": 2046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5374481580873387e-05,
+      "loss": 1.1114,
+      "step": 2047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5372041961454013e-05,
+      "loss": 0.9586,
+      "step": 2048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5369602342034643e-05,
+      "loss": 0.9378,
+      "step": 2049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5367162722615273e-05,
+      "loss": 0.9358,
+      "step": 2050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5364723103195903e-05,
+      "loss": 0.9375,
+      "step": 2051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5362283483776532e-05,
+      "loss": 0.9763,
+      "step": 2052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5359843864357162e-05,
+      "loss": 0.9744,
+      "step": 2053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.535740424493779e-05,
+      "loss": 0.9564,
+      "step": 2054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5354964625518422e-05,
+      "loss": 0.9759,
+      "step": 2055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.535252500609905e-05,
+      "loss": 0.9534,
+      "step": 2056
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5350085386679678e-05,
+      "loss": 0.9931,
+      "step": 2057
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5347645767260308e-05,
+      "loss": 0.9524,
+      "step": 2058
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5345206147840937e-05,
+      "loss": 0.9002,
+      "step": 2059
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5342766528421567e-05,
+      "loss": 0.9684,
+      "step": 2060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5340326909002197e-05,
+      "loss": 0.9428,
+      "step": 2061
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5337887289582827e-05,
+      "loss": 0.9109,
+      "step": 2062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5335447670163456e-05,
+      "loss": 0.9407,
+      "step": 2063
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5333008050744086e-05,
+      "loss": 0.9783,
+      "step": 2064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5330568431324716e-05,
+      "loss": 0.9578,
+      "step": 2065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5328128811905345e-05,
+      "loss": 0.9695,
+      "step": 2066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5325689192485972e-05,
+      "loss": 0.9364,
+      "step": 2067
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.53232495730666e-05,
+      "loss": 0.9618,
+      "step": 2068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5320809953647235e-05,
+      "loss": 0.9012,
+      "step": 2069
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.531837033422786e-05,
+      "loss": 0.9886,
+      "step": 2070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.531593071480849e-05,
+      "loss": 0.9847,
+      "step": 2071
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.531349109538912e-05,
+      "loss": 0.9563,
+      "step": 2072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.531105147596975e-05,
+      "loss": 1.0017,
+      "step": 2073
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.530861185655038e-05,
+      "loss": 0.9669,
+      "step": 2074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.530617223713101e-05,
+      "loss": 0.9361,
+      "step": 2075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5303732617711636e-05,
+      "loss": 0.9625,
+      "step": 2076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5301292998292266e-05,
+      "loss": 0.9078,
+      "step": 2077
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.52988533788729e-05,
+      "loss": 0.959,
+      "step": 2078
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5296413759453525e-05,
+      "loss": 0.9731,
+      "step": 2079
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5293974140034155e-05,
+      "loss": 0.8883,
+      "step": 2080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5291534520614785e-05,
+      "loss": 0.9368,
+      "step": 2081
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5289094901195415e-05,
+      "loss": 0.9371,
+      "step": 2082
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5286655281776044e-05,
+      "loss": 1.035,
+      "step": 2083
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5284215662356674e-05,
+      "loss": 0.9907,
+      "step": 2084
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5281776042937304e-05,
+      "loss": 0.9175,
+      "step": 2085
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5279336423517934e-05,
+      "loss": 0.9634,
+      "step": 2086
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5276896804098563e-05,
+      "loss": 0.9908,
+      "step": 2087
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5274457184679193e-05,
+      "loss": 0.9433,
+      "step": 2088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.527201756525982e-05,
+      "loss": 1.0154,
+      "step": 2089
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.526957794584045e-05,
+      "loss": 1.0094,
+      "step": 2090
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.9708256125450134,
+      "eval_runtime": 203.5221,
+      "eval_samples_per_second": 82.541,
+      "eval_steps_per_second": 10.318,
+      "step": 2090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.526713832642108e-05,
+      "loss": 0.9263,
+      "step": 2091
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.526469870700171e-05,
+      "loss": 0.9145,
+      "step": 2092
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.526225908758234e-05,
+      "loss": 0.977,
+      "step": 2093
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5259819468162968e-05,
+      "loss": 0.9751,
+      "step": 2094
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5257379848743596e-05,
+      "loss": 0.9118,
+      "step": 2095
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5254940229324226e-05,
+      "loss": 0.971,
+      "step": 2096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5252500609904857e-05,
+      "loss": 0.9398,
+      "step": 2097
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5250060990485484e-05,
+      "loss": 0.9798,
+      "step": 2098
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5247621371066115e-05,
+      "loss": 0.9357,
+      "step": 2099
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5245181751646745e-05,
+      "loss": 0.877,
+      "step": 2100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5242742132227373e-05,
+      "loss": 0.9484,
+      "step": 2101
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5240302512808003e-05,
+      "loss": 0.8969,
+      "step": 2102
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5237862893388632e-05,
+      "loss": 0.9863,
+      "step": 2103
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5235423273969264e-05,
+      "loss": 1.0163,
+      "step": 2104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.523298365454989e-05,
+      "loss": 1.0221,
+      "step": 2105
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5230544035130522e-05,
+      "loss": 0.9516,
+      "step": 2106
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5228104415711151e-05,
+      "loss": 0.9694,
+      "step": 2107
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.522566479629178e-05,
+      "loss": 0.978,
+      "step": 2108
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.522322517687241e-05,
+      "loss": 1.0208,
+      "step": 2109
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5220785557453039e-05,
+      "loss": 0.9715,
+      "step": 2110
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5218345938033667e-05,
+      "loss": 1.0433,
+      "step": 2111
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5215906318614297e-05,
+      "loss": 0.9908,
+      "step": 2112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5213466699194928e-05,
+      "loss": 0.9965,
+      "step": 2113
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5211027079775556e-05,
+      "loss": 1.002,
+      "step": 2114
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5208587460356186e-05,
+      "loss": 1.0256,
+      "step": 2115
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5206147840936816e-05,
+      "loss": 1.0065,
+      "step": 2116
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5203708221517444e-05,
+      "loss": 0.944,
+      "step": 2117
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5201268602098074e-05,
+      "loss": 0.9972,
+      "step": 2118
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5198828982678703e-05,
+      "loss": 1.0086,
+      "step": 2119
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5196389363259331e-05,
+      "loss": 1.0272,
+      "step": 2120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5193949743839961e-05,
+      "loss": 0.9928,
+      "step": 2121
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5191510124420593e-05,
+      "loss": 1.0016,
+      "step": 2122
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5189070505001222e-05,
+      "loss": 0.9219,
+      "step": 2123
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.518663088558185e-05,
+      "loss": 0.9259,
+      "step": 2124
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.518419126616248e-05,
+      "loss": 0.9662,
+      "step": 2125
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.518175164674311e-05,
+      "loss": 1.0096,
+      "step": 2126
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5179312027323738e-05,
+      "loss": 1.0271,
+      "step": 2127
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5176872407904368e-05,
+      "loss": 0.9209,
+      "step": 2128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5174432788484999e-05,
+      "loss": 0.9963,
+      "step": 2129
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5171993169065627e-05,
+      "loss": 0.9921,
+      "step": 2130
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5169553549646257e-05,
+      "loss": 0.976,
+      "step": 2131
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5167113930226887e-05,
+      "loss": 0.8505,
+      "step": 2132
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5164674310807515e-05,
+      "loss": 0.796,
+      "step": 2133
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5162234691388144e-05,
+      "loss": 0.8105,
+      "step": 2134
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5159795071968774e-05,
+      "loss": 0.7939,
+      "step": 2135
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5157355452549402e-05,
+      "loss": 0.8035,
+      "step": 2136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5154915833130032e-05,
+      "loss": 0.8228,
+      "step": 2137
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5152476213710663e-05,
+      "loss": 0.8158,
+      "step": 2138
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5150036594291291e-05,
+      "loss": 0.8181,
+      "step": 2139
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5147596974871921e-05,
+      "loss": 0.7992,
+      "step": 2140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5145157355452551e-05,
+      "loss": 0.7986,
+      "step": 2141
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5142717736033179e-05,
+      "loss": 0.8063,
+      "step": 2142
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5140278116613809e-05,
+      "loss": 0.7671,
+      "step": 2143
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5137838497194438e-05,
+      "loss": 0.8825,
+      "step": 2144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.513539887777507e-05,
+      "loss": 0.8485,
+      "step": 2145
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5132959258355698e-05,
+      "loss": 0.8137,
+      "step": 2146
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5130519638936328e-05,
+      "loss": 0.8443,
+      "step": 2147
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5128080019516957e-05,
+      "loss": 0.8697,
+      "step": 2148
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5125640400097585e-05,
+      "loss": 0.7823,
+      "step": 2149
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5123200780678215e-05,
+      "loss": 0.8003,
+      "step": 2150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5120761161258845e-05,
+      "loss": 0.7662,
+      "step": 2151
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5118321541839473e-05,
+      "loss": 0.813,
+      "step": 2152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5115881922420104e-05,
+      "loss": 0.7912,
+      "step": 2153
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5113442303000734e-05,
+      "loss": 0.8158,
+      "step": 2154
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5111002683581362e-05,
+      "loss": 0.7329,
+      "step": 2155
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5108563064161992e-05,
+      "loss": 0.7763,
+      "step": 2156
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5106123444742622e-05,
+      "loss": 0.8634,
+      "step": 2157
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.510368382532325e-05,
+      "loss": 0.7596,
+      "step": 2158
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.510124420590388e-05,
+      "loss": 0.8489,
+      "step": 2159
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.509880458648451e-05,
+      "loss": 0.7739,
+      "step": 2160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5096364967065137e-05,
+      "loss": 0.8217,
+      "step": 2161
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5093925347645769e-05,
+      "loss": 0.7945,
+      "step": 2162
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5091485728226399e-05,
+      "loss": 0.7955,
+      "step": 2163
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5089046108807028e-05,
+      "loss": 0.7833,
+      "step": 2164
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5086606489387656e-05,
+      "loss": 0.7649,
+      "step": 2165
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5084166869968286e-05,
+      "loss": 0.8016,
+      "step": 2166
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5081727250548916e-05,
+      "loss": 0.746,
+      "step": 2167
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5079287631129544e-05,
+      "loss": 0.7749,
+      "step": 2168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5076848011710175e-05,
+      "loss": 0.8125,
+      "step": 2169
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5074408392290805e-05,
+      "loss": 0.835,
+      "step": 2170
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5071968772871433e-05,
+      "loss": 0.7235,
+      "step": 2171
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5069529153452063e-05,
+      "loss": 0.71,
+      "step": 2172
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5067089534032693e-05,
+      "loss": 0.747,
+      "step": 2173
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.506464991461332e-05,
+      "loss": 0.741,
+      "step": 2174
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.506221029519395e-05,
+      "loss": 0.8093,
+      "step": 2175
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5059770675774582e-05,
+      "loss": 0.8463,
+      "step": 2176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5057331056355208e-05,
+      "loss": 0.7629,
+      "step": 2177
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.505489143693584e-05,
+      "loss": 0.872,
+      "step": 2178
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.505245181751647e-05,
+      "loss": 0.7581,
+      "step": 2179
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5050012198097097e-05,
+      "loss": 0.7907,
+      "step": 2180
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5047572578677727e-05,
+      "loss": 0.7964,
+      "step": 2181
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5045132959258357e-05,
+      "loss": 0.8131,
+      "step": 2182
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5042693339838987e-05,
+      "loss": 0.8623,
+      "step": 2183
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5040253720419615e-05,
+      "loss": 0.756,
+      "step": 2184
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5037814101000246e-05,
+      "loss": 0.7574,
+      "step": 2185
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5035374481580876e-05,
+      "loss": 0.7919,
+      "step": 2186
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5032934862161504e-05,
+      "loss": 0.7781,
+      "step": 2187
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5030495242742134e-05,
+      "loss": 0.7704,
+      "step": 2188
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5028055623322763e-05,
+      "loss": 0.8159,
+      "step": 2189
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5025616003903391e-05,
+      "loss": 0.7711,
+      "step": 2190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5023176384484021e-05,
+      "loss": 0.7899,
+      "step": 2191
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5020736765064653e-05,
+      "loss": 0.7325,
+      "step": 2192
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5018297145645279e-05,
+      "loss": 0.6852,
+      "step": 2193
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.501585752622591e-05,
+      "loss": 0.7983,
+      "step": 2194
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.501341790680654e-05,
+      "loss": 0.7508,
+      "step": 2195
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5010978287387168e-05,
+      "loss": 0.8001,
+      "step": 2196
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5008538667967798e-05,
+      "loss": 0.7869,
+      "step": 2197
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5006099048548428e-05,
+      "loss": 0.7747,
+      "step": 2198
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5003659429129056e-05,
+      "loss": 0.7251,
+      "step": 2199
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5001219809709686e-05,
+      "loss": 0.7938,
+      "step": 2200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4998780190290317e-05,
+      "loss": 0.7307,
+      "step": 2201
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4996340570870947e-05,
+      "loss": 0.7841,
+      "step": 2202
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4993900951451575e-05,
+      "loss": 0.8038,
+      "step": 2203
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4991461332032204e-05,
+      "loss": 0.7602,
+      "step": 2204
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4989021712612834e-05,
+      "loss": 0.7738,
+      "step": 2205
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4986582093193462e-05,
+      "loss": 0.7801,
+      "step": 2206
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4984142473774092e-05,
+      "loss": 0.7678,
+      "step": 2207
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4981702854354723e-05,
+      "loss": 0.8312,
+      "step": 2208
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.497926323493535e-05,
+      "loss": 0.7604,
+      "step": 2209
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4976823615515981e-05,
+      "loss": 0.7457,
+      "step": 2210
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4974383996096611e-05,
+      "loss": 0.7382,
+      "step": 2211
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4971944376677239e-05,
+      "loss": 0.8208,
+      "step": 2212
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4969504757257869e-05,
+      "loss": 0.8146,
+      "step": 2213
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4967065137838499e-05,
+      "loss": 0.8211,
+      "step": 2214
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4964625518419127e-05,
+      "loss": 0.8117,
+      "step": 2215
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4962185898999756e-05,
+      "loss": 0.8559,
+      "step": 2216
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4959746279580388e-05,
+      "loss": 0.8021,
+      "step": 2217
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4957306660161016e-05,
+      "loss": 0.7409,
+      "step": 2218
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4954867040741646e-05,
+      "loss": 0.8099,
+      "step": 2219
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4952427421322275e-05,
+      "loss": 0.8011,
+      "step": 2220
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4949987801902905e-05,
+      "loss": 0.7515,
+      "step": 2221
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4947548182483533e-05,
+      "loss": 0.812,
+      "step": 2222
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4945108563064163e-05,
+      "loss": 0.7743,
+      "step": 2223
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4942668943644794e-05,
+      "loss": 0.7954,
+      "step": 2224
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.494022932422542e-05,
+      "loss": 0.7965,
+      "step": 2225
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4937789704806052e-05,
+      "loss": 0.78,
+      "step": 2226
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4935350085386682e-05,
+      "loss": 0.8972,
+      "step": 2227
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.493291046596731e-05,
+      "loss": 0.7716,
+      "step": 2228
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.493047084654794e-05,
+      "loss": 0.7781,
+      "step": 2229
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.492803122712857e-05,
+      "loss": 0.8223,
+      "step": 2230
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4925591607709197e-05,
+      "loss": 0.7642,
+      "step": 2231
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4923151988289827e-05,
+      "loss": 0.7927,
+      "step": 2232
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4920712368870459e-05,
+      "loss": 0.8057,
+      "step": 2233
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4918272749451087e-05,
+      "loss": 0.7372,
+      "step": 2234
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4915833130031716e-05,
+      "loss": 0.7804,
+      "step": 2235
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4913393510612346e-05,
+      "loss": 0.7894,
+      "step": 2236
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4910953891192974e-05,
+      "loss": 0.8058,
+      "step": 2237
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4908514271773604e-05,
+      "loss": 0.7901,
+      "step": 2238
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4906074652354234e-05,
+      "loss": 0.788,
+      "step": 2239
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4903635032934865e-05,
+      "loss": 0.7778,
+      "step": 2240
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4901195413515493e-05,
+      "loss": 0.8027,
+      "step": 2241
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4898755794096123e-05,
+      "loss": 0.796,
+      "step": 2242
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4896316174676753e-05,
+      "loss": 0.7928,
+      "step": 2243
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.489387655525738e-05,
+      "loss": 0.8006,
+      "step": 2244
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.489143693583801e-05,
+      "loss": 0.7476,
+      "step": 2245
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.488899731641864e-05,
+      "loss": 0.8182,
+      "step": 2246
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4886557696999268e-05,
+      "loss": 0.8303,
+      "step": 2247
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4884118077579898e-05,
+      "loss": 0.8137,
+      "step": 2248
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.488167845816053e-05,
+      "loss": 0.7892,
+      "step": 2249
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4879238838741157e-05,
+      "loss": 0.77,
+      "step": 2250
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4876799219321787e-05,
+      "loss": 0.7782,
+      "step": 2251
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4874359599902417e-05,
+      "loss": 0.8284,
+      "step": 2252
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4871919980483045e-05,
+      "loss": 0.7948,
+      "step": 2253
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4869480361063675e-05,
+      "loss": 0.7556,
+      "step": 2254
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4867040741644305e-05,
+      "loss": 0.7845,
+      "step": 2255
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4864601122224933e-05,
+      "loss": 0.7723,
+      "step": 2256
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4862161502805564e-05,
+      "loss": 0.7967,
+      "step": 2257
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4859721883386194e-05,
+      "loss": 0.7965,
+      "step": 2258
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4857282263966823e-05,
+      "loss": 0.7922,
+      "step": 2259
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4854842644547452e-05,
+      "loss": 0.7912,
+      "step": 2260
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4852403025128081e-05,
+      "loss": 0.7603,
+      "step": 2261
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4849963405708711e-05,
+      "loss": 0.8525,
+      "step": 2262
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4847523786289339e-05,
+      "loss": 0.7904,
+      "step": 2263
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.484508416686997e-05,
+      "loss": 0.7339,
+      "step": 2264
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.48426445474506e-05,
+      "loss": 0.8016,
+      "step": 2265
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4840204928031228e-05,
+      "loss": 0.7327,
+      "step": 2266
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4837765308611858e-05,
+      "loss": 0.8033,
+      "step": 2267
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4835325689192488e-05,
+      "loss": 0.8951,
+      "step": 2268
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4832886069773116e-05,
+      "loss": 0.8513,
+      "step": 2269
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4830446450353746e-05,
+      "loss": 0.7799,
+      "step": 2270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4828006830934375e-05,
+      "loss": 0.7953,
+      "step": 2271
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4825567211515003e-05,
+      "loss": 0.7732,
+      "step": 2272
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4823127592095635e-05,
+      "loss": 0.7691,
+      "step": 2273
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4820687972676265e-05,
+      "loss": 0.8202,
+      "step": 2274
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4818248353256893e-05,
+      "loss": 0.7976,
+      "step": 2275
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4815808733837522e-05,
+      "loss": 0.7721,
+      "step": 2276
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4813369114418152e-05,
+      "loss": 0.8014,
+      "step": 2277
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.481092949499878e-05,
+      "loss": 0.784,
+      "step": 2278
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.480848987557941e-05,
+      "loss": 0.7672,
+      "step": 2279
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4806050256160041e-05,
+      "loss": 0.7595,
+      "step": 2280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4803610636740671e-05,
+      "loss": 0.8616,
+      "step": 2281
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4801171017321299e-05,
+      "loss": 0.8347,
+      "step": 2282
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4798731397901929e-05,
+      "loss": 0.8108,
+      "step": 2283
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4796291778482559e-05,
+      "loss": 0.7874,
+      "step": 2284
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4793852159063187e-05,
+      "loss": 0.8433,
+      "step": 2285
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4791412539643816e-05,
+      "loss": 0.7259,
+      "step": 2286
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4788972920224446e-05,
+      "loss": 0.7232,
+      "step": 2287
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4786533300805074e-05,
+      "loss": 0.8164,
+      "step": 2288
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4784093681385706e-05,
+      "loss": 0.7599,
+      "step": 2289
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4781654061966335e-05,
+      "loss": 0.8173,
+      "step": 2290
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4779214442546963e-05,
+      "loss": 0.7555,
+      "step": 2291
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4776774823127593e-05,
+      "loss": 0.7998,
+      "step": 2292
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4774335203708223e-05,
+      "loss": 0.7727,
+      "step": 2293
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4771895584288851e-05,
+      "loss": 0.7744,
+      "step": 2294
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.476945596486948e-05,
+      "loss": 0.7687,
+      "step": 2295
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4767016345450112e-05,
+      "loss": 0.8626,
+      "step": 2296
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4764576726030739e-05,
+      "loss": 0.7721,
+      "step": 2297
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.476213710661137e-05,
+      "loss": 0.7683,
+      "step": 2298
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4759697487192e-05,
+      "loss": 0.8651,
+      "step": 2299
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.475725786777263e-05,
+      "loss": 0.7584,
+      "step": 2300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4754818248353258e-05,
+      "loss": 0.8241,
+      "step": 2301
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4752378628933887e-05,
+      "loss": 0.7775,
+      "step": 2302
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4749939009514519e-05,
+      "loss": 0.8136,
+      "step": 2303
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4747499390095145e-05,
+      "loss": 0.8137,
+      "step": 2304
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4745059770675776e-05,
+      "loss": 0.7679,
+      "step": 2305
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4742620151256406e-05,
+      "loss": 0.8502,
+      "step": 2306
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4740180531837034e-05,
+      "loss": 0.802,
+      "step": 2307
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4737740912417664e-05,
+      "loss": 0.7953,
+      "step": 2308
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4735301292998294e-05,
+      "loss": 0.8007,
+      "step": 2309
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4732861673578922e-05,
+      "loss": 0.7901,
+      "step": 2310
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4730422054159552e-05,
+      "loss": 0.8323,
+      "step": 2311
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4727982434740183e-05,
+      "loss": 0.7738,
+      "step": 2312
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4725542815320811e-05,
+      "loss": 0.7902,
+      "step": 2313
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.472310319590144e-05,
+      "loss": 0.8083,
+      "step": 2314
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.472066357648207e-05,
+      "loss": 0.7701,
+      "step": 2315
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4718223957062699e-05,
+      "loss": 0.7525,
+      "step": 2316
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4715784337643328e-05,
+      "loss": 0.8772,
+      "step": 2317
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4713344718223958e-05,
+      "loss": 0.7415,
+      "step": 2318
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.471090509880459e-05,
+      "loss": 0.8118,
+      "step": 2319
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4708465479385216e-05,
+      "loss": 0.8329,
+      "step": 2320
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4706025859965847e-05,
+      "loss": 0.7167,
+      "step": 2321
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4703586240546477e-05,
+      "loss": 0.7808,
+      "step": 2322
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4701146621127105e-05,
+      "loss": 0.7301,
+      "step": 2323
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4698707001707735e-05,
+      "loss": 0.7784,
+      "step": 2324
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4696267382288365e-05,
+      "loss": 0.755,
+      "step": 2325
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4693827762868993e-05,
+      "loss": 0.7657,
+      "step": 2326
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4691388143449622e-05,
+      "loss": 0.7829,
+      "step": 2327
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4688948524030254e-05,
+      "loss": 0.7637,
+      "step": 2328
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4686508904610882e-05,
+      "loss": 0.8505,
+      "step": 2329
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4684069285191512e-05,
+      "loss": 0.7666,
+      "step": 2330
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4681629665772141e-05,
+      "loss": 0.7978,
+      "step": 2331
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.467919004635277e-05,
+      "loss": 0.791,
+      "step": 2332
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.46767504269334e-05,
+      "loss": 0.836,
+      "step": 2333
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4674310807514029e-05,
+      "loss": 0.7737,
+      "step": 2334
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4671871188094657e-05,
+      "loss": 0.83,
+      "step": 2335
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4669431568675287e-05,
+      "loss": 0.7124,
+      "step": 2336
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4666991949255918e-05,
+      "loss": 0.7338,
+      "step": 2337
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4664552329836548e-05,
+      "loss": 0.8423,
+      "step": 2338
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4662112710417176e-05,
+      "loss": 0.789,
+      "step": 2339
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4659673090997806e-05,
+      "loss": 0.7891,
+      "step": 2340
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4657233471578435e-05,
+      "loss": 0.7617,
+      "step": 2341
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4654793852159063e-05,
+      "loss": 0.846,
+      "step": 2342
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4652354232739693e-05,
+      "loss": 0.7905,
+      "step": 2343
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4649914613320325e-05,
+      "loss": 0.7988,
+      "step": 2344
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4647474993900953e-05,
+      "loss": 0.7812,
+      "step": 2345
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4645035374481582e-05,
+      "loss": 0.7596,
+      "step": 2346
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4642595755062212e-05,
+      "loss": 0.7799,
+      "step": 2347
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.464015613564284e-05,
+      "loss": 0.759,
+      "step": 2348
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463771651622347e-05,
+      "loss": 0.8343,
+      "step": 2349
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.46352768968041e-05,
+      "loss": 0.7537,
+      "step": 2350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4632837277384728e-05,
+      "loss": 0.7769,
+      "step": 2351
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.463039765796536e-05,
+      "loss": 0.8175,
+      "step": 2352
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4627958038545989e-05,
+      "loss": 0.876,
+      "step": 2353
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4625518419126617e-05,
+      "loss": 0.827,
+      "step": 2354
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4623078799707247e-05,
+      "loss": 0.7784,
+      "step": 2355
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4620639180287877e-05,
+      "loss": 0.8638,
+      "step": 2356
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4618199560868506e-05,
+      "loss": 0.7988,
+      "step": 2357
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4615759941449134e-05,
+      "loss": 0.7442,
+      "step": 2358
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4613320322029764e-05,
+      "loss": 0.8352,
+      "step": 2359
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4610880702610395e-05,
+      "loss": 0.8444,
+      "step": 2360
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4608441083191024e-05,
+      "loss": 0.8172,
+      "step": 2361
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4606001463771653e-05,
+      "loss": 0.8077,
+      "step": 2362
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4603561844352283e-05,
+      "loss": 0.7648,
+      "step": 2363
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4601122224932911e-05,
+      "loss": 0.8203,
+      "step": 2364
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.459868260551354e-05,
+      "loss": 0.7874,
+      "step": 2365
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.459624298609417e-05,
+      "loss": 0.7542,
+      "step": 2366
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4593803366674799e-05,
+      "loss": 0.7886,
+      "step": 2367
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.459136374725543e-05,
+      "loss": 0.7881,
+      "step": 2368
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.458892412783606e-05,
+      "loss": 0.7515,
+      "step": 2369
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4586484508416688e-05,
+      "loss": 0.7822,
+      "step": 2370
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4584044888997318e-05,
+      "loss": 0.8073,
+      "step": 2371
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4581605269577947e-05,
+      "loss": 0.8163,
+      "step": 2372
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4579165650158575e-05,
+      "loss": 0.7687,
+      "step": 2373
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4576726030739205e-05,
+      "loss": 0.7735,
+      "step": 2374
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4574286411319835e-05,
+      "loss": 0.7605,
+      "step": 2375
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4571846791900466e-05,
+      "loss": 0.8205,
+      "step": 2376
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4569407172481094e-05,
+      "loss": 0.6821,
+      "step": 2377
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4566967553061724e-05,
+      "loss": 0.7567,
+      "step": 2378
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4564527933642354e-05,
+      "loss": 0.7979,
+      "step": 2379
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4562088314222982e-05,
+      "loss": 0.7783,
+      "step": 2380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4559648694803612e-05,
+      "loss": 0.7652,
+      "step": 2381
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4557209075384241e-05,
+      "loss": 0.8032,
+      "step": 2382
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.455476945596487e-05,
+      "loss": 0.8021,
+      "step": 2383
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4552329836545501e-05,
+      "loss": 0.7462,
+      "step": 2384
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.454989021712613e-05,
+      "loss": 0.7973,
+      "step": 2385
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4547450597706759e-05,
+      "loss": 0.7686,
+      "step": 2386
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4545010978287388e-05,
+      "loss": 0.7734,
+      "step": 2387
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4542571358868018e-05,
+      "loss": 0.7135,
+      "step": 2388
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4540131739448646e-05,
+      "loss": 0.8268,
+      "step": 2389
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4537692120029276e-05,
+      "loss": 0.83,
+      "step": 2390
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4535252500609907e-05,
+      "loss": 0.7968,
+      "step": 2391
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4532812881190534e-05,
+      "loss": 0.7426,
+      "step": 2392
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4530373261771165e-05,
+      "loss": 0.7698,
+      "step": 2393
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4527933642351795e-05,
+      "loss": 0.7482,
+      "step": 2394
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4525494022932425e-05,
+      "loss": 0.7666,
+      "step": 2395
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4523054403513053e-05,
+      "loss": 0.824,
+      "step": 2396
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4520614784093682e-05,
+      "loss": 0.7491,
+      "step": 2397
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4518175164674312e-05,
+      "loss": 0.8978,
+      "step": 2398
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.451573554525494e-05,
+      "loss": 0.739,
+      "step": 2399
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4513295925835572e-05,
+      "loss": 0.8194,
+      "step": 2400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4510856306416201e-05,
+      "loss": 0.7868,
+      "step": 2401
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.450841668699683e-05,
+      "loss": 0.809,
+      "step": 2402
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.450597706757746e-05,
+      "loss": 0.768,
+      "step": 2403
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4503537448158089e-05,
+      "loss": 0.8458,
+      "step": 2404
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4501097828738717e-05,
+      "loss": 0.7817,
+      "step": 2405
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4498658209319347e-05,
+      "loss": 0.7664,
+      "step": 2406
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4496218589899978e-05,
+      "loss": 0.7476,
+      "step": 2407
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4493778970480605e-05,
+      "loss": 0.758,
+      "step": 2408
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4491339351061236e-05,
+      "loss": 0.7449,
+      "step": 2409
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4488899731641866e-05,
+      "loss": 0.71,
+      "step": 2410
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4486460112222494e-05,
+      "loss": 0.7711,
+      "step": 2411
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4484020492803124e-05,
+      "loss": 0.7698,
+      "step": 2412
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4481580873383753e-05,
+      "loss": 0.8235,
+      "step": 2413
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4479141253964381e-05,
+      "loss": 0.7851,
+      "step": 2414
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4476701634545011e-05,
+      "loss": 0.8586,
+      "step": 2415
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4474262015125643e-05,
+      "loss": 0.7913,
+      "step": 2416
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4471822395706272e-05,
+      "loss": 0.7604,
+      "step": 2417
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.44693827762869e-05,
+      "loss": 0.7552,
+      "step": 2418
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.446694315686753e-05,
+      "loss": 0.7761,
+      "step": 2419
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.446450353744816e-05,
+      "loss": 0.8058,
+      "step": 2420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4462063918028788e-05,
+      "loss": 0.8162,
+      "step": 2421
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4459624298609418e-05,
+      "loss": 0.7974,
+      "step": 2422
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4457184679190049e-05,
+      "loss": 0.8338,
+      "step": 2423
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4454745059770675e-05,
+      "loss": 0.7639,
+      "step": 2424
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4452305440351307e-05,
+      "loss": 0.7855,
+      "step": 2425
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4449865820931937e-05,
+      "loss": 0.7177,
+      "step": 2426
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4447426201512565e-05,
+      "loss": 0.7558,
+      "step": 2427
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4444986582093194e-05,
+      "loss": 0.8499,
+      "step": 2428
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4442546962673824e-05,
+      "loss": 0.7558,
+      "step": 2429
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4440107343254452e-05,
+      "loss": 0.7461,
+      "step": 2430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4437667723835082e-05,
+      "loss": 0.8099,
+      "step": 2431
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4435228104415713e-05,
+      "loss": 0.7288,
+      "step": 2432
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4432788484996341e-05,
+      "loss": 0.7861,
+      "step": 2433
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4430348865576971e-05,
+      "loss": 0.8123,
+      "step": 2434
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4427909246157601e-05,
+      "loss": 0.8465,
+      "step": 2435
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.442546962673823e-05,
+      "loss": 0.7658,
+      "step": 2436
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4423030007318859e-05,
+      "loss": 0.7899,
+      "step": 2437
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4420590387899488e-05,
+      "loss": 0.8013,
+      "step": 2438
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.441815076848012e-05,
+      "loss": 0.8873,
+      "step": 2439
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4415711149060748e-05,
+      "loss": 0.8218,
+      "step": 2440
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4413271529641378e-05,
+      "loss": 0.8177,
+      "step": 2441
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4410831910222007e-05,
+      "loss": 0.8173,
+      "step": 2442
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4408392290802635e-05,
+      "loss": 0.8127,
+      "step": 2443
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4405952671383265e-05,
+      "loss": 0.8282,
+      "step": 2444
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4403513051963895e-05,
+      "loss": 0.8567,
+      "step": 2445
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4401073432544523e-05,
+      "loss": 0.732,
+      "step": 2446
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4398633813125153e-05,
+      "loss": 0.8033,
+      "step": 2447
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4396194193705784e-05,
+      "loss": 0.8098,
+      "step": 2448
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4393754574286412e-05,
+      "loss": 0.8569,
+      "step": 2449
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4391314954867042e-05,
+      "loss": 0.8134,
+      "step": 2450
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4388875335447672e-05,
+      "loss": 0.7554,
+      "step": 2451
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.43864357160283e-05,
+      "loss": 0.8456,
+      "step": 2452
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.438399609660893e-05,
+      "loss": 0.7931,
+      "step": 2453
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.438155647718956e-05,
+      "loss": 0.7379,
+      "step": 2454
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.437911685777019e-05,
+      "loss": 0.8206,
+      "step": 2455
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4376677238350819e-05,
+      "loss": 0.784,
+      "step": 2456
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4374237618931449e-05,
+      "loss": 0.8139,
+      "step": 2457
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4371797999512078e-05,
+      "loss": 0.7869,
+      "step": 2458
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4369358380092706e-05,
+      "loss": 0.7589,
+      "step": 2459
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4366918760673336e-05,
+      "loss": 0.729,
+      "step": 2460
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4364479141253966e-05,
+      "loss": 0.7649,
+      "step": 2461
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4362039521834594e-05,
+      "loss": 0.8495,
+      "step": 2462
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4359599902415224e-05,
+      "loss": 0.8124,
+      "step": 2463
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4357160282995855e-05,
+      "loss": 0.8107,
+      "step": 2464
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4354720663576483e-05,
+      "loss": 0.8124,
+      "step": 2465
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4352281044157113e-05,
+      "loss": 0.7923,
+      "step": 2466
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4349841424737743e-05,
+      "loss": 0.8089,
+      "step": 2467
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.434740180531837e-05,
+      "loss": 0.7972,
+      "step": 2468
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4344962185899e-05,
+      "loss": 0.7889,
+      "step": 2469
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.434252256647963e-05,
+      "loss": 0.7038,
+      "step": 2470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4340082947060258e-05,
+      "loss": 0.7746,
+      "step": 2471
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.433764332764089e-05,
+      "loss": 0.7887,
+      "step": 2472
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.433520370822152e-05,
+      "loss": 0.7148,
+      "step": 2473
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4332764088802149e-05,
+      "loss": 0.7703,
+      "step": 2474
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4330324469382777e-05,
+      "loss": 0.7904,
+      "step": 2475
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4327884849963407e-05,
+      "loss": 0.7673,
+      "step": 2476
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4325445230544037e-05,
+      "loss": 0.7746,
+      "step": 2477
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4323005611124665e-05,
+      "loss": 0.7865,
+      "step": 2478
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4320565991705296e-05,
+      "loss": 0.798,
+      "step": 2479
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4318126372285926e-05,
+      "loss": 0.7744,
+      "step": 2480
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4315686752866554e-05,
+      "loss": 0.8108,
+      "step": 2481
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4313247133447184e-05,
+      "loss": 0.7601,
+      "step": 2482
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4310807514027813e-05,
+      "loss": 0.7837,
+      "step": 2483
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4308367894608441e-05,
+      "loss": 0.8227,
+      "step": 2484
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4305928275189071e-05,
+      "loss": 0.7915,
+      "step": 2485
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4303488655769701e-05,
+      "loss": 0.725,
+      "step": 2486
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4301049036350329e-05,
+      "loss": 0.8344,
+      "step": 2487
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.429860941693096e-05,
+      "loss": 0.7788,
+      "step": 2488
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.429616979751159e-05,
+      "loss": 0.83,
+      "step": 2489
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4293730178092218e-05,
+      "loss": 0.7953,
+      "step": 2490
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4291290558672848e-05,
+      "loss": 0.7586,
+      "step": 2491
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4288850939253478e-05,
+      "loss": 0.7805,
+      "step": 2492
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4286411319834107e-05,
+      "loss": 0.8067,
+      "step": 2493
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4283971700414735e-05,
+      "loss": 0.8168,
+      "step": 2494
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4281532080995367e-05,
+      "loss": 0.814,
+      "step": 2495
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4279092461575997e-05,
+      "loss": 0.8326,
+      "step": 2496
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4276652842156625e-05,
+      "loss": 0.7632,
+      "step": 2497
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4274213222737254e-05,
+      "loss": 0.7104,
+      "step": 2498
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4271773603317884e-05,
+      "loss": 0.7799,
+      "step": 2499
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4269333983898512e-05,
+      "loss": 0.7906,
+      "step": 2500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4266894364479142e-05,
+      "loss": 0.8087,
+      "step": 2501
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4264454745059772e-05,
+      "loss": 0.7894,
+      "step": 2502
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.42620151256404e-05,
+      "loss": 0.8021,
+      "step": 2503
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4259575506221031e-05,
+      "loss": 0.7851,
+      "step": 2504
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4257135886801661e-05,
+      "loss": 0.7733,
+      "step": 2505
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4254696267382289e-05,
+      "loss": 0.8012,
+      "step": 2506
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4252256647962919e-05,
+      "loss": 0.8505,
+      "step": 2507
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4249817028543549e-05,
+      "loss": 0.7375,
+      "step": 2508
+    },
+    {
+      "epoch": 1.18,
+      "eval_loss": 0.9797788858413696,
+      "eval_runtime": 204.2919,
+      "eval_samples_per_second": 82.23,
+      "eval_steps_per_second": 10.279,
+      "step": 2508
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4247377409124177e-05,
+      "loss": 0.7439,
+      "step": 2509
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4244937789704806e-05,
+      "loss": 0.8199,
+      "step": 2510
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4242498170285438e-05,
+      "loss": 0.7665,
+      "step": 2511
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4240058550866068e-05,
+      "loss": 0.7305,
+      "step": 2512
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4237618931446696e-05,
+      "loss": 0.7826,
+      "step": 2513
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4235179312027325e-05,
+      "loss": 0.7434,
+      "step": 2514
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4232739692607955e-05,
+      "loss": 0.789,
+      "step": 2515
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4230300073188583e-05,
+      "loss": 0.8151,
+      "step": 2516
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4227860453769213e-05,
+      "loss": 0.8112,
+      "step": 2517
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4225420834349844e-05,
+      "loss": 0.7915,
+      "step": 2518
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.422298121493047e-05,
+      "loss": 0.7625,
+      "step": 2519
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4220541595511102e-05,
+      "loss": 0.6523,
+      "step": 2520
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4218101976091732e-05,
+      "loss": 0.7416,
+      "step": 2521
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.421566235667236e-05,
+      "loss": 0.7952,
+      "step": 2522
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.421322273725299e-05,
+      "loss": 0.8139,
+      "step": 2523
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.421078311783362e-05,
+      "loss": 0.8007,
+      "step": 2524
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4208343498414247e-05,
+      "loss": 0.7393,
+      "step": 2525
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4205903878994877e-05,
+      "loss": 0.9483,
+      "step": 2526
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4203464259575509e-05,
+      "loss": 0.8035,
+      "step": 2527
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4201024640156137e-05,
+      "loss": 0.7722,
+      "step": 2528
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4198585020736766e-05,
+      "loss": 0.8397,
+      "step": 2529
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4196145401317396e-05,
+      "loss": 0.9576,
+      "step": 2530
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4193705781898026e-05,
+      "loss": 0.8613,
+      "step": 2531
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4191266162478654e-05,
+      "loss": 0.9023,
+      "step": 2532
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4188826543059284e-05,
+      "loss": 0.8136,
+      "step": 2533
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4186386923639915e-05,
+      "loss": 0.8522,
+      "step": 2534
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4183947304220541e-05,
+      "loss": 0.8357,
+      "step": 2535
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4181507684801173e-05,
+      "loss": 0.7874,
+      "step": 2536
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4179068065381803e-05,
+      "loss": 0.8839,
+      "step": 2537
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.417662844596243e-05,
+      "loss": 0.8438,
+      "step": 2538
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.417418882654306e-05,
+      "loss": 0.8515,
+      "step": 2539
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.417174920712369e-05,
+      "loss": 0.8131,
+      "step": 2540
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4169309587704318e-05,
+      "loss": 0.8289,
+      "step": 2541
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4166869968284948e-05,
+      "loss": 0.8308,
+      "step": 2542
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.416443034886558e-05,
+      "loss": 0.7707,
+      "step": 2543
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4161990729446207e-05,
+      "loss": 0.7763,
+      "step": 2544
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4159551110026837e-05,
+      "loss": 0.8123,
+      "step": 2545
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4157111490607467e-05,
+      "loss": 0.8378,
+      "step": 2546
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4154671871188095e-05,
+      "loss": 0.8388,
+      "step": 2547
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4152232251768725e-05,
+      "loss": 0.809,
+      "step": 2548
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4149792632349354e-05,
+      "loss": 0.8174,
+      "step": 2549
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4147353012929983e-05,
+      "loss": 0.827,
+      "step": 2550
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4144913393510612e-05,
+      "loss": 0.8552,
+      "step": 2551
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4142473774091244e-05,
+      "loss": 0.7965,
+      "step": 2552
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4140034154671873e-05,
+      "loss": 0.8067,
+      "step": 2553
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4137594535252502e-05,
+      "loss": 0.779,
+      "step": 2554
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4135154915833131e-05,
+      "loss": 0.8005,
+      "step": 2555
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4132715296413761e-05,
+      "loss": 0.765,
+      "step": 2556
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4130275676994389e-05,
+      "loss": 0.7255,
+      "step": 2557
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4127836057575019e-05,
+      "loss": 0.8506,
+      "step": 2558
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.412539643815565e-05,
+      "loss": 0.7712,
+      "step": 2559
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4122956818736278e-05,
+      "loss": 0.7692,
+      "step": 2560
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4120517199316908e-05,
+      "loss": 0.8087,
+      "step": 2561
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4118077579897538e-05,
+      "loss": 0.8407,
+      "step": 2562
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4115637960478166e-05,
+      "loss": 0.7795,
+      "step": 2563
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4113198341058796e-05,
+      "loss": 0.7444,
+      "step": 2564
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4110758721639425e-05,
+      "loss": 0.7839,
+      "step": 2565
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4108319102220053e-05,
+      "loss": 0.7653,
+      "step": 2566
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4105879482800685e-05,
+      "loss": 0.7394,
+      "step": 2567
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4103439863381315e-05,
+      "loss": 0.7602,
+      "step": 2568
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4101000243961943e-05,
+      "loss": 0.7241,
+      "step": 2569
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4098560624542572e-05,
+      "loss": 0.7744,
+      "step": 2570
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4096121005123202e-05,
+      "loss": 0.8259,
+      "step": 2571
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4093681385703832e-05,
+      "loss": 0.7902,
+      "step": 2572
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.409124176628446e-05,
+      "loss": 0.7668,
+      "step": 2573
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.408880214686509e-05,
+      "loss": 0.8228,
+      "step": 2574
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4086362527445721e-05,
+      "loss": 0.8142,
+      "step": 2575
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4083922908026349e-05,
+      "loss": 0.8048,
+      "step": 2576
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4081483288606979e-05,
+      "loss": 0.8711,
+      "step": 2577
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4079043669187609e-05,
+      "loss": 0.7743,
+      "step": 2578
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4076604049768237e-05,
+      "loss": 0.7737,
+      "step": 2579
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4074164430348866e-05,
+      "loss": 0.7723,
+      "step": 2580
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4071724810929496e-05,
+      "loss": 0.7794,
+      "step": 2581
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4069285191510124e-05,
+      "loss": 0.9116,
+      "step": 2582
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4066845572090756e-05,
+      "loss": 0.7625,
+      "step": 2583
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4064405952671385e-05,
+      "loss": 0.79,
+      "step": 2584
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4061966333252013e-05,
+      "loss": 0.7291,
+      "step": 2585
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4059526713832643e-05,
+      "loss": 0.804,
+      "step": 2586
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4057087094413273e-05,
+      "loss": 0.8107,
+      "step": 2587
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4054647474993901e-05,
+      "loss": 0.8389,
+      "step": 2588
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.405220785557453e-05,
+      "loss": 0.8002,
+      "step": 2589
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.404976823615516e-05,
+      "loss": 0.8336,
+      "step": 2590
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4047328616735792e-05,
+      "loss": 0.784,
+      "step": 2591
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.404488899731642e-05,
+      "loss": 0.8363,
+      "step": 2592
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.404244937789705e-05,
+      "loss": 0.8019,
+      "step": 2593
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.404000975847768e-05,
+      "loss": 0.7532,
+      "step": 2594
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4037570139058307e-05,
+      "loss": 0.8227,
+      "step": 2595
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4035130519638937e-05,
+      "loss": 0.7551,
+      "step": 2596
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4032690900219567e-05,
+      "loss": 0.7737,
+      "step": 2597
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4030251280800195e-05,
+      "loss": 0.7879,
+      "step": 2598
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4027811661380826e-05,
+      "loss": 0.7829,
+      "step": 2599
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4025372041961456e-05,
+      "loss": 0.8452,
+      "step": 2600
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.4022932422542084e-05,
+      "loss": 0.8223,
+      "step": 2601
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.4020492803122714e-05,
+      "loss": 0.9005,
+      "step": 2602
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.4018053183703344e-05,
+      "loss": 0.7915,
+      "step": 2603
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.4015613564283972e-05,
+      "loss": 0.8099,
+      "step": 2604
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.4013173944864602e-05,
+      "loss": 0.8536,
+      "step": 2605
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.4010734325445233e-05,
+      "loss": 0.7644,
+      "step": 2606
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.400829470602586e-05,
+      "loss": 0.7865,
+      "step": 2607
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.400585508660649e-05,
+      "loss": 0.8611,
+      "step": 2608
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.400341546718712e-05,
+      "loss": 0.848,
+      "step": 2609
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.400097584776775e-05,
+      "loss": 0.7521,
+      "step": 2610
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3998536228348378e-05,
+      "loss": 0.8358,
+      "step": 2611
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3996096608929008e-05,
+      "loss": 0.7832,
+      "step": 2612
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3993656989509638e-05,
+      "loss": 0.8216,
+      "step": 2613
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3991217370090266e-05,
+      "loss": 0.7885,
+      "step": 2614
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3988777750670897e-05,
+      "loss": 0.8216,
+      "step": 2615
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3986338131251527e-05,
+      "loss": 0.834,
+      "step": 2616
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3983898511832155e-05,
+      "loss": 0.8499,
+      "step": 2617
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3981458892412785e-05,
+      "loss": 0.7611,
+      "step": 2618
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3979019272993415e-05,
+      "loss": 0.8934,
+      "step": 2619
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3976579653574043e-05,
+      "loss": 0.8001,
+      "step": 2620
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3974140034154672e-05,
+      "loss": 0.8039,
+      "step": 2621
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3971700414735304e-05,
+      "loss": 0.8268,
+      "step": 2622
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.396926079531593e-05,
+      "loss": 0.7975,
+      "step": 2623
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3966821175896562e-05,
+      "loss": 0.8424,
+      "step": 2624
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3964381556477191e-05,
+      "loss": 0.7389,
+      "step": 2625
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.396194193705782e-05,
+      "loss": 0.741,
+      "step": 2626
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3959502317638449e-05,
+      "loss": 0.7226,
+      "step": 2627
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3957062698219079e-05,
+      "loss": 0.7665,
+      "step": 2628
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3954623078799709e-05,
+      "loss": 0.7212,
+      "step": 2629
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3952183459380337e-05,
+      "loss": 0.7461,
+      "step": 2630
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3949743839960968e-05,
+      "loss": 0.8513,
+      "step": 2631
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3947304220541598e-05,
+      "loss": 0.8199,
+      "step": 2632
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3944864601122226e-05,
+      "loss": 0.7514,
+      "step": 2633
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3942424981702856e-05,
+      "loss": 0.7991,
+      "step": 2634
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3939985362283485e-05,
+      "loss": 0.7977,
+      "step": 2635
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3937545742864113e-05,
+      "loss": 0.8693,
+      "step": 2636
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3935106123444743e-05,
+      "loss": 0.7985,
+      "step": 2637
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3932666504025375e-05,
+      "loss": 0.7725,
+      "step": 2638
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3930226884606001e-05,
+      "loss": 0.7867,
+      "step": 2639
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3927787265186632e-05,
+      "loss": 0.8008,
+      "step": 2640
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3925347645767262e-05,
+      "loss": 0.734,
+      "step": 2641
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.392290802634789e-05,
+      "loss": 0.7888,
+      "step": 2642
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.392046840692852e-05,
+      "loss": 0.7796,
+      "step": 2643
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.391802878750915e-05,
+      "loss": 0.7828,
+      "step": 2644
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3915589168089778e-05,
+      "loss": 0.8312,
+      "step": 2645
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3913149548670408e-05,
+      "loss": 0.7717,
+      "step": 2646
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3910709929251039e-05,
+      "loss": 0.803,
+      "step": 2647
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3908270309831669e-05,
+      "loss": 0.8286,
+      "step": 2648
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3905830690412297e-05,
+      "loss": 0.8447,
+      "step": 2649
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3903391070992926e-05,
+      "loss": 0.8163,
+      "step": 2650
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3900951451573556e-05,
+      "loss": 0.7647,
+      "step": 2651
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3898511832154184e-05,
+      "loss": 0.7701,
+      "step": 2652
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3896072212734814e-05,
+      "loss": 0.7852,
+      "step": 2653
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3893632593315445e-05,
+      "loss": 0.7366,
+      "step": 2654
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3891192973896074e-05,
+      "loss": 0.7924,
+      "step": 2655
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3888753354476703e-05,
+      "loss": 0.7389,
+      "step": 2656
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3886313735057333e-05,
+      "loss": 0.7407,
+      "step": 2657
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3883874115637961e-05,
+      "loss": 0.8298,
+      "step": 2658
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.388143449621859e-05,
+      "loss": 0.838,
+      "step": 2659
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.387899487679922e-05,
+      "loss": 0.7837,
+      "step": 2660
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3876555257379849e-05,
+      "loss": 0.7698,
+      "step": 2661
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3874115637960478e-05,
+      "loss": 0.7695,
+      "step": 2662
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.387167601854111e-05,
+      "loss": 0.7456,
+      "step": 2663
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3869236399121738e-05,
+      "loss": 0.759,
+      "step": 2664
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3866796779702368e-05,
+      "loss": 0.7379,
+      "step": 2665
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3864357160282997e-05,
+      "loss": 0.8009,
+      "step": 2666
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3861917540863627e-05,
+      "loss": 0.7649,
+      "step": 2667
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3859477921444255e-05,
+      "loss": 0.8028,
+      "step": 2668
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3857038302024885e-05,
+      "loss": 0.8329,
+      "step": 2669
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3854598682605516e-05,
+      "loss": 0.8647,
+      "step": 2670
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3852159063186144e-05,
+      "loss": 0.8448,
+      "step": 2671
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3849719443766774e-05,
+      "loss": 0.8163,
+      "step": 2672
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3847279824347404e-05,
+      "loss": 0.7956,
+      "step": 2673
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3844840204928032e-05,
+      "loss": 0.7967,
+      "step": 2674
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3842400585508662e-05,
+      "loss": 0.736,
+      "step": 2675
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3839960966089291e-05,
+      "loss": 0.711,
+      "step": 2676
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.383752134666992e-05,
+      "loss": 0.8307,
+      "step": 2677
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.383508172725055e-05,
+      "loss": 0.781,
+      "step": 2678
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.383264210783118e-05,
+      "loss": 0.7784,
+      "step": 2679
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3830202488411809e-05,
+      "loss": 0.7992,
+      "step": 2680
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3827762868992438e-05,
+      "loss": 0.7372,
+      "step": 2681
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3825323249573068e-05,
+      "loss": 0.796,
+      "step": 2682
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3822883630153696e-05,
+      "loss": 0.7592,
+      "step": 2683
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3820444010734326e-05,
+      "loss": 0.8098,
+      "step": 2684
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3818004391314956e-05,
+      "loss": 0.7945,
+      "step": 2685
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3815564771895584e-05,
+      "loss": 0.7631,
+      "step": 2686
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3813125152476215e-05,
+      "loss": 0.7913,
+      "step": 2687
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3810685533056845e-05,
+      "loss": 0.7312,
+      "step": 2688
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3808245913637475e-05,
+      "loss": 0.8373,
+      "step": 2689
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3805806294218103e-05,
+      "loss": 0.7756,
+      "step": 2690
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3803366674798732e-05,
+      "loss": 0.8281,
+      "step": 2691
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3800927055379362e-05,
+      "loss": 0.7778,
+      "step": 2692
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.379848743595999e-05,
+      "loss": 0.7472,
+      "step": 2693
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3796047816540622e-05,
+      "loss": 0.7596,
+      "step": 2694
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3793608197121251e-05,
+      "loss": 0.7803,
+      "step": 2695
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.379116857770188e-05,
+      "loss": 0.7496,
+      "step": 2696
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.378872895828251e-05,
+      "loss": 0.781,
+      "step": 2697
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3786289338863139e-05,
+      "loss": 0.8475,
+      "step": 2698
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3783849719443767e-05,
+      "loss": 0.785,
+      "step": 2699
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3781410100024397e-05,
+      "loss": 0.8051,
+      "step": 2700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3778970480605027e-05,
+      "loss": 0.7636,
+      "step": 2701
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3776530861185655e-05,
+      "loss": 0.7995,
+      "step": 2702
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3774091241766286e-05,
+      "loss": 0.827,
+      "step": 2703
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3771651622346916e-05,
+      "loss": 0.7723,
+      "step": 2704
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3769212002927544e-05,
+      "loss": 0.813,
+      "step": 2705
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3766772383508174e-05,
+      "loss": 0.7931,
+      "step": 2706
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3764332764088803e-05,
+      "loss": 0.8178,
+      "step": 2707
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3761893144669433e-05,
+      "loss": 0.9285,
+      "step": 2708
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3759453525250061e-05,
+      "loss": 0.7789,
+      "step": 2709
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3757013905830693e-05,
+      "loss": 0.8243,
+      "step": 2710
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3754574286411322e-05,
+      "loss": 0.8487,
+      "step": 2711
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.375213466699195e-05,
+      "loss": 0.8103,
+      "step": 2712
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.374969504757258e-05,
+      "loss": 0.8104,
+      "step": 2713
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.374725542815321e-05,
+      "loss": 0.7907,
+      "step": 2714
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3744815808733838e-05,
+      "loss": 0.8519,
+      "step": 2715
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3742376189314468e-05,
+      "loss": 0.8373,
+      "step": 2716
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3739936569895097e-05,
+      "loss": 0.7371,
+      "step": 2717
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3737496950475725e-05,
+      "loss": 0.8823,
+      "step": 2718
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3735057331056357e-05,
+      "loss": 0.7659,
+      "step": 2719
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3732617711636987e-05,
+      "loss": 0.7997,
+      "step": 2720
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3730178092217615e-05,
+      "loss": 0.8376,
+      "step": 2721
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3727738472798244e-05,
+      "loss": 0.8322,
+      "step": 2722
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3725298853378874e-05,
+      "loss": 0.7715,
+      "step": 2723
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3722859233959502e-05,
+      "loss": 0.8124,
+      "step": 2724
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3720419614540132e-05,
+      "loss": 0.7281,
+      "step": 2725
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3717979995120763e-05,
+      "loss": 0.876,
+      "step": 2726
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3715540375701393e-05,
+      "loss": 0.7455,
+      "step": 2727
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3713100756282021e-05,
+      "loss": 0.7922,
+      "step": 2728
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3710661136862651e-05,
+      "loss": 0.7518,
+      "step": 2729
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.370822151744328e-05,
+      "loss": 0.7835,
+      "step": 2730
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3705781898023909e-05,
+      "loss": 0.7984,
+      "step": 2731
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3703342278604538e-05,
+      "loss": 0.7885,
+      "step": 2732
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.370090265918517e-05,
+      "loss": 0.8623,
+      "step": 2733
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3698463039765796e-05,
+      "loss": 0.7565,
+      "step": 2734
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3696023420346428e-05,
+      "loss": 0.7506,
+      "step": 2735
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3693583800927057e-05,
+      "loss": 0.7919,
+      "step": 2736
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3691144181507685e-05,
+      "loss": 0.8027,
+      "step": 2737
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3688704562088315e-05,
+      "loss": 0.8252,
+      "step": 2738
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3686264942668945e-05,
+      "loss": 0.7528,
+      "step": 2739
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3683825323249573e-05,
+      "loss": 0.84,
+      "step": 2740
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3681385703830203e-05,
+      "loss": 0.8082,
+      "step": 2741
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3678946084410834e-05,
+      "loss": 0.873,
+      "step": 2742
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3676506464991462e-05,
+      "loss": 0.7871,
+      "step": 2743
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3674066845572092e-05,
+      "loss": 0.7924,
+      "step": 2744
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3671627226152722e-05,
+      "loss": 0.7078,
+      "step": 2745
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3669187606733351e-05,
+      "loss": 0.8132,
+      "step": 2746
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.366674798731398e-05,
+      "loss": 0.7467,
+      "step": 2747
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.366430836789461e-05,
+      "loss": 0.8094,
+      "step": 2748
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.366186874847524e-05,
+      "loss": 0.7214,
+      "step": 2749
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3659429129055867e-05,
+      "loss": 0.7655,
+      "step": 2750
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3656989509636498e-05,
+      "loss": 0.756,
+      "step": 2751
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3654549890217128e-05,
+      "loss": 0.7487,
+      "step": 2752
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3652110270797756e-05,
+      "loss": 0.7779,
+      "step": 2753
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3649670651378386e-05,
+      "loss": 0.7575,
+      "step": 2754
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3647231031959016e-05,
+      "loss": 0.8137,
+      "step": 2755
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3644791412539644e-05,
+      "loss": 0.7966,
+      "step": 2756
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3642351793120274e-05,
+      "loss": 0.8789,
+      "step": 2757
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3639912173700905e-05,
+      "loss": 0.8315,
+      "step": 2758
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3637472554281533e-05,
+      "loss": 0.8253,
+      "step": 2759
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3635032934862163e-05,
+      "loss": 0.7603,
+      "step": 2760
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3632593315442793e-05,
+      "loss": 0.8157,
+      "step": 2761
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.363015369602342e-05,
+      "loss": 0.865,
+      "step": 2762
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.362771407660405e-05,
+      "loss": 0.9092,
+      "step": 2763
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.362527445718468e-05,
+      "loss": 0.7544,
+      "step": 2764
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3622834837765312e-05,
+      "loss": 0.8204,
+      "step": 2765
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3620395218345938e-05,
+      "loss": 0.7765,
+      "step": 2766
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.361795559892657e-05,
+      "loss": 0.7549,
+      "step": 2767
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3615515979507199e-05,
+      "loss": 0.7732,
+      "step": 2768
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3613076360087827e-05,
+      "loss": 0.7825,
+      "step": 2769
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3610636740668457e-05,
+      "loss": 0.833,
+      "step": 2770
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3608197121249087e-05,
+      "loss": 0.7856,
+      "step": 2771
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3605757501829715e-05,
+      "loss": 0.8641,
+      "step": 2772
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3603317882410344e-05,
+      "loss": 0.803,
+      "step": 2773
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3600878262990976e-05,
+      "loss": 0.7275,
+      "step": 2774
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3598438643571604e-05,
+      "loss": 0.8448,
+      "step": 2775
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3595999024152234e-05,
+      "loss": 0.7558,
+      "step": 2776
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3593559404732863e-05,
+      "loss": 0.7824,
+      "step": 2777
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3591119785313491e-05,
+      "loss": 0.7895,
+      "step": 2778
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3588680165894121e-05,
+      "loss": 0.808,
+      "step": 2779
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3586240546474751e-05,
+      "loss": 0.754,
+      "step": 2780
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3583800927055379e-05,
+      "loss": 0.8421,
+      "step": 2781
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.358136130763601e-05,
+      "loss": 0.8413,
+      "step": 2782
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.357892168821664e-05,
+      "loss": 0.7585,
+      "step": 2783
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.357648206879727e-05,
+      "loss": 0.7724,
+      "step": 2784
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3574042449377898e-05,
+      "loss": 0.8658,
+      "step": 2785
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3571602829958528e-05,
+      "loss": 0.7963,
+      "step": 2786
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3569163210539157e-05,
+      "loss": 0.7994,
+      "step": 2787
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3566723591119785e-05,
+      "loss": 0.8553,
+      "step": 2788
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3564283971700415e-05,
+      "loss": 0.7612,
+      "step": 2789
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3561844352281047e-05,
+      "loss": 0.8144,
+      "step": 2790
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3559404732861675e-05,
+      "loss": 0.7867,
+      "step": 2791
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3556965113442304e-05,
+      "loss": 0.7701,
+      "step": 2792
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3554525494022934e-05,
+      "loss": 0.8142,
+      "step": 2793
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3552085874603562e-05,
+      "loss": 0.7349,
+      "step": 2794
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3549646255184192e-05,
+      "loss": 0.7912,
+      "step": 2795
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3547206635764822e-05,
+      "loss": 0.8405,
+      "step": 2796
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.354476701634545e-05,
+      "loss": 0.7837,
+      "step": 2797
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3542327396926081e-05,
+      "loss": 0.7978,
+      "step": 2798
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3539887777506711e-05,
+      "loss": 0.7904,
+      "step": 2799
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3537448158087339e-05,
+      "loss": 0.7893,
+      "step": 2800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3535008538667969e-05,
+      "loss": 0.7957,
+      "step": 2801
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3532568919248599e-05,
+      "loss": 0.7811,
+      "step": 2802
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3530129299829228e-05,
+      "loss": 0.8167,
+      "step": 2803
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3527689680409856e-05,
+      "loss": 0.8052,
+      "step": 2804
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3525250060990486e-05,
+      "loss": 0.886,
+      "step": 2805
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3522810441571117e-05,
+      "loss": 0.7626,
+      "step": 2806
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3520370822151746e-05,
+      "loss": 0.8149,
+      "step": 2807
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3517931202732375e-05,
+      "loss": 0.7568,
+      "step": 2808
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3515491583313005e-05,
+      "loss": 0.7459,
+      "step": 2809
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3513051963893633e-05,
+      "loss": 0.8147,
+      "step": 2810
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3510612344474263e-05,
+      "loss": 0.8088,
+      "step": 2811
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3508172725054893e-05,
+      "loss": 0.8899,
+      "step": 2812
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.350573310563552e-05,
+      "loss": 0.7683,
+      "step": 2813
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3503293486216152e-05,
+      "loss": 0.8264,
+      "step": 2814
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3500853866796782e-05,
+      "loss": 0.778,
+      "step": 2815
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.349841424737741e-05,
+      "loss": 0.8022,
+      "step": 2816
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.349597462795804e-05,
+      "loss": 0.7897,
+      "step": 2817
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.349353500853867e-05,
+      "loss": 0.8067,
+      "step": 2818
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3491095389119297e-05,
+      "loss": 0.8652,
+      "step": 2819
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3488655769699927e-05,
+      "loss": 0.7618,
+      "step": 2820
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3486216150280559e-05,
+      "loss": 0.7522,
+      "step": 2821
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3483776530861185e-05,
+      "loss": 0.7733,
+      "step": 2822
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3481336911441816e-05,
+      "loss": 0.7957,
+      "step": 2823
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3478897292022446e-05,
+      "loss": 0.7912,
+      "step": 2824
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3476457672603076e-05,
+      "loss": 0.784,
+      "step": 2825
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3474018053183704e-05,
+      "loss": 0.8036,
+      "step": 2826
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3471578433764334e-05,
+      "loss": 0.7744,
+      "step": 2827
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3469138814344963e-05,
+      "loss": 0.8178,
+      "step": 2828
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3466699194925591e-05,
+      "loss": 0.7534,
+      "step": 2829
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3464259575506223e-05,
+      "loss": 0.8546,
+      "step": 2830
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3461819956086853e-05,
+      "loss": 0.8247,
+      "step": 2831
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.345938033666748e-05,
+      "loss": 0.8163,
+      "step": 2832
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.345694071724811e-05,
+      "loss": 0.7565,
+      "step": 2833
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.345450109782874e-05,
+      "loss": 0.7262,
+      "step": 2834
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3452061478409368e-05,
+      "loss": 0.7696,
+      "step": 2835
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3449621858989998e-05,
+      "loss": 0.737,
+      "step": 2836
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.344718223957063e-05,
+      "loss": 0.7545,
+      "step": 2837
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3444742620151256e-05,
+      "loss": 0.7705,
+      "step": 2838
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3442303000731887e-05,
+      "loss": 0.811,
+      "step": 2839
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3439863381312517e-05,
+      "loss": 0.7825,
+      "step": 2840
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3437423761893145e-05,
+      "loss": 0.8897,
+      "step": 2841
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3434984142473775e-05,
+      "loss": 0.7241,
+      "step": 2842
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3432544523054404e-05,
+      "loss": 0.7424,
+      "step": 2843
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3430104903635034e-05,
+      "loss": 0.7624,
+      "step": 2844
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3427665284215662e-05,
+      "loss": 0.7903,
+      "step": 2845
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3425225664796294e-05,
+      "loss": 0.7953,
+      "step": 2846
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3422786045376923e-05,
+      "loss": 0.8621,
+      "step": 2847
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3420346425957552e-05,
+      "loss": 0.8824,
+      "step": 2848
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3417906806538181e-05,
+      "loss": 0.777,
+      "step": 2849
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3415467187118811e-05,
+      "loss": 0.7757,
+      "step": 2850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3413027567699439e-05,
+      "loss": 0.8293,
+      "step": 2851
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3410587948280069e-05,
+      "loss": 0.7779,
+      "step": 2852
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.34081483288607e-05,
+      "loss": 0.797,
+      "step": 2853
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3405708709441327e-05,
+      "loss": 0.7835,
+      "step": 2854
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3403269090021958e-05,
+      "loss": 0.8289,
+      "step": 2855
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3400829470602588e-05,
+      "loss": 0.7675,
+      "step": 2856
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3398389851183216e-05,
+      "loss": 0.7442,
+      "step": 2857
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3395950231763846e-05,
+      "loss": 0.777,
+      "step": 2858
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3393510612344475e-05,
+      "loss": 0.8064,
+      "step": 2859
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3391070992925103e-05,
+      "loss": 0.7791,
+      "step": 2860
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3388631373505733e-05,
+      "loss": 0.8214,
+      "step": 2861
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3386191754086365e-05,
+      "loss": 0.8479,
+      "step": 2862
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3383752134666994e-05,
+      "loss": 0.8251,
+      "step": 2863
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3381312515247622e-05,
+      "loss": 0.8218,
+      "step": 2864
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3378872895828252e-05,
+      "loss": 0.7717,
+      "step": 2865
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3376433276408882e-05,
+      "loss": 0.8332,
+      "step": 2866
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.337399365698951e-05,
+      "loss": 0.8079,
+      "step": 2867
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.337155403757014e-05,
+      "loss": 0.7458,
+      "step": 2868
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3369114418150771e-05,
+      "loss": 0.7555,
+      "step": 2869
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3366674798731399e-05,
+      "loss": 0.8303,
+      "step": 2870
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3364235179312029e-05,
+      "loss": 0.7852,
+      "step": 2871
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3361795559892659e-05,
+      "loss": 0.7985,
+      "step": 2872
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3359355940473287e-05,
+      "loss": 0.805,
+      "step": 2873
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3356916321053916e-05,
+      "loss": 0.7713,
+      "step": 2874
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3354476701634546e-05,
+      "loss": 0.8665,
+      "step": 2875
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3352037082215174e-05,
+      "loss": 0.7728,
+      "step": 2876
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3349597462795804e-05,
+      "loss": 0.7871,
+      "step": 2877
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3347157843376435e-05,
+      "loss": 0.7043,
+      "step": 2878
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3344718223957063e-05,
+      "loss": 0.7436,
+      "step": 2879
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3342278604537693e-05,
+      "loss": 0.7566,
+      "step": 2880
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3339838985118323e-05,
+      "loss": 0.812,
+      "step": 2881
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3337399365698953e-05,
+      "loss": 0.7689,
+      "step": 2882
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.333495974627958e-05,
+      "loss": 0.7716,
+      "step": 2883
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.333252012686021e-05,
+      "loss": 0.8171,
+      "step": 2884
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3330080507440842e-05,
+      "loss": 0.7649,
+      "step": 2885
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.332764088802147e-05,
+      "loss": 0.8474,
+      "step": 2886
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.33252012686021e-05,
+      "loss": 0.7402,
+      "step": 2887
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.332276164918273e-05,
+      "loss": 0.7709,
+      "step": 2888
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3320322029763357e-05,
+      "loss": 0.8571,
+      "step": 2889
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3317882410343987e-05,
+      "loss": 0.8091,
+      "step": 2890
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3315442790924617e-05,
+      "loss": 0.8342,
+      "step": 2891
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3313003171505245e-05,
+      "loss": 0.7346,
+      "step": 2892
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3310563552085875e-05,
+      "loss": 0.8282,
+      "step": 2893
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3308123932666506e-05,
+      "loss": 0.8023,
+      "step": 2894
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3305684313247134e-05,
+      "loss": 0.7739,
+      "step": 2895
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3303244693827764e-05,
+      "loss": 0.7713,
+      "step": 2896
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3300805074408394e-05,
+      "loss": 0.8161,
+      "step": 2897
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3298365454989022e-05,
+      "loss": 0.7211,
+      "step": 2898
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3295925835569652e-05,
+      "loss": 0.7874,
+      "step": 2899
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3293486216150281e-05,
+      "loss": 0.7603,
+      "step": 2900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3291046596730913e-05,
+      "loss": 0.7927,
+      "step": 2901
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.328860697731154e-05,
+      "loss": 0.7912,
+      "step": 2902
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.328616735789217e-05,
+      "loss": 0.8038,
+      "step": 2903
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.32837277384728e-05,
+      "loss": 0.7673,
+      "step": 2904
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3281288119053428e-05,
+      "loss": 0.7887,
+      "step": 2905
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3278848499634058e-05,
+      "loss": 0.7412,
+      "step": 2906
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3276408880214688e-05,
+      "loss": 0.7763,
+      "step": 2907
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3273969260795316e-05,
+      "loss": 0.8799,
+      "step": 2908
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3271529641375947e-05,
+      "loss": 0.8137,
+      "step": 2909
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3269090021956577e-05,
+      "loss": 0.7298,
+      "step": 2910
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3266650402537205e-05,
+      "loss": 0.8321,
+      "step": 2911
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3264210783117835e-05,
+      "loss": 0.7807,
+      "step": 2912
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3261771163698465e-05,
+      "loss": 0.7315,
+      "step": 2913
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3259331544279093e-05,
+      "loss": 0.7718,
+      "step": 2914
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3256891924859722e-05,
+      "loss": 0.8437,
+      "step": 2915
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3254452305440352e-05,
+      "loss": 0.8308,
+      "step": 2916
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.325201268602098e-05,
+      "loss": 0.8021,
+      "step": 2917
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3249573066601612e-05,
+      "loss": 0.7566,
+      "step": 2918
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3247133447182241e-05,
+      "loss": 0.7954,
+      "step": 2919
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3244693827762871e-05,
+      "loss": 0.7743,
+      "step": 2920
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3242254208343499e-05,
+      "loss": 0.8168,
+      "step": 2921
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3239814588924129e-05,
+      "loss": 0.7699,
+      "step": 2922
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3237374969504759e-05,
+      "loss": 0.8201,
+      "step": 2923
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3234935350085387e-05,
+      "loss": 0.7877,
+      "step": 2924
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3232495730666018e-05,
+      "loss": 0.7861,
+      "step": 2925
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3230056111246648e-05,
+      "loss": 0.8079,
+      "step": 2926
+    },
+    {
+      "epoch": 1.38,
+      "eval_loss": 0.9687467813491821,
+      "eval_runtime": 205.352,
+      "eval_samples_per_second": 81.806,
+      "eval_steps_per_second": 10.226,
+      "step": 2926
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3227616491827276e-05,
+      "loss": 0.7821,
+      "step": 2927
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3225176872407906e-05,
+      "loss": 0.7537,
+      "step": 2928
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3222737252988535e-05,
+      "loss": 0.7365,
+      "step": 2929
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3220297633569163e-05,
+      "loss": 0.7912,
+      "step": 2930
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3217858014149793e-05,
+      "loss": 0.7958,
+      "step": 2931
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3215418394730423e-05,
+      "loss": 0.8025,
+      "step": 2932
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3212978775311051e-05,
+      "loss": 0.865,
+      "step": 2933
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3210539155891682e-05,
+      "loss": 0.7945,
+      "step": 2934
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3208099536472312e-05,
+      "loss": 0.7286,
+      "step": 2935
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.320565991705294e-05,
+      "loss": 0.8579,
+      "step": 2936
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.320322029763357e-05,
+      "loss": 0.7475,
+      "step": 2937
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.32007806782142e-05,
+      "loss": 0.7815,
+      "step": 2938
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.319834105879483e-05,
+      "loss": 0.7862,
+      "step": 2939
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3195901439375457e-05,
+      "loss": 0.7955,
+      "step": 2940
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3193461819956089e-05,
+      "loss": 0.7902,
+      "step": 2941
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3191022200536719e-05,
+      "loss": 0.7679,
+      "step": 2942
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3188582581117347e-05,
+      "loss": 0.765,
+      "step": 2943
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3186142961697976e-05,
+      "loss": 0.7926,
+      "step": 2944
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3183703342278606e-05,
+      "loss": 0.7768,
+      "step": 2945
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3181263722859234e-05,
+      "loss": 0.8166,
+      "step": 2946
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3178824103439864e-05,
+      "loss": 0.7479,
+      "step": 2947
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3176384484020495e-05,
+      "loss": 0.8221,
+      "step": 2948
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3173944864601122e-05,
+      "loss": 0.7713,
+      "step": 2949
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3171505245181753e-05,
+      "loss": 0.7434,
+      "step": 2950
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3169065625762383e-05,
+      "loss": 0.8204,
+      "step": 2951
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3166626006343011e-05,
+      "loss": 0.7836,
+      "step": 2952
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.316418638692364e-05,
+      "loss": 0.7486,
+      "step": 2953
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.316174676750427e-05,
+      "loss": 0.7869,
+      "step": 2954
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3159307148084899e-05,
+      "loss": 0.7731,
+      "step": 2955
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3156867528665528e-05,
+      "loss": 0.7742,
+      "step": 2956
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.315442790924616e-05,
+      "loss": 0.7691,
+      "step": 2957
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3151988289826788e-05,
+      "loss": 0.7575,
+      "step": 2958
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3149548670407418e-05,
+      "loss": 0.7285,
+      "step": 2959
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3147109050988047e-05,
+      "loss": 0.8282,
+      "step": 2960
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3144669431568677e-05,
+      "loss": 0.7908,
+      "step": 2961
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3142229812149305e-05,
+      "loss": 0.8529,
+      "step": 2962
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3139790192729935e-05,
+      "loss": 0.7869,
+      "step": 2963
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3137350573310566e-05,
+      "loss": 0.8177,
+      "step": 2964
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3134910953891193e-05,
+      "loss": 0.7345,
+      "step": 2965
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3132471334471824e-05,
+      "loss": 0.8247,
+      "step": 2966
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3130031715052454e-05,
+      "loss": 0.7927,
+      "step": 2967
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3127592095633082e-05,
+      "loss": 0.8259,
+      "step": 2968
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3125152476213712e-05,
+      "loss": 0.7531,
+      "step": 2969
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3122712856794341e-05,
+      "loss": 0.736,
+      "step": 2970
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.312027323737497e-05,
+      "loss": 0.7795,
+      "step": 2971
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3117833617955599e-05,
+      "loss": 0.7917,
+      "step": 2972
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.311539399853623e-05,
+      "loss": 0.8169,
+      "step": 2973
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3112954379116859e-05,
+      "loss": 0.7814,
+      "step": 2974
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3110514759697488e-05,
+      "loss": 0.7598,
+      "step": 2975
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3108075140278118e-05,
+      "loss": 0.8176,
+      "step": 2976
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3105635520858746e-05,
+      "loss": 0.7635,
+      "step": 2977
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3103195901439376e-05,
+      "loss": 0.7615,
+      "step": 2978
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3100756282020006e-05,
+      "loss": 0.7893,
+      "step": 2979
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3098316662600637e-05,
+      "loss": 0.77,
+      "step": 2980
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3095877043181263e-05,
+      "loss": 0.8254,
+      "step": 2981
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3093437423761895e-05,
+      "loss": 0.7744,
+      "step": 2982
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3090997804342525e-05,
+      "loss": 0.8066,
+      "step": 2983
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3088558184923153e-05,
+      "loss": 0.7772,
+      "step": 2984
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3086118565503782e-05,
+      "loss": 0.7521,
+      "step": 2985
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3083678946084412e-05,
+      "loss": 0.7588,
+      "step": 2986
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.308123932666504e-05,
+      "loss": 0.7818,
+      "step": 2987
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.307879970724567e-05,
+      "loss": 0.8049,
+      "step": 2988
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3076360087826301e-05,
+      "loss": 0.7836,
+      "step": 2989
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.307392046840693e-05,
+      "loss": 0.7902,
+      "step": 2990
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.307148084898756e-05,
+      "loss": 0.7374,
+      "step": 2991
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3069041229568189e-05,
+      "loss": 0.8195,
+      "step": 2992
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3066601610148817e-05,
+      "loss": 0.7111,
+      "step": 2993
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3064161990729447e-05,
+      "loss": 0.8254,
+      "step": 2994
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3061722371310076e-05,
+      "loss": 0.7773,
+      "step": 2995
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3059282751890705e-05,
+      "loss": 0.7666,
+      "step": 2996
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3056843132471336e-05,
+      "loss": 0.8069,
+      "step": 2997
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3054403513051966e-05,
+      "loss": 0.8094,
+      "step": 2998
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3051963893632595e-05,
+      "loss": 0.7498,
+      "step": 2999
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3049524274213224e-05,
+      "loss": 0.8193,
+      "step": 3000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3047084654793853e-05,
+      "loss": 0.7557,
+      "step": 3001
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3044645035374483e-05,
+      "loss": 0.8345,
+      "step": 3002
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3042205415955111e-05,
+      "loss": 0.8027,
+      "step": 3003
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.303976579653574e-05,
+      "loss": 0.7732,
+      "step": 3004
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3037326177116372e-05,
+      "loss": 0.7671,
+      "step": 3005
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3034886557697e-05,
+      "loss": 0.7457,
+      "step": 3006
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.303244693827763e-05,
+      "loss": 0.7767,
+      "step": 3007
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.303000731885826e-05,
+      "loss": 0.77,
+      "step": 3008
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3027567699438888e-05,
+      "loss": 0.7803,
+      "step": 3009
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3025128080019518e-05,
+      "loss": 0.7735,
+      "step": 3010
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3022688460600147e-05,
+      "loss": 0.8026,
+      "step": 3011
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3020248841180775e-05,
+      "loss": 0.8146,
+      "step": 3012
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3017809221761407e-05,
+      "loss": 0.8334,
+      "step": 3013
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3015369602342037e-05,
+      "loss": 0.7282,
+      "step": 3014
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3012929982922665e-05,
+      "loss": 0.7594,
+      "step": 3015
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3010490363503294e-05,
+      "loss": 0.8271,
+      "step": 3016
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3008050744083924e-05,
+      "loss": 0.7468,
+      "step": 3017
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3005611124664554e-05,
+      "loss": 0.8364,
+      "step": 3018
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.3003171505245182e-05,
+      "loss": 0.8142,
+      "step": 3019
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.3000731885825812e-05,
+      "loss": 0.7795,
+      "step": 3020
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2998292266406443e-05,
+      "loss": 0.7855,
+      "step": 3021
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2995852646987071e-05,
+      "loss": 0.7942,
+      "step": 3022
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2993413027567701e-05,
+      "loss": 0.7547,
+      "step": 3023
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.299097340814833e-05,
+      "loss": 0.7913,
+      "step": 3024
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2988533788728959e-05,
+      "loss": 0.8094,
+      "step": 3025
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2986094169309588e-05,
+      "loss": 0.7435,
+      "step": 3026
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2983654549890218e-05,
+      "loss": 0.7904,
+      "step": 3027
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2981214930470846e-05,
+      "loss": 0.7962,
+      "step": 3028
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2978775311051478e-05,
+      "loss": 0.8035,
+      "step": 3029
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2976335691632107e-05,
+      "loss": 0.7873,
+      "step": 3030
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2973896072212735e-05,
+      "loss": 0.7985,
+      "step": 3031
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2971456452793365e-05,
+      "loss": 0.7903,
+      "step": 3032
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2969016833373995e-05,
+      "loss": 0.7284,
+      "step": 3033
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2966577213954623e-05,
+      "loss": 0.7328,
+      "step": 3034
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2964137594535253e-05,
+      "loss": 0.7673,
+      "step": 3035
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2961697975115884e-05,
+      "loss": 0.7258,
+      "step": 3036
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2959258355696514e-05,
+      "loss": 0.7946,
+      "step": 3037
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2956818736277142e-05,
+      "loss": 0.8332,
+      "step": 3038
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2954379116857772e-05,
+      "loss": 0.8197,
+      "step": 3039
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2951939497438401e-05,
+      "loss": 0.7818,
+      "step": 3040
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.294949987801903e-05,
+      "loss": 0.8143,
+      "step": 3041
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.294706025859966e-05,
+      "loss": 0.7932,
+      "step": 3042
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2944620639180289e-05,
+      "loss": 0.8272,
+      "step": 3043
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2942181019760917e-05,
+      "loss": 0.7555,
+      "step": 3044
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2939741400341548e-05,
+      "loss": 0.7343,
+      "step": 3045
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2937301780922178e-05,
+      "loss": 0.8183,
+      "step": 3046
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2934862161502806e-05,
+      "loss": 0.7219,
+      "step": 3047
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2932422542083436e-05,
+      "loss": 0.7289,
+      "step": 3048
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2929982922664066e-05,
+      "loss": 0.7808,
+      "step": 3049
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2927543303244694e-05,
+      "loss": 0.7853,
+      "step": 3050
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2925103683825324e-05,
+      "loss": 0.799,
+      "step": 3051
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2922664064405955e-05,
+      "loss": 0.77,
+      "step": 3052
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2920224444986581e-05,
+      "loss": 0.7585,
+      "step": 3053
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2917784825567213e-05,
+      "loss": 0.7457,
+      "step": 3054
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2915345206147843e-05,
+      "loss": 0.752,
+      "step": 3055
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2912905586728472e-05,
+      "loss": 0.8158,
+      "step": 3056
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.29104659673091e-05,
+      "loss": 0.7804,
+      "step": 3057
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.290802634788973e-05,
+      "loss": 0.7413,
+      "step": 3058
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.290558672847036e-05,
+      "loss": 0.7865,
+      "step": 3059
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2903147109050988e-05,
+      "loss": 0.7817,
+      "step": 3060
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.290070748963162e-05,
+      "loss": 0.8097,
+      "step": 3061
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2898267870212249e-05,
+      "loss": 0.7522,
+      "step": 3062
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2895828250792877e-05,
+      "loss": 0.7763,
+      "step": 3063
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2893388631373507e-05,
+      "loss": 0.7361,
+      "step": 3064
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2890949011954137e-05,
+      "loss": 0.7765,
+      "step": 3065
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2888509392534765e-05,
+      "loss": 0.8421,
+      "step": 3066
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2886069773115394e-05,
+      "loss": 0.7755,
+      "step": 3067
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2883630153696026e-05,
+      "loss": 0.757,
+      "step": 3068
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2881190534276652e-05,
+      "loss": 0.8383,
+      "step": 3069
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2878750914857284e-05,
+      "loss": 0.7569,
+      "step": 3070
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2876311295437913e-05,
+      "loss": 0.8219,
+      "step": 3071
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2873871676018541e-05,
+      "loss": 0.808,
+      "step": 3072
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2871432056599171e-05,
+      "loss": 0.8244,
+      "step": 3073
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2868992437179801e-05,
+      "loss": 0.7662,
+      "step": 3074
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2866552817760432e-05,
+      "loss": 0.8562,
+      "step": 3075
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2864113198341059e-05,
+      "loss": 0.7755,
+      "step": 3076
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.286167357892169e-05,
+      "loss": 0.726,
+      "step": 3077
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.285923395950232e-05,
+      "loss": 0.7918,
+      "step": 3078
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2856794340082948e-05,
+      "loss": 0.7529,
+      "step": 3079
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2854354720663578e-05,
+      "loss": 0.724,
+      "step": 3080
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2851915101244207e-05,
+      "loss": 0.784,
+      "step": 3081
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2849475481824835e-05,
+      "loss": 0.8262,
+      "step": 3082
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2847035862405465e-05,
+      "loss": 0.8412,
+      "step": 3083
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2844596242986097e-05,
+      "loss": 0.756,
+      "step": 3084
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2842156623566725e-05,
+      "loss": 0.7362,
+      "step": 3085
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2839717004147354e-05,
+      "loss": 0.7399,
+      "step": 3086
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2837277384727984e-05,
+      "loss": 0.7702,
+      "step": 3087
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2834837765308612e-05,
+      "loss": 0.7546,
+      "step": 3088
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2832398145889242e-05,
+      "loss": 0.7631,
+      "step": 3089
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2829958526469872e-05,
+      "loss": 0.8161,
+      "step": 3090
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.28275189070505e-05,
+      "loss": 0.7763,
+      "step": 3091
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.282507928763113e-05,
+      "loss": 0.7879,
+      "step": 3092
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2822639668211761e-05,
+      "loss": 0.7656,
+      "step": 3093
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2820200048792389e-05,
+      "loss": 0.8129,
+      "step": 3094
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2817760429373019e-05,
+      "loss": 0.7657,
+      "step": 3095
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2815320809953648e-05,
+      "loss": 0.8049,
+      "step": 3096
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2812881190534278e-05,
+      "loss": 0.7778,
+      "step": 3097
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2810441571114906e-05,
+      "loss": 0.7545,
+      "step": 3098
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2808001951695536e-05,
+      "loss": 0.8461,
+      "step": 3099
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2805562332276167e-05,
+      "loss": 0.8025,
+      "step": 3100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2803122712856796e-05,
+      "loss": 0.7224,
+      "step": 3101
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2800683093437425e-05,
+      "loss": 0.7656,
+      "step": 3102
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2798243474018055e-05,
+      "loss": 0.761,
+      "step": 3103
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2795803854598683e-05,
+      "loss": 0.7968,
+      "step": 3104
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2793364235179313e-05,
+      "loss": 0.8193,
+      "step": 3105
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2790924615759943e-05,
+      "loss": 0.7691,
+      "step": 3106
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.278848499634057e-05,
+      "loss": 0.7897,
+      "step": 3107
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.27860453769212e-05,
+      "loss": 0.8155,
+      "step": 3108
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2783605757501832e-05,
+      "loss": 0.7891,
+      "step": 3109
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.278116613808246e-05,
+      "loss": 0.7967,
+      "step": 3110
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.277872651866309e-05,
+      "loss": 0.8114,
+      "step": 3111
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.277628689924372e-05,
+      "loss": 0.8565,
+      "step": 3112
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2773847279824347e-05,
+      "loss": 0.7209,
+      "step": 3113
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2771407660404977e-05,
+      "loss": 0.7484,
+      "step": 3114
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2768968040985607e-05,
+      "loss": 0.7869,
+      "step": 3115
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2766528421566238e-05,
+      "loss": 0.6623,
+      "step": 3116
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2764088802146866e-05,
+      "loss": 0.7724,
+      "step": 3117
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2761649182727496e-05,
+      "loss": 0.8091,
+      "step": 3118
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2759209563308126e-05,
+      "loss": 0.7794,
+      "step": 3119
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2756769943888754e-05,
+      "loss": 0.7856,
+      "step": 3120
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2754330324469384e-05,
+      "loss": 0.7529,
+      "step": 3121
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2751890705050013e-05,
+      "loss": 0.763,
+      "step": 3122
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2749451085630641e-05,
+      "loss": 0.7422,
+      "step": 3123
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2747011466211273e-05,
+      "loss": 0.7507,
+      "step": 3124
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2744571846791903e-05,
+      "loss": 0.7825,
+      "step": 3125
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.274213222737253e-05,
+      "loss": 0.7489,
+      "step": 3126
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.273969260795316e-05,
+      "loss": 0.752,
+      "step": 3127
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.273725298853379e-05,
+      "loss": 0.717,
+      "step": 3128
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2734813369114418e-05,
+      "loss": 0.81,
+      "step": 3129
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2732373749695048e-05,
+      "loss": 0.8296,
+      "step": 3130
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2729934130275678e-05,
+      "loss": 0.7316,
+      "step": 3131
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2727494510856306e-05,
+      "loss": 0.8096,
+      "step": 3132
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2725054891436937e-05,
+      "loss": 0.7683,
+      "step": 3133
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2722615272017567e-05,
+      "loss": 0.7861,
+      "step": 3134
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2720175652598197e-05,
+      "loss": 0.801,
+      "step": 3135
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2717736033178825e-05,
+      "loss": 0.7575,
+      "step": 3136
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2715296413759454e-05,
+      "loss": 0.8269,
+      "step": 3137
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2712856794340084e-05,
+      "loss": 0.81,
+      "step": 3138
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2710417174920712e-05,
+      "loss": 0.8036,
+      "step": 3139
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2707977555501344e-05,
+      "loss": 0.8082,
+      "step": 3140
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2705537936081973e-05,
+      "loss": 0.7805,
+      "step": 3141
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2703098316662601e-05,
+      "loss": 0.7802,
+      "step": 3142
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2700658697243231e-05,
+      "loss": 0.7872,
+      "step": 3143
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2698219077823861e-05,
+      "loss": 0.7557,
+      "step": 3144
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2695779458404489e-05,
+      "loss": 0.7777,
+      "step": 3145
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2693339838985119e-05,
+      "loss": 0.7729,
+      "step": 3146
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2690900219565749e-05,
+      "loss": 0.7496,
+      "step": 3147
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2688460600146377e-05,
+      "loss": 0.7391,
+      "step": 3148
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2686020980727008e-05,
+      "loss": 0.7472,
+      "step": 3149
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2683581361307638e-05,
+      "loss": 0.7513,
+      "step": 3150
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2681141741888266e-05,
+      "loss": 0.7225,
+      "step": 3151
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2678702122468896e-05,
+      "loss": 0.8257,
+      "step": 3152
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2676262503049525e-05,
+      "loss": 0.7588,
+      "step": 3153
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2673822883630155e-05,
+      "loss": 0.7392,
+      "step": 3154
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2671383264210783e-05,
+      "loss": 0.7957,
+      "step": 3155
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2668943644791415e-05,
+      "loss": 0.7287,
+      "step": 3156
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2666504025372044e-05,
+      "loss": 0.7757,
+      "step": 3157
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2664064405952672e-05,
+      "loss": 0.7904,
+      "step": 3158
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2661624786533302e-05,
+      "loss": 0.7955,
+      "step": 3159
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2659185167113932e-05,
+      "loss": 0.7762,
+      "step": 3160
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.265674554769456e-05,
+      "loss": 0.7727,
+      "step": 3161
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.265430592827519e-05,
+      "loss": 0.7623,
+      "step": 3162
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2651866308855821e-05,
+      "loss": 0.833,
+      "step": 3163
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2649426689436447e-05,
+      "loss": 0.7623,
+      "step": 3164
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2646987070017079e-05,
+      "loss": 0.7382,
+      "step": 3165
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2644547450597709e-05,
+      "loss": 0.7506,
+      "step": 3166
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2642107831178337e-05,
+      "loss": 0.7367,
+      "step": 3167
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2639668211758966e-05,
+      "loss": 0.7415,
+      "step": 3168
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2637228592339596e-05,
+      "loss": 0.7432,
+      "step": 3169
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2634788972920224e-05,
+      "loss": 0.8381,
+      "step": 3170
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2632349353500854e-05,
+      "loss": 0.7797,
+      "step": 3171
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2629909734081485e-05,
+      "loss": 0.7766,
+      "step": 3172
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2627470114662115e-05,
+      "loss": 0.8194,
+      "step": 3173
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2625030495242743e-05,
+      "loss": 0.7753,
+      "step": 3174
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2622590875823373e-05,
+      "loss": 0.7725,
+      "step": 3175
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2620151256404003e-05,
+      "loss": 0.8484,
+      "step": 3176
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.261771163698463e-05,
+      "loss": 0.8304,
+      "step": 3177
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.261527201756526e-05,
+      "loss": 0.7574,
+      "step": 3178
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2612832398145892e-05,
+      "loss": 0.7998,
+      "step": 3179
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2610392778726518e-05,
+      "loss": 0.8013,
+      "step": 3180
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.260795315930715e-05,
+      "loss": 0.7338,
+      "step": 3181
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.260551353988778e-05,
+      "loss": 0.7744,
+      "step": 3182
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2603073920468407e-05,
+      "loss": 0.7981,
+      "step": 3183
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2600634301049037e-05,
+      "loss": 0.7557,
+      "step": 3184
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2598194681629667e-05,
+      "loss": 0.7396,
+      "step": 3185
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2595755062210295e-05,
+      "loss": 0.8047,
+      "step": 3186
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2593315442790925e-05,
+      "loss": 0.7861,
+      "step": 3187
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2590875823371556e-05,
+      "loss": 0.8227,
+      "step": 3188
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2588436203952184e-05,
+      "loss": 0.775,
+      "step": 3189
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2585996584532814e-05,
+      "loss": 0.7565,
+      "step": 3190
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2583556965113444e-05,
+      "loss": 0.889,
+      "step": 3191
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2581117345694073e-05,
+      "loss": 0.7537,
+      "step": 3192
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2578677726274702e-05,
+      "loss": 0.8059,
+      "step": 3193
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2576238106855331e-05,
+      "loss": 0.7972,
+      "step": 3194
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2573798487435963e-05,
+      "loss": 0.7857,
+      "step": 3195
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2571358868016589e-05,
+      "loss": 0.7456,
+      "step": 3196
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.256891924859722e-05,
+      "loss": 0.7893,
+      "step": 3197
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.256647962917785e-05,
+      "loss": 0.7096,
+      "step": 3198
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2564040009758478e-05,
+      "loss": 0.7379,
+      "step": 3199
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2561600390339108e-05,
+      "loss": 0.8149,
+      "step": 3200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2559160770919738e-05,
+      "loss": 0.7347,
+      "step": 3201
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2556721151500366e-05,
+      "loss": 0.7792,
+      "step": 3202
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2554281532080996e-05,
+      "loss": 0.7462,
+      "step": 3203
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2551841912661627e-05,
+      "loss": 0.753,
+      "step": 3204
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2549402293242255e-05,
+      "loss": 0.7363,
+      "step": 3205
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2546962673822885e-05,
+      "loss": 0.7819,
+      "step": 3206
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2544523054403515e-05,
+      "loss": 0.7875,
+      "step": 3207
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2542083434984143e-05,
+      "loss": 0.7957,
+      "step": 3208
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2539643815564772e-05,
+      "loss": 0.7498,
+      "step": 3209
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2537204196145402e-05,
+      "loss": 0.7638,
+      "step": 3210
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2534764576726034e-05,
+      "loss": 0.7945,
+      "step": 3211
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2532324957306662e-05,
+      "loss": 0.8437,
+      "step": 3212
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2529885337887291e-05,
+      "loss": 0.8161,
+      "step": 3213
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2527445718467921e-05,
+      "loss": 0.7698,
+      "step": 3214
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2525006099048549e-05,
+      "loss": 0.7867,
+      "step": 3215
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2522566479629179e-05,
+      "loss": 0.7662,
+      "step": 3216
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2520126860209809e-05,
+      "loss": 0.8368,
+      "step": 3217
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2517687240790437e-05,
+      "loss": 0.7017,
+      "step": 3218
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2515247621371066e-05,
+      "loss": 0.7543,
+      "step": 3219
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2512808001951698e-05,
+      "loss": 0.8134,
+      "step": 3220
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2510368382532326e-05,
+      "loss": 0.817,
+      "step": 3221
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2507928763112956e-05,
+      "loss": 0.7824,
+      "step": 3222
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2505489143693585e-05,
+      "loss": 0.7854,
+      "step": 3223
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2503049524274213e-05,
+      "loss": 0.7215,
+      "step": 3224
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2500609904854843e-05,
+      "loss": 0.8315,
+      "step": 3225
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2498170285435473e-05,
+      "loss": 0.8475,
+      "step": 3226
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2495730666016101e-05,
+      "loss": 0.7365,
+      "step": 3227
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2493291046596732e-05,
+      "loss": 0.7654,
+      "step": 3228
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2490851427177362e-05,
+      "loss": 0.8177,
+      "step": 3229
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.248841180775799e-05,
+      "loss": 0.7902,
+      "step": 3230
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.248597218833862e-05,
+      "loss": 0.8387,
+      "step": 3231
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.248353256891925e-05,
+      "loss": 0.7703,
+      "step": 3232
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.248109294949988e-05,
+      "loss": 0.8587,
+      "step": 3233
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2478653330080507e-05,
+      "loss": 0.7341,
+      "step": 3234
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2476213710661139e-05,
+      "loss": 0.7245,
+      "step": 3235
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2473774091241769e-05,
+      "loss": 0.7928,
+      "step": 3236
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2471334471822397e-05,
+      "loss": 0.8514,
+      "step": 3237
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2468894852403026e-05,
+      "loss": 0.8643,
+      "step": 3238
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2466455232983656e-05,
+      "loss": 0.7721,
+      "step": 3239
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2464015613564284e-05,
+      "loss": 0.8008,
+      "step": 3240
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2461575994144914e-05,
+      "loss": 0.8045,
+      "step": 3241
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2459136374725544e-05,
+      "loss": 0.7802,
+      "step": 3242
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2456696755306172e-05,
+      "loss": 0.8511,
+      "step": 3243
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2454257135886803e-05,
+      "loss": 0.7596,
+      "step": 3244
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2451817516467433e-05,
+      "loss": 0.7413,
+      "step": 3245
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2449377897048061e-05,
+      "loss": 0.8085,
+      "step": 3246
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.244693827762869e-05,
+      "loss": 0.7979,
+      "step": 3247
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.244449865820932e-05,
+      "loss": 0.7377,
+      "step": 3248
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2442059038789949e-05,
+      "loss": 0.7781,
+      "step": 3249
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2439619419370578e-05,
+      "loss": 0.8195,
+      "step": 3250
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.243717979995121e-05,
+      "loss": 0.7192,
+      "step": 3251
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.243474018053184e-05,
+      "loss": 0.7578,
+      "step": 3252
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2432300561112468e-05,
+      "loss": 0.8248,
+      "step": 3253
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2429860941693097e-05,
+      "loss": 0.7735,
+      "step": 3254
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2427421322273727e-05,
+      "loss": 0.7585,
+      "step": 3255
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2424981702854355e-05,
+      "loss": 0.8444,
+      "step": 3256
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2422542083434985e-05,
+      "loss": 0.8021,
+      "step": 3257
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2420102464015615e-05,
+      "loss": 0.735,
+      "step": 3258
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2417662844596243e-05,
+      "loss": 0.7632,
+      "step": 3259
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2415223225176874e-05,
+      "loss": 0.7725,
+      "step": 3260
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2412783605757504e-05,
+      "loss": 0.8111,
+      "step": 3261
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2410343986338132e-05,
+      "loss": 0.7595,
+      "step": 3262
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2407904366918762e-05,
+      "loss": 0.7836,
+      "step": 3263
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2405464747499391e-05,
+      "loss": 0.7692,
+      "step": 3264
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.240302512808002e-05,
+      "loss": 0.8088,
+      "step": 3265
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2400585508660649e-05,
+      "loss": 0.7258,
+      "step": 3266
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.239814588924128e-05,
+      "loss": 0.8298,
+      "step": 3267
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2395706269821907e-05,
+      "loss": 0.8157,
+      "step": 3268
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2393266650402538e-05,
+      "loss": 0.7296,
+      "step": 3269
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2390827030983168e-05,
+      "loss": 0.7462,
+      "step": 3270
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2388387411563798e-05,
+      "loss": 0.7829,
+      "step": 3271
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2385947792144426e-05,
+      "loss": 0.7285,
+      "step": 3272
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2383508172725056e-05,
+      "loss": 0.758,
+      "step": 3273
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2381068553305687e-05,
+      "loss": 0.8172,
+      "step": 3274
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2378628933886313e-05,
+      "loss": 0.803,
+      "step": 3275
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2376189314466945e-05,
+      "loss": 0.745,
+      "step": 3276
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2373749695047575e-05,
+      "loss": 0.7976,
+      "step": 3277
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2371310075628203e-05,
+      "loss": 0.7621,
+      "step": 3278
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2368870456208832e-05,
+      "loss": 0.7611,
+      "step": 3279
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2366430836789462e-05,
+      "loss": 0.782,
+      "step": 3280
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.236399121737009e-05,
+      "loss": 0.7614,
+      "step": 3281
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.236155159795072e-05,
+      "loss": 0.7649,
+      "step": 3282
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2359111978531351e-05,
+      "loss": 0.7948,
+      "step": 3283
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2356672359111978e-05,
+      "loss": 0.7389,
+      "step": 3284
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.235423273969261e-05,
+      "loss": 0.763,
+      "step": 3285
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2351793120273239e-05,
+      "loss": 0.8224,
+      "step": 3286
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2349353500853867e-05,
+      "loss": 0.7579,
+      "step": 3287
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2346913881434497e-05,
+      "loss": 0.831,
+      "step": 3288
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2344474262015126e-05,
+      "loss": 0.7777,
+      "step": 3289
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2342034642595758e-05,
+      "loss": 0.6894,
+      "step": 3290
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2339595023176384e-05,
+      "loss": 0.781,
+      "step": 3291
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2337155403757016e-05,
+      "loss": 0.7749,
+      "step": 3292
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2334715784337645e-05,
+      "loss": 0.7758,
+      "step": 3293
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2332276164918274e-05,
+      "loss": 0.7983,
+      "step": 3294
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2329836545498903e-05,
+      "loss": 0.8203,
+      "step": 3295
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2327396926079533e-05,
+      "loss": 0.8062,
+      "step": 3296
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2324957306660161e-05,
+      "loss": 0.8668,
+      "step": 3297
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.232251768724079e-05,
+      "loss": 0.8069,
+      "step": 3298
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2320078067821422e-05,
+      "loss": 0.7302,
+      "step": 3299
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.231763844840205e-05,
+      "loss": 0.8666,
+      "step": 3300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.231519882898268e-05,
+      "loss": 0.787,
+      "step": 3301
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.231275920956331e-05,
+      "loss": 0.7921,
+      "step": 3302
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2310319590143938e-05,
+      "loss": 0.7846,
+      "step": 3303
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2307879970724568e-05,
+      "loss": 0.8369,
+      "step": 3304
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2305440351305197e-05,
+      "loss": 0.8009,
+      "step": 3305
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2303000731885825e-05,
+      "loss": 0.7967,
+      "step": 3306
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2300561112466455e-05,
+      "loss": 0.7582,
+      "step": 3307
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2298121493047087e-05,
+      "loss": 0.7839,
+      "step": 3308
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2295681873627716e-05,
+      "loss": 0.7836,
+      "step": 3309
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2293242254208344e-05,
+      "loss": 0.8207,
+      "step": 3310
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2290802634788974e-05,
+      "loss": 0.835,
+      "step": 3311
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2288363015369604e-05,
+      "loss": 0.7652,
+      "step": 3312
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2285923395950232e-05,
+      "loss": 0.7283,
+      "step": 3313
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2283483776530862e-05,
+      "loss": 0.7599,
+      "step": 3314
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2281044157111493e-05,
+      "loss": 0.8143,
+      "step": 3315
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2278604537692121e-05,
+      "loss": 0.7811,
+      "step": 3316
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2276164918272751e-05,
+      "loss": 0.8286,
+      "step": 3317
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.227372529885338e-05,
+      "loss": 0.7449,
+      "step": 3318
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2271285679434009e-05,
+      "loss": 0.8004,
+      "step": 3319
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2268846060014638e-05,
+      "loss": 0.7977,
+      "step": 3320
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2266406440595268e-05,
+      "loss": 0.8228,
+      "step": 3321
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2263966821175896e-05,
+      "loss": 0.7388,
+      "step": 3322
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2261527201756528e-05,
+      "loss": 0.7911,
+      "step": 3323
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2259087582337157e-05,
+      "loss": 0.8307,
+      "step": 3324
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2256647962917785e-05,
+      "loss": 0.773,
+      "step": 3325
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2254208343498415e-05,
+      "loss": 0.7692,
+      "step": 3326
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2251768724079045e-05,
+      "loss": 0.7469,
+      "step": 3327
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2249329104659675e-05,
+      "loss": 0.8291,
+      "step": 3328
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2246889485240303e-05,
+      "loss": 0.7741,
+      "step": 3329
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2244449865820932e-05,
+      "loss": 0.8301,
+      "step": 3330
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2242010246401564e-05,
+      "loss": 0.7724,
+      "step": 3331
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2239570626982192e-05,
+      "loss": 0.8003,
+      "step": 3332
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2237131007562822e-05,
+      "loss": 0.8381,
+      "step": 3333
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2234691388143451e-05,
+      "loss": 0.7782,
+      "step": 3334
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.223225176872408e-05,
+      "loss": 0.7872,
+      "step": 3335
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.222981214930471e-05,
+      "loss": 0.7902,
+      "step": 3336
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2227372529885339e-05,
+      "loss": 0.8051,
+      "step": 3337
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2224932910465967e-05,
+      "loss": 0.6954,
+      "step": 3338
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2222493291046598e-05,
+      "loss": 0.7092,
+      "step": 3339
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2220053671627228e-05,
+      "loss": 0.7937,
+      "step": 3340
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2217614052207856e-05,
+      "loss": 0.7669,
+      "step": 3341
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2215174432788486e-05,
+      "loss": 0.7517,
+      "step": 3342
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2212734813369116e-05,
+      "loss": 0.7919,
+      "step": 3343
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2210295193949744e-05,
+      "loss": 0.8169,
+      "step": 3344
+    },
+    {
+      "epoch": 1.58,
+      "eval_loss": 0.9610660076141357,
+      "eval_runtime": 205.4211,
+      "eval_samples_per_second": 81.778,
+      "eval_steps_per_second": 10.223,
+      "step": 3344
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2207855574530374e-05,
+      "loss": 0.8316,
+      "step": 3345
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2205415955111003e-05,
+      "loss": 0.7924,
+      "step": 3346
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2202976335691635e-05,
+      "loss": 0.7583,
+      "step": 3347
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2200536716272263e-05,
+      "loss": 0.7455,
+      "step": 3348
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2198097096852893e-05,
+      "loss": 0.8116,
+      "step": 3349
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2195657477433522e-05,
+      "loss": 0.7113,
+      "step": 3350
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.219321785801415e-05,
+      "loss": 0.8272,
+      "step": 3351
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.219077823859478e-05,
+      "loss": 0.7863,
+      "step": 3352
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.218833861917541e-05,
+      "loss": 0.8086,
+      "step": 3353
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2185898999756038e-05,
+      "loss": 0.7868,
+      "step": 3354
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.218345938033667e-05,
+      "loss": 0.7597,
+      "step": 3355
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2181019760917299e-05,
+      "loss": 0.7862,
+      "step": 3356
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2178580141497927e-05,
+      "loss": 0.7951,
+      "step": 3357
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2176140522078557e-05,
+      "loss": 0.7726,
+      "step": 3358
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2173700902659187e-05,
+      "loss": 0.8025,
+      "step": 3359
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2171261283239815e-05,
+      "loss": 0.7807,
+      "step": 3360
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2168821663820444e-05,
+      "loss": 0.7834,
+      "step": 3361
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2166382044401076e-05,
+      "loss": 0.7764,
+      "step": 3362
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2163942424981702e-05,
+      "loss": 0.8277,
+      "step": 3363
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2161502805562334e-05,
+      "loss": 0.8045,
+      "step": 3364
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2159063186142963e-05,
+      "loss": 0.7261,
+      "step": 3365
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2156623566723591e-05,
+      "loss": 0.753,
+      "step": 3366
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2154183947304221e-05,
+      "loss": 0.8114,
+      "step": 3367
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2151744327884851e-05,
+      "loss": 0.7691,
+      "step": 3368
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.214930470846548e-05,
+      "loss": 0.7874,
+      "step": 3369
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2146865089046109e-05,
+      "loss": 0.7615,
+      "step": 3370
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.214442546962674e-05,
+      "loss": 0.7917,
+      "step": 3371
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.214198585020737e-05,
+      "loss": 0.7317,
+      "step": 3372
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2139546230787998e-05,
+      "loss": 0.7959,
+      "step": 3373
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2137106611368628e-05,
+      "loss": 0.8021,
+      "step": 3374
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2134666991949257e-05,
+      "loss": 0.7888,
+      "step": 3375
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2132227372529885e-05,
+      "loss": 0.7643,
+      "step": 3376
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2129787753110515e-05,
+      "loss": 0.7701,
+      "step": 3377
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2127348133691147e-05,
+      "loss": 0.752,
+      "step": 3378
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2124908514271773e-05,
+      "loss": 0.8348,
+      "step": 3379
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2122468894852404e-05,
+      "loss": 0.7836,
+      "step": 3380
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2120029275433034e-05,
+      "loss": 0.8112,
+      "step": 3381
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2117589656013662e-05,
+      "loss": 0.8193,
+      "step": 3382
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2115150036594292e-05,
+      "loss": 0.7686,
+      "step": 3383
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2112710417174922e-05,
+      "loss": 0.7632,
+      "step": 3384
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.211027079775555e-05,
+      "loss": 0.8657,
+      "step": 3385
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.210783117833618e-05,
+      "loss": 0.7643,
+      "step": 3386
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2105391558916811e-05,
+      "loss": 0.7477,
+      "step": 3387
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.210295193949744e-05,
+      "loss": 0.8582,
+      "step": 3388
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2100512320078069e-05,
+      "loss": 0.7696,
+      "step": 3389
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2098072700658698e-05,
+      "loss": 0.7597,
+      "step": 3390
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2095633081239328e-05,
+      "loss": 0.8148,
+      "step": 3391
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2093193461819956e-05,
+      "loss": 0.7971,
+      "step": 3392
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2090753842400586e-05,
+      "loss": 0.776,
+      "step": 3393
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2088314222981217e-05,
+      "loss": 0.7965,
+      "step": 3394
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2085874603561844e-05,
+      "loss": 0.754,
+      "step": 3395
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2083434984142475e-05,
+      "loss": 0.7758,
+      "step": 3396
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2080995364723105e-05,
+      "loss": 0.7523,
+      "step": 3397
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2078555745303733e-05,
+      "loss": 0.7933,
+      "step": 3398
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2076116125884363e-05,
+      "loss": 0.7881,
+      "step": 3399
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2073676506464993e-05,
+      "loss": 0.8461,
+      "step": 3400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.207123688704562e-05,
+      "loss": 0.7471,
+      "step": 3401
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.206879726762625e-05,
+      "loss": 0.7835,
+      "step": 3402
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2066357648206882e-05,
+      "loss": 0.7523,
+      "step": 3403
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.206391802878751e-05,
+      "loss": 0.8011,
+      "step": 3404
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.206147840936814e-05,
+      "loss": 0.7549,
+      "step": 3405
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.205903878994877e-05,
+      "loss": 0.7947,
+      "step": 3406
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2056599170529399e-05,
+      "loss": 0.7684,
+      "step": 3407
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2054159551110027e-05,
+      "loss": 0.7477,
+      "step": 3408
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2051719931690657e-05,
+      "loss": 0.8162,
+      "step": 3409
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2049280312271288e-05,
+      "loss": 0.7104,
+      "step": 3410
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2046840692851916e-05,
+      "loss": 0.7385,
+      "step": 3411
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2044401073432546e-05,
+      "loss": 0.8041,
+      "step": 3412
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2041961454013176e-05,
+      "loss": 0.7762,
+      "step": 3413
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2039521834593804e-05,
+      "loss": 0.7718,
+      "step": 3414
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2037082215174434e-05,
+      "loss": 0.7084,
+      "step": 3415
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2034642595755063e-05,
+      "loss": 0.8234,
+      "step": 3416
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2032202976335691e-05,
+      "loss": 0.7683,
+      "step": 3417
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2029763356916321e-05,
+      "loss": 0.7924,
+      "step": 3418
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2027323737496953e-05,
+      "loss": 0.8176,
+      "step": 3419
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.202488411807758e-05,
+      "loss": 0.7736,
+      "step": 3420
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.202244449865821e-05,
+      "loss": 0.7362,
+      "step": 3421
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.202000487923884e-05,
+      "loss": 0.8917,
+      "step": 3422
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2017565259819468e-05,
+      "loss": 0.7457,
+      "step": 3423
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2015125640400098e-05,
+      "loss": 0.7488,
+      "step": 3424
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2012686020980728e-05,
+      "loss": 0.7914,
+      "step": 3425
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2010246401561359e-05,
+      "loss": 0.8048,
+      "step": 3426
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2007806782141987e-05,
+      "loss": 0.7934,
+      "step": 3427
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2005367162722617e-05,
+      "loss": 0.838,
+      "step": 3428
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2002927543303247e-05,
+      "loss": 0.7792,
+      "step": 3429
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2000487923883875e-05,
+      "loss": 0.7756,
+      "step": 3430
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.1998048304464504e-05,
+      "loss": 0.8605,
+      "step": 3431
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.1995608685045134e-05,
+      "loss": 0.8761,
+      "step": 3432
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.1993169065625762e-05,
+      "loss": 0.7562,
+      "step": 3433
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.1990729446206392e-05,
+      "loss": 0.8412,
+      "step": 3434
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.1988289826787023e-05,
+      "loss": 0.7086,
+      "step": 3435
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1985850207367651e-05,
+      "loss": 0.7205,
+      "step": 3436
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1983410587948281e-05,
+      "loss": 0.7366,
+      "step": 3437
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1980970968528911e-05,
+      "loss": 0.7515,
+      "step": 3438
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1978531349109539e-05,
+      "loss": 0.7955,
+      "step": 3439
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1976091729690169e-05,
+      "loss": 0.7591,
+      "step": 3440
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1973652110270798e-05,
+      "loss": 0.7916,
+      "step": 3441
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1971212490851427e-05,
+      "loss": 0.7291,
+      "step": 3442
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1968772871432058e-05,
+      "loss": 0.7134,
+      "step": 3443
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1966333252012688e-05,
+      "loss": 0.8016,
+      "step": 3444
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1963893632593317e-05,
+      "loss": 0.7883,
+      "step": 3445
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1961454013173946e-05,
+      "loss": 0.8393,
+      "step": 3446
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1959014393754575e-05,
+      "loss": 0.76,
+      "step": 3447
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1956574774335205e-05,
+      "loss": 0.866,
+      "step": 3448
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1954135154915833e-05,
+      "loss": 0.781,
+      "step": 3449
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1951695535496464e-05,
+      "loss": 0.8069,
+      "step": 3450
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1949255916077094e-05,
+      "loss": 0.7708,
+      "step": 3451
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1946816296657722e-05,
+      "loss": 0.7873,
+      "step": 3452
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1944376677238352e-05,
+      "loss": 0.762,
+      "step": 3453
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1941937057818982e-05,
+      "loss": 0.7503,
+      "step": 3454
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.193949743839961e-05,
+      "loss": 0.812,
+      "step": 3455
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.193705781898024e-05,
+      "loss": 0.7719,
+      "step": 3456
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.193461819956087e-05,
+      "loss": 0.7945,
+      "step": 3457
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1932178580141497e-05,
+      "loss": 0.7751,
+      "step": 3458
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1929738960722129e-05,
+      "loss": 0.7816,
+      "step": 3459
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1927299341302759e-05,
+      "loss": 0.8021,
+      "step": 3460
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1924859721883387e-05,
+      "loss": 0.7499,
+      "step": 3461
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1922420102464016e-05,
+      "loss": 0.7916,
+      "step": 3462
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1919980483044646e-05,
+      "loss": 0.7998,
+      "step": 3463
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1917540863625276e-05,
+      "loss": 0.733,
+      "step": 3464
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1915101244205904e-05,
+      "loss": 0.8019,
+      "step": 3465
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1912661624786535e-05,
+      "loss": 0.8003,
+      "step": 3466
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1910222005367165e-05,
+      "loss": 0.721,
+      "step": 3467
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1907782385947793e-05,
+      "loss": 0.8166,
+      "step": 3468
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1905342766528423e-05,
+      "loss": 0.7632,
+      "step": 3469
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1902903147109053e-05,
+      "loss": 0.8391,
+      "step": 3470
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.190046352768968e-05,
+      "loss": 0.7676,
+      "step": 3471
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.189802390827031e-05,
+      "loss": 0.704,
+      "step": 3472
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.189558428885094e-05,
+      "loss": 0.7865,
+      "step": 3473
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1893144669431568e-05,
+      "loss": 0.8298,
+      "step": 3474
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.18907050500122e-05,
+      "loss": 0.771,
+      "step": 3475
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.188826543059283e-05,
+      "loss": 0.8032,
+      "step": 3476
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1885825811173457e-05,
+      "loss": 0.7901,
+      "step": 3477
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1883386191754087e-05,
+      "loss": 0.7707,
+      "step": 3478
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1880946572334717e-05,
+      "loss": 0.8075,
+      "step": 3479
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1878506952915345e-05,
+      "loss": 0.6856,
+      "step": 3480
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1876067333495975e-05,
+      "loss": 0.762,
+      "step": 3481
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1873627714076606e-05,
+      "loss": 0.7123,
+      "step": 3482
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1871188094657236e-05,
+      "loss": 0.7971,
+      "step": 3483
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1868748475237864e-05,
+      "loss": 0.7671,
+      "step": 3484
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1866308855818494e-05,
+      "loss": 0.7855,
+      "step": 3485
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1863869236399123e-05,
+      "loss": 0.7493,
+      "step": 3486
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1861429616979751e-05,
+      "loss": 0.7985,
+      "step": 3487
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1858989997560381e-05,
+      "loss": 0.7805,
+      "step": 3488
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1856550378141013e-05,
+      "loss": 0.841,
+      "step": 3489
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1854110758721639e-05,
+      "loss": 0.7674,
+      "step": 3490
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.185167113930227e-05,
+      "loss": 0.7249,
+      "step": 3491
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.18492315198829e-05,
+      "loss": 0.8068,
+      "step": 3492
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1846791900463528e-05,
+      "loss": 0.7455,
+      "step": 3493
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1844352281044158e-05,
+      "loss": 0.7724,
+      "step": 3494
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1841912661624788e-05,
+      "loss": 0.8025,
+      "step": 3495
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1839473042205416e-05,
+      "loss": 0.7858,
+      "step": 3496
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1837033422786046e-05,
+      "loss": 0.8315,
+      "step": 3497
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1834593803366677e-05,
+      "loss": 0.8248,
+      "step": 3498
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1832154183947305e-05,
+      "loss": 0.8264,
+      "step": 3499
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1829714564527935e-05,
+      "loss": 0.8128,
+      "step": 3500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1827274945108565e-05,
+      "loss": 0.7607,
+      "step": 3501
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1824835325689193e-05,
+      "loss": 0.7802,
+      "step": 3502
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1822395706269822e-05,
+      "loss": 0.7874,
+      "step": 3503
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1819956086850452e-05,
+      "loss": 0.7831,
+      "step": 3504
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1817516467431084e-05,
+      "loss": 0.7547,
+      "step": 3505
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.181507684801171e-05,
+      "loss": 0.8177,
+      "step": 3506
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1812637228592341e-05,
+      "loss": 0.7368,
+      "step": 3507
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1810197609172971e-05,
+      "loss": 0.8014,
+      "step": 3508
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1807757989753599e-05,
+      "loss": 0.7729,
+      "step": 3509
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1805318370334229e-05,
+      "loss": 0.7507,
+      "step": 3510
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1802878750914859e-05,
+      "loss": 0.8,
+      "step": 3511
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1800439131495487e-05,
+      "loss": 0.7461,
+      "step": 3512
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1797999512076116e-05,
+      "loss": 0.7962,
+      "step": 3513
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1795559892656748e-05,
+      "loss": 0.7705,
+      "step": 3514
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1793120273237376e-05,
+      "loss": 0.8248,
+      "step": 3515
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1790680653818006e-05,
+      "loss": 0.8065,
+      "step": 3516
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1788241034398635e-05,
+      "loss": 0.7526,
+      "step": 3517
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1785801414979263e-05,
+      "loss": 0.7667,
+      "step": 3518
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1783361795559893e-05,
+      "loss": 0.764,
+      "step": 3519
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1780922176140523e-05,
+      "loss": 0.7713,
+      "step": 3520
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1778482556721151e-05,
+      "loss": 0.73,
+      "step": 3521
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.177604293730178e-05,
+      "loss": 0.7492,
+      "step": 3522
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1773603317882412e-05,
+      "loss": 0.7626,
+      "step": 3523
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1771163698463042e-05,
+      "loss": 0.7869,
+      "step": 3524
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.176872407904367e-05,
+      "loss": 0.8008,
+      "step": 3525
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.17662844596243e-05,
+      "loss": 0.7947,
+      "step": 3526
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.176384484020493e-05,
+      "loss": 0.8188,
+      "step": 3527
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1761405220785557e-05,
+      "loss": 0.7747,
+      "step": 3528
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1758965601366187e-05,
+      "loss": 0.7439,
+      "step": 3529
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1756525981946819e-05,
+      "loss": 0.7445,
+      "step": 3530
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1754086362527447e-05,
+      "loss": 0.7669,
+      "step": 3531
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1751646743108076e-05,
+      "loss": 0.8176,
+      "step": 3532
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1749207123688706e-05,
+      "loss": 0.78,
+      "step": 3533
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1746767504269334e-05,
+      "loss": 0.7899,
+      "step": 3534
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1744327884849964e-05,
+      "loss": 0.8035,
+      "step": 3535
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1741888265430594e-05,
+      "loss": 0.7383,
+      "step": 3536
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1739448646011222e-05,
+      "loss": 0.7719,
+      "step": 3537
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1737009026591853e-05,
+      "loss": 0.7219,
+      "step": 3538
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1734569407172483e-05,
+      "loss": 0.8065,
+      "step": 3539
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1732129787753111e-05,
+      "loss": 0.7625,
+      "step": 3540
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.172969016833374e-05,
+      "loss": 0.7577,
+      "step": 3541
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.172725054891437e-05,
+      "loss": 0.7553,
+      "step": 3542
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1724810929495e-05,
+      "loss": 0.7234,
+      "step": 3543
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1722371310075628e-05,
+      "loss": 0.7774,
+      "step": 3544
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1719931690656258e-05,
+      "loss": 0.7674,
+      "step": 3545
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.171749207123689e-05,
+      "loss": 0.7765,
+      "step": 3546
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1715052451817518e-05,
+      "loss": 0.8789,
+      "step": 3547
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1712612832398147e-05,
+      "loss": 0.7791,
+      "step": 3548
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1710173212978777e-05,
+      "loss": 0.7767,
+      "step": 3549
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1707733593559405e-05,
+      "loss": 0.8096,
+      "step": 3550
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1705293974140035e-05,
+      "loss": 0.7995,
+      "step": 3551
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1702854354720665e-05,
+      "loss": 0.7753,
+      "step": 3552
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1700414735301293e-05,
+      "loss": 0.8028,
+      "step": 3553
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1697975115881924e-05,
+      "loss": 0.7329,
+      "step": 3554
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1695535496462554e-05,
+      "loss": 0.7647,
+      "step": 3555
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1693095877043182e-05,
+      "loss": 0.8205,
+      "step": 3556
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1690656257623812e-05,
+      "loss": 0.7695,
+      "step": 3557
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1688216638204441e-05,
+      "loss": 0.783,
+      "step": 3558
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.168577701878507e-05,
+      "loss": 0.8031,
+      "step": 3559
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1683337399365699e-05,
+      "loss": 0.7347,
+      "step": 3560
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1680897779946329e-05,
+      "loss": 0.7797,
+      "step": 3561
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.167845816052696e-05,
+      "loss": 0.8016,
+      "step": 3562
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1676018541107588e-05,
+      "loss": 0.7979,
+      "step": 3563
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1673578921688218e-05,
+      "loss": 0.8035,
+      "step": 3564
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1671139302268848e-05,
+      "loss": 0.7823,
+      "step": 3565
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1668699682849476e-05,
+      "loss": 0.7881,
+      "step": 3566
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1666260063430106e-05,
+      "loss": 0.7594,
+      "step": 3567
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1663820444010735e-05,
+      "loss": 0.8102,
+      "step": 3568
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1661380824591363e-05,
+      "loss": 0.7885,
+      "step": 3569
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1658941205171995e-05,
+      "loss": 0.7653,
+      "step": 3570
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1656501585752625e-05,
+      "loss": 0.7707,
+      "step": 3571
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1654061966333253e-05,
+      "loss": 0.7103,
+      "step": 3572
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1651622346913882e-05,
+      "loss": 0.7867,
+      "step": 3573
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1649182727494512e-05,
+      "loss": 0.7654,
+      "step": 3574
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.164674310807514e-05,
+      "loss": 0.808,
+      "step": 3575
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.164430348865577e-05,
+      "loss": 0.8111,
+      "step": 3576
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1641863869236401e-05,
+      "loss": 0.7844,
+      "step": 3577
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1639424249817028e-05,
+      "loss": 0.8334,
+      "step": 3578
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.163698463039766e-05,
+      "loss": 0.8267,
+      "step": 3579
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1634545010978289e-05,
+      "loss": 0.7293,
+      "step": 3580
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1632105391558919e-05,
+      "loss": 0.8526,
+      "step": 3581
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1629665772139547e-05,
+      "loss": 0.7327,
+      "step": 3582
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1627226152720176e-05,
+      "loss": 0.8158,
+      "step": 3583
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1624786533300806e-05,
+      "loss": 0.7775,
+      "step": 3584
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1622346913881434e-05,
+      "loss": 0.8304,
+      "step": 3585
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1619907294462066e-05,
+      "loss": 0.7385,
+      "step": 3586
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1617467675042695e-05,
+      "loss": 0.8075,
+      "step": 3587
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1615028055623323e-05,
+      "loss": 0.7527,
+      "step": 3588
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1612588436203953e-05,
+      "loss": 0.7647,
+      "step": 3589
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1610148816784583e-05,
+      "loss": 0.8567,
+      "step": 3590
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1607709197365211e-05,
+      "loss": 0.7385,
+      "step": 3591
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.160526957794584e-05,
+      "loss": 0.7749,
+      "step": 3592
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1602829958526472e-05,
+      "loss": 0.7573,
+      "step": 3593
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1600390339107099e-05,
+      "loss": 0.7667,
+      "step": 3594
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.159795071968773e-05,
+      "loss": 0.8436,
+      "step": 3595
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.159551110026836e-05,
+      "loss": 0.8145,
+      "step": 3596
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1593071480848988e-05,
+      "loss": 0.7807,
+      "step": 3597
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1590631861429618e-05,
+      "loss": 0.7303,
+      "step": 3598
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1588192242010247e-05,
+      "loss": 0.8017,
+      "step": 3599
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1585752622590877e-05,
+      "loss": 0.8246,
+      "step": 3600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1583313003171505e-05,
+      "loss": 0.7558,
+      "step": 3601
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1580873383752137e-05,
+      "loss": 0.7279,
+      "step": 3602
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1578433764332766e-05,
+      "loss": 0.8216,
+      "step": 3603
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1575994144913394e-05,
+      "loss": 0.8444,
+      "step": 3604
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1573554525494024e-05,
+      "loss": 0.8277,
+      "step": 3605
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1571114906074654e-05,
+      "loss": 0.7632,
+      "step": 3606
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1568675286655282e-05,
+      "loss": 0.7631,
+      "step": 3607
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1566235667235912e-05,
+      "loss": 0.7318,
+      "step": 3608
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1563796047816543e-05,
+      "loss": 0.768,
+      "step": 3609
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.156135642839717e-05,
+      "loss": 0.7862,
+      "step": 3610
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.15589168089778e-05,
+      "loss": 0.8648,
+      "step": 3611
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.155647718955843e-05,
+      "loss": 0.7716,
+      "step": 3612
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1554037570139059e-05,
+      "loss": 0.7575,
+      "step": 3613
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1551597950719688e-05,
+      "loss": 0.8368,
+      "step": 3614
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1549158331300318e-05,
+      "loss": 0.7635,
+      "step": 3615
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1546718711880946e-05,
+      "loss": 0.73,
+      "step": 3616
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1544279092461576e-05,
+      "loss": 0.786,
+      "step": 3617
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1541839473042207e-05,
+      "loss": 0.7854,
+      "step": 3618
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1539399853622837e-05,
+      "loss": 0.7598,
+      "step": 3619
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1536960234203465e-05,
+      "loss": 0.7692,
+      "step": 3620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1534520614784095e-05,
+      "loss": 0.7845,
+      "step": 3621
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1532080995364725e-05,
+      "loss": 0.7602,
+      "step": 3622
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1529641375945353e-05,
+      "loss": 0.7129,
+      "step": 3623
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1527201756525982e-05,
+      "loss": 0.8076,
+      "step": 3624
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1524762137106614e-05,
+      "loss": 0.774,
+      "step": 3625
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1522322517687242e-05,
+      "loss": 0.7854,
+      "step": 3626
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1519882898267872e-05,
+      "loss": 0.831,
+      "step": 3627
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1517443278848501e-05,
+      "loss": 0.796,
+      "step": 3628
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.151500365942913e-05,
+      "loss": 0.74,
+      "step": 3629
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.151256404000976e-05,
+      "loss": 0.7854,
+      "step": 3630
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1510124420590389e-05,
+      "loss": 0.746,
+      "step": 3631
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1507684801171017e-05,
+      "loss": 0.7572,
+      "step": 3632
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1505245181751647e-05,
+      "loss": 0.7294,
+      "step": 3633
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1502805562332278e-05,
+      "loss": 0.8279,
+      "step": 3634
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1500365942912906e-05,
+      "loss": 0.8119,
+      "step": 3635
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1497926323493536e-05,
+      "loss": 0.7627,
+      "step": 3636
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1495486704074166e-05,
+      "loss": 0.7869,
+      "step": 3637
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1493047084654794e-05,
+      "loss": 0.826,
+      "step": 3638
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1490607465235424e-05,
+      "loss": 0.7967,
+      "step": 3639
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1488167845816053e-05,
+      "loss": 0.8138,
+      "step": 3640
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1485728226396685e-05,
+      "loss": 0.8789,
+      "step": 3641
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1483288606977313e-05,
+      "loss": 0.7441,
+      "step": 3642
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1480848987557942e-05,
+      "loss": 0.8503,
+      "step": 3643
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1478409368138572e-05,
+      "loss": 0.7499,
+      "step": 3644
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.14759697487192e-05,
+      "loss": 0.7741,
+      "step": 3645
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.147353012929983e-05,
+      "loss": 0.7693,
+      "step": 3646
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.147109050988046e-05,
+      "loss": 0.8055,
+      "step": 3647
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1468650890461088e-05,
+      "loss": 0.7843,
+      "step": 3648
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1466211271041718e-05,
+      "loss": 0.8049,
+      "step": 3649
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1463771651622349e-05,
+      "loss": 0.7905,
+      "step": 3650
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1461332032202977e-05,
+      "loss": 0.7636,
+      "step": 3651
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1458892412783607e-05,
+      "loss": 0.7856,
+      "step": 3652
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1456452793364237e-05,
+      "loss": 0.7767,
+      "step": 3653
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1454013173944865e-05,
+      "loss": 0.7919,
+      "step": 3654
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1451573554525494e-05,
+      "loss": 0.8014,
+      "step": 3655
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1449133935106124e-05,
+      "loss": 0.8217,
+      "step": 3656
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1446694315686752e-05,
+      "loss": 0.7001,
+      "step": 3657
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1444254696267384e-05,
+      "loss": 0.7682,
+      "step": 3658
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1441815076848013e-05,
+      "loss": 0.7775,
+      "step": 3659
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1439375457428643e-05,
+      "loss": 0.7358,
+      "step": 3660
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1436935838009271e-05,
+      "loss": 0.7962,
+      "step": 3661
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1434496218589901e-05,
+      "loss": 0.8551,
+      "step": 3662
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.143205659917053e-05,
+      "loss": 0.7368,
+      "step": 3663
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1429616979751159e-05,
+      "loss": 0.8591,
+      "step": 3664
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.142717736033179e-05,
+      "loss": 0.8064,
+      "step": 3665
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.142473774091242e-05,
+      "loss": 0.8385,
+      "step": 3666
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1422298121493048e-05,
+      "loss": 0.7103,
+      "step": 3667
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1419858502073678e-05,
+      "loss": 0.8148,
+      "step": 3668
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1417418882654307e-05,
+      "loss": 0.7715,
+      "step": 3669
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1414979263234935e-05,
+      "loss": 0.7537,
+      "step": 3670
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1412539643815565e-05,
+      "loss": 0.792,
+      "step": 3671
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1410100024396195e-05,
+      "loss": 0.8244,
+      "step": 3672
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1407660404976823e-05,
+      "loss": 0.7507,
+      "step": 3673
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1405220785557454e-05,
+      "loss": 0.8218,
+      "step": 3674
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1402781166138084e-05,
+      "loss": 0.8135,
+      "step": 3675
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1400341546718712e-05,
+      "loss": 0.847,
+      "step": 3676
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1397901927299342e-05,
+      "loss": 0.8135,
+      "step": 3677
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1395462307879972e-05,
+      "loss": 0.7491,
+      "step": 3678
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1393022688460601e-05,
+      "loss": 0.7667,
+      "step": 3679
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.139058306904123e-05,
+      "loss": 0.8115,
+      "step": 3680
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1388143449621861e-05,
+      "loss": 0.7872,
+      "step": 3681
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.138570383020249e-05,
+      "loss": 0.7645,
+      "step": 3682
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1383264210783119e-05,
+      "loss": 0.7755,
+      "step": 3683
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1380824591363748e-05,
+      "loss": 0.752,
+      "step": 3684
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1378384971944378e-05,
+      "loss": 0.8431,
+      "step": 3685
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1375945352525006e-05,
+      "loss": 0.7662,
+      "step": 3686
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1373505733105636e-05,
+      "loss": 0.8444,
+      "step": 3687
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1371066113686266e-05,
+      "loss": 0.7287,
+      "step": 3688
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1368626494266894e-05,
+      "loss": 0.7847,
+      "step": 3689
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1366186874847525e-05,
+      "loss": 0.7824,
+      "step": 3690
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1363747255428155e-05,
+      "loss": 0.8116,
+      "step": 3691
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1361307636008783e-05,
+      "loss": 0.7302,
+      "step": 3692
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1358868016589413e-05,
+      "loss": 0.7973,
+      "step": 3693
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1356428397170043e-05,
+      "loss": 0.7719,
+      "step": 3694
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.135398877775067e-05,
+      "loss": 0.7832,
+      "step": 3695
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.13515491583313e-05,
+      "loss": 0.7565,
+      "step": 3696
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1349109538911932e-05,
+      "loss": 0.6976,
+      "step": 3697
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1346669919492561e-05,
+      "loss": 0.7631,
+      "step": 3698
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.134423030007319e-05,
+      "loss": 0.7707,
+      "step": 3699
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.134179068065382e-05,
+      "loss": 0.7701,
+      "step": 3700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1339351061234449e-05,
+      "loss": 0.7638,
+      "step": 3701
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1336911441815077e-05,
+      "loss": 0.8047,
+      "step": 3702
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1334471822395707e-05,
+      "loss": 0.8129,
+      "step": 3703
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1332032202976338e-05,
+      "loss": 0.7611,
+      "step": 3704
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1329592583556965e-05,
+      "loss": 0.7681,
+      "step": 3705
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1327152964137596e-05,
+      "loss": 0.7136,
+      "step": 3706
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1324713344718226e-05,
+      "loss": 0.7704,
+      "step": 3707
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1322273725298854e-05,
+      "loss": 0.8082,
+      "step": 3708
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1319834105879484e-05,
+      "loss": 0.7006,
+      "step": 3709
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1317394486460113e-05,
+      "loss": 0.8046,
+      "step": 3710
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1314954867040741e-05,
+      "loss": 0.7494,
+      "step": 3711
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1312515247621371e-05,
+      "loss": 0.7732,
+      "step": 3712
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1310075628202003e-05,
+      "loss": 0.7811,
+      "step": 3713
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.130763600878263e-05,
+      "loss": 0.8529,
+      "step": 3714
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.130519638936326e-05,
+      "loss": 0.7344,
+      "step": 3715
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.130275676994389e-05,
+      "loss": 0.7699,
+      "step": 3716
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.130031715052452e-05,
+      "loss": 0.7041,
+      "step": 3717
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1297877531105148e-05,
+      "loss": 0.7744,
+      "step": 3718
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1295437911685778e-05,
+      "loss": 0.7086,
+      "step": 3719
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1292998292266409e-05,
+      "loss": 0.7662,
+      "step": 3720
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1290558672847035e-05,
+      "loss": 0.7466,
+      "step": 3721
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1288119053427667e-05,
+      "loss": 0.7513,
+      "step": 3722
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1285679434008297e-05,
+      "loss": 0.8146,
+      "step": 3723
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1283239814588925e-05,
+      "loss": 0.7353,
+      "step": 3724
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1280800195169554e-05,
+      "loss": 0.8131,
+      "step": 3725
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1278360575750184e-05,
+      "loss": 0.8346,
+      "step": 3726
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1275920956330812e-05,
+      "loss": 0.805,
+      "step": 3727
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1273481336911442e-05,
+      "loss": 0.7808,
+      "step": 3728
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1271041717492073e-05,
+      "loss": 0.7638,
+      "step": 3729
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1268602098072701e-05,
+      "loss": 0.8022,
+      "step": 3730
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1266162478653331e-05,
+      "loss": 0.8162,
+      "step": 3731
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1263722859233961e-05,
+      "loss": 0.794,
+      "step": 3732
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1261283239814589e-05,
+      "loss": 0.7552,
+      "step": 3733
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1258843620395219e-05,
+      "loss": 0.7692,
+      "step": 3734
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1256404000975848e-05,
+      "loss": 0.8108,
+      "step": 3735
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.125396438155648e-05,
+      "loss": 0.8039,
+      "step": 3736
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1251524762137106e-05,
+      "loss": 0.802,
+      "step": 3737
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1249085142717738e-05,
+      "loss": 0.7835,
+      "step": 3738
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1246645523298367e-05,
+      "loss": 0.7951,
+      "step": 3739
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1244205903878996e-05,
+      "loss": 0.8631,
+      "step": 3740
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1241766284459625e-05,
+      "loss": 0.7817,
+      "step": 3741
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1239326665040255e-05,
+      "loss": 0.8077,
+      "step": 3742
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1236887045620883e-05,
+      "loss": 0.769,
+      "step": 3743
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1234447426201513e-05,
+      "loss": 0.7847,
+      "step": 3744
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1232007806782144e-05,
+      "loss": 0.8164,
+      "step": 3745
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1229568187362772e-05,
+      "loss": 0.7539,
+      "step": 3746
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1227128567943402e-05,
+      "loss": 0.8123,
+      "step": 3747
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1224688948524032e-05,
+      "loss": 0.7968,
+      "step": 3748
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.122224932910466e-05,
+      "loss": 0.7771,
+      "step": 3749
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.121980970968529e-05,
+      "loss": 0.7267,
+      "step": 3750
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.121737009026592e-05,
+      "loss": 0.7904,
+      "step": 3751
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1214930470846547e-05,
+      "loss": 0.8157,
+      "step": 3752
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1212490851427179e-05,
+      "loss": 0.7827,
+      "step": 3753
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1210051232007809e-05,
+      "loss": 0.837,
+      "step": 3754
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1207611612588438e-05,
+      "loss": 0.7661,
+      "step": 3755
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1205171993169066e-05,
+      "loss": 0.8154,
+      "step": 3756
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1202732373749696e-05,
+      "loss": 0.803,
+      "step": 3757
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1200292754330326e-05,
+      "loss": 0.7552,
+      "step": 3758
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1197853134910954e-05,
+      "loss": 0.7381,
+      "step": 3759
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1195413515491584e-05,
+      "loss": 0.7808,
+      "step": 3760
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1192973896072215e-05,
+      "loss": 0.7554,
+      "step": 3761
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1190534276652843e-05,
+      "loss": 0.7668,
+      "step": 3762
+    },
+    {
+      "epoch": 1.78,
+      "eval_loss": 0.9531322717666626,
+      "eval_runtime": 205.7806,
+      "eval_samples_per_second": 81.635,
+      "eval_steps_per_second": 10.205,
+      "step": 3762
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1188094657233473e-05,
+      "loss": 0.8041,
+      "step": 3763
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1185655037814103e-05,
+      "loss": 0.794,
+      "step": 3764
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.118321541839473e-05,
+      "loss": 0.7808,
+      "step": 3765
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.118077579897536e-05,
+      "loss": 0.7843,
+      "step": 3766
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.117833617955599e-05,
+      "loss": 0.772,
+      "step": 3767
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1175896560136618e-05,
+      "loss": 0.7825,
+      "step": 3768
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.117345694071725e-05,
+      "loss": 0.744,
+      "step": 3769
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.117101732129788e-05,
+      "loss": 0.7473,
+      "step": 3770
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1168577701878507e-05,
+      "loss": 0.8212,
+      "step": 3771
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1166138082459137e-05,
+      "loss": 0.7681,
+      "step": 3772
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1163698463039767e-05,
+      "loss": 0.7507,
+      "step": 3773
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1161258843620395e-05,
+      "loss": 0.7444,
+      "step": 3774
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1158819224201025e-05,
+      "loss": 0.7302,
+      "step": 3775
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1156379604781654e-05,
+      "loss": 0.7508,
+      "step": 3776
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1153939985362286e-05,
+      "loss": 0.7665,
+      "step": 3777
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1151500365942914e-05,
+      "loss": 0.8158,
+      "step": 3778
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1149060746523544e-05,
+      "loss": 0.8033,
+      "step": 3779
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1146621127104173e-05,
+      "loss": 0.7965,
+      "step": 3780
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1144181507684801e-05,
+      "loss": 0.7448,
+      "step": 3781
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1141741888265431e-05,
+      "loss": 0.7706,
+      "step": 3782
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1139302268846061e-05,
+      "loss": 0.8075,
+      "step": 3783
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1136862649426689e-05,
+      "loss": 0.7841,
+      "step": 3784
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.113442303000732e-05,
+      "loss": 0.7711,
+      "step": 3785
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.113198341058795e-05,
+      "loss": 0.7421,
+      "step": 3786
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1129543791168578e-05,
+      "loss": 0.7822,
+      "step": 3787
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1127104171749208e-05,
+      "loss": 0.8186,
+      "step": 3788
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1124664552329838e-05,
+      "loss": 0.7669,
+      "step": 3789
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1122224932910466e-05,
+      "loss": 0.8241,
+      "step": 3790
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1119785313491096e-05,
+      "loss": 0.7655,
+      "step": 3791
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1117345694071727e-05,
+      "loss": 0.7679,
+      "step": 3792
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1114906074652353e-05,
+      "loss": 0.7892,
+      "step": 3793
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1112466455232985e-05,
+      "loss": 0.8212,
+      "step": 3794
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1110026835813615e-05,
+      "loss": 0.75,
+      "step": 3795
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1107587216394244e-05,
+      "loss": 0.8274,
+      "step": 3796
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1105147596974872e-05,
+      "loss": 0.7847,
+      "step": 3797
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1102707977555502e-05,
+      "loss": 0.7959,
+      "step": 3798
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1100268358136132e-05,
+      "loss": 0.77,
+      "step": 3799
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.109782873871676e-05,
+      "loss": 0.7725,
+      "step": 3800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1095389119297391e-05,
+      "loss": 0.8035,
+      "step": 3801
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1092949499878021e-05,
+      "loss": 0.7596,
+      "step": 3802
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1090509880458649e-05,
+      "loss": 0.7611,
+      "step": 3803
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1088070261039279e-05,
+      "loss": 0.7971,
+      "step": 3804
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1085630641619909e-05,
+      "loss": 0.8162,
+      "step": 3805
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1083191022200537e-05,
+      "loss": 0.7843,
+      "step": 3806
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1080751402781166e-05,
+      "loss": 0.8018,
+      "step": 3807
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1078311783361798e-05,
+      "loss": 0.7933,
+      "step": 3808
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1075872163942424e-05,
+      "loss": 0.7198,
+      "step": 3809
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1073432544523056e-05,
+      "loss": 0.7554,
+      "step": 3810
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1070992925103685e-05,
+      "loss": 0.8005,
+      "step": 3811
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1068553305684313e-05,
+      "loss": 0.8135,
+      "step": 3812
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1066113686264943e-05,
+      "loss": 0.8046,
+      "step": 3813
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1063674066845573e-05,
+      "loss": 0.7564,
+      "step": 3814
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1061234447426203e-05,
+      "loss": 0.8222,
+      "step": 3815
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.105879482800683e-05,
+      "loss": 0.7822,
+      "step": 3816
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1056355208587462e-05,
+      "loss": 0.808,
+      "step": 3817
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1053915589168092e-05,
+      "loss": 0.7348,
+      "step": 3818
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.105147596974872e-05,
+      "loss": 0.709,
+      "step": 3819
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.104903635032935e-05,
+      "loss": 0.768,
+      "step": 3820
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.104659673090998e-05,
+      "loss": 0.8019,
+      "step": 3821
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1044157111490607e-05,
+      "loss": 0.7346,
+      "step": 3822
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1041717492071237e-05,
+      "loss": 0.7712,
+      "step": 3823
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1039277872651869e-05,
+      "loss": 0.8056,
+      "step": 3824
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1036838253232495e-05,
+      "loss": 0.8503,
+      "step": 3825
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1034398633813126e-05,
+      "loss": 0.7586,
+      "step": 3826
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1031959014393756e-05,
+      "loss": 0.7938,
+      "step": 3827
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1029519394974384e-05,
+      "loss": 0.767,
+      "step": 3828
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1027079775555014e-05,
+      "loss": 0.7749,
+      "step": 3829
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1024640156135644e-05,
+      "loss": 0.7486,
+      "step": 3830
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1022200536716272e-05,
+      "loss": 0.7394,
+      "step": 3831
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1019760917296901e-05,
+      "loss": 0.741,
+      "step": 3832
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1017321297877533e-05,
+      "loss": 0.834,
+      "step": 3833
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1014881678458163e-05,
+      "loss": 0.7941,
+      "step": 3834
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.101244205903879e-05,
+      "loss": 0.7755,
+      "step": 3835
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.101000243961942e-05,
+      "loss": 0.8439,
+      "step": 3836
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.100756282020005e-05,
+      "loss": 0.7827,
+      "step": 3837
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1005123200780678e-05,
+      "loss": 0.8179,
+      "step": 3838
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1002683581361308e-05,
+      "loss": 0.7708,
+      "step": 3839
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.100024396194194e-05,
+      "loss": 0.7774,
+      "step": 3840
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0997804342522568e-05,
+      "loss": 0.7698,
+      "step": 3841
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0995364723103197e-05,
+      "loss": 0.7542,
+      "step": 3842
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0992925103683827e-05,
+      "loss": 0.7588,
+      "step": 3843
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0990485484264455e-05,
+      "loss": 0.8115,
+      "step": 3844
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0988045864845085e-05,
+      "loss": 0.7593,
+      "step": 3845
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0985606245425715e-05,
+      "loss": 0.7552,
+      "step": 3846
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0983166626006343e-05,
+      "loss": 0.8004,
+      "step": 3847
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0980727006586972e-05,
+      "loss": 0.7647,
+      "step": 3848
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0978287387167604e-05,
+      "loss": 0.8039,
+      "step": 3849
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0975847767748232e-05,
+      "loss": 0.7689,
+      "step": 3850
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0973408148328862e-05,
+      "loss": 0.7956,
+      "step": 3851
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0970968528909491e-05,
+      "loss": 0.7871,
+      "step": 3852
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0968528909490121e-05,
+      "loss": 0.7514,
+      "step": 3853
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0966089290070749e-05,
+      "loss": 0.7606,
+      "step": 3854
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0963649670651379e-05,
+      "loss": 0.7985,
+      "step": 3855
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.096121005123201e-05,
+      "loss": 0.751,
+      "step": 3856
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0958770431812638e-05,
+      "loss": 0.7922,
+      "step": 3857
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0956330812393268e-05,
+      "loss": 0.7874,
+      "step": 3858
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0953891192973898e-05,
+      "loss": 0.8349,
+      "step": 3859
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0951451573554526e-05,
+      "loss": 0.7708,
+      "step": 3860
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0949011954135156e-05,
+      "loss": 0.7576,
+      "step": 3861
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0946572334715785e-05,
+      "loss": 0.7272,
+      "step": 3862
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0944132715296413e-05,
+      "loss": 0.8312,
+      "step": 3863
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0941693095877043e-05,
+      "loss": 0.7827,
+      "step": 3864
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0939253476457675e-05,
+      "loss": 0.7705,
+      "step": 3865
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0936813857038303e-05,
+      "loss": 0.7065,
+      "step": 3866
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0934374237618932e-05,
+      "loss": 0.8148,
+      "step": 3867
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0931934618199562e-05,
+      "loss": 0.7809,
+      "step": 3868
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.092949499878019e-05,
+      "loss": 0.7046,
+      "step": 3869
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.092705537936082e-05,
+      "loss": 0.7861,
+      "step": 3870
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.092461575994145e-05,
+      "loss": 0.7658,
+      "step": 3871
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0922176140522081e-05,
+      "loss": 0.8025,
+      "step": 3872
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.091973652110271e-05,
+      "loss": 0.8051,
+      "step": 3873
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0917296901683339e-05,
+      "loss": 0.8011,
+      "step": 3874
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0914857282263969e-05,
+      "loss": 0.7922,
+      "step": 3875
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0912417662844597e-05,
+      "loss": 0.7968,
+      "step": 3876
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0909978043425226e-05,
+      "loss": 0.7716,
+      "step": 3877
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0907538424005856e-05,
+      "loss": 0.7799,
+      "step": 3878
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0905098804586484e-05,
+      "loss": 0.8007,
+      "step": 3879
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0902659185167116e-05,
+      "loss": 0.7577,
+      "step": 3880
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0900219565747745e-05,
+      "loss": 0.7339,
+      "step": 3881
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0897779946328373e-05,
+      "loss": 0.7774,
+      "step": 3882
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0895340326909003e-05,
+      "loss": 0.7798,
+      "step": 3883
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0892900707489633e-05,
+      "loss": 0.7564,
+      "step": 3884
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0890461088070261e-05,
+      "loss": 0.7825,
+      "step": 3885
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.088802146865089e-05,
+      "loss": 0.7827,
+      "step": 3886
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.088558184923152e-05,
+      "loss": 0.7126,
+      "step": 3887
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0883142229812149e-05,
+      "loss": 0.7575,
+      "step": 3888
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.088070261039278e-05,
+      "loss": 0.8231,
+      "step": 3889
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.087826299097341e-05,
+      "loss": 0.8055,
+      "step": 3890
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.087582337155404e-05,
+      "loss": 0.7472,
+      "step": 3891
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0873383752134668e-05,
+      "loss": 0.7487,
+      "step": 3892
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0870944132715297e-05,
+      "loss": 0.7668,
+      "step": 3893
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0868504513295927e-05,
+      "loss": 0.7311,
+      "step": 3894
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0866064893876555e-05,
+      "loss": 0.8611,
+      "step": 3895
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0863625274457187e-05,
+      "loss": 0.8222,
+      "step": 3896
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0861185655037816e-05,
+      "loss": 0.7216,
+      "step": 3897
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0858746035618444e-05,
+      "loss": 0.7558,
+      "step": 3898
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0856306416199074e-05,
+      "loss": 0.7744,
+      "step": 3899
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0853866796779704e-05,
+      "loss": 0.7828,
+      "step": 3900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0851427177360332e-05,
+      "loss": 0.7606,
+      "step": 3901
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0848987557940962e-05,
+      "loss": 0.7857,
+      "step": 3902
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0846547938521591e-05,
+      "loss": 0.8067,
+      "step": 3903
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.084410831910222e-05,
+      "loss": 0.7488,
+      "step": 3904
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.084166869968285e-05,
+      "loss": 0.7687,
+      "step": 3905
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.083922908026348e-05,
+      "loss": 0.8103,
+      "step": 3906
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0836789460844109e-05,
+      "loss": 0.7363,
+      "step": 3907
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0834349841424738e-05,
+      "loss": 0.6865,
+      "step": 3908
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0831910222005368e-05,
+      "loss": 0.7715,
+      "step": 3909
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0829470602585996e-05,
+      "loss": 0.8487,
+      "step": 3910
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0827030983166626e-05,
+      "loss": 0.7587,
+      "step": 3911
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0824591363747257e-05,
+      "loss": 0.7787,
+      "step": 3912
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0822151744327887e-05,
+      "loss": 0.7568,
+      "step": 3913
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0819712124908515e-05,
+      "loss": 0.8048,
+      "step": 3914
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0817272505489145e-05,
+      "loss": 0.8395,
+      "step": 3915
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0814832886069775e-05,
+      "loss": 0.7275,
+      "step": 3916
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0812393266650403e-05,
+      "loss": 0.685,
+      "step": 3917
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0809953647231032e-05,
+      "loss": 0.7919,
+      "step": 3918
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0807514027811664e-05,
+      "loss": 0.8115,
+      "step": 3919
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.080507440839229e-05,
+      "loss": 0.835,
+      "step": 3920
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0802634788972922e-05,
+      "loss": 0.7761,
+      "step": 3921
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0800195169553551e-05,
+      "loss": 0.8131,
+      "step": 3922
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.079775555013418e-05,
+      "loss": 0.7675,
+      "step": 3923
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.079531593071481e-05,
+      "loss": 0.7573,
+      "step": 3924
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0792876311295439e-05,
+      "loss": 0.785,
+      "step": 3925
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0790436691876067e-05,
+      "loss": 0.7086,
+      "step": 3926
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0787997072456697e-05,
+      "loss": 0.7841,
+      "step": 3927
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0785557453037328e-05,
+      "loss": 0.8524,
+      "step": 3928
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0783117833617956e-05,
+      "loss": 0.8025,
+      "step": 3929
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0780678214198586e-05,
+      "loss": 0.7453,
+      "step": 3930
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0778238594779216e-05,
+      "loss": 0.7284,
+      "step": 3931
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0775798975359845e-05,
+      "loss": 0.8063,
+      "step": 3932
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0773359355940473e-05,
+      "loss": 0.7418,
+      "step": 3933
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0770919736521103e-05,
+      "loss": 0.7314,
+      "step": 3934
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0768480117101735e-05,
+      "loss": 0.7572,
+      "step": 3935
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0766040497682361e-05,
+      "loss": 0.7872,
+      "step": 3936
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0763600878262992e-05,
+      "loss": 0.783,
+      "step": 3937
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0761161258843622e-05,
+      "loss": 0.7446,
+      "step": 3938
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.075872163942425e-05,
+      "loss": 0.7728,
+      "step": 3939
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.075628202000488e-05,
+      "loss": 0.7458,
+      "step": 3940
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.075384240058551e-05,
+      "loss": 0.7312,
+      "step": 3941
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0751402781166138e-05,
+      "loss": 0.7118,
+      "step": 3942
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0748963161746768e-05,
+      "loss": 0.7576,
+      "step": 3943
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0746523542327399e-05,
+      "loss": 0.8231,
+      "step": 3944
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0744083922908027e-05,
+      "loss": 0.7675,
+      "step": 3945
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0741644303488657e-05,
+      "loss": 0.8067,
+      "step": 3946
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0739204684069287e-05,
+      "loss": 0.7498,
+      "step": 3947
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0736765064649915e-05,
+      "loss": 0.7548,
+      "step": 3948
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0734325445230544e-05,
+      "loss": 0.7822,
+      "step": 3949
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0731885825811174e-05,
+      "loss": 0.7746,
+      "step": 3950
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0729446206391806e-05,
+      "loss": 0.7838,
+      "step": 3951
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0727006586972432e-05,
+      "loss": 0.7251,
+      "step": 3952
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0724566967553063e-05,
+      "loss": 0.7609,
+      "step": 3953
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0722127348133693e-05,
+      "loss": 0.7191,
+      "step": 3954
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0719687728714321e-05,
+      "loss": 0.7844,
+      "step": 3955
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.071724810929495e-05,
+      "loss": 0.7887,
+      "step": 3956
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.071480848987558e-05,
+      "loss": 0.7586,
+      "step": 3957
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0712368870456209e-05,
+      "loss": 0.8464,
+      "step": 3958
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0709929251036838e-05,
+      "loss": 0.7457,
+      "step": 3959
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.070748963161747e-05,
+      "loss": 0.7768,
+      "step": 3960
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0705050012198098e-05,
+      "loss": 0.7944,
+      "step": 3961
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0702610392778728e-05,
+      "loss": 0.6991,
+      "step": 3962
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0700170773359357e-05,
+      "loss": 0.7324,
+      "step": 3963
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0697731153939985e-05,
+      "loss": 0.7415,
+      "step": 3964
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0695291534520615e-05,
+      "loss": 0.7169,
+      "step": 3965
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0692851915101245e-05,
+      "loss": 0.794,
+      "step": 3966
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0690412295681873e-05,
+      "loss": 0.7381,
+      "step": 3967
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0687972676262504e-05,
+      "loss": 0.7895,
+      "step": 3968
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0685533056843134e-05,
+      "loss": 0.7939,
+      "step": 3969
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0683093437423764e-05,
+      "loss": 0.7846,
+      "step": 3970
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0680653818004392e-05,
+      "loss": 0.7927,
+      "step": 3971
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0678214198585022e-05,
+      "loss": 0.7652,
+      "step": 3972
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0675774579165651e-05,
+      "loss": 0.7227,
+      "step": 3973
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.067333495974628e-05,
+      "loss": 0.8565,
+      "step": 3974
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.067089534032691e-05,
+      "loss": 0.7607,
+      "step": 3975
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.066845572090754e-05,
+      "loss": 0.8054,
+      "step": 3976
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0666016101488169e-05,
+      "loss": 0.8229,
+      "step": 3977
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0663576482068798e-05,
+      "loss": 0.782,
+      "step": 3978
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0661136862649428e-05,
+      "loss": 0.7363,
+      "step": 3979
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0658697243230056e-05,
+      "loss": 0.7957,
+      "step": 3980
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0656257623810686e-05,
+      "loss": 0.7463,
+      "step": 3981
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0653818004391316e-05,
+      "loss": 0.7315,
+      "step": 3982
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0651378384971944e-05,
+      "loss": 0.8033,
+      "step": 3983
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0648938765552575e-05,
+      "loss": 0.8018,
+      "step": 3984
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0646499146133205e-05,
+      "loss": 0.764,
+      "step": 3985
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0644059526713833e-05,
+      "loss": 0.8169,
+      "step": 3986
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0641619907294463e-05,
+      "loss": 0.7924,
+      "step": 3987
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0639180287875092e-05,
+      "loss": 0.7938,
+      "step": 3988
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0636740668455722e-05,
+      "loss": 0.7351,
+      "step": 3989
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.063430104903635e-05,
+      "loss": 0.7568,
+      "step": 3990
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.063186142961698e-05,
+      "loss": 0.7729,
+      "step": 3991
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0629421810197611e-05,
+      "loss": 0.778,
+      "step": 3992
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.062698219077824e-05,
+      "loss": 0.7513,
+      "step": 3993
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.062454257135887e-05,
+      "loss": 0.7529,
+      "step": 3994
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0622102951939499e-05,
+      "loss": 0.7991,
+      "step": 3995
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0619663332520127e-05,
+      "loss": 0.7871,
+      "step": 3996
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0617223713100757e-05,
+      "loss": 0.8996,
+      "step": 3997
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0614784093681387e-05,
+      "loss": 0.8064,
+      "step": 3998
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0612344474262015e-05,
+      "loss": 0.832,
+      "step": 3999
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0609904854842646e-05,
+      "loss": 0.7605,
+      "step": 4000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0607465235423276e-05,
+      "loss": 0.656,
+      "step": 4001
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0605025616003904e-05,
+      "loss": 0.742,
+      "step": 4002
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0602585996584534e-05,
+      "loss": 0.7596,
+      "step": 4003
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0600146377165163e-05,
+      "loss": 0.7847,
+      "step": 4004
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0597706757745791e-05,
+      "loss": 0.7435,
+      "step": 4005
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0595267138326421e-05,
+      "loss": 0.7864,
+      "step": 4006
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0592827518907053e-05,
+      "loss": 0.7542,
+      "step": 4007
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0590387899487682e-05,
+      "loss": 0.7559,
+      "step": 4008
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.058794828006831e-05,
+      "loss": 0.7282,
+      "step": 4009
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.058550866064894e-05,
+      "loss": 0.7662,
+      "step": 4010
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.058306904122957e-05,
+      "loss": 0.8185,
+      "step": 4011
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0580629421810198e-05,
+      "loss": 0.7936,
+      "step": 4012
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0578189802390828e-05,
+      "loss": 0.781,
+      "step": 4013
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0575750182971457e-05,
+      "loss": 0.783,
+      "step": 4014
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0573310563552085e-05,
+      "loss": 0.7559,
+      "step": 4015
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0570870944132717e-05,
+      "loss": 0.7334,
+      "step": 4016
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0568431324713347e-05,
+      "loss": 0.7615,
+      "step": 4017
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0565991705293975e-05,
+      "loss": 0.714,
+      "step": 4018
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0563552085874604e-05,
+      "loss": 0.7361,
+      "step": 4019
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0561112466455234e-05,
+      "loss": 0.7545,
+      "step": 4020
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0558672847035862e-05,
+      "loss": 0.7376,
+      "step": 4021
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0556233227616492e-05,
+      "loss": 0.847,
+      "step": 4022
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0553793608197123e-05,
+      "loss": 0.7875,
+      "step": 4023
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.055135398877775e-05,
+      "loss": 0.7465,
+      "step": 4024
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0548914369358381e-05,
+      "loss": 0.8315,
+      "step": 4025
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0546474749939011e-05,
+      "loss": 0.7886,
+      "step": 4026
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.054403513051964e-05,
+      "loss": 0.8302,
+      "step": 4027
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0541595511100269e-05,
+      "loss": 0.8382,
+      "step": 4028
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0539155891680898e-05,
+      "loss": 0.7712,
+      "step": 4029
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0536716272261528e-05,
+      "loss": 0.7652,
+      "step": 4030
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0534276652842156e-05,
+      "loss": 0.7536,
+      "step": 4031
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0531837033422788e-05,
+      "loss": 0.7678,
+      "step": 4032
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0529397414003417e-05,
+      "loss": 0.7483,
+      "step": 4033
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0526957794584045e-05,
+      "loss": 0.7624,
+      "step": 4034
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0524518175164675e-05,
+      "loss": 0.7483,
+      "step": 4035
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0522078555745305e-05,
+      "loss": 0.8092,
+      "step": 4036
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0519638936325933e-05,
+      "loss": 0.751,
+      "step": 4037
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0517199316906563e-05,
+      "loss": 0.7957,
+      "step": 4038
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0514759697487194e-05,
+      "loss": 0.7996,
+      "step": 4039
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.051232007806782e-05,
+      "loss": 0.7666,
+      "step": 4040
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0509880458648452e-05,
+      "loss": 0.7335,
+      "step": 4041
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0507440839229082e-05,
+      "loss": 0.8035,
+      "step": 4042
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.050500121980971e-05,
+      "loss": 0.7661,
+      "step": 4043
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.050256160039034e-05,
+      "loss": 0.7639,
+      "step": 4044
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.050012198097097e-05,
+      "loss": 0.7876,
+      "step": 4045
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0497682361551597e-05,
+      "loss": 0.7725,
+      "step": 4046
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0495242742132227e-05,
+      "loss": 0.7491,
+      "step": 4047
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0492803122712859e-05,
+      "loss": 0.7916,
+      "step": 4048
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0490363503293488e-05,
+      "loss": 0.7814,
+      "step": 4049
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0487923883874116e-05,
+      "loss": 0.8741,
+      "step": 4050
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0485484264454746e-05,
+      "loss": 0.8399,
+      "step": 4051
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0483044645035376e-05,
+      "loss": 0.7846,
+      "step": 4052
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0480605025616004e-05,
+      "loss": 0.8127,
+      "step": 4053
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0478165406196634e-05,
+      "loss": 0.7963,
+      "step": 4054
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0475725786777265e-05,
+      "loss": 0.7399,
+      "step": 4055
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0473286167357893e-05,
+      "loss": 0.7437,
+      "step": 4056
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0470846547938523e-05,
+      "loss": 0.7871,
+      "step": 4057
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0468406928519153e-05,
+      "loss": 0.7416,
+      "step": 4058
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.046596730909978e-05,
+      "loss": 0.7252,
+      "step": 4059
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.046352768968041e-05,
+      "loss": 0.7587,
+      "step": 4060
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.046108807026104e-05,
+      "loss": 0.74,
+      "step": 4061
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0458648450841668e-05,
+      "loss": 0.8814,
+      "step": 4062
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0456208831422298e-05,
+      "loss": 0.7481,
+      "step": 4063
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.045376921200293e-05,
+      "loss": 0.763,
+      "step": 4064
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0451329592583557e-05,
+      "loss": 0.7479,
+      "step": 4065
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0448889973164187e-05,
+      "loss": 0.7455,
+      "step": 4066
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0446450353744817e-05,
+      "loss": 0.7655,
+      "step": 4067
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0444010734325447e-05,
+      "loss": 0.7836,
+      "step": 4068
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0441571114906075e-05,
+      "loss": 0.7476,
+      "step": 4069
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0439131495486704e-05,
+      "loss": 0.736,
+      "step": 4070
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0436691876067336e-05,
+      "loss": 0.758,
+      "step": 4071
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0434252256647964e-05,
+      "loss": 0.7326,
+      "step": 4072
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0431812637228594e-05,
+      "loss": 0.7982,
+      "step": 4073
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0429373017809223e-05,
+      "loss": 0.8306,
+      "step": 4074
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0426933398389851e-05,
+      "loss": 0.7123,
+      "step": 4075
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0424493778970481e-05,
+      "loss": 0.765,
+      "step": 4076
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0422054159551111e-05,
+      "loss": 0.7563,
+      "step": 4077
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0419614540131739e-05,
+      "loss": 0.8104,
+      "step": 4078
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0417174920712369e-05,
+      "loss": 0.759,
+      "step": 4079
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0414735301293e-05,
+      "loss": 0.7715,
+      "step": 4080
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0412295681873628e-05,
+      "loss": 0.7645,
+      "step": 4081
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0409856062454258e-05,
+      "loss": 0.7714,
+      "step": 4082
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0407416443034888e-05,
+      "loss": 0.8347,
+      "step": 4083
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0404976823615516e-05,
+      "loss": 0.783,
+      "step": 4084
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0402537204196146e-05,
+      "loss": 0.8759,
+      "step": 4085
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0400097584776775e-05,
+      "loss": 0.7499,
+      "step": 4086
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0397657965357407e-05,
+      "loss": 0.7721,
+      "step": 4087
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0395218345938035e-05,
+      "loss": 0.7793,
+      "step": 4088
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0392778726518664e-05,
+      "loss": 0.7972,
+      "step": 4089
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0390339107099294e-05,
+      "loss": 0.8022,
+      "step": 4090
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0387899487679922e-05,
+      "loss": 0.7963,
+      "step": 4091
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0385459868260552e-05,
+      "loss": 0.8878,
+      "step": 4092
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0383020248841182e-05,
+      "loss": 0.7021,
+      "step": 4093
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.038058062942181e-05,
+      "loss": 0.7449,
+      "step": 4094
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0378141010002441e-05,
+      "loss": 0.7453,
+      "step": 4095
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0375701390583071e-05,
+      "loss": 0.7683,
+      "step": 4096
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0373261771163699e-05,
+      "loss": 0.7855,
+      "step": 4097
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0370822151744329e-05,
+      "loss": 0.7162,
+      "step": 4098
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0368382532324959e-05,
+      "loss": 0.766,
+      "step": 4099
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0365942912905587e-05,
+      "loss": 0.7934,
+      "step": 4100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0363503293486216e-05,
+      "loss": 0.7963,
+      "step": 4101
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0361063674066846e-05,
+      "loss": 0.8187,
+      "step": 4102
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0358624054647474e-05,
+      "loss": 0.7714,
+      "step": 4103
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0356184435228106e-05,
+      "loss": 0.7542,
+      "step": 4104
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0353744815808735e-05,
+      "loss": 0.7837,
+      "step": 4105
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0351305196389365e-05,
+      "loss": 0.8211,
+      "step": 4106
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0348865576969993e-05,
+      "loss": 0.8041,
+      "step": 4107
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0346425957550623e-05,
+      "loss": 0.7803,
+      "step": 4108
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0343986338131253e-05,
+      "loss": 0.84,
+      "step": 4109
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.034154671871188e-05,
+      "loss": 0.7677,
+      "step": 4110
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0339107099292512e-05,
+      "loss": 0.7595,
+      "step": 4111
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0336667479873142e-05,
+      "loss": 0.7106,
+      "step": 4112
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.033422786045377e-05,
+      "loss": 0.6995,
+      "step": 4113
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.03317882410344e-05,
+      "loss": 0.779,
+      "step": 4114
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.032934862161503e-05,
+      "loss": 0.8005,
+      "step": 4115
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0326909002195657e-05,
+      "loss": 0.7543,
+      "step": 4116
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0324469382776287e-05,
+      "loss": 0.8242,
+      "step": 4117
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0322029763356917e-05,
+      "loss": 0.816,
+      "step": 4118
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0319590143937545e-05,
+      "loss": 0.7583,
+      "step": 4119
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0317150524518176e-05,
+      "loss": 0.7892,
+      "step": 4120
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0314710905098806e-05,
+      "loss": 0.7654,
+      "step": 4121
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0312271285679434e-05,
+      "loss": 0.7858,
+      "step": 4122
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0309831666260064e-05,
+      "loss": 0.8219,
+      "step": 4123
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0307392046840694e-05,
+      "loss": 0.7575,
+      "step": 4124
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0304952427421323e-05,
+      "loss": 0.8447,
+      "step": 4125
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0302512808001951e-05,
+      "loss": 0.8187,
+      "step": 4126
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0300073188582583e-05,
+      "loss": 0.77,
+      "step": 4127
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0297633569163213e-05,
+      "loss": 0.7746,
+      "step": 4128
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.029519394974384e-05,
+      "loss": 0.7738,
+      "step": 4129
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.029275433032447e-05,
+      "loss": 0.8197,
+      "step": 4130
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.02903147109051e-05,
+      "loss": 0.812,
+      "step": 4131
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0287875091485728e-05,
+      "loss": 0.8045,
+      "step": 4132
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0285435472066358e-05,
+      "loss": 0.7776,
+      "step": 4133
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.028299585264699e-05,
+      "loss": 0.7415,
+      "step": 4134
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0280556233227616e-05,
+      "loss": 0.728,
+      "step": 4135
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0278116613808247e-05,
+      "loss": 0.7303,
+      "step": 4136
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0275676994388877e-05,
+      "loss": 0.8157,
+      "step": 4137
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0273237374969505e-05,
+      "loss": 0.7551,
+      "step": 4138
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0270797755550135e-05,
+      "loss": 0.77,
+      "step": 4139
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0268358136130765e-05,
+      "loss": 0.7925,
+      "step": 4140
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0265918516711393e-05,
+      "loss": 0.7474,
+      "step": 4141
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0263478897292022e-05,
+      "loss": 0.7405,
+      "step": 4142
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0261039277872654e-05,
+      "loss": 0.7743,
+      "step": 4143
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0258599658453283e-05,
+      "loss": 0.7512,
+      "step": 4144
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0256160039033912e-05,
+      "loss": 0.7828,
+      "step": 4145
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0253720419614541e-05,
+      "loss": 0.7254,
+      "step": 4146
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0251280800195171e-05,
+      "loss": 0.7549,
+      "step": 4147
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0248841180775799e-05,
+      "loss": 0.8049,
+      "step": 4148
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0246401561356429e-05,
+      "loss": 0.7564,
+      "step": 4149
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.024396194193706e-05,
+      "loss": 0.789,
+      "step": 4150
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0241522322517687e-05,
+      "loss": 0.8299,
+      "step": 4151
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0239082703098318e-05,
+      "loss": 0.7734,
+      "step": 4152
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0236643083678948e-05,
+      "loss": 0.7472,
+      "step": 4153
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0234203464259576e-05,
+      "loss": 0.7557,
+      "step": 4154
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0231763844840206e-05,
+      "loss": 0.7917,
+      "step": 4155
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0229324225420835e-05,
+      "loss": 0.7981,
+      "step": 4156
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0226884606001463e-05,
+      "loss": 0.8109,
+      "step": 4157
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0224444986582093e-05,
+      "loss": 0.7867,
+      "step": 4158
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0222005367162725e-05,
+      "loss": 0.8035,
+      "step": 4159
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0219565747743353e-05,
+      "loss": 0.7338,
+      "step": 4160
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0217126128323982e-05,
+      "loss": 0.8029,
+      "step": 4161
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0214686508904612e-05,
+      "loss": 0.8245,
+      "step": 4162
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0212246889485242e-05,
+      "loss": 0.79,
+      "step": 4163
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.020980727006587e-05,
+      "loss": 0.7256,
+      "step": 4164
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.02073676506465e-05,
+      "loss": 0.7169,
+      "step": 4165
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0204928031227131e-05,
+      "loss": 0.6971,
+      "step": 4166
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0202488411807757e-05,
+      "loss": 0.7189,
+      "step": 4167
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0200048792388389e-05,
+      "loss": 0.7887,
+      "step": 4168
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0197609172969019e-05,
+      "loss": 0.7697,
+      "step": 4169
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0195169553549647e-05,
+      "loss": 0.7566,
+      "step": 4170
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0192729934130276e-05,
+      "loss": 0.7888,
+      "step": 4171
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0190290314710906e-05,
+      "loss": 0.7562,
+      "step": 4172
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0187850695291534e-05,
+      "loss": 0.8274,
+      "step": 4173
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0185411075872164e-05,
+      "loss": 0.7455,
+      "step": 4174
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0182971456452795e-05,
+      "loss": 0.747,
+      "step": 4175
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0180531837033423e-05,
+      "loss": 0.7665,
+      "step": 4176
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0178092217614053e-05,
+      "loss": 0.7767,
+      "step": 4177
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0175652598194683e-05,
+      "loss": 0.7662,
+      "step": 4178
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0173212978775311e-05,
+      "loss": 0.7774,
+      "step": 4179
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.017077335935594e-05,
+      "loss": 0.7454,
+      "step": 4180
+    },
+    {
+      "epoch": 1.98,
+      "eval_loss": 0.947238028049469,
+      "eval_runtime": 205.6888,
+      "eval_samples_per_second": 81.672,
+      "eval_steps_per_second": 10.21,
+      "step": 4180
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.016833373993657e-05,
+      "loss": 0.7417,
+      "step": 4181
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0165894120517199e-05,
+      "loss": 0.7492,
+      "step": 4182
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.016345450109783e-05,
+      "loss": 0.8685,
+      "step": 4183
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.016101488167846e-05,
+      "loss": 0.8506,
+      "step": 4184
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.015857526225909e-05,
+      "loss": 0.714,
+      "step": 4185
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0156135642839718e-05,
+      "loss": 0.7563,
+      "step": 4186
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0153696023420347e-05,
+      "loss": 0.7901,
+      "step": 4187
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0151256404000977e-05,
+      "loss": 0.7607,
+      "step": 4188
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0148816784581605e-05,
+      "loss": 0.7985,
+      "step": 4189
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0146377165162235e-05,
+      "loss": 0.8016,
+      "step": 4190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0143937545742866e-05,
+      "loss": 0.7297,
+      "step": 4191
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0141497926323494e-05,
+      "loss": 0.7448,
+      "step": 4192
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0139058306904124e-05,
+      "loss": 0.7331,
+      "step": 4193
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0136618687484754e-05,
+      "loss": 0.8412,
+      "step": 4194
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0134179068065382e-05,
+      "loss": 0.7796,
+      "step": 4195
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0131739448646012e-05,
+      "loss": 0.8364,
+      "step": 4196
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0129299829226641e-05,
+      "loss": 0.7829,
+      "step": 4197
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.012686020980727e-05,
+      "loss": 0.7664,
+      "step": 4198
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.01244205903879e-05,
+      "loss": 0.8241,
+      "step": 4199
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.012198097096853e-05,
+      "loss": 0.8042,
+      "step": 4200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0119541351549159e-05,
+      "loss": 0.723,
+      "step": 4201
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0117101732129788e-05,
+      "loss": 0.7499,
+      "step": 4202
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0114662112710418e-05,
+      "loss": 0.7925,
+      "step": 4203
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0112222493291048e-05,
+      "loss": 0.7429,
+      "step": 4204
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0109782873871676e-05,
+      "loss": 0.7645,
+      "step": 4205
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0107343254452306e-05,
+      "loss": 0.7062,
+      "step": 4206
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0104903635032937e-05,
+      "loss": 0.8287,
+      "step": 4207
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0102464015613565e-05,
+      "loss": 0.7336,
+      "step": 4208
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0100024396194195e-05,
+      "loss": 0.7416,
+      "step": 4209
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0097584776774825e-05,
+      "loss": 0.8157,
+      "step": 4210
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0095145157355453e-05,
+      "loss": 0.7743,
+      "step": 4211
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0092705537936082e-05,
+      "loss": 0.7317,
+      "step": 4212
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0090265918516712e-05,
+      "loss": 0.8092,
+      "step": 4213
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.008782629909734e-05,
+      "loss": 0.8227,
+      "step": 4214
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0085386679677972e-05,
+      "loss": 0.7655,
+      "step": 4215
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0082947060258601e-05,
+      "loss": 0.747,
+      "step": 4216
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.008050744083923e-05,
+      "loss": 0.7413,
+      "step": 4217
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.007806782141986e-05,
+      "loss": 0.7435,
+      "step": 4218
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0075628202000489e-05,
+      "loss": 0.7554,
+      "step": 4219
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0073188582581117e-05,
+      "loss": 0.7286,
+      "step": 4220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0070748963161747e-05,
+      "loss": 0.8038,
+      "step": 4221
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0068309343742378e-05,
+      "loss": 0.794,
+      "step": 4222
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0065869724323008e-05,
+      "loss": 0.73,
+      "step": 4223
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0063430104903636e-05,
+      "loss": 0.7215,
+      "step": 4224
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0060990485484266e-05,
+      "loss": 0.7617,
+      "step": 4225
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0058550866064895e-05,
+      "loss": 0.7459,
+      "step": 4226
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0056111246645523e-05,
+      "loss": 0.7648,
+      "step": 4227
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0053671627226153e-05,
+      "loss": 0.8007,
+      "step": 4228
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0051232007806783e-05,
+      "loss": 0.8343,
+      "step": 4229
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0048792388387411e-05,
+      "loss": 0.7685,
+      "step": 4230
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0046352768968042e-05,
+      "loss": 0.8042,
+      "step": 4231
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0043913149548672e-05,
+      "loss": 0.7727,
+      "step": 4232
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.00414735301293e-05,
+      "loss": 0.8041,
+      "step": 4233
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.003903391070993e-05,
+      "loss": 0.745,
+      "step": 4234
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.003659429129056e-05,
+      "loss": 0.8031,
+      "step": 4235
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0034154671871188e-05,
+      "loss": 0.7685,
+      "step": 4236
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0031715052451818e-05,
+      "loss": 0.8033,
+      "step": 4237
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0029275433032449e-05,
+      "loss": 0.7677,
+      "step": 4238
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0026835813613075e-05,
+      "loss": 0.7693,
+      "step": 4239
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0024396194193707e-05,
+      "loss": 0.7139,
+      "step": 4240
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0021956574774337e-05,
+      "loss": 0.7923,
+      "step": 4241
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0019516955354966e-05,
+      "loss": 0.7319,
+      "step": 4242
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0017077335935594e-05,
+      "loss": 0.756,
+      "step": 4243
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0014637716516224e-05,
+      "loss": 0.7469,
+      "step": 4244
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0012198097096855e-05,
+      "loss": 0.7496,
+      "step": 4245
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0009758477677482e-05,
+      "loss": 0.7426,
+      "step": 4246
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0007318858258113e-05,
+      "loss": 0.8016,
+      "step": 4247
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0004879238838743e-05,
+      "loss": 0.7496,
+      "step": 4248
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0002439619419371e-05,
+      "loss": 0.7563,
+      "step": 4249
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1e-05,
+      "loss": 0.777,
+      "step": 4250
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.99756038058063e-06,
+      "loss": 0.7229,
+      "step": 4251
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.99512076116126e-06,
+      "loss": 0.717,
+      "step": 4252
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.992681141741888e-06,
+      "loss": 0.8087,
+      "step": 4253
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.990241522322518e-06,
+      "loss": 0.7,
+      "step": 4254
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.987801902903148e-06,
+      "loss": 0.7893,
+      "step": 4255
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.985362283483778e-06,
+      "loss": 0.7702,
+      "step": 4256
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.982922664064407e-06,
+      "loss": 0.7294,
+      "step": 4257
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.980483044645037e-06,
+      "loss": 0.8357,
+      "step": 4258
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.978043425225665e-06,
+      "loss": 0.7711,
+      "step": 4259
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.975603805806295e-06,
+      "loss": 0.8059,
+      "step": 4260
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.973164186386925e-06,
+      "loss": 0.7555,
+      "step": 4261
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.970724566967553e-06,
+      "loss": 0.7812,
+      "step": 4262
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.968284947548184e-06,
+      "loss": 0.5521,
+      "step": 4263
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.965845328128812e-06,
+      "loss": 0.5535,
+      "step": 4264
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.963405708709442e-06,
+      "loss": 0.6107,
+      "step": 4265
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.960966089290072e-06,
+      "loss": 0.6826,
+      "step": 4266
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.958526469870701e-06,
+      "loss": 0.5559,
+      "step": 4267
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.956086850451331e-06,
+      "loss": 0.5581,
+      "step": 4268
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.95364723103196e-06,
+      "loss": 0.6115,
+      "step": 4269
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.951207611612589e-06,
+      "loss": 0.6072,
+      "step": 4270
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.948767992193219e-06,
+      "loss": 0.5814,
+      "step": 4271
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.946328372773848e-06,
+      "loss": 0.6495,
+      "step": 4272
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.943888753354476e-06,
+      "loss": 0.5808,
+      "step": 4273
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.941449133935108e-06,
+      "loss": 0.5581,
+      "step": 4274
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.939009514515736e-06,
+      "loss": 0.6254,
+      "step": 4275
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.936569895096366e-06,
+      "loss": 0.5567,
+      "step": 4276
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.934130275676995e-06,
+      "loss": 0.5785,
+      "step": 4277
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.931690656257623e-06,
+      "loss": 0.5577,
+      "step": 4278
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.929251036838255e-06,
+      "loss": 0.5449,
+      "step": 4279
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.926811417418883e-06,
+      "loss": 0.5688,
+      "step": 4280
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.924371797999513e-06,
+      "loss": 0.655,
+      "step": 4281
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.921932178580142e-06,
+      "loss": 0.5622,
+      "step": 4282
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.919492559160772e-06,
+      "loss": 0.5671,
+      "step": 4283
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.917052939741402e-06,
+      "loss": 0.6014,
+      "step": 4284
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.91461332032203e-06,
+      "loss": 0.5837,
+      "step": 4285
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.91217370090266e-06,
+      "loss": 0.5807,
+      "step": 4286
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.90973408148329e-06,
+      "loss": 0.5501,
+      "step": 4287
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.90729446206392e-06,
+      "loss": 0.5456,
+      "step": 4288
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.904854842644547e-06,
+      "loss": 0.5891,
+      "step": 4289
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.902415223225179e-06,
+      "loss": 0.5826,
+      "step": 4290
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.899975603805807e-06,
+      "loss": 0.6134,
+      "step": 4291
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.897535984386437e-06,
+      "loss": 0.6074,
+      "step": 4292
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.895096364967066e-06,
+      "loss": 0.5565,
+      "step": 4293
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.892656745547696e-06,
+      "loss": 0.5999,
+      "step": 4294
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.890217126128326e-06,
+      "loss": 0.6111,
+      "step": 4295
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.887777506708954e-06,
+      "loss": 0.5832,
+      "step": 4296
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.885337887289584e-06,
+      "loss": 0.6122,
+      "step": 4297
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.882898267870213e-06,
+      "loss": 0.594,
+      "step": 4298
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.880458648450843e-06,
+      "loss": 0.6169,
+      "step": 4299
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.878019029031471e-06,
+      "loss": 0.6235,
+      "step": 4300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.8755794096121e-06,
+      "loss": 0.5829,
+      "step": 4301
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.87313979019273e-06,
+      "loss": 0.6036,
+      "step": 4302
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.87070017077336e-06,
+      "loss": 0.5767,
+      "step": 4303
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.86826055135399e-06,
+      "loss": 0.5804,
+      "step": 4304
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.865820931934618e-06,
+      "loss": 0.5938,
+      "step": 4305
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.86338131251525e-06,
+      "loss": 0.6536,
+      "step": 4306
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.860941693095878e-06,
+      "loss": 0.6073,
+      "step": 4307
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.858502073676507e-06,
+      "loss": 0.6049,
+      "step": 4308
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.856062454257137e-06,
+      "loss": 0.5693,
+      "step": 4309
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.853622834837767e-06,
+      "loss": 0.6299,
+      "step": 4310
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.851183215418395e-06,
+      "loss": 0.5812,
+      "step": 4311
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.848743595999025e-06,
+      "loss": 0.5933,
+      "step": 4312
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.846303976579654e-06,
+      "loss": 0.5627,
+      "step": 4313
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.843864357160284e-06,
+      "loss": 0.5486,
+      "step": 4314
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.841424737740914e-06,
+      "loss": 0.6073,
+      "step": 4315
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.838985118321542e-06,
+      "loss": 0.627,
+      "step": 4316
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.836545498902172e-06,
+      "loss": 0.5479,
+      "step": 4317
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.834105879482801e-06,
+      "loss": 0.6116,
+      "step": 4318
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.831666260063431e-06,
+      "loss": 0.5573,
+      "step": 4319
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.829226640644061e-06,
+      "loss": 0.5908,
+      "step": 4320
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.826787021224689e-06,
+      "loss": 0.5554,
+      "step": 4321
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.824347401805319e-06,
+      "loss": 0.568,
+      "step": 4322
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.821907782385948e-06,
+      "loss": 0.6023,
+      "step": 4323
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.819468162966578e-06,
+      "loss": 0.5731,
+      "step": 4324
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.817028543547208e-06,
+      "loss": 0.6064,
+      "step": 4325
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.814588924127838e-06,
+      "loss": 0.583,
+      "step": 4326
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.812149304708466e-06,
+      "loss": 0.5748,
+      "step": 4327
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.809709685289095e-06,
+      "loss": 0.5618,
+      "step": 4328
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.807270065869725e-06,
+      "loss": 0.5597,
+      "step": 4329
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.804830446450353e-06,
+      "loss": 0.5568,
+      "step": 4330
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.802390827030985e-06,
+      "loss": 0.6289,
+      "step": 4331
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.799951207611613e-06,
+      "loss": 0.5362,
+      "step": 4332
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.797511588192244e-06,
+      "loss": 0.5432,
+      "step": 4333
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.795071968772872e-06,
+      "loss": 0.5597,
+      "step": 4334
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.792632349353502e-06,
+      "loss": 0.5963,
+      "step": 4335
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.790192729934132e-06,
+      "loss": 0.5916,
+      "step": 4336
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.78775311051476e-06,
+      "loss": 0.5563,
+      "step": 4337
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.78531349109539e-06,
+      "loss": 0.5983,
+      "step": 4338
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.78287387167602e-06,
+      "loss": 0.59,
+      "step": 4339
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.780434252256649e-06,
+      "loss": 0.5338,
+      "step": 4340
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.777994632837277e-06,
+      "loss": 0.5728,
+      "step": 4341
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.775555013417909e-06,
+      "loss": 0.6063,
+      "step": 4342
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.773115393998537e-06,
+      "loss": 0.5437,
+      "step": 4343
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.770675774579166e-06,
+      "loss": 0.6859,
+      "step": 4344
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.768236155159796e-06,
+      "loss": 0.5776,
+      "step": 4345
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.765796535740426e-06,
+      "loss": 0.5538,
+      "step": 4346
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.763356916321056e-06,
+      "loss": 0.6154,
+      "step": 4347
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.760917296901684e-06,
+      "loss": 0.6133,
+      "step": 4348
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.758477677482313e-06,
+      "loss": 0.5972,
+      "step": 4349
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.756038058062943e-06,
+      "loss": 0.6207,
+      "step": 4350
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.753598438643573e-06,
+      "loss": 0.5629,
+      "step": 4351
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.751158819224203e-06,
+      "loss": 0.5856,
+      "step": 4352
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.74871919980483e-06,
+      "loss": 0.5518,
+      "step": 4353
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.74627958038546e-06,
+      "loss": 0.6005,
+      "step": 4354
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.74383996096609e-06,
+      "loss": 0.5981,
+      "step": 4355
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.74140034154672e-06,
+      "loss": 0.5326,
+      "step": 4356
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.738960722127348e-06,
+      "loss": 0.592,
+      "step": 4357
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.73652110270798e-06,
+      "loss": 0.5625,
+      "step": 4358
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.734081483288607e-06,
+      "loss": 0.5139,
+      "step": 4359
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.731641863869237e-06,
+      "loss": 0.5749,
+      "step": 4360
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.729202244449867e-06,
+      "loss": 0.5742,
+      "step": 4361
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.726762625030497e-06,
+      "loss": 0.6127,
+      "step": 4362
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.724323005611126e-06,
+      "loss": 0.6246,
+      "step": 4363
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.721883386191754e-06,
+      "loss": 0.612,
+      "step": 4364
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.719443766772384e-06,
+      "loss": 0.5884,
+      "step": 4365
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.717004147353014e-06,
+      "loss": 0.566,
+      "step": 4366
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.714564527933644e-06,
+      "loss": 0.601,
+      "step": 4367
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.712124908514272e-06,
+      "loss": 0.5973,
+      "step": 4368
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.709685289094901e-06,
+      "loss": 0.557,
+      "step": 4369
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.707245669675531e-06,
+      "loss": 0.5707,
+      "step": 4370
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.704806050256161e-06,
+      "loss": 0.51,
+      "step": 4371
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.70236643083679e-06,
+      "loss": 0.5762,
+      "step": 4372
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.699926811417419e-06,
+      "loss": 0.5612,
+      "step": 4373
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.69748719199805e-06,
+      "loss": 0.5903,
+      "step": 4374
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.695047572578678e-06,
+      "loss": 0.5705,
+      "step": 4375
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.692607953159308e-06,
+      "loss": 0.5691,
+      "step": 4376
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.690168333739938e-06,
+      "loss": 0.6262,
+      "step": 4377
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.687728714320567e-06,
+      "loss": 0.5248,
+      "step": 4378
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.685289094901195e-06,
+      "loss": 0.575,
+      "step": 4379
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.682849475481825e-06,
+      "loss": 0.5798,
+      "step": 4380
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.680409856062455e-06,
+      "loss": 0.618,
+      "step": 4381
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.677970236643085e-06,
+      "loss": 0.6195,
+      "step": 4382
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.675530617223714e-06,
+      "loss": 0.6199,
+      "step": 4383
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.673090997804343e-06,
+      "loss": 0.6133,
+      "step": 4384
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.670651378384974e-06,
+      "loss": 0.6032,
+      "step": 4385
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.668211758965602e-06,
+      "loss": 0.6087,
+      "step": 4386
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.665772139546232e-06,
+      "loss": 0.6104,
+      "step": 4387
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.663332520126861e-06,
+      "loss": 0.5844,
+      "step": 4388
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.66089290070749e-06,
+      "loss": 0.5918,
+      "step": 4389
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.65845328128812e-06,
+      "loss": 0.5497,
+      "step": 4390
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.656013661868749e-06,
+      "loss": 0.5703,
+      "step": 4391
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.653574042449379e-06,
+      "loss": 0.6008,
+      "step": 4392
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.651134423030009e-06,
+      "loss": 0.6015,
+      "step": 4393
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.648694803610638e-06,
+      "loss": 0.6328,
+      "step": 4394
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.646255184191266e-06,
+      "loss": 0.6465,
+      "step": 4395
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.643815564771896e-06,
+      "loss": 0.573,
+      "step": 4396
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.641375945352526e-06,
+      "loss": 0.5557,
+      "step": 4397
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.638936325933156e-06,
+      "loss": 0.6061,
+      "step": 4398
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.636496706513785e-06,
+      "loss": 0.5645,
+      "step": 4399
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.634057087094413e-06,
+      "loss": 0.5407,
+      "step": 4400
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 8348,
+  "num_train_epochs": 4,
+  "save_steps": 4400,
+  "total_flos": 921218140078080.0,
+  "trial_name": null,
+  "trial_params": null
+}