diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5377 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 892,
+  "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": 12.227,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 12.7771,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 12.7177,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 12.635,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 12.2989,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-07,
+      "loss": 12.6181,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2e-06,
+      "loss": 11.9585,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-06,
+      "loss": 11.24,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-06,
+      "loss": 11.5953,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999999999999997e-06,
+      "loss": 11.647,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-06,
+      "loss": 11.2743,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-06,
+      "loss": 10.1301,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8e-06,
+      "loss": 9.871,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.399999999999999e-06,
+      "loss": 9.482,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 8.7618,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.599999999999999e-06,
+      "loss": 8.6365,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.2e-06,
+      "loss": 8.26,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.799999999999998e-06,
+      "loss": 8.037,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.4e-06,
+      "loss": 7.6623,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.999999999999999e-06,
+      "loss": 7.4867,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.6e-06,
+      "loss": 7.2931,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.02e-05,
+      "loss": 7.1014,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0799999999999998e-05,
+      "loss": 7.0479,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.14e-05,
+      "loss": 6.7736,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 6.5709,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.26e-05,
+      "loss": 6.5562,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3199999999999997e-05,
+      "loss": 6.3255,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3799999999999998e-05,
+      "loss": 6.015,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.44e-05,
+      "loss": 5.8947,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 6.0453,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5599999999999996e-05,
+      "loss": 6.0951,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6199999999999997e-05,
+      "loss": 5.7709,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.68e-05,
+      "loss": 5.7683,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.74e-05,
+      "loss": 5.394,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 5.5852,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8599999999999998e-05,
+      "loss": 5.3502,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.92e-05,
+      "loss": 5.4977,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.98e-05,
+      "loss": 5.4443,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.04e-05,
+      "loss": 5.6272,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1e-05,
+      "loss": 5.5404,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1599999999999996e-05,
+      "loss": 5.3339,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2199999999999998e-05,
+      "loss": 5.4734,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.28e-05,
+      "loss": 5.1924,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.34e-05,
+      "loss": 5.3261,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 5.4006,
+      "step": 45
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 5.4764,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.52e-05,
+      "loss": 5.5291,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.52e-05,
+      "loss": 5.6644,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5799999999999997e-05,
+      "loss": 5.616,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.6399999999999995e-05,
+      "loss": 5.2866,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 6.3282,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7599999999999997e-05,
+      "loss": 6.1156,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 5.6303,
+      "step": 53
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.88e-05,
+      "loss": 5.3366,
+      "step": 54
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.94e-05,
+      "loss": 5.1893,
+      "step": 55
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 5.4874,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.06e-05,
+      "loss": 5.4816,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.119999999999999e-05,
+      "loss": 5.5773,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1799999999999994e-05,
+      "loss": 5.2495,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2399999999999995e-05,
+      "loss": 5.3377,
+      "step": 60
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 5.2889,
+      "step": 61
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.36e-05,
+      "loss": 5.246,
+      "step": 62
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.42e-05,
+      "loss": 5.0918,
+      "step": 63
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.48e-05,
+      "loss": 5.3254,
+      "step": 64
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.539999999999999e-05,
+      "loss": 4.9528,
+      "step": 65
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 5.1956,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.6599999999999995e-05,
+      "loss": 5.034,
+      "step": 67
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.7199999999999996e-05,
+      "loss": 4.983,
+      "step": 68
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.78e-05,
+      "loss": 4.9528,
+      "step": 69
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.84e-05,
+      "loss": 5.06,
+      "step": 70
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9e-05,
+      "loss": 5.174,
+      "step": 71
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.96e-05,
+      "loss": 5.1699,
+      "step": 72
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.02e-05,
+      "loss": 4.9884,
+      "step": 73
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.08e-05,
+      "loss": 5.0604,
+      "step": 74
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.14e-05,
+      "loss": 4.9204,
+      "step": 75
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.2e-05,
+      "loss": 5.0904,
+      "step": 76
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.259999999999999e-05,
+      "loss": 4.9985,
+      "step": 77
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.319999999999999e-05,
+      "loss": 4.9279,
+      "step": 78
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.3799999999999994e-05,
+      "loss": 4.8253,
+      "step": 79
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.4399999999999995e-05,
+      "loss": 4.9923,
+      "step": 80
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 5.0213,
+      "step": 81
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.56e-05,
+      "loss": 4.8089,
+      "step": 82
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.62e-05,
+      "loss": 5.0773,
+      "step": 83
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.68e-05,
+      "loss": 5.011,
+      "step": 84
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7399999999999993e-05,
+      "loss": 4.9196,
+      "step": 85
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 4.8375,
+      "step": 86
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8599999999999995e-05,
+      "loss": 4.9236,
+      "step": 87
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9199999999999997e-05,
+      "loss": 4.9146,
+      "step": 88
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.98e-05,
+      "loss": 4.9634,
+      "step": 89
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.04e-05,
+      "loss": 4.835,
+      "step": 90
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.1e-05,
+      "loss": 5.1032,
+      "step": 91
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.1599999999999994e-05,
+      "loss": 4.987,
+      "step": 92
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.2199999999999995e-05,
+      "loss": 4.5916,
+      "step": 93
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.279999999999999e-05,
+      "loss": 4.8832,
+      "step": 94
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.339999999999999e-05,
+      "loss": 5.1372,
+      "step": 95
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 5.1099,
+      "step": 96
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.459999999999999e-05,
+      "loss": 4.8033,
+      "step": 97
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.519999999999999e-05,
+      "loss": 5.1438,
+      "step": 98
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.5799999999999994e-05,
+      "loss": 4.911,
+      "step": 99
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 4.5011,
+      "step": 100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 6.252,
+      "step": 101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.76e-05,
+      "loss": 6.0714,
+      "step": 102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.82e-05,
+      "loss": 5.5859,
+      "step": 103
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.88e-05,
+      "loss": 5.1686,
+      "step": 104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.94e-05,
+      "loss": 4.9099,
+      "step": 105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 5.1412,
+      "step": 106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.0599999999999996e-05,
+      "loss": 5.0822,
+      "step": 107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.12e-05,
+      "loss": 5.2151,
+      "step": 108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.18e-05,
+      "loss": 4.7549,
+      "step": 109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.239999999999999e-05,
+      "loss": 4.9275,
+      "step": 110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.299999999999999e-05,
+      "loss": 5.0012,
+      "step": 111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.359999999999999e-05,
+      "loss": 4.9506,
+      "step": 112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.419999999999999e-05,
+      "loss": 4.8381,
+      "step": 113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.479999999999999e-05,
+      "loss": 4.7752,
+      "step": 114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.539999999999999e-05,
+      "loss": 4.8818,
+      "step": 115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 4.6591,
+      "step": 116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.659999999999999e-05,
+      "loss": 4.9267,
+      "step": 117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.72e-05,
+      "loss": 4.8077,
+      "step": 118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.78e-05,
+      "loss": 4.9312,
+      "step": 119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.84e-05,
+      "loss": 4.9596,
+      "step": 120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.9e-05,
+      "loss": 4.7884,
+      "step": 121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.96e-05,
+      "loss": 5.0901,
+      "step": 122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.02e-05,
+      "loss": 4.6784,
+      "step": 123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.079999999999999e-05,
+      "loss": 4.9247,
+      "step": 124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.139999999999999e-05,
+      "loss": 4.9386,
+      "step": 125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 4.8603,
+      "step": 126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.259999999999999e-05,
+      "loss": 4.9604,
+      "step": 127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.319999999999999e-05,
+      "loss": 4.8704,
+      "step": 128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.379999999999999e-05,
+      "loss": 4.7229,
+      "step": 129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.439999999999999e-05,
+      "loss": 4.9599,
+      "step": 130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.5e-05,
+      "loss": 4.8136,
+      "step": 131
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.56e-05,
+      "loss": 4.8715,
+      "step": 132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.62e-05,
+      "loss": 4.8147,
+      "step": 133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.68e-05,
+      "loss": 4.9103,
+      "step": 134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.74e-05,
+      "loss": 4.6857,
+      "step": 135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.8e-05,
+      "loss": 4.7652,
+      "step": 136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.86e-05,
+      "loss": 5.0094,
+      "step": 137
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.92e-05,
+      "loss": 4.7424,
+      "step": 138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.98e-05,
+      "loss": 4.8443,
+      "step": 139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.04e-05,
+      "loss": 4.6487,
+      "step": 140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.1e-05,
+      "loss": 5.0035,
+      "step": 141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.16e-05,
+      "loss": 5.0282,
+      "step": 142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.22e-05,
+      "loss": 4.9758,
+      "step": 143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.28e-05,
+      "loss": 4.7827,
+      "step": 144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.34e-05,
+      "loss": 4.8904,
+      "step": 145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.4e-05,
+      "loss": 4.902,
+      "step": 146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.459999999999998e-05,
+      "loss": 5.0951,
+      "step": 147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.519999999999998e-05,
+      "loss": 4.8262,
+      "step": 148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.579999999999998e-05,
+      "loss": 5.1813,
+      "step": 149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.639999999999999e-05,
+      "loss": 4.3312,
+      "step": 150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 5.5749,
+      "step": 151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.759999999999999e-05,
+      "loss": 5.5083,
+      "step": 152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.819999999999999e-05,
+      "loss": 5.3511,
+      "step": 153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.879999999999999e-05,
+      "loss": 4.9529,
+      "step": 154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.939999999999999e-05,
+      "loss": 4.7879,
+      "step": 155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 4.8424,
+      "step": 156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.059999999999999e-05,
+      "loss": 5.061,
+      "step": 157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.12e-05,
+      "loss": 5.0679,
+      "step": 158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.18e-05,
+      "loss": 5.194,
+      "step": 159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.24e-05,
+      "loss": 4.8643,
+      "step": 160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.3e-05,
+      "loss": 4.8178,
+      "step": 161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.36e-05,
+      "loss": 4.9475,
+      "step": 162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419999999999999e-05,
+      "loss": 4.7963,
+      "step": 163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.479999999999999e-05,
+      "loss": 4.9378,
+      "step": 164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.539999999999999e-05,
+      "loss": 4.8338,
+      "step": 165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 4.8404,
+      "step": 166
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.659999999999999e-05,
+      "loss": 4.8793,
+      "step": 167
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.719999999999999e-05,
+      "loss": 4.7065,
+      "step": 168
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.779999999999999e-05,
+      "loss": 4.7181,
+      "step": 169
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.839999999999999e-05,
+      "loss": 4.744,
+      "step": 170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.9e-05,
+      "loss": 5.0243,
+      "step": 171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.96e-05,
+      "loss": 4.9735,
+      "step": 172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001002,
+      "loss": 4.8479,
+      "step": 173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001008,
+      "loss": 4.6975,
+      "step": 174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001014,
+      "loss": 4.7985,
+      "step": 175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000102,
+      "loss": 4.7022,
+      "step": 176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001026,
+      "loss": 4.7026,
+      "step": 177
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010319999999999999,
+      "loss": 4.6673,
+      "step": 178
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010379999999999999,
+      "loss": 4.8146,
+      "step": 179
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010439999999999999,
+      "loss": 4.8315,
+      "step": 180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010499999999999999,
+      "loss": 4.8464,
+      "step": 181
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00010559999999999998,
+      "loss": 4.7597,
+      "step": 182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010619999999999998,
+      "loss": 4.8652,
+      "step": 183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010679999999999998,
+      "loss": 4.8472,
+      "step": 184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010739999999999998,
+      "loss": 4.764,
+      "step": 185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 4.712,
+      "step": 186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010859999999999998,
+      "loss": 4.8689,
+      "step": 187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010919999999999998,
+      "loss": 4.5683,
+      "step": 188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010979999999999999,
+      "loss": 4.7651,
+      "step": 189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00011039999999999999,
+      "loss": 4.7211,
+      "step": 190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00011099999999999999,
+      "loss": 4.7803,
+      "step": 191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00011159999999999999,
+      "loss": 4.818,
+      "step": 192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00011219999999999999,
+      "loss": 4.99,
+      "step": 193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00011279999999999999,
+      "loss": 4.5846,
+      "step": 194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00011339999999999999,
+      "loss": 4.6508,
+      "step": 195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 5.0115,
+      "step": 196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001146,
+      "loss": 4.8395,
+      "step": 197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001152,
+      "loss": 4.77,
+      "step": 198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001158,
+      "loss": 4.9126,
+      "step": 199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001164,
+      "loss": 4.8374,
+      "step": 200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000117,
+      "loss": 5.5324,
+      "step": 201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001176,
+      "loss": 5.4875,
+      "step": 202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001182,
+      "loss": 5.1839,
+      "step": 203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001188,
+      "loss": 5.0931,
+      "step": 204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001194,
+      "loss": 4.8868,
+      "step": 205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 4.8339,
+      "step": 206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012059999999999999,
+      "loss": 4.8605,
+      "step": 207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012119999999999999,
+      "loss": 4.715,
+      "step": 208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012179999999999999,
+      "loss": 4.8493,
+      "step": 209
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001224,
+      "loss": 4.8462,
+      "step": 210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00012299999999999998,
+      "loss": 4.7033,
+      "step": 211
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001236,
+      "loss": 4.8332,
+      "step": 212
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00012419999999999998,
+      "loss": 4.7906,
+      "step": 213
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00012479999999999997,
+      "loss": 4.8212,
+      "step": 214
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00012539999999999999,
+      "loss": 4.7387,
+      "step": 215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 4.8769,
+      "step": 216
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001266,
+      "loss": 4.8184,
+      "step": 217
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00012719999999999997,
+      "loss": 4.6971,
+      "step": 218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001278,
+      "loss": 4.7483,
+      "step": 219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00012839999999999998,
+      "loss": 4.8553,
+      "step": 220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000129,
+      "loss": 4.7291,
+      "step": 221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00012959999999999998,
+      "loss": 4.9811,
+      "step": 222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001302,
+      "loss": 4.9486,
+      "step": 223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00013079999999999998,
+      "loss": 4.9872,
+      "step": 224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001314,
+      "loss": 5.0354,
+      "step": 225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 5.1846,
+      "step": 226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001326,
+      "loss": 4.7989,
+      "step": 227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00013319999999999999,
+      "loss": 4.8591,
+      "step": 228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001338,
+      "loss": 4.8231,
+      "step": 229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001344,
+      "loss": 4.8881,
+      "step": 230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000135,
+      "loss": 4.7428,
+      "step": 231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001356,
+      "loss": 4.9219,
+      "step": 232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001362,
+      "loss": 4.7862,
+      "step": 233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001368,
+      "loss": 4.8917,
+      "step": 234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001374,
+      "loss": 4.9955,
+      "step": 235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000138,
+      "loss": 4.8297,
+      "step": 236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001386,
+      "loss": 4.9536,
+      "step": 237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001392,
+      "loss": 5.0285,
+      "step": 238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00013979999999999998,
+      "loss": 4.7294,
+      "step": 239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001404,
+      "loss": 4.9076,
+      "step": 240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00014099999999999998,
+      "loss": 4.7713,
+      "step": 241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00014159999999999997,
+      "loss": 4.9299,
+      "step": 242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001422,
+      "loss": 4.8412,
+      "step": 243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00014279999999999997,
+      "loss": 4.6361,
+      "step": 244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001434,
+      "loss": 5.1148,
+      "step": 245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 4.8241,
+      "step": 246
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001446,
+      "loss": 5.0455,
+      "step": 247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00014519999999999998,
+      "loss": 4.9531,
+      "step": 248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001458,
+      "loss": 4.8026,
+      "step": 249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00014639999999999998,
+      "loss": 4.4284,
+      "step": 250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000147,
+      "loss": 6.0484,
+      "step": 251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00014759999999999998,
+      "loss": 5.6329,
+      "step": 252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001482,
+      "loss": 5.3109,
+      "step": 253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00014879999999999998,
+      "loss": 5.2561,
+      "step": 254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001494,
+      "loss": 4.9578,
+      "step": 255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015,
+      "loss": 4.9713,
+      "step": 256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015059999999999997,
+      "loss": 4.9373,
+      "step": 257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001512,
+      "loss": 4.8036,
+      "step": 258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015179999999999998,
+      "loss": 4.8621,
+      "step": 259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001524,
+      "loss": 4.985,
+      "step": 260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015299999999999998,
+      "loss": 4.937,
+      "step": 261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001536,
+      "loss": 4.6815,
+      "step": 262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015419999999999998,
+      "loss": 4.8316,
+      "step": 263
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001548,
+      "loss": 4.7113,
+      "step": 264
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015539999999999998,
+      "loss": 4.8462,
+      "step": 265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000156,
+      "loss": 4.9109,
+      "step": 266
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015659999999999998,
+      "loss": 4.7336,
+      "step": 267
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001572,
+      "loss": 4.9493,
+      "step": 268
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001578,
+      "loss": 4.8003,
+      "step": 269
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001584,
+      "loss": 4.9139,
+      "step": 270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000159,
+      "loss": 4.5881,
+      "step": 271
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001596,
+      "loss": 4.6624,
+      "step": 272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001602,
+      "loss": 4.6558,
+      "step": 273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001608,
+      "loss": 4.6816,
+      "step": 274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001614,
+      "loss": 4.8323,
+      "step": 275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000162,
+      "loss": 4.8204,
+      "step": 276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001626,
+      "loss": 4.8373,
+      "step": 277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001632,
+      "loss": 4.7286,
+      "step": 278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001638,
+      "loss": 4.7064,
+      "step": 279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001644,
+      "loss": 4.7758,
+      "step": 280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000165,
+      "loss": 4.5942,
+      "step": 281
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001656,
+      "loss": 4.8,
+      "step": 282
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001662,
+      "loss": 4.5438,
+      "step": 283
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001668,
+      "loss": 4.8274,
+      "step": 284
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001674,
+      "loss": 4.8702,
+      "step": 285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000168,
+      "loss": 4.7105,
+      "step": 286
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001686,
+      "loss": 4.8917,
+      "step": 287
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016919999999999997,
+      "loss": 4.9552,
+      "step": 288
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016979999999999998,
+      "loss": 4.8993,
+      "step": 289
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017039999999999997,
+      "loss": 4.8569,
+      "step": 290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017099999999999998,
+      "loss": 4.7941,
+      "step": 291
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017159999999999997,
+      "loss": 4.6413,
+      "step": 292
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017219999999999998,
+      "loss": 4.8278,
+      "step": 293
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017279999999999997,
+      "loss": 4.8177,
+      "step": 294
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017339999999999996,
+      "loss": 4.6866,
+      "step": 295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 4.8502,
+      "step": 296
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017459999999999996,
+      "loss": 4.8619,
+      "step": 297
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017519999999999998,
+      "loss": 4.6715,
+      "step": 298
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017579999999999996,
+      "loss": 5.0034,
+      "step": 299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017639999999999998,
+      "loss": 4.7177,
+      "step": 300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017699999999999997,
+      "loss": 5.8594,
+      "step": 301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017759999999999998,
+      "loss": 5.746,
+      "step": 302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017819999999999997,
+      "loss": 5.4407,
+      "step": 303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017879999999999998,
+      "loss": 5.2606,
+      "step": 304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017939999999999997,
+      "loss": 4.9303,
+      "step": 305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 5.0291,
+      "step": 306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018059999999999997,
+      "loss": 5.0468,
+      "step": 307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018119999999999999,
+      "loss": 4.974,
+      "step": 308
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018179999999999997,
+      "loss": 4.9039,
+      "step": 309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001824,
+      "loss": 4.9634,
+      "step": 310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018299999999999998,
+      "loss": 5.0115,
+      "step": 311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001836,
+      "loss": 4.955,
+      "step": 312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018419999999999998,
+      "loss": 4.9483,
+      "step": 313
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001848,
+      "loss": 4.8552,
+      "step": 314
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018539999999999998,
+      "loss": 4.7023,
+      "step": 315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000186,
+      "loss": 4.8409,
+      "step": 316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018659999999999998,
+      "loss": 4.7218,
+      "step": 317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001872,
+      "loss": 4.7217,
+      "step": 318
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018779999999999998,
+      "loss": 4.7875,
+      "step": 319
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018839999999999997,
+      "loss": 4.6982,
+      "step": 320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018899999999999999,
+      "loss": 4.7792,
+      "step": 321
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018959999999999997,
+      "loss": 4.7917,
+      "step": 322
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001902,
+      "loss": 4.7345,
+      "step": 323
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019079999999999998,
+      "loss": 4.8833,
+      "step": 324
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001914,
+      "loss": 4.8808,
+      "step": 325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 4.8451,
+      "step": 326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001926,
+      "loss": 4.6829,
+      "step": 327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019319999999999998,
+      "loss": 4.6917,
+      "step": 328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001938,
+      "loss": 4.9283,
+      "step": 329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019439999999999998,
+      "loss": 4.931,
+      "step": 330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000195,
+      "loss": 4.7855,
+      "step": 331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019559999999999998,
+      "loss": 4.7239,
+      "step": 332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001962,
+      "loss": 4.7628,
+      "step": 333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019679999999999999,
+      "loss": 4.8131,
+      "step": 334
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001974,
+      "loss": 4.9196,
+      "step": 335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000198,
+      "loss": 4.7766,
+      "step": 336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001986,
+      "loss": 4.6247,
+      "step": 337
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001992,
+      "loss": 4.9321,
+      "step": 338
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001998,
+      "loss": 4.6726,
+      "step": 339
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002004,
+      "loss": 4.7444,
+      "step": 340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000201,
+      "loss": 4.7374,
+      "step": 341
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002016,
+      "loss": 4.7581,
+      "step": 342
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002022,
+      "loss": 4.7161,
+      "step": 343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002028,
+      "loss": 4.8216,
+      "step": 344
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00020339999999999998,
+      "loss": 4.7525,
+      "step": 345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000204,
+      "loss": 4.7387,
+      "step": 346
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00020459999999999999,
+      "loss": 4.7731,
+      "step": 347
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002052,
+      "loss": 4.8147,
+      "step": 348
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002058,
+      "loss": 4.8416,
+      "step": 349
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00020639999999999998,
+      "loss": 4.6396,
+      "step": 350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00020699999999999996,
+      "loss": 5.9008,
+      "step": 351
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00020759999999999998,
+      "loss": 5.739,
+      "step": 352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00020819999999999996,
+      "loss": 5.3717,
+      "step": 353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00020879999999999998,
+      "loss": 5.1989,
+      "step": 354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00020939999999999997,
+      "loss": 5.1891,
+      "step": 355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 4.8991,
+      "step": 356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00021059999999999997,
+      "loss": 4.9887,
+      "step": 357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00021119999999999996,
+      "loss": 4.885,
+      "step": 358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00021179999999999997,
+      "loss": 4.9382,
+      "step": 359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00021239999999999996,
+      "loss": 4.9564,
+      "step": 360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 4.7946,
+      "step": 361
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021359999999999996,
+      "loss": 4.8095,
+      "step": 362
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021419999999999998,
+      "loss": 4.7538,
+      "step": 363
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021479999999999996,
+      "loss": 4.9914,
+      "step": 364
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021539999999999998,
+      "loss": 4.8159,
+      "step": 365
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 4.678,
+      "step": 366
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021659999999999998,
+      "loss": 4.7531,
+      "step": 367
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021719999999999997,
+      "loss": 4.7875,
+      "step": 368
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021779999999999998,
+      "loss": 4.8139,
+      "step": 369
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021839999999999997,
+      "loss": 4.8691,
+      "step": 370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 4.7698,
+      "step": 371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00021959999999999997,
+      "loss": 5.0252,
+      "step": 372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00022019999999999999,
+      "loss": 4.7653,
+      "step": 373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00022079999999999997,
+      "loss": 4.9352,
+      "step": 374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002214,
+      "loss": 4.7104,
+      "step": 375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 4.7958,
+      "step": 376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002226,
+      "loss": 4.5804,
+      "step": 377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00022319999999999998,
+      "loss": 4.9133,
+      "step": 378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002238,
+      "loss": 4.8053,
+      "step": 379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00022439999999999998,
+      "loss": 4.8824,
+      "step": 380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000225,
+      "loss": 4.8564,
+      "step": 381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00022559999999999998,
+      "loss": 4.7516,
+      "step": 382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00022619999999999997,
+      "loss": 4.8333,
+      "step": 383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00022679999999999998,
+      "loss": 4.8387,
+      "step": 384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00022739999999999997,
+      "loss": 4.6767,
+      "step": 385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 4.7029,
+      "step": 386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00022859999999999997,
+      "loss": 4.7181,
+      "step": 387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002292,
+      "loss": 4.6297,
+      "step": 388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00022979999999999997,
+      "loss": 4.7326,
+      "step": 389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002304,
+      "loss": 4.8956,
+      "step": 390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023099999999999998,
+      "loss": 4.8496,
+      "step": 391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002316,
+      "loss": 4.6732,
+      "step": 392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023219999999999998,
+      "loss": 4.7438,
+      "step": 393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002328,
+      "loss": 4.9382,
+      "step": 394
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00023339999999999998,
+      "loss": 5.0041,
+      "step": 395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000234,
+      "loss": 4.7323,
+      "step": 396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023459999999999998,
+      "loss": 4.6038,
+      "step": 397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002352,
+      "loss": 4.7239,
+      "step": 398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023579999999999999,
+      "loss": 4.6071,
+      "step": 399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002364,
+      "loss": 4.5138,
+      "step": 400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000237,
+      "loss": 5.6771,
+      "step": 401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002376,
+      "loss": 5.4054,
+      "step": 402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002382,
+      "loss": 5.1976,
+      "step": 403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002388,
+      "loss": 5.0373,
+      "step": 404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002394,
+      "loss": 5.1013,
+      "step": 405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 4.9948,
+      "step": 406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002406,
+      "loss": 4.889,
+      "step": 407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024119999999999998,
+      "loss": 4.7801,
+      "step": 408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002418,
+      "loss": 4.9982,
+      "step": 409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024239999999999998,
+      "loss": 4.8896,
+      "step": 410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000243,
+      "loss": 4.7741,
+      "step": 411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024359999999999999,
+      "loss": 4.8769,
+      "step": 412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024419999999999997,
+      "loss": 4.8612,
+      "step": 413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002448,
+      "loss": 4.6607,
+      "step": 414
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00024539999999999995,
+      "loss": 4.6997,
+      "step": 415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 4.7932,
+      "step": 416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002466,
+      "loss": 4.6429,
+      "step": 417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002472,
+      "loss": 4.8403,
+      "step": 418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00024779999999999995,
+      "loss": 4.825,
+      "step": 419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00024839999999999997,
+      "loss": 4.8656,
+      "step": 420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000249,
+      "loss": 4.8104,
+      "step": 421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00024959999999999994,
+      "loss": 4.8133,
+      "step": 422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00025019999999999996,
+      "loss": 4.7329,
+      "step": 423
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00025079999999999997,
+      "loss": 4.8088,
+      "step": 424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002514,
+      "loss": 4.8696,
+      "step": 425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 4.8655,
+      "step": 426
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00025259999999999996,
+      "loss": 4.6672,
+      "step": 427
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002532,
+      "loss": 4.8857,
+      "step": 428
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002538,
+      "loss": 4.7625,
+      "step": 429
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00025439999999999995,
+      "loss": 4.923,
+      "step": 430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00025499999999999996,
+      "loss": 4.9942,
+      "step": 431
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002556,
+      "loss": 4.8094,
+      "step": 432
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002562,
+      "loss": 4.5657,
+      "step": 433
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025679999999999995,
+      "loss": 4.7551,
+      "step": 434
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025739999999999997,
+      "loss": 4.6967,
+      "step": 435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000258,
+      "loss": 4.9245,
+      "step": 436
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002586,
+      "loss": 4.9264,
+      "step": 437
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025919999999999996,
+      "loss": 4.842,
+      "step": 438
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025979999999999997,
+      "loss": 4.7254,
+      "step": 439
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002604,
+      "loss": 4.6844,
+      "step": 440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000261,
+      "loss": 4.8309,
+      "step": 441
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00026159999999999996,
+      "loss": 4.7051,
+      "step": 442
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002622,
+      "loss": 4.7382,
+      "step": 443
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002628,
+      "loss": 4.9942,
+      "step": 444
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00026339999999999995,
+      "loss": 4.6426,
+      "step": 445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 4.45,
+      "step": 446
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002646,
+      "loss": 4.5201,
+      "step": 447
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002652,
+      "loss": 4.6927,
+      "step": 448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00026579999999999996,
+      "loss": 4.4609,
+      "step": 449
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00026639999999999997,
+      "loss": 4.6057,
+      "step": 450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000267,
+      "loss": 5.8256,
+      "step": 451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002676,
+      "loss": 5.6904,
+      "step": 452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00026819999999999996,
+      "loss": 5.6517,
+      "step": 453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002688,
+      "loss": 5.0165,
+      "step": 454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002694,
+      "loss": 5.1745,
+      "step": 455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027,
+      "loss": 5.1104,
+      "step": 456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027059999999999996,
+      "loss": 4.9597,
+      "step": 457
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002712,
+      "loss": 4.9,
+      "step": 458
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002718,
+      "loss": 4.9072,
+      "step": 459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002724,
+      "loss": 5.0076,
+      "step": 460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027299999999999997,
+      "loss": 4.6326,
+      "step": 461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002736,
+      "loss": 4.7496,
+      "step": 462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002742,
+      "loss": 4.7875,
+      "step": 463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002748,
+      "loss": 4.6762,
+      "step": 464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027539999999999997,
+      "loss": 4.8027,
+      "step": 465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000276,
+      "loss": 4.7373,
+      "step": 466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002766,
+      "loss": 4.7553,
+      "step": 467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002772,
+      "loss": 4.6693,
+      "step": 468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002778,
+      "loss": 4.784,
+      "step": 469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002784,
+      "loss": 4.6701,
+      "step": 470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000279,
+      "loss": 4.6513,
+      "step": 471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027959999999999997,
+      "loss": 4.6554,
+      "step": 472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002802,
+      "loss": 4.6965,
+      "step": 473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002808,
+      "loss": 4.6918,
+      "step": 474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00028139999999999996,
+      "loss": 4.7398,
+      "step": 475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 4.7939,
+      "step": 476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002826,
+      "loss": 4.8769,
+      "step": 477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00028319999999999994,
+      "loss": 4.6449,
+      "step": 478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00028379999999999996,
+      "loss": 4.4692,
+      "step": 479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002844,
+      "loss": 4.7562,
+      "step": 480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000285,
+      "loss": 4.6195,
+      "step": 481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00028559999999999995,
+      "loss": 4.5762,
+      "step": 482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00028619999999999996,
+      "loss": 4.778,
+      "step": 483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002868,
+      "loss": 4.7613,
+      "step": 484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00028739999999999994,
+      "loss": 4.8086,
+      "step": 485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 4.6631,
+      "step": 486
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00028859999999999997,
+      "loss": 4.6249,
+      "step": 487
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002892,
+      "loss": 4.6237,
+      "step": 488
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00028979999999999994,
+      "loss": 4.7287,
+      "step": 489
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029039999999999996,
+      "loss": 4.8408,
+      "step": 490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029099999999999997,
+      "loss": 4.5833,
+      "step": 491
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002916,
+      "loss": 4.7231,
+      "step": 492
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029219999999999995,
+      "loss": 4.7464,
+      "step": 493
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029279999999999996,
+      "loss": 4.6638,
+      "step": 494
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002934,
+      "loss": 4.7317,
+      "step": 495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000294,
+      "loss": 4.4279,
+      "step": 496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029459999999999995,
+      "loss": 4.1769,
+      "step": 497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029519999999999997,
+      "loss": 4.9004,
+      "step": 498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002958,
+      "loss": 4.5785,
+      "step": 499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002964,
+      "loss": 3.9983,
+      "step": 500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029699999999999996,
+      "loss": 6.0793,
+      "step": 501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029759999999999997,
+      "loss": 5.4207,
+      "step": 502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002982,
+      "loss": 5.3129,
+      "step": 503
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002988,
+      "loss": 5.1749,
+      "step": 504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029939999999999996,
+      "loss": 5.035,
+      "step": 505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003,
+      "loss": 5.0972,
+      "step": 506
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000299234693877551,
+      "loss": 4.9699,
+      "step": 507
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029846938775510205,
+      "loss": 4.8904,
+      "step": 508
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029770408163265304,
+      "loss": 4.9586,
+      "step": 509
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002969387755102041,
+      "loss": 4.8407,
+      "step": 510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029617346938775506,
+      "loss": 4.7549,
+      "step": 511
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002954081632653061,
+      "loss": 4.6833,
+      "step": 512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002946428571428571,
+      "loss": 5.0416,
+      "step": 513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002938775510204081,
+      "loss": 4.7472,
+      "step": 514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029311224489795917,
+      "loss": 4.7657,
+      "step": 515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002923469387755102,
+      "loss": 4.5764,
+      "step": 516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002915816326530612,
+      "loss": 4.8957,
+      "step": 517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029081632653061223,
+      "loss": 4.7626,
+      "step": 518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002900510204081632,
+      "loss": 4.7969,
+      "step": 519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00028928571428571425,
+      "loss": 4.7956,
+      "step": 520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002885204081632653,
+      "loss": 4.7104,
+      "step": 521
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002877551020408163,
+      "loss": 4.661,
+      "step": 522
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002869897959183673,
+      "loss": 4.5696,
+      "step": 523
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00028622448979591836,
+      "loss": 4.6741,
+      "step": 524
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002854591836734694,
+      "loss": 4.7646,
+      "step": 525
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002846938775510204,
+      "loss": 4.5753,
+      "step": 526
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00028392857142857137,
+      "loss": 4.64,
+      "step": 527
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002831632653061224,
+      "loss": 4.7073,
+      "step": 528
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00028239795918367345,
+      "loss": 4.6321,
+      "step": 529
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002816326530612245,
+      "loss": 4.9232,
+      "step": 530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00028086734693877547,
+      "loss": 4.6835,
+      "step": 531
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002801020408163265,
+      "loss": 4.4473,
+      "step": 532
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00027933673469387755,
+      "loss": 4.5455,
+      "step": 533
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00027857142857142854,
+      "loss": 4.5822,
+      "step": 534
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002778061224489796,
+      "loss": 4.577,
+      "step": 535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00027704081632653056,
+      "loss": 4.7212,
+      "step": 536
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002762755102040816,
+      "loss": 4.7135,
+      "step": 537
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00027551020408163264,
+      "loss": 4.7634,
+      "step": 538
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002747448979591837,
+      "loss": 4.529,
+      "step": 539
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00027397959183673466,
+      "loss": 4.6949,
+      "step": 540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002732142857142857,
+      "loss": 4.6912,
+      "step": 541
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002724489795918367,
+      "loss": 4.7091,
+      "step": 542
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00027168367346938773,
+      "loss": 4.7672,
+      "step": 543
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002709183673469387,
+      "loss": 4.4937,
+      "step": 544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00027015306122448975,
+      "loss": 4.7518,
+      "step": 545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002693877551020408,
+      "loss": 4.5006,
+      "step": 546
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026862244897959183,
+      "loss": 4.599,
+      "step": 547
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026785714285714287,
+      "loss": 4.5588,
+      "step": 548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00026709183673469386,
+      "loss": 4.0995,
+      "step": 549
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002663265306122449,
+      "loss": 4.6646,
+      "step": 550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002655612244897959,
+      "loss": 5.6484,
+      "step": 551
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002647959183673469,
+      "loss": 5.9121,
+      "step": 552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002640306122448979,
+      "loss": 5.6086,
+      "step": 553
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00026326530612244894,
+      "loss": 5.2566,
+      "step": 554
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002625,
+      "loss": 5.0084,
+      "step": 555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000261734693877551,
+      "loss": 4.9042,
+      "step": 556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000260969387755102,
+      "loss": 4.7538,
+      "step": 557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00026020408163265305,
+      "loss": 4.7717,
+      "step": 558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00025943877551020403,
+      "loss": 4.7679,
+      "step": 559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00025867346938775507,
+      "loss": 4.6943,
+      "step": 560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002579081632653061,
+      "loss": 4.8734,
+      "step": 561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002571428571428571,
+      "loss": 4.6494,
+      "step": 562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00025637755102040814,
+      "loss": 4.6779,
+      "step": 563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002556122448979592,
+      "loss": 4.6811,
+      "step": 564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002548469387755102,
+      "loss": 4.7522,
+      "step": 565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002540816326530612,
+      "loss": 4.6224,
+      "step": 566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002533163265306122,
+      "loss": 4.6819,
+      "step": 567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002525510204081632,
+      "loss": 4.5401,
+      "step": 568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00025178571428571426,
+      "loss": 4.6458,
+      "step": 569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002510204081632653,
+      "loss": 4.6042,
+      "step": 570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002502551020408163,
+      "loss": 4.4918,
+      "step": 571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00024948979591836733,
+      "loss": 4.6797,
+      "step": 572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00024872448979591837,
+      "loss": 4.5863,
+      "step": 573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00024795918367346935,
+      "loss": 4.4528,
+      "step": 574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002471938775510204,
+      "loss": 4.6643,
+      "step": 575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002464285714285714,
+      "loss": 4.5213,
+      "step": 576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002456632653061224,
+      "loss": 4.6087,
+      "step": 577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00024489795918367346,
+      "loss": 4.6959,
+      "step": 578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00024413265306122447,
+      "loss": 4.3075,
+      "step": 579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00024336734693877548,
+      "loss": 4.7373,
+      "step": 580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002426020408163265,
+      "loss": 4.7056,
+      "step": 581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00024183673469387753,
+      "loss": 4.4479,
+      "step": 582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00024107142857142857,
+      "loss": 4.4347,
+      "step": 583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00024030612244897956,
+      "loss": 4.5181,
+      "step": 584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00023954081632653057,
+      "loss": 4.6854,
+      "step": 585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002387755102040816,
+      "loss": 4.5102,
+      "step": 586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00023801020408163265,
+      "loss": 4.6516,
+      "step": 587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00023724489795918366,
+      "loss": 4.4988,
+      "step": 588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00023647959183673467,
+      "loss": 4.391,
+      "step": 589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00023571428571428569,
+      "loss": 4.4491,
+      "step": 590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00023494897959183673,
+      "loss": 4.6666,
+      "step": 591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00023418367346938774,
+      "loss": 4.5857,
+      "step": 592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00023341836734693875,
+      "loss": 4.7841,
+      "step": 593
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00023265306122448976,
+      "loss": 4.3905,
+      "step": 594
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002318877551020408,
+      "loss": 4.4626,
+      "step": 595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00023112244897959181,
+      "loss": 4.122,
+      "step": 596
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00023035714285714285,
+      "loss": 4.2741,
+      "step": 597
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00022959183673469384,
+      "loss": 4.2433,
+      "step": 598
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00022882653061224488,
+      "loss": 3.9966,
+      "step": 599
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002280612244897959,
+      "loss": 3.7098,
+      "step": 600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00022729591836734693,
+      "loss": 5.4903,
+      "step": 601
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00022653061224489791,
+      "loss": 5.1487,
+      "step": 602
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00022576530612244895,
+      "loss": 5.1314,
+      "step": 603
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000225,
+      "loss": 4.9014,
+      "step": 604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000224234693877551,
+      "loss": 4.8637,
+      "step": 605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00022346938775510205,
+      "loss": 4.6627,
+      "step": 606
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00022270408163265303,
+      "loss": 4.731,
+      "step": 607
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00022193877551020407,
+      "loss": 4.7518,
+      "step": 608
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00022117346938775508,
+      "loss": 4.653,
+      "step": 609
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00022040816326530612,
+      "loss": 4.6627,
+      "step": 610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002196428571428571,
+      "loss": 4.5396,
+      "step": 611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00021887755102040815,
+      "loss": 4.7246,
+      "step": 612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00021811224489795916,
+      "loss": 4.5917,
+      "step": 613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002173469387755102,
+      "loss": 4.6016,
+      "step": 614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002165816326530612,
+      "loss": 4.6397,
+      "step": 615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00021581632653061222,
+      "loss": 4.568,
+      "step": 616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00021505102040816324,
+      "loss": 4.4928,
+      "step": 617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00021428571428571427,
+      "loss": 4.4731,
+      "step": 618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002135204081632653,
+      "loss": 4.4815,
+      "step": 619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002127551020408163,
+      "loss": 4.4908,
+      "step": 620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002119897959183673,
+      "loss": 4.4912,
+      "step": 621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021122448979591835,
+      "loss": 4.507,
+      "step": 622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002104591836734694,
+      "loss": 4.6002,
+      "step": 623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002096938775510204,
+      "loss": 4.4304,
+      "step": 624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002089285714285714,
+      "loss": 4.62,
+      "step": 625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00020816326530612243,
+      "loss": 4.6484,
+      "step": 626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00020739795918367347,
+      "loss": 4.5712,
+      "step": 627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00020663265306122448,
+      "loss": 4.4835,
+      "step": 628
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002058673469387755,
+      "loss": 4.4776,
+      "step": 629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002051020408163265,
+      "loss": 4.6577,
+      "step": 630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020433673469387754,
+      "loss": 4.5518,
+      "step": 631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020357142857142856,
+      "loss": 4.3805,
+      "step": 632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020280612244897957,
+      "loss": 4.5106,
+      "step": 633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020204081632653058,
+      "loss": 4.5316,
+      "step": 634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020127551020408162,
+      "loss": 4.5212,
+      "step": 635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020051020408163263,
+      "loss": 4.5112,
+      "step": 636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019974489795918367,
+      "loss": 4.41,
+      "step": 637
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019897959183673466,
+      "loss": 4.341,
+      "step": 638
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001982142857142857,
+      "loss": 4.5112,
+      "step": 639
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001974489795918367,
+      "loss": 4.2144,
+      "step": 640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019668367346938775,
+      "loss": 4.4248,
+      "step": 641
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019591836734693873,
+      "loss": 4.7221,
+      "step": 642
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019515306122448977,
+      "loss": 4.5176,
+      "step": 643
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001943877551020408,
+      "loss": 4.2739,
+      "step": 644
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019362244897959182,
+      "loss": 4.1045,
+      "step": 645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019285714285714286,
+      "loss": 4.2817,
+      "step": 646
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019209183673469385,
+      "loss": 4.0479,
+      "step": 647
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001913265306122449,
+      "loss": 3.954,
+      "step": 648
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001905612244897959,
+      "loss": 4.2274,
+      "step": 649
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018979591836734694,
+      "loss": 3.7388,
+      "step": 650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018903061224489793,
+      "loss": 5.5663,
+      "step": 651
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018826530612244896,
+      "loss": 5.492,
+      "step": 652
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018749999999999998,
+      "loss": 5.2419,
+      "step": 653
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018673469387755102,
+      "loss": 5.0356,
+      "step": 654
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018596938775510203,
+      "loss": 4.8648,
+      "step": 655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018520408163265304,
+      "loss": 4.7763,
+      "step": 656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018443877551020405,
+      "loss": 4.5477,
+      "step": 657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001836734693877551,
+      "loss": 4.8061,
+      "step": 658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001829081632653061,
+      "loss": 4.5695,
+      "step": 659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018214285714285712,
+      "loss": 4.649,
+      "step": 660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018137755102040813,
+      "loss": 4.5107,
+      "step": 661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018061224489795917,
+      "loss": 4.3821,
+      "step": 662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001798469387755102,
+      "loss": 4.6153,
+      "step": 663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017908163265306122,
+      "loss": 4.4183,
+      "step": 664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001783163265306122,
+      "loss": 4.3385,
+      "step": 665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017755102040816325,
+      "loss": 4.667,
+      "step": 666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017678571428571428,
+      "loss": 4.4584,
+      "step": 667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001760204081632653,
+      "loss": 4.5485,
+      "step": 668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001752551020408163,
+      "loss": 4.6394,
+      "step": 669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017448979591836732,
+      "loss": 4.4904,
+      "step": 670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017372448979591836,
+      "loss": 4.3553,
+      "step": 671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017295918367346937,
+      "loss": 4.3642,
+      "step": 672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001721938775510204,
+      "loss": 4.3629,
+      "step": 673
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001714285714285714,
+      "loss": 4.3463,
+      "step": 674
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017066326530612244,
+      "loss": 4.3872,
+      "step": 675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016989795918367345,
+      "loss": 4.4303,
+      "step": 676
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001691326530612245,
+      "loss": 4.3736,
+      "step": 677
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016836734693877547,
+      "loss": 4.1379,
+      "step": 678
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016760204081632651,
+      "loss": 4.2715,
+      "step": 679
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016683673469387753,
+      "loss": 4.3948,
+      "step": 680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016607142857142857,
+      "loss": 4.5069,
+      "step": 681
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016530612244897955,
+      "loss": 4.2144,
+      "step": 682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001645408163265306,
+      "loss": 4.3885,
+      "step": 683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016377551020408163,
+      "loss": 4.4772,
+      "step": 684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016301020408163264,
+      "loss": 4.4007,
+      "step": 685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016224489795918368,
+      "loss": 4.3419,
+      "step": 686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016147959183673467,
+      "loss": 4.4038,
+      "step": 687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001607142857142857,
+      "loss": 4.4789,
+      "step": 688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00015994897959183672,
+      "loss": 4.3293,
+      "step": 689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00015918367346938776,
+      "loss": 4.5397,
+      "step": 690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00015841836734693874,
+      "loss": 4.1725,
+      "step": 691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00015765306122448978,
+      "loss": 4.2432,
+      "step": 692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001568877551020408,
+      "loss": 4.1948,
+      "step": 693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00015612244897959183,
+      "loss": 4.4819,
+      "step": 694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00015535714285714285,
+      "loss": 4.1618,
+      "step": 695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00015459183673469386,
+      "loss": 4.1592,
+      "step": 696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00015382653061224487,
+      "loss": 4.1156,
+      "step": 697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001530612244897959,
+      "loss": 3.7754,
+      "step": 698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00015229591836734692,
+      "loss": 3.8484,
+      "step": 699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00015153061224489794,
+      "loss": 3.4047,
+      "step": 700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00015076530612244895,
+      "loss": 5.5129,
+      "step": 701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00015,
+      "loss": 5.6794,
+      "step": 702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014923469387755103,
+      "loss": 5.2709,
+      "step": 703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014846938775510204,
+      "loss": 4.8596,
+      "step": 704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014770408163265305,
+      "loss": 4.7869,
+      "step": 705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014693877551020406,
+      "loss": 4.9212,
+      "step": 706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001461734693877551,
+      "loss": 4.759,
+      "step": 707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014540816326530611,
+      "loss": 4.7732,
+      "step": 708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014464285714285713,
+      "loss": 4.6,
+      "step": 709
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014387755102040814,
+      "loss": 4.5738,
+      "step": 710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014311224489795918,
+      "loss": 4.4732,
+      "step": 711
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001423469387755102,
+      "loss": 4.6386,
+      "step": 712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001415816326530612,
+      "loss": 4.4168,
+      "step": 713
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014081632653061224,
+      "loss": 4.6063,
+      "step": 714
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014005102040816326,
+      "loss": 4.5329,
+      "step": 715
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013928571428571427,
+      "loss": 4.3883,
+      "step": 716
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013852040816326528,
+      "loss": 4.8677,
+      "step": 717
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013775510204081632,
+      "loss": 4.4869,
+      "step": 718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013698979591836733,
+      "loss": 4.223,
+      "step": 719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013622448979591834,
+      "loss": 4.4263,
+      "step": 720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013545918367346936,
+      "loss": 4.3025,
+      "step": 721
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001346938775510204,
+      "loss": 4.473,
+      "step": 722
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013392857142857144,
+      "loss": 4.2456,
+      "step": 723
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013316326530612245,
+      "loss": 4.4833,
+      "step": 724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013239795918367346,
+      "loss": 4.361,
+      "step": 725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013163265306122447,
+      "loss": 4.4559,
+      "step": 726
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001308673469387755,
+      "loss": 4.3644,
+      "step": 727
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013010204081632652,
+      "loss": 4.3202,
+      "step": 728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012933673469387754,
+      "loss": 4.1825,
+      "step": 729
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012857142857142855,
+      "loss": 4.4168,
+      "step": 730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001278061224489796,
+      "loss": 4.4075,
+      "step": 731
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001270408163265306,
+      "loss": 4.3405,
+      "step": 732
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001262755102040816,
+      "loss": 4.2644,
+      "step": 733
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012551020408163265,
+      "loss": 4.3358,
+      "step": 734
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012474489795918366,
+      "loss": 4.1384,
+      "step": 735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012397959183673468,
+      "loss": 4.2409,
+      "step": 736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001232142857142857,
+      "loss": 4.3069,
+      "step": 737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012244897959183673,
+      "loss": 4.3357,
+      "step": 738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012168367346938774,
+      "loss": 4.2431,
+      "step": 739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012091836734693877,
+      "loss": 4.4386,
+      "step": 740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012015306122448978,
+      "loss": 4.3386,
+      "step": 741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001193877551020408,
+      "loss": 4.1781,
+      "step": 742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00011862244897959183,
+      "loss": 3.9827,
+      "step": 743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00011785714285714284,
+      "loss": 4.0762,
+      "step": 744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011709183673469387,
+      "loss": 4.0074,
+      "step": 745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011632653061224488,
+      "loss": 4.3218,
+      "step": 746
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011556122448979591,
+      "loss": 3.7867,
+      "step": 747
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011479591836734692,
+      "loss": 3.7808,
+      "step": 748
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011403061224489795,
+      "loss": 3.8772,
+      "step": 749
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011326530612244896,
+      "loss": 3.5621,
+      "step": 750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001125,
+      "loss": 5.3619,
+      "step": 751
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011173469387755102,
+      "loss": 5.3502,
+      "step": 752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011096938775510204,
+      "loss": 5.1276,
+      "step": 753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011020408163265306,
+      "loss": 4.7787,
+      "step": 754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010943877551020407,
+      "loss": 4.8037,
+      "step": 755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001086734693877551,
+      "loss": 4.7847,
+      "step": 756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010790816326530611,
+      "loss": 4.5984,
+      "step": 757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010714285714285714,
+      "loss": 4.7408,
+      "step": 758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010637755102040815,
+      "loss": 4.5638,
+      "step": 759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010561224489795918,
+      "loss": 4.5597,
+      "step": 760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001048469387755102,
+      "loss": 4.4244,
+      "step": 761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010408163265306121,
+      "loss": 4.4303,
+      "step": 762
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010331632653061224,
+      "loss": 4.4948,
+      "step": 763
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010255102040816325,
+      "loss": 4.2372,
+      "step": 764
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010178571428571428,
+      "loss": 4.3861,
+      "step": 765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010102040816326529,
+      "loss": 4.324,
+      "step": 766
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010025510204081632,
+      "loss": 4.2721,
+      "step": 767
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.948979591836733e-05,
+      "loss": 4.3372,
+      "step": 768
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.872448979591835e-05,
+      "loss": 4.4393,
+      "step": 769
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.795918367346937e-05,
+      "loss": 4.4343,
+      "step": 770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.71938775510204e-05,
+      "loss": 4.2603,
+      "step": 771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.642857142857143e-05,
+      "loss": 4.3608,
+      "step": 772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.566326530612244e-05,
+      "loss": 4.41,
+      "step": 773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.489795918367347e-05,
+      "loss": 4.0719,
+      "step": 774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.413265306122448e-05,
+      "loss": 4.0283,
+      "step": 775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.336734693877551e-05,
+      "loss": 4.2473,
+      "step": 776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.260204081632652e-05,
+      "loss": 4.2656,
+      "step": 777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.183673469387755e-05,
+      "loss": 4.3672,
+      "step": 778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.107142857142856e-05,
+      "loss": 4.0381,
+      "step": 779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.030612244897958e-05,
+      "loss": 4.1203,
+      "step": 780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.954081632653061e-05,
+      "loss": 4.1021,
+      "step": 781
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.877551020408162e-05,
+      "loss": 4.1532,
+      "step": 782
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.801020408163265e-05,
+      "loss": 4.2019,
+      "step": 783
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.724489795918366e-05,
+      "loss": 4.1272,
+      "step": 784
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.647959183673469e-05,
+      "loss": 4.1634,
+      "step": 785
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.57142857142857e-05,
+      "loss": 4.087,
+      "step": 786
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.494897959183672e-05,
+      "loss": 4.1283,
+      "step": 787
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.418367346938774e-05,
+      "loss": 4.217,
+      "step": 788
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.341836734693876e-05,
+      "loss": 4.1432,
+      "step": 789
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.265306122448978e-05,
+      "loss": 4.1936,
+      "step": 790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.188775510204081e-05,
+      "loss": 4.1346,
+      "step": 791
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.112244897959184e-05,
+      "loss": 3.8563,
+      "step": 792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.035714285714285e-05,
+      "loss": 4.0926,
+      "step": 793
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.959183673469388e-05,
+      "loss": 4.1567,
+      "step": 794
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.882653061224489e-05,
+      "loss": 3.996,
+      "step": 795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.806122448979592e-05,
+      "loss": 3.9794,
+      "step": 796
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.729591836734693e-05,
+      "loss": 3.8768,
+      "step": 797
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.653061224489796e-05,
+      "loss": 3.707,
+      "step": 798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.576530612244897e-05,
+      "loss": 3.739,
+      "step": 799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.5e-05,
+      "loss": 3.3022,
+      "step": 800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.423469387755102e-05,
+      "loss": 5.0181,
+      "step": 801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.346938775510203e-05,
+      "loss": 4.8254,
+      "step": 802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.270408163265306e-05,
+      "loss": 4.9372,
+      "step": 803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.193877551020407e-05,
+      "loss": 4.7835,
+      "step": 804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.11734693877551e-05,
+      "loss": 4.4648,
+      "step": 805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.040816326530612e-05,
+      "loss": 4.5644,
+      "step": 806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.964285714285713e-05,
+      "loss": 4.474,
+      "step": 807
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.887755102040816e-05,
+      "loss": 4.2169,
+      "step": 808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.811224489795917e-05,
+      "loss": 4.39,
+      "step": 809
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.73469387755102e-05,
+      "loss": 4.1697,
+      "step": 810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.658163265306122e-05,
+      "loss": 4.4591,
+      "step": 811
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.581632653061224e-05,
+      "loss": 4.5694,
+      "step": 812
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.505102040816326e-05,
+      "loss": 4.1565,
+      "step": 813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.428571428571427e-05,
+      "loss": 4.2417,
+      "step": 814
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.35204081632653e-05,
+      "loss": 4.1501,
+      "step": 815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.275510204081633e-05,
+      "loss": 4.1563,
+      "step": 816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.198979591836734e-05,
+      "loss": 4.1727,
+      "step": 817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.122448979591836e-05,
+      "loss": 4.1451,
+      "step": 818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.045918367346938e-05,
+      "loss": 4.0258,
+      "step": 819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.96938775510204e-05,
+      "loss": 4.1776,
+      "step": 820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.892857142857142e-05,
+      "loss": 4.0993,
+      "step": 821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.816326530612244e-05,
+      "loss": 4.0599,
+      "step": 822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.739795918367346e-05,
+      "loss": 4.0424,
+      "step": 823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.663265306122448e-05,
+      "loss": 4.1469,
+      "step": 824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.586734693877551e-05,
+      "loss": 4.3691,
+      "step": 825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.510204081632653e-05,
+      "loss": 4.2156,
+      "step": 826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.433673469387755e-05,
+      "loss": 4.0954,
+      "step": 827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.357142857142857e-05,
+      "loss": 3.9803,
+      "step": 828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.280612244897959e-05,
+      "loss": 3.9851,
+      "step": 829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.204081632653061e-05,
+      "loss": 4.1883,
+      "step": 830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.1275510204081626e-05,
+      "loss": 3.9742,
+      "step": 831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.0510204081632645e-05,
+      "loss": 3.9406,
+      "step": 832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9744897959183664e-05,
+      "loss": 4.1378,
+      "step": 833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.897959183673468e-05,
+      "loss": 4.0522,
+      "step": 834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8214285714285716e-05,
+      "loss": 4.0633,
+      "step": 835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7448979591836735e-05,
+      "loss": 4.0434,
+      "step": 836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6683673469387754e-05,
+      "loss": 3.9876,
+      "step": 837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.591836734693877e-05,
+      "loss": 3.9931,
+      "step": 838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.515306122448979e-05,
+      "loss": 3.9029,
+      "step": 839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.438775510204081e-05,
+      "loss": 3.8351,
+      "step": 840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.362244897959183e-05,
+      "loss": 4.0748,
+      "step": 841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.285714285714285e-05,
+      "loss": 3.8285,
+      "step": 842
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.209183673469387e-05,
+      "loss": 4.0758,
+      "step": 843
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.132653061224489e-05,
+      "loss": 3.8493,
+      "step": 844
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.056122448979592e-05,
+      "loss": 3.7991,
+      "step": 845
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.979591836734694e-05,
+      "loss": 3.7989,
+      "step": 846
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.903061224489796e-05,
+      "loss": 3.4765,
+      "step": 847
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.826530612244898e-05,
+      "loss": 3.5615,
+      "step": 848
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.75e-05,
+      "loss": 3.3629,
+      "step": 849
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6734693877551016e-05,
+      "loss": 3.0621,
+      "step": 850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5969387755102035e-05,
+      "loss": 4.6279,
+      "step": 851
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.520408163265306e-05,
+      "loss": 4.5867,
+      "step": 852
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.443877551020408e-05,
+      "loss": 4.7184,
+      "step": 853
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.36734693877551e-05,
+      "loss": 4.6184,
+      "step": 854
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.290816326530612e-05,
+      "loss": 4.4629,
+      "step": 855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.214285714285714e-05,
+      "loss": 4.4819,
+      "step": 856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.137755102040816e-05,
+      "loss": 4.665,
+      "step": 857
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.061224489795918e-05,
+      "loss": 4.4192,
+      "step": 858
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.98469387755102e-05,
+      "loss": 4.201,
+      "step": 859
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.908163265306122e-05,
+      "loss": 4.3568,
+      "step": 860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.831632653061224e-05,
+      "loss": 4.41,
+      "step": 861
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7551020408163265e-05,
+      "loss": 4.1868,
+      "step": 862
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6785714285714284e-05,
+      "loss": 4.2379,
+      "step": 863
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6020408163265303e-05,
+      "loss": 4.401,
+      "step": 864
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5255102040816323e-05,
+      "loss": 4.1638,
+      "step": 865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.448979591836734e-05,
+      "loss": 3.9931,
+      "step": 866
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3724489795918367e-05,
+      "loss": 4.201,
+      "step": 867
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2959183673469387e-05,
+      "loss": 4.0319,
+      "step": 868
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2193877551020406e-05,
+      "loss": 3.9311,
+      "step": 869
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1428571428571425e-05,
+      "loss": 4.2493,
+      "step": 870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0663265306122444e-05,
+      "loss": 3.9831,
+      "step": 871
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.989795918367347e-05,
+      "loss": 4.0472,
+      "step": 872
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.913265306122449e-05,
+      "loss": 4.1426,
+      "step": 873
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8367346938775508e-05,
+      "loss": 4.0451,
+      "step": 874
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.760204081632653e-05,
+      "loss": 4.1395,
+      "step": 875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.683673469387755e-05,
+      "loss": 3.9079,
+      "step": 876
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.607142857142857e-05,
+      "loss": 3.9106,
+      "step": 877
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.530612244897959e-05,
+      "loss": 4.1012,
+      "step": 878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.454081632653061e-05,
+      "loss": 4.0773,
+      "step": 879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3775510204081633e-05,
+      "loss": 4.2916,
+      "step": 880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3010204081632652e-05,
+      "loss": 3.8016,
+      "step": 881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.224489795918367e-05,
+      "loss": 4.0127,
+      "step": 882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1479591836734693e-05,
+      "loss": 3.8414,
+      "step": 883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0714285714285712e-05,
+      "loss": 3.7901,
+      "step": 884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.948979591836735e-06,
+      "loss": 3.7644,
+      "step": 885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.183673469387754e-06,
+      "loss": 3.8546,
+      "step": 886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.418367346938775e-06,
+      "loss": 3.6187,
+      "step": 887
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.653061224489796e-06,
+      "loss": 3.7715,
+      "step": 888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.887755102040816e-06,
+      "loss": 3.748,
+      "step": 889
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.122448979591835e-06,
+      "loss": 3.2923,
+      "step": 890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.357142857142856e-06,
+      "loss": 3.4304,
+      "step": 891
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.591836734693877e-06,
+      "loss": 3.0179,
+      "step": 892
+    },
+    {
+      "epoch": 1.0,
+      "step": 892,
+      "total_flos": 0.0,
+      "train_loss": 4.8473983449786235,
+      "train_runtime": 5009.4893,
+      "train_samples_per_second": 5.697,
+      "train_steps_per_second": 0.178
+    }
+  ],
+  "max_steps": 892,
+  "num_train_epochs": 1,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}