diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,32227 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 5352,
+  "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": 5.0347,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 5.1538,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-07,
+      "loss": 5.3522,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-06,
+      "loss": 4.806,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-06,
+      "loss": 4.9661,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-06,
+      "loss": 4.9227,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999999999999997e-06,
+      "loss": 4.8098,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-06,
+      "loss": 4.9628,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-06,
+      "loss": 5.1179,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8e-06,
+      "loss": 4.8084,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.399999999999999e-06,
+      "loss": 4.8764,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 4.8278,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.599999999999999e-06,
+      "loss": 4.9859,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2e-06,
+      "loss": 4.6472,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.799999999999998e-06,
+      "loss": 4.7015,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.4e-06,
+      "loss": 4.8147,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.999999999999999e-06,
+      "loss": 4.8334,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.6e-06,
+      "loss": 4.6314,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.02e-05,
+      "loss": 4.6675,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0799999999999998e-05,
+      "loss": 4.8608,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.14e-05,
+      "loss": 4.5897,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 5.0104,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.26e-05,
+      "loss": 4.7388,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3199999999999997e-05,
+      "loss": 4.4948,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3799999999999998e-05,
+      "loss": 4.6334,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.44e-05,
+      "loss": 4.7224,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 4.546,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5599999999999996e-05,
+      "loss": 4.3284,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6199999999999997e-05,
+      "loss": 4.6369,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.68e-05,
+      "loss": 4.4973,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.74e-05,
+      "loss": 4.3774,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 4.7679,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8599999999999998e-05,
+      "loss": 4.485,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.92e-05,
+      "loss": 4.3252,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.98e-05,
+      "loss": 4.5852,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.04e-05,
+      "loss": 4.7339,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1e-05,
+      "loss": 4.4773,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1599999999999996e-05,
+      "loss": 4.6284,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2199999999999998e-05,
+      "loss": 4.7114,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.28e-05,
+      "loss": 4.6472,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.34e-05,
+      "loss": 4.4312,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 4.3292,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 4.5945,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.52e-05,
+      "loss": 4.446,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5799999999999997e-05,
+      "loss": 4.6805,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6399999999999995e-05,
+      "loss": 4.6733,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6399999999999995e-05,
+      "loss": 4.7317,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 4.968,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7599999999999997e-05,
+      "loss": 4.9883,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 4.9689,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.88e-05,
+      "loss": 4.4177,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.94e-05,
+      "loss": 4.5179,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 4.5067,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.06e-05,
+      "loss": 4.3917,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.119999999999999e-05,
+      "loss": 4.4934,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1799999999999994e-05,
+      "loss": 4.531,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2399999999999995e-05,
+      "loss": 4.4738,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 4.3973,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.36e-05,
+      "loss": 4.4008,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.42e-05,
+      "loss": 4.4024,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.48e-05,
+      "loss": 4.5088,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.539999999999999e-05,
+      "loss": 4.4178,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 4.7433,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6599999999999995e-05,
+      "loss": 4.2205,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7199999999999996e-05,
+      "loss": 4.4178,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.78e-05,
+      "loss": 4.5211,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.84e-05,
+      "loss": 4.5313,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9e-05,
+      "loss": 4.5319,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.96e-05,
+      "loss": 4.4488,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.02e-05,
+      "loss": 4.3718,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.08e-05,
+      "loss": 4.4829,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.14e-05,
+      "loss": 4.1671,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2e-05,
+      "loss": 4.5087,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.259999999999999e-05,
+      "loss": 4.2754,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.319999999999999e-05,
+      "loss": 4.7867,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3799999999999994e-05,
+      "loss": 4.579,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.4399999999999995e-05,
+      "loss": 4.3593,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 4.3001,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.56e-05,
+      "loss": 4.3357,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.62e-05,
+      "loss": 4.6133,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.68e-05,
+      "loss": 4.2947,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7399999999999993e-05,
+      "loss": 4.4702,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 4.2645,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8599999999999995e-05,
+      "loss": 4.394,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9199999999999997e-05,
+      "loss": 4.323,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98e-05,
+      "loss": 4.3164,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.04e-05,
+      "loss": 4.7098,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1e-05,
+      "loss": 4.2797,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1599999999999994e-05,
+      "loss": 4.634,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2199999999999995e-05,
+      "loss": 4.1559,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.279999999999999e-05,
+      "loss": 4.6111,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.339999999999999e-05,
+      "loss": 4.2188,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 4.1351,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.459999999999999e-05,
+      "loss": 4.53,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.519999999999999e-05,
+      "loss": 4.7714,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.5799999999999994e-05,
+      "loss": 4.4468,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 4.6357,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 4.5265,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.76e-05,
+      "loss": 4.8286,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.82e-05,
+      "loss": 4.7722,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.88e-05,
+      "loss": 4.4797,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.94e-05,
+      "loss": 4.4386,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 4.4645,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.0599999999999996e-05,
+      "loss": 4.3597,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.12e-05,
+      "loss": 4.4285,
+      "step": 105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.18e-05,
+      "loss": 4.4002,
+      "step": 106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.239999999999999e-05,
+      "loss": 4.4149,
+      "step": 107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.299999999999999e-05,
+      "loss": 4.5237,
+      "step": 108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.359999999999999e-05,
+      "loss": 4.3944,
+      "step": 109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.419999999999999e-05,
+      "loss": 4.4581,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.479999999999999e-05,
+      "loss": 4.3656,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.539999999999999e-05,
+      "loss": 4.242,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 4.4439,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.659999999999999e-05,
+      "loss": 4.1988,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.72e-05,
+      "loss": 4.1937,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.78e-05,
+      "loss": 4.3388,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.84e-05,
+      "loss": 4.5135,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.9e-05,
+      "loss": 4.3099,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.96e-05,
+      "loss": 4.4612,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.02e-05,
+      "loss": 4.5471,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.079999999999999e-05,
+      "loss": 4.5253,
+      "step": 121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.139999999999999e-05,
+      "loss": 4.4407,
+      "step": 122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 4.4566,
+      "step": 123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.259999999999999e-05,
+      "loss": 4.4137,
+      "step": 124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.319999999999999e-05,
+      "loss": 4.1583,
+      "step": 125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.379999999999999e-05,
+      "loss": 4.7014,
+      "step": 126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.439999999999999e-05,
+      "loss": 4.4383,
+      "step": 127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.5e-05,
+      "loss": 4.4343,
+      "step": 128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.56e-05,
+      "loss": 4.2959,
+      "step": 129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.62e-05,
+      "loss": 4.4316,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.68e-05,
+      "loss": 4.6903,
+      "step": 131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.74e-05,
+      "loss": 4.4366,
+      "step": 132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.8e-05,
+      "loss": 4.4498,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.86e-05,
+      "loss": 4.4776,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.92e-05,
+      "loss": 4.3126,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.98e-05,
+      "loss": 4.6324,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.04e-05,
+      "loss": 4.2774,
+      "step": 137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.1e-05,
+      "loss": 4.5811,
+      "step": 138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.16e-05,
+      "loss": 4.6809,
+      "step": 139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.22e-05,
+      "loss": 4.3594,
+      "step": 140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.28e-05,
+      "loss": 4.6116,
+      "step": 141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.34e-05,
+      "loss": 4.7769,
+      "step": 142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.4e-05,
+      "loss": 4.4567,
+      "step": 143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.459999999999998e-05,
+      "loss": 4.5362,
+      "step": 144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.519999999999998e-05,
+      "loss": 4.8371,
+      "step": 145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.579999999999998e-05,
+      "loss": 4.7281,
+      "step": 146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.639999999999999e-05,
+      "loss": 4.6324,
+      "step": 147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 4.6977,
+      "step": 148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.759999999999999e-05,
+      "loss": 4.9014,
+      "step": 149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.819999999999999e-05,
+      "loss": 4.3144,
+      "step": 150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.879999999999999e-05,
+      "loss": 4.4341,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.939999999999999e-05,
+      "loss": 5.0036,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 4.2174,
+      "step": 153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.059999999999999e-05,
+      "loss": 4.6666,
+      "step": 154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.12e-05,
+      "loss": 4.4909,
+      "step": 155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.18e-05,
+      "loss": 4.7215,
+      "step": 156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.24e-05,
+      "loss": 4.6221,
+      "step": 157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.3e-05,
+      "loss": 4.602,
+      "step": 158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.36e-05,
+      "loss": 4.6643,
+      "step": 159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.419999999999999e-05,
+      "loss": 4.7788,
+      "step": 160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.479999999999999e-05,
+      "loss": 4.5957,
+      "step": 161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.539999999999999e-05,
+      "loss": 4.872,
+      "step": 162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 4.5996,
+      "step": 163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.659999999999999e-05,
+      "loss": 4.8318,
+      "step": 164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.719999999999999e-05,
+      "loss": 4.5114,
+      "step": 165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.779999999999999e-05,
+      "loss": 4.7531,
+      "step": 166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.839999999999999e-05,
+      "loss": 4.4511,
+      "step": 167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9e-05,
+      "loss": 4.2934,
+      "step": 168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96e-05,
+      "loss": 4.3232,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001002,
+      "loss": 4.469,
+      "step": 170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001008,
+      "loss": 4.6166,
+      "step": 171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001014,
+      "loss": 4.739,
+      "step": 172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000102,
+      "loss": 4.4606,
+      "step": 173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001026,
+      "loss": 4.6898,
+      "step": 174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00010319999999999999,
+      "loss": 4.9279,
+      "step": 175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00010379999999999999,
+      "loss": 4.4113,
+      "step": 176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00010439999999999999,
+      "loss": 4.5888,
+      "step": 177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00010499999999999999,
+      "loss": 4.4544,
+      "step": 178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00010559999999999998,
+      "loss": 4.2751,
+      "step": 179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00010619999999999998,
+      "loss": 4.9034,
+      "step": 180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00010679999999999998,
+      "loss": 4.7315,
+      "step": 181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00010739999999999998,
+      "loss": 4.4414,
+      "step": 182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 4.6995,
+      "step": 183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00010859999999999998,
+      "loss": 4.6077,
+      "step": 184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00010919999999999998,
+      "loss": 4.6827,
+      "step": 185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00010979999999999999,
+      "loss": 4.6928,
+      "step": 186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00011039999999999999,
+      "loss": 4.554,
+      "step": 187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00011099999999999999,
+      "loss": 4.6813,
+      "step": 188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00011159999999999999,
+      "loss": 4.8874,
+      "step": 189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00011219999999999999,
+      "loss": 4.6254,
+      "step": 190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00011279999999999999,
+      "loss": 4.8119,
+      "step": 191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00011339999999999999,
+      "loss": 4.553,
+      "step": 192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 4.8291,
+      "step": 193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001146,
+      "loss": 4.5065,
+      "step": 194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001152,
+      "loss": 4.8022,
+      "step": 195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001158,
+      "loss": 5.3007,
+      "step": 196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001164,
+      "loss": 4.7359,
+      "step": 197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000117,
+      "loss": 4.6245,
+      "step": 198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001176,
+      "loss": 4.7074,
+      "step": 199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001182,
+      "loss": 4.5744,
+      "step": 200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001188,
+      "loss": 4.6738,
+      "step": 201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001194,
+      "loss": 4.5338,
+      "step": 202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 4.6975,
+      "step": 203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00012059999999999999,
+      "loss": 4.7613,
+      "step": 204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00012119999999999999,
+      "loss": 4.4438,
+      "step": 205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00012179999999999999,
+      "loss": 4.9233,
+      "step": 206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001224,
+      "loss": 4.743,
+      "step": 207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00012299999999999998,
+      "loss": 4.6789,
+      "step": 208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001236,
+      "loss": 4.4835,
+      "step": 209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00012419999999999998,
+      "loss": 4.7349,
+      "step": 210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00012479999999999997,
+      "loss": 4.5732,
+      "step": 211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00012539999999999999,
+      "loss": 4.8291,
+      "step": 212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 4.5537,
+      "step": 213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001266,
+      "loss": 4.3187,
+      "step": 214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00012719999999999997,
+      "loss": 5.2484,
+      "step": 215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001278,
+      "loss": 4.2393,
+      "step": 216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00012839999999999998,
+      "loss": 4.6971,
+      "step": 217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000129,
+      "loss": 4.547,
+      "step": 218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00012959999999999998,
+      "loss": 4.9665,
+      "step": 219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001302,
+      "loss": 5.2846,
+      "step": 220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00013079999999999998,
+      "loss": 4.6881,
+      "step": 221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001314,
+      "loss": 5.1414,
+      "step": 222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 4.5517,
+      "step": 223
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001326,
+      "loss": 4.63,
+      "step": 224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013319999999999999,
+      "loss": 4.8383,
+      "step": 225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001338,
+      "loss": 4.9055,
+      "step": 226
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001344,
+      "loss": 4.7254,
+      "step": 227
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000135,
+      "loss": 4.613,
+      "step": 228
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001356,
+      "loss": 4.6208,
+      "step": 229
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001362,
+      "loss": 4.6031,
+      "step": 230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001368,
+      "loss": 4.6104,
+      "step": 231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001374,
+      "loss": 4.9044,
+      "step": 232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000138,
+      "loss": 4.8058,
+      "step": 233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001386,
+      "loss": 4.6189,
+      "step": 234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001392,
+      "loss": 4.5056,
+      "step": 235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013979999999999998,
+      "loss": 5.1172,
+      "step": 236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001404,
+      "loss": 4.6722,
+      "step": 237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014099999999999998,
+      "loss": 4.8817,
+      "step": 238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014159999999999997,
+      "loss": 6.3849,
+      "step": 239
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001422,
+      "loss": 4.6616,
+      "step": 240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00014279999999999997,
+      "loss": 4.7094,
+      "step": 241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001434,
+      "loss": 4.6329,
+      "step": 242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 7.603,
+      "step": 243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001446,
+      "loss": 5.8798,
+      "step": 244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00014519999999999998,
+      "loss": 4.8108,
+      "step": 245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001458,
+      "loss": 4.8132,
+      "step": 246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00014639999999999998,
+      "loss": 5.4812,
+      "step": 247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000147,
+      "loss": 4.8593,
+      "step": 248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00014759999999999998,
+      "loss": 4.6082,
+      "step": 249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001482,
+      "loss": 5.5023,
+      "step": 250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00014879999999999998,
+      "loss": 4.7813,
+      "step": 251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001494,
+      "loss": 5.4587,
+      "step": 252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00015,
+      "loss": 4.9109,
+      "step": 253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00015059999999999997,
+      "loss": 4.8091,
+      "step": 254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001512,
+      "loss": 4.6819,
+      "step": 255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00015179999999999998,
+      "loss": 4.7197,
+      "step": 256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001524,
+      "loss": 4.6961,
+      "step": 257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00015299999999999998,
+      "loss": 5.8124,
+      "step": 258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001536,
+      "loss": 4.9215,
+      "step": 259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00015419999999999998,
+      "loss": 4.4569,
+      "step": 260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001548,
+      "loss": 5.2456,
+      "step": 261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00015539999999999998,
+      "loss": 5.2503,
+      "step": 262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000156,
+      "loss": 5.1344,
+      "step": 263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00015659999999999998,
+      "loss": 4.8289,
+      "step": 264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001572,
+      "loss": 4.7765,
+      "step": 265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001578,
+      "loss": 4.4726,
+      "step": 266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001584,
+      "loss": 5.0556,
+      "step": 267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000159,
+      "loss": 5.0821,
+      "step": 268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001596,
+      "loss": 4.8935,
+      "step": 269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001602,
+      "loss": 4.4007,
+      "step": 270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001608,
+      "loss": 4.6542,
+      "step": 271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001614,
+      "loss": 4.6102,
+      "step": 272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000162,
+      "loss": 5.2424,
+      "step": 273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001626,
+      "loss": 6.8519,
+      "step": 274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001632,
+      "loss": 4.8167,
+      "step": 275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001638,
+      "loss": 6.3422,
+      "step": 276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001644,
+      "loss": 4.7503,
+      "step": 277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000165,
+      "loss": 4.8717,
+      "step": 278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001656,
+      "loss": 4.6356,
+      "step": 279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001662,
+      "loss": 4.5977,
+      "step": 280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001668,
+      "loss": 5.3127,
+      "step": 281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001674,
+      "loss": 5.5068,
+      "step": 282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000168,
+      "loss": 5.3972,
+      "step": 283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001686,
+      "loss": 4.9133,
+      "step": 284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00016919999999999997,
+      "loss": 4.7082,
+      "step": 285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00016979999999999998,
+      "loss": 5.1508,
+      "step": 286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017039999999999997,
+      "loss": 4.7987,
+      "step": 287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017099999999999998,
+      "loss": 4.9981,
+      "step": 288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017159999999999997,
+      "loss": 4.756,
+      "step": 289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017219999999999998,
+      "loss": 5.5313,
+      "step": 290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017279999999999997,
+      "loss": 4.6752,
+      "step": 291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017339999999999996,
+      "loss": 5.9425,
+      "step": 292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 4.8854,
+      "step": 293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017459999999999996,
+      "loss": 4.9277,
+      "step": 294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017519999999999998,
+      "loss": 4.7325,
+      "step": 295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017579999999999996,
+      "loss": 6.6008,
+      "step": 296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017639999999999998,
+      "loss": 5.068,
+      "step": 297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017699999999999997,
+      "loss": 4.9548,
+      "step": 298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017759999999999998,
+      "loss": 4.7012,
+      "step": 299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017819999999999997,
+      "loss": 6.2599,
+      "step": 300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017879999999999998,
+      "loss": 5.1843,
+      "step": 301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017939999999999997,
+      "loss": 5.3553,
+      "step": 302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 5.4725,
+      "step": 303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018059999999999997,
+      "loss": 4.6136,
+      "step": 304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018119999999999999,
+      "loss": 5.6863,
+      "step": 305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018179999999999997,
+      "loss": 4.8248,
+      "step": 306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001824,
+      "loss": 4.87,
+      "step": 307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018299999999999998,
+      "loss": 6.5341,
+      "step": 308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001836,
+      "loss": 4.9969,
+      "step": 309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018419999999999998,
+      "loss": 5.457,
+      "step": 310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001848,
+      "loss": 4.7291,
+      "step": 311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018539999999999998,
+      "loss": 4.9024,
+      "step": 312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000186,
+      "loss": 4.7124,
+      "step": 313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018659999999999998,
+      "loss": 5.2463,
+      "step": 314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001872,
+      "loss": 4.8806,
+      "step": 315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018779999999999998,
+      "loss": 4.8438,
+      "step": 316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018839999999999997,
+      "loss": 4.6296,
+      "step": 317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018899999999999999,
+      "loss": 4.4458,
+      "step": 318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018959999999999997,
+      "loss": 4.9277,
+      "step": 319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001902,
+      "loss": 5.5921,
+      "step": 320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019079999999999998,
+      "loss": 5.632,
+      "step": 321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001914,
+      "loss": 5.2905,
+      "step": 322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 5.4334,
+      "step": 323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001926,
+      "loss": 4.6342,
+      "step": 324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019319999999999998,
+      "loss": 4.7252,
+      "step": 325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001938,
+      "loss": 4.8713,
+      "step": 326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019439999999999998,
+      "loss": 4.7396,
+      "step": 327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000195,
+      "loss": 5.1595,
+      "step": 328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019559999999999998,
+      "loss": 4.8263,
+      "step": 329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001962,
+      "loss": 5.2953,
+      "step": 330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019679999999999999,
+      "loss": 4.9895,
+      "step": 331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001974,
+      "loss": 5.143,
+      "step": 332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000198,
+      "loss": 5.9161,
+      "step": 333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001986,
+      "loss": 5.0185,
+      "step": 334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001992,
+      "loss": 5.2889,
+      "step": 335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001998,
+      "loss": 5.2346,
+      "step": 336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002004,
+      "loss": 4.719,
+      "step": 337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000201,
+      "loss": 4.7736,
+      "step": 338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002016,
+      "loss": 4.769,
+      "step": 339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002022,
+      "loss": 4.848,
+      "step": 340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002028,
+      "loss": 4.5971,
+      "step": 341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00020339999999999998,
+      "loss": 4.7998,
+      "step": 342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000204,
+      "loss": 5.1519,
+      "step": 343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00020459999999999999,
+      "loss": 4.3911,
+      "step": 344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002052,
+      "loss": 5.3684,
+      "step": 345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002058,
+      "loss": 4.7293,
+      "step": 346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00020639999999999998,
+      "loss": 5.4307,
+      "step": 347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00020699999999999996,
+      "loss": 5.1887,
+      "step": 348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00020759999999999998,
+      "loss": 6.1516,
+      "step": 349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00020819999999999996,
+      "loss": 6.2343,
+      "step": 350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00020879999999999998,
+      "loss": 5.7203,
+      "step": 351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00020939999999999997,
+      "loss": 5.0998,
+      "step": 352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 4.895,
+      "step": 353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021059999999999997,
+      "loss": 4.8778,
+      "step": 354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021119999999999996,
+      "loss": 4.6386,
+      "step": 355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021179999999999997,
+      "loss": 5.0082,
+      "step": 356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021239999999999996,
+      "loss": 5.2028,
+      "step": 357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 5.1243,
+      "step": 358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021359999999999996,
+      "loss": 5.8256,
+      "step": 359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021419999999999998,
+      "loss": 4.9463,
+      "step": 360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021479999999999996,
+      "loss": 5.1464,
+      "step": 361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021539999999999998,
+      "loss": 4.5296,
+      "step": 362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 5.1842,
+      "step": 363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021659999999999998,
+      "loss": 5.1303,
+      "step": 364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021719999999999997,
+      "loss": 5.388,
+      "step": 365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00021779999999999998,
+      "loss": 5.2611,
+      "step": 366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00021839999999999997,
+      "loss": 4.7396,
+      "step": 367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 4.7492,
+      "step": 368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00021959999999999997,
+      "loss": 5.4778,
+      "step": 369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022019999999999999,
+      "loss": 5.056,
+      "step": 370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022079999999999997,
+      "loss": 4.6836,
+      "step": 371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002214,
+      "loss": 5.4872,
+      "step": 372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 4.9566,
+      "step": 373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002226,
+      "loss": 5.092,
+      "step": 374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022319999999999998,
+      "loss": 4.637,
+      "step": 375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002238,
+      "loss": 4.7808,
+      "step": 376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022439999999999998,
+      "loss": 4.8161,
+      "step": 377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000225,
+      "loss": 4.6822,
+      "step": 378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022559999999999998,
+      "loss": 4.7936,
+      "step": 379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022619999999999997,
+      "loss": 5.2234,
+      "step": 380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022679999999999998,
+      "loss": 4.4151,
+      "step": 381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022739999999999997,
+      "loss": 4.8784,
+      "step": 382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 4.7343,
+      "step": 383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00022859999999999997,
+      "loss": 5.8847,
+      "step": 384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002292,
+      "loss": 4.9225,
+      "step": 385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00022979999999999997,
+      "loss": 5.4836,
+      "step": 386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002304,
+      "loss": 5.7179,
+      "step": 387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00023099999999999998,
+      "loss": 4.5246,
+      "step": 388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002316,
+      "loss": 4.8167,
+      "step": 389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00023219999999999998,
+      "loss": 6.1876,
+      "step": 390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002328,
+      "loss": 4.5322,
+      "step": 391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00023339999999999998,
+      "loss": 5.5851,
+      "step": 392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000234,
+      "loss": 4.9948,
+      "step": 393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00023459999999999998,
+      "loss": 4.8198,
+      "step": 394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002352,
+      "loss": 5.6014,
+      "step": 395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00023579999999999999,
+      "loss": 6.2024,
+      "step": 396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002364,
+      "loss": 6.2467,
+      "step": 397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000237,
+      "loss": 7.4593,
+      "step": 398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002376,
+      "loss": 7.6539,
+      "step": 399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002382,
+      "loss": 5.2311,
+      "step": 400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002388,
+      "loss": 4.9665,
+      "step": 401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002394,
+      "loss": 4.9601,
+      "step": 402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 4.7942,
+      "step": 403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002406,
+      "loss": 5.179,
+      "step": 404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024119999999999998,
+      "loss": 4.7939,
+      "step": 405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002418,
+      "loss": 5.3789,
+      "step": 406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024239999999999998,
+      "loss": 5.0828,
+      "step": 407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000243,
+      "loss": 5.3537,
+      "step": 408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024359999999999999,
+      "loss": 5.5745,
+      "step": 409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024419999999999997,
+      "loss": 5.2492,
+      "step": 410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002448,
+      "loss": 5.2284,
+      "step": 411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024539999999999995,
+      "loss": 4.9789,
+      "step": 412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 5.5324,
+      "step": 413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002466,
+      "loss": 5.1439,
+      "step": 414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002472,
+      "loss": 5.3242,
+      "step": 415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024779999999999995,
+      "loss": 4.8789,
+      "step": 416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024839999999999997,
+      "loss": 5.8162,
+      "step": 417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000249,
+      "loss": 5.0106,
+      "step": 418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00024959999999999994,
+      "loss": 5.0066,
+      "step": 419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025019999999999996,
+      "loss": 5.0835,
+      "step": 420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025079999999999997,
+      "loss": 4.714,
+      "step": 421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002514,
+      "loss": 5.1203,
+      "step": 422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 5.3753,
+      "step": 423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025259999999999996,
+      "loss": 5.5015,
+      "step": 424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002532,
+      "loss": 4.9811,
+      "step": 425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002538,
+      "loss": 5.1797,
+      "step": 426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025439999999999995,
+      "loss": 6.1654,
+      "step": 427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025499999999999996,
+      "loss": 7.3542,
+      "step": 428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002556,
+      "loss": 5.0921,
+      "step": 429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002562,
+      "loss": 4.8804,
+      "step": 430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025679999999999995,
+      "loss": 5.1751,
+      "step": 431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025739999999999997,
+      "loss": 5.051,
+      "step": 432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000258,
+      "loss": 4.8449,
+      "step": 433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002586,
+      "loss": 4.986,
+      "step": 434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025919999999999996,
+      "loss": 5.1099,
+      "step": 435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025979999999999997,
+      "loss": 5.3317,
+      "step": 436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002604,
+      "loss": 4.9233,
+      "step": 437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000261,
+      "loss": 5.0732,
+      "step": 438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00026159999999999996,
+      "loss": 4.6898,
+      "step": 439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002622,
+      "loss": 4.9492,
+      "step": 440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002628,
+      "loss": 5.7867,
+      "step": 441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00026339999999999995,
+      "loss": 5.7642,
+      "step": 442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 4.8011,
+      "step": 443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002646,
+      "loss": 5.5959,
+      "step": 444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002652,
+      "loss": 5.1056,
+      "step": 445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00026579999999999996,
+      "loss": 6.8447,
+      "step": 446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00026639999999999997,
+      "loss": 5.5555,
+      "step": 447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000267,
+      "loss": 5.435,
+      "step": 448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002676,
+      "loss": 6.6167,
+      "step": 449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00026819999999999996,
+      "loss": 5.7015,
+      "step": 450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002688,
+      "loss": 5.3358,
+      "step": 451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002694,
+      "loss": 6.0488,
+      "step": 452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00027,
+      "loss": 6.6422,
+      "step": 453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00027059999999999996,
+      "loss": 4.9632,
+      "step": 454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002712,
+      "loss": 5.068,
+      "step": 455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002718,
+      "loss": 4.8883,
+      "step": 456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002724,
+      "loss": 4.8911,
+      "step": 457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00027299999999999997,
+      "loss": 5.7467,
+      "step": 458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002736,
+      "loss": 4.6221,
+      "step": 459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002742,
+      "loss": 5.5107,
+      "step": 460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002748,
+      "loss": 5.5784,
+      "step": 461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00027539999999999997,
+      "loss": 4.8506,
+      "step": 462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000276,
+      "loss": 5.7129,
+      "step": 463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002766,
+      "loss": 4.9862,
+      "step": 464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002772,
+      "loss": 4.7395,
+      "step": 465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002778,
+      "loss": 5.6962,
+      "step": 466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002784,
+      "loss": 5.1175,
+      "step": 467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000279,
+      "loss": 5.1618,
+      "step": 468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00027959999999999997,
+      "loss": 4.87,
+      "step": 469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002802,
+      "loss": 4.9536,
+      "step": 470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002808,
+      "loss": 5.5207,
+      "step": 471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00028139999999999996,
+      "loss": 4.9375,
+      "step": 472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 4.826,
+      "step": 473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002826,
+      "loss": 5.6,
+      "step": 474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028319999999999994,
+      "loss": 5.0958,
+      "step": 475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028379999999999996,
+      "loss": 5.4091,
+      "step": 476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002844,
+      "loss": 5.4463,
+      "step": 477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000285,
+      "loss": 4.8026,
+      "step": 478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028559999999999995,
+      "loss": 5.39,
+      "step": 479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028619999999999996,
+      "loss": 5.9881,
+      "step": 480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002868,
+      "loss": 4.923,
+      "step": 481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028739999999999994,
+      "loss": 6.0273,
+      "step": 482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 5.3764,
+      "step": 483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028859999999999997,
+      "loss": 5.3997,
+      "step": 484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002892,
+      "loss": 5.4994,
+      "step": 485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028979999999999994,
+      "loss": 6.0153,
+      "step": 486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029039999999999996,
+      "loss": 5.2351,
+      "step": 487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029099999999999997,
+      "loss": 5.2671,
+      "step": 488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002916,
+      "loss": 5.4524,
+      "step": 489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029219999999999995,
+      "loss": 4.9553,
+      "step": 490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029279999999999996,
+      "loss": 4.853,
+      "step": 491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002934,
+      "loss": 5.7447,
+      "step": 492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000294,
+      "loss": 5.068,
+      "step": 493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029459999999999995,
+      "loss": 6.2985,
+      "step": 494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029519999999999997,
+      "loss": 4.9962,
+      "step": 495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002958,
+      "loss": 5.313,
+      "step": 496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002964,
+      "loss": 6.5489,
+      "step": 497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029699999999999996,
+      "loss": 7.621,
+      "step": 498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029759999999999997,
+      "loss": 5.363,
+      "step": 499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002982,
+      "loss": 4.799,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 5.451260089874268,
+      "eval_runtime": 966.8313,
+      "eval_samples_per_second": 2.733,
+      "eval_steps_per_second": 0.342,
+      "eval_wer": 1.9663823879412932,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002988,
+      "loss": 6.2281,
+      "step": 501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029939999999999996,
+      "loss": 4.9097,
+      "step": 502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003,
+      "loss": 5.8719,
+      "step": 503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002999381698268755,
+      "loss": 4.9623,
+      "step": 504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000299876339653751,
+      "loss": 5.8877,
+      "step": 505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002998145094806265,
+      "loss": 5.0653,
+      "step": 506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029975267930750204,
+      "loss": 5.8915,
+      "step": 507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029969084913437753,
+      "loss": 5.1228,
+      "step": 508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002996290189612531,
+      "loss": 5.802,
+      "step": 509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029956718878812856,
+      "loss": 5.4614,
+      "step": 510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002995053586150041,
+      "loss": 5.0334,
+      "step": 511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002994435284418796,
+      "loss": 5.3914,
+      "step": 512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029938169826875514,
+      "loss": 5.8218,
+      "step": 513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029931986809563063,
+      "loss": 4.657,
+      "step": 514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002992580379225061,
+      "loss": 4.781,
+      "step": 515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029919620774938166,
+      "loss": 7.1354,
+      "step": 516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002991343775762572,
+      "loss": 5.8064,
+      "step": 517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002990725474031327,
+      "loss": 5.4236,
+      "step": 518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029901071723000824,
+      "loss": 4.9984,
+      "step": 519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029894888705688373,
+      "loss": 5.5457,
+      "step": 520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029888705688375927,
+      "loss": 4.6279,
+      "step": 521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029882522671063476,
+      "loss": 5.4678,
+      "step": 522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002987633965375103,
+      "loss": 5.4336,
+      "step": 523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002987015663643858,
+      "loss": 5.1198,
+      "step": 524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002986397361912613,
+      "loss": 6.8515,
+      "step": 525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029857790601813683,
+      "loss": 5.1662,
+      "step": 526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029851607584501237,
+      "loss": 5.8174,
+      "step": 527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029845424567188786,
+      "loss": 4.7318,
+      "step": 528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029839241549876335,
+      "loss": 5.1465,
+      "step": 529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002983305853256389,
+      "loss": 5.4406,
+      "step": 530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002982687551525144,
+      "loss": 4.689,
+      "step": 531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029820692497938993,
+      "loss": 4.6031,
+      "step": 532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981450948062654,
+      "loss": 4.9763,
+      "step": 533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029808326463314096,
+      "loss": 4.7341,
+      "step": 534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029802143446001645,
+      "loss": 5.6098,
+      "step": 535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000297959604286892,
+      "loss": 5.0063,
+      "step": 536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002978977741137675,
+      "loss": 4.9367,
+      "step": 537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000297835943940643,
+      "loss": 5.2729,
+      "step": 538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002977741137675185,
+      "loss": 5.0129,
+      "step": 539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029771228359439406,
+      "loss": 5.2093,
+      "step": 540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029765045342126955,
+      "loss": 5.4632,
+      "step": 541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029758862324814504,
+      "loss": 4.9785,
+      "step": 542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002975267930750206,
+      "loss": 6.3607,
+      "step": 543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029746496290189607,
+      "loss": 4.8114,
+      "step": 544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002974031327287716,
+      "loss": 4.7121,
+      "step": 545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029734130255564716,
+      "loss": 5.5902,
+      "step": 546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029727947238252265,
+      "loss": 6.2206,
+      "step": 547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029721764220939814,
+      "loss": 7.2627,
+      "step": 548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002971558120362737,
+      "loss": 6.7123,
+      "step": 549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002970939818631492,
+      "loss": 5.1086,
+      "step": 550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002970321516900247,
+      "loss": 5.7206,
+      "step": 551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002969703215169002,
+      "loss": 6.7092,
+      "step": 552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029690849134377575,
+      "loss": 6.3015,
+      "step": 553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029684666117065124,
+      "loss": 4.9949,
+      "step": 554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002967848309975268,
+      "loss": 5.0623,
+      "step": 555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029672300082440227,
+      "loss": 5.6326,
+      "step": 556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002966611706512778,
+      "loss": 5.275,
+      "step": 557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002965993404781533,
+      "loss": 5.0093,
+      "step": 558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029653751030502885,
+      "loss": 4.7437,
+      "step": 559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029647568013190434,
+      "loss": 4.9625,
+      "step": 560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029641384995877983,
+      "loss": 5.4984,
+      "step": 561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029635201978565537,
+      "loss": 5.6253,
+      "step": 562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002962901896125309,
+      "loss": 4.7852,
+      "step": 563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002962283594394064,
+      "loss": 5.3771,
+      "step": 564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002961665292662819,
+      "loss": 5.7434,
+      "step": 565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029610469909315744,
+      "loss": 4.8438,
+      "step": 566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002960428689200329,
+      "loss": 5.0875,
+      "step": 567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029598103874690847,
+      "loss": 5.2043,
+      "step": 568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000295919208573784,
+      "loss": 4.8769,
+      "step": 569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002958573784006595,
+      "loss": 5.3198,
+      "step": 570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000295795548227535,
+      "loss": 5.4131,
+      "step": 571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029573371805441054,
+      "loss": 5.2971,
+      "step": 572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000295671887881286,
+      "loss": 6.4192,
+      "step": 573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002956100577081615,
+      "loss": 5.0931,
+      "step": 574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029554822753503706,
+      "loss": 4.8497,
+      "step": 575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002954863973619126,
+      "loss": 5.1916,
+      "step": 576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002954245671887881,
+      "loss": 5.3191,
+      "step": 577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029536273701566364,
+      "loss": 5.1317,
+      "step": 578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002953009068425391,
+      "loss": 5.3397,
+      "step": 579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029523907666941467,
+      "loss": 6.177,
+      "step": 580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029517724649629016,
+      "loss": 4.8234,
+      "step": 581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002951154163231657,
+      "loss": 5.3559,
+      "step": 582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002950535861500412,
+      "loss": 5.2198,
+      "step": 583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002949917559769167,
+      "loss": 5.2803,
+      "step": 584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002949299258037922,
+      "loss": 6.2425,
+      "step": 585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029486809563066777,
+      "loss": 5.2988,
+      "step": 586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029480626545754326,
+      "loss": 5.6357,
+      "step": 587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029474443528441875,
+      "loss": 5.1015,
+      "step": 588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002946826051112943,
+      "loss": 5.8464,
+      "step": 589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002946207749381698,
+      "loss": 5.358,
+      "step": 590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002945589447650453,
+      "loss": 4.9242,
+      "step": 591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029449711459192087,
+      "loss": 5.2489,
+      "step": 592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029443528441879636,
+      "loss": 6.1619,
+      "step": 593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029437345424567185,
+      "loss": 5.1924,
+      "step": 594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002943116240725474,
+      "loss": 4.9266,
+      "step": 595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002942497938994229,
+      "loss": 5.0016,
+      "step": 596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029418796372629837,
+      "loss": 7.6006,
+      "step": 597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002941261335531739,
+      "loss": 6.8589,
+      "step": 598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029406430338004946,
+      "loss": 5.2786,
+      "step": 599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029400247320692495,
+      "loss": 5.415,
+      "step": 600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002939406430338005,
+      "loss": 4.8837,
+      "step": 601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000293878812860676,
+      "loss": 5.3328,
+      "step": 602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029381698268755147,
+      "loss": 4.8891,
+      "step": 603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000293755152514427,
+      "loss": 4.9667,
+      "step": 604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029369332234130256,
+      "loss": 5.4072,
+      "step": 605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029363149216817805,
+      "loss": 4.9442,
+      "step": 606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029356966199505354,
+      "loss": 6.1916,
+      "step": 607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002935078318219291,
+      "loss": 4.9482,
+      "step": 608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029344600164880457,
+      "loss": 5.8162,
+      "step": 609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002933841714756801,
+      "loss": 5.1522,
+      "step": 610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002933223413025556,
+      "loss": 5.4065,
+      "step": 611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029326051112943115,
+      "loss": 5.0334,
+      "step": 612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029319868095630664,
+      "loss": 5.0196,
+      "step": 613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002931368507831822,
+      "loss": 5.3866,
+      "step": 614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002930750206100577,
+      "loss": 5.3787,
+      "step": 615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002930131904369332,
+      "loss": 4.8984,
+      "step": 616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002929513602638087,
+      "loss": 5.6667,
+      "step": 617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029288953009068425,
+      "loss": 4.7201,
+      "step": 618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029282769991755974,
+      "loss": 4.8588,
+      "step": 619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002927658697444352,
+      "loss": 6.2014,
+      "step": 620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029270403957131077,
+      "loss": 5.2276,
+      "step": 621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002926422093981863,
+      "loss": 4.8498,
+      "step": 622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002925803792250618,
+      "loss": 5.0932,
+      "step": 623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029251854905193735,
+      "loss": 7.2175,
+      "step": 624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029245671887881283,
+      "loss": 5.2835,
+      "step": 625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002923948887056883,
+      "loss": 4.8726,
+      "step": 626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029233305853256387,
+      "loss": 4.8157,
+      "step": 627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002922712283594394,
+      "loss": 6.234,
+      "step": 628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002922093981863149,
+      "loss": 6.5592,
+      "step": 629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002921475680131904,
+      "loss": 4.8927,
+      "step": 630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029208573784006593,
+      "loss": 5.2312,
+      "step": 631
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002920239076669414,
+      "loss": 5.3573,
+      "step": 632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029196207749381697,
+      "loss": 5.084,
+      "step": 633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029190024732069246,
+      "loss": 5.2281,
+      "step": 634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000291838417147568,
+      "loss": 4.9629,
+      "step": 635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002917765869744435,
+      "loss": 4.7146,
+      "step": 636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029171475680131903,
+      "loss": 4.913,
+      "step": 637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002916529266281945,
+      "loss": 4.621,
+      "step": 638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029159109645507,
+      "loss": 5.0082,
+      "step": 639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029152926628194556,
+      "loss": 6.3035,
+      "step": 640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002914674361088211,
+      "loss": 6.0562,
+      "step": 641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002914056059356966,
+      "loss": 5.4795,
+      "step": 642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002913437757625721,
+      "loss": 4.916,
+      "step": 643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002912819455894476,
+      "loss": 4.9765,
+      "step": 644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029122011541632317,
+      "loss": 4.6902,
+      "step": 645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029115828524319866,
+      "loss": 5.0,
+      "step": 646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002910964550700742,
+      "loss": 6.1168,
+      "step": 647
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002910346248969497,
+      "loss": 6.3845,
+      "step": 648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002909727947238252,
+      "loss": 5.2678,
+      "step": 649
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002909109645507007,
+      "loss": 5.2474,
+      "step": 650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029084913437757627,
+      "loss": 5.3689,
+      "step": 651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029078730420445176,
+      "loss": 4.9006,
+      "step": 652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029072547403132724,
+      "loss": 5.1356,
+      "step": 653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002906636438582028,
+      "loss": 4.8688,
+      "step": 654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002906018136850783,
+      "loss": 5.2982,
+      "step": 655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002905399835119538,
+      "loss": 4.8107,
+      "step": 656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002904781533388293,
+      "loss": 4.7908,
+      "step": 657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029041632316570485,
+      "loss": 6.76,
+      "step": 658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029035449299258034,
+      "loss": 5.2828,
+      "step": 659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002902926628194559,
+      "loss": 4.9841,
+      "step": 660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002902308326463314,
+      "loss": 5.5075,
+      "step": 661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029016900247320687,
+      "loss": 4.95,
+      "step": 662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002901071723000824,
+      "loss": 4.8843,
+      "step": 663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029004534212695795,
+      "loss": 5.4184,
+      "step": 664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028998351195383344,
+      "loss": 5.1116,
+      "step": 665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028992168178070893,
+      "loss": 5.9981,
+      "step": 666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002898598516075845,
+      "loss": 4.9382,
+      "step": 667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028979802143445997,
+      "loss": 5.7551,
+      "step": 668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002897361912613355,
+      "loss": 4.7493,
+      "step": 669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028967436108821105,
+      "loss": 5.0395,
+      "step": 670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028961253091508654,
+      "loss": 5.2752,
+      "step": 671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028955070074196203,
+      "loss": 5.9185,
+      "step": 672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002894888705688376,
+      "loss": 4.8457,
+      "step": 673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002894270403957131,
+      "loss": 4.9783,
+      "step": 674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002893652102225886,
+      "loss": 4.91,
+      "step": 675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002893033800494641,
+      "loss": 5.0751,
+      "step": 676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028924154987633964,
+      "loss": 4.7223,
+      "step": 677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028917971970321513,
+      "loss": 4.8771,
+      "step": 678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002891178895300907,
+      "loss": 5.2208,
+      "step": 679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028905605935696617,
+      "loss": 5.9935,
+      "step": 680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002889942291838417,
+      "loss": 5.2525,
+      "step": 681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002889323990107172,
+      "loss": 4.8554,
+      "step": 682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028887056883759274,
+      "loss": 5.6744,
+      "step": 683
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028880873866446823,
+      "loss": 5.191,
+      "step": 684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002887469084913437,
+      "loss": 5.2159,
+      "step": 685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028868507831821927,
+      "loss": 5.39,
+      "step": 686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002886232481450948,
+      "loss": 4.7996,
+      "step": 687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002885614179719703,
+      "loss": 4.9891,
+      "step": 688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002884995877988458,
+      "loss": 5.0524,
+      "step": 689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028843775762572133,
+      "loss": 4.8859,
+      "step": 690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002883759274525968,
+      "loss": 4.7275,
+      "step": 691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028831409727947236,
+      "loss": 5.5271,
+      "step": 692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028825226710634785,
+      "loss": 5.316,
+      "step": 693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002881904369332234,
+      "loss": 4.986,
+      "step": 694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002881286067600989,
+      "loss": 5.0691,
+      "step": 695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028806677658697443,
+      "loss": 5.0126,
+      "step": 696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002880049464138499,
+      "loss": 4.6673,
+      "step": 697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002879431162407254,
+      "loss": 4.9013,
+      "step": 698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028788128606760095,
+      "loss": 5.8476,
+      "step": 699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002878194558944765,
+      "loss": 5.1243,
+      "step": 700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000287757625721352,
+      "loss": 5.119,
+      "step": 701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028769579554822753,
+      "loss": 4.9844,
+      "step": 702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000287633965375103,
+      "loss": 5.4307,
+      "step": 703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028757213520197856,
+      "loss": 5.9045,
+      "step": 704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028751030502885405,
+      "loss": 5.8109,
+      "step": 705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002874484748557296,
+      "loss": 5.8098,
+      "step": 706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002873866446826051,
+      "loss": 5.6664,
+      "step": 707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002873248145094806,
+      "loss": 5.3156,
+      "step": 708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002872629843363561,
+      "loss": 5.0711,
+      "step": 709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028720115416323166,
+      "loss": 5.1298,
+      "step": 710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028713932399010715,
+      "loss": 5.0136,
+      "step": 711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028707749381698264,
+      "loss": 4.9275,
+      "step": 712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002870156636438582,
+      "loss": 5.0506,
+      "step": 713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002869538334707337,
+      "loss": 5.2233,
+      "step": 714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002868920032976092,
+      "loss": 5.6857,
+      "step": 715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002868301731244847,
+      "loss": 4.6578,
+      "step": 716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028676834295136025,
+      "loss": 4.8152,
+      "step": 717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028670651277823574,
+      "loss": 5.0952,
+      "step": 718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002866446826051113,
+      "loss": 4.9838,
+      "step": 719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002865828524319868,
+      "loss": 4.7918,
+      "step": 720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028652102225886226,
+      "loss": 5.3484,
+      "step": 721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002864591920857378,
+      "loss": 5.0334,
+      "step": 722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028639736191261335,
+      "loss": 4.9461,
+      "step": 723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028633553173948884,
+      "loss": 5.3516,
+      "step": 724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002862737015663644,
+      "loss": 5.1519,
+      "step": 725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002862118713932399,
+      "loss": 5.2874,
+      "step": 726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028615004122011536,
+      "loss": 4.7864,
+      "step": 727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002860882110469909,
+      "loss": 5.3114,
+      "step": 728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028602638087386645,
+      "loss": 4.8014,
+      "step": 729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028596455070074194,
+      "loss": 5.182,
+      "step": 730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028590272052761743,
+      "loss": 4.8042,
+      "step": 731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000285840890354493,
+      "loss": 4.7048,
+      "step": 732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002857790601813685,
+      "loss": 5.9609,
+      "step": 733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000285717230008244,
+      "loss": 4.9273,
+      "step": 734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002856553998351195,
+      "loss": 5.3523,
+      "step": 735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028559356966199504,
+      "loss": 4.8498,
+      "step": 736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028553173948887053,
+      "loss": 5.0884,
+      "step": 737
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002854699093157461,
+      "loss": 5.1514,
+      "step": 738
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028540807914262156,
+      "loss": 5.9656,
+      "step": 739
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002853462489694971,
+      "loss": 4.987,
+      "step": 740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002852844187963726,
+      "loss": 5.8286,
+      "step": 741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00028522258862324814,
+      "loss": 4.9612,
+      "step": 742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00028516075845012363,
+      "loss": 6.1989,
+      "step": 743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002850989282769991,
+      "loss": 5.1364,
+      "step": 744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00028503709810387466,
+      "loss": 6.5852,
+      "step": 745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002849752679307502,
+      "loss": 5.208,
+      "step": 746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002849134377576257,
+      "loss": 5.2926,
+      "step": 747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002848516075845012,
+      "loss": 5.2483,
+      "step": 748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00028478977741137673,
+      "loss": 4.8284,
+      "step": 749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002847279472382522,
+      "loss": 5.1513,
+      "step": 750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00028466611706512776,
+      "loss": 5.6949,
+      "step": 751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002846042868920033,
+      "loss": 5.3198,
+      "step": 752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002845424567188788,
+      "loss": 5.1494,
+      "step": 753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002844806265457543,
+      "loss": 5.4803,
+      "step": 754
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00028441879637262983,
+      "loss": 4.925,
+      "step": 755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002843569661995053,
+      "loss": 5.5812,
+      "step": 756
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00028429513602638086,
+      "loss": 5.9013,
+      "step": 757
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00028423330585325635,
+      "loss": 5.1175,
+      "step": 758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002841714756801319,
+      "loss": 4.9526,
+      "step": 759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002841096455070074,
+      "loss": 4.9495,
+      "step": 760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00028404781533388293,
+      "loss": 5.5628,
+      "step": 761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002839859851607584,
+      "loss": 5.055,
+      "step": 762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00028392415498763396,
+      "loss": 4.8701,
+      "step": 763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00028386232481450945,
+      "loss": 4.7375,
+      "step": 764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000283800494641385,
+      "loss": 4.7783,
+      "step": 765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002837386644682605,
+      "loss": 5.4166,
+      "step": 766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000283676834295136,
+      "loss": 5.1005,
+      "step": 767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002836150041220115,
+      "loss": 5.0015,
+      "step": 768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00028355317394888706,
+      "loss": 5.5799,
+      "step": 769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00028349134377576255,
+      "loss": 4.8968,
+      "step": 770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00028342951360263804,
+      "loss": 5.4965,
+      "step": 771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002833676834295136,
+      "loss": 4.8652,
+      "step": 772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00028330585325638907,
+      "loss": 4.9641,
+      "step": 773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002832440230832646,
+      "loss": 5.2281,
+      "step": 774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00028318219291014016,
+      "loss": 4.7379,
+      "step": 775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00028312036273701565,
+      "loss": 5.0607,
+      "step": 776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028305853256389114,
+      "loss": 5.0183,
+      "step": 777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002829967023907667,
+      "loss": 5.0654,
+      "step": 778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028293487221764217,
+      "loss": 4.9868,
+      "step": 779
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028287304204451766,
+      "loss": 4.9887,
+      "step": 780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002828112118713932,
+      "loss": 4.8849,
+      "step": 781
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028274938169826875,
+      "loss": 5.0583,
+      "step": 782
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028268755152514424,
+      "loss": 5.2482,
+      "step": 783
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002826257213520198,
+      "loss": 4.8583,
+      "step": 784
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028256389117889527,
+      "loss": 5.333,
+      "step": 785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028250206100577076,
+      "loss": 5.7919,
+      "step": 786
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002824402308326463,
+      "loss": 5.3446,
+      "step": 787
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028237840065952185,
+      "loss": 5.4577,
+      "step": 788
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028231657048639734,
+      "loss": 5.3624,
+      "step": 789
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028225474031327283,
+      "loss": 5.8872,
+      "step": 790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028219291014014837,
+      "loss": 5.6629,
+      "step": 791
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028213107996702386,
+      "loss": 5.6396,
+      "step": 792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002820692497938994,
+      "loss": 5.8536,
+      "step": 793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002820074196207749,
+      "loss": 5.1191,
+      "step": 794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028194558944765044,
+      "loss": 5.6243,
+      "step": 795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002818837592745259,
+      "loss": 6.8198,
+      "step": 796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028182192910140147,
+      "loss": 5.8785,
+      "step": 797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000281760098928277,
+      "loss": 5.2595,
+      "step": 798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002816982687551525,
+      "loss": 5.1714,
+      "step": 799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000281636438582028,
+      "loss": 5.2275,
+      "step": 800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028157460840890354,
+      "loss": 5.793,
+      "step": 801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000281512778235779,
+      "loss": 5.7473,
+      "step": 802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002814509480626545,
+      "loss": 5.6947,
+      "step": 803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028138911788953006,
+      "loss": 5.2635,
+      "step": 804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002813272877164056,
+      "loss": 5.4362,
+      "step": 805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002812654575432811,
+      "loss": 5.2913,
+      "step": 806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028120362737015664,
+      "loss": 5.0803,
+      "step": 807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002811417971970321,
+      "loss": 4.7245,
+      "step": 808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002810799670239076,
+      "loss": 5.0516,
+      "step": 809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028101813685078316,
+      "loss": 5.272,
+      "step": 810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002809563066776587,
+      "loss": 4.9901,
+      "step": 811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002808944765045342,
+      "loss": 5.3345,
+      "step": 812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002808326463314097,
+      "loss": 5.5047,
+      "step": 813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002807708161582852,
+      "loss": 4.9683,
+      "step": 814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002807089859851607,
+      "loss": 4.804,
+      "step": 815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028064715581203626,
+      "loss": 4.8909,
+      "step": 816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028058532563891175,
+      "loss": 5.2764,
+      "step": 817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002805234954657873,
+      "loss": 5.3709,
+      "step": 818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002804616652926628,
+      "loss": 4.9774,
+      "step": 819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002803998351195383,
+      "loss": 5.0658,
+      "step": 820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002803380049464138,
+      "loss": 5.0624,
+      "step": 821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002802761747732893,
+      "loss": 5.1805,
+      "step": 822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028021434460016485,
+      "loss": 6.2101,
+      "step": 823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002801525144270404,
+      "loss": 5.104,
+      "step": 824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002800906842539159,
+      "loss": 5.049,
+      "step": 825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028002885408079137,
+      "loss": 5.1928,
+      "step": 826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002799670239076669,
+      "loss": 5.4952,
+      "step": 827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00027990519373454246,
+      "loss": 5.1276,
+      "step": 828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00027984336356141795,
+      "loss": 5.2754,
+      "step": 829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002797815333882935,
+      "loss": 5.0364,
+      "step": 830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000279719703215169,
+      "loss": 5.0416,
+      "step": 831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027965787304204447,
+      "loss": 5.4829,
+      "step": 832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027959604286892,
+      "loss": 5.1105,
+      "step": 833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027953421269579556,
+      "loss": 4.6162,
+      "step": 834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027947238252267105,
+      "loss": 5.2166,
+      "step": 835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027941055234954654,
+      "loss": 4.962,
+      "step": 836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002793487221764221,
+      "loss": 5.1047,
+      "step": 837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027928689200329757,
+      "loss": 5.075,
+      "step": 838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002792250618301731,
+      "loss": 4.9826,
+      "step": 839
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002791632316570486,
+      "loss": 5.0707,
+      "step": 840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027910140148392415,
+      "loss": 5.8912,
+      "step": 841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027903957131079964,
+      "loss": 4.8758,
+      "step": 842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002789777411376752,
+      "loss": 4.9275,
+      "step": 843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027891591096455067,
+      "loss": 5.0259,
+      "step": 844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027885408079142616,
+      "loss": 4.6065,
+      "step": 845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002787922506183017,
+      "loss": 4.6332,
+      "step": 846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027873042044517725,
+      "loss": 4.8051,
+      "step": 847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00027866859027205274,
+      "loss": 5.109,
+      "step": 848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002786067600989282,
+      "loss": 5.0631,
+      "step": 849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00027854492992580377,
+      "loss": 7.3654,
+      "step": 850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00027848309975267926,
+      "loss": 5.1474,
+      "step": 851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002784212695795548,
+      "loss": 5.1119,
+      "step": 852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00027835943940643035,
+      "loss": 5.1925,
+      "step": 853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00027829760923330583,
+      "loss": 5.1712,
+      "step": 854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002782357790601813,
+      "loss": 5.187,
+      "step": 855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00027817394888705687,
+      "loss": 5.579,
+      "step": 856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002781121187139324,
+      "loss": 4.9342,
+      "step": 857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002780502885408079,
+      "loss": 5.1602,
+      "step": 858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002779884583676834,
+      "loss": 5.3623,
+      "step": 859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00027792662819455893,
+      "loss": 5.6963,
+      "step": 860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002778647980214344,
+      "loss": 4.5597,
+      "step": 861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00027780296784830997,
+      "loss": 5.3215,
+      "step": 862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00027774113767518546,
+      "loss": 6.3289,
+      "step": 863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000277679307502061,
+      "loss": 5.0031,
+      "step": 864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002776174773289365,
+      "loss": 5.6062,
+      "step": 865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00027755564715581203,
+      "loss": 5.0658,
+      "step": 866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002774938169826875,
+      "loss": 5.1153,
+      "step": 867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000277431986809563,
+      "loss": 5.3629,
+      "step": 868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00027737015663643856,
+      "loss": 4.9396,
+      "step": 869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002773083264633141,
+      "loss": 4.9218,
+      "step": 870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002772464962901896,
+      "loss": 5.0613,
+      "step": 871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002771846661170651,
+      "loss": 5.0563,
+      "step": 872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002771228359439406,
+      "loss": 4.9426,
+      "step": 873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002770610057708161,
+      "loss": 5.0223,
+      "step": 874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00027699917559769166,
+      "loss": 5.1232,
+      "step": 875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002769373454245672,
+      "loss": 5.7038,
+      "step": 876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002768755152514427,
+      "loss": 5.418,
+      "step": 877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002768136850783182,
+      "loss": 4.9436,
+      "step": 878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002767518549051937,
+      "loss": 4.9434,
+      "step": 879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002766900247320692,
+      "loss": 4.9558,
+      "step": 880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002766281945589447,
+      "loss": 5.3522,
+      "step": 881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00027656636438582024,
+      "loss": 4.7351,
+      "step": 882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002765045342126958,
+      "loss": 5.3871,
+      "step": 883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002764427040395713,
+      "loss": 5.7166,
+      "step": 884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002763808738664468,
+      "loss": 5.4573,
+      "step": 885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002763190436933223,
+      "loss": 5.4306,
+      "step": 886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00027625721352019786,
+      "loss": 5.5954,
+      "step": 887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00027619538334707334,
+      "loss": 4.7391,
+      "step": 888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002761335531739489,
+      "loss": 5.8233,
+      "step": 889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002760717230008244,
+      "loss": 5.5343,
+      "step": 890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00027600989282769987,
+      "loss": 4.9222,
+      "step": 891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002759480626545754,
+      "loss": 5.0679,
+      "step": 892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00027588623248145095,
+      "loss": 6.3717,
+      "step": 893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00027582440230832644,
+      "loss": 4.9716,
+      "step": 894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00027576257213520193,
+      "loss": 5.208,
+      "step": 895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002757007419620775,
+      "loss": 5.3652,
+      "step": 896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00027563891178895297,
+      "loss": 4.8286,
+      "step": 897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002755770816158285,
+      "loss": 6.7559,
+      "step": 898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000275515251442704,
+      "loss": 6.7759,
+      "step": 899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00027545342126957954,
+      "loss": 5.7652,
+      "step": 900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027539159109645503,
+      "loss": 4.9373,
+      "step": 901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002753297609233306,
+      "loss": 5.4103,
+      "step": 902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027526793075020607,
+      "loss": 5.9478,
+      "step": 903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027520610057708156,
+      "loss": 5.0821,
+      "step": 904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002751442704039571,
+      "loss": 5.1597,
+      "step": 905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027508244023083264,
+      "loss": 5.3728,
+      "step": 906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027502061005770813,
+      "loss": 5.317,
+      "step": 907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002749587798845837,
+      "loss": 4.7962,
+      "step": 908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027489694971145917,
+      "loss": 5.0814,
+      "step": 909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027483511953833466,
+      "loss": 5.0159,
+      "step": 910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002747732893652102,
+      "loss": 5.3736,
+      "step": 911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027471145919208574,
+      "loss": 5.0327,
+      "step": 912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027464962901896123,
+      "loss": 5.3104,
+      "step": 913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002745877988458367,
+      "loss": 4.908,
+      "step": 914
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027452596867271227,
+      "loss": 5.3872,
+      "step": 915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002744641384995878,
+      "loss": 5.1436,
+      "step": 916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002744023083264633,
+      "loss": 4.6757,
+      "step": 917
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002743404781533388,
+      "loss": 5.5533,
+      "step": 918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027427864798021433,
+      "loss": 4.9847,
+      "step": 919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002742168178070898,
+      "loss": 4.6189,
+      "step": 920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027415498763396536,
+      "loss": 4.9507,
+      "step": 921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027409315746084085,
+      "loss": 4.8455,
+      "step": 922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002740313272877164,
+      "loss": 4.9744,
+      "step": 923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002739694971145919,
+      "loss": 5.5476,
+      "step": 924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027390766694146743,
+      "loss": 5.2253,
+      "step": 925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002738458367683429,
+      "loss": 5.9283,
+      "step": 926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002737840065952184,
+      "loss": 4.8417,
+      "step": 927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027372217642209395,
+      "loss": 4.9368,
+      "step": 928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002736603462489695,
+      "loss": 4.6865,
+      "step": 929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000273598516075845,
+      "loss": 4.9849,
+      "step": 930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002735366859027205,
+      "loss": 5.1776,
+      "step": 931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000273474855729596,
+      "loss": 5.1977,
+      "step": 932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002734130255564715,
+      "loss": 5.0894,
+      "step": 933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027335119538334705,
+      "loss": 5.0541,
+      "step": 934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002732893652102226,
+      "loss": 5.3406,
+      "step": 935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002732275350370981,
+      "loss": 5.2157,
+      "step": 936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002731657048639736,
+      "loss": 5.2462,
+      "step": 937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002731038746908491,
+      "loss": 4.7258,
+      "step": 938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002730420445177246,
+      "loss": 5.0443,
+      "step": 939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027298021434460015,
+      "loss": 5.2481,
+      "step": 940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027291838417147564,
+      "loss": 5.1747,
+      "step": 941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002728565539983512,
+      "loss": 4.8337,
+      "step": 942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002727947238252267,
+      "loss": 6.0124,
+      "step": 943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002727328936521022,
+      "loss": 5.3778,
+      "step": 944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002726710634789777,
+      "loss": 5.0874,
+      "step": 945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027260923330585325,
+      "loss": 5.0156,
+      "step": 946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027254740313272874,
+      "loss": 5.0765,
+      "step": 947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002724855729596043,
+      "loss": 4.9047,
+      "step": 948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002724237427864798,
+      "loss": 5.8466,
+      "step": 949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027236191261335526,
+      "loss": 4.4749,
+      "step": 950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002723000824402308,
+      "loss": 5.052,
+      "step": 951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027223825226710635,
+      "loss": 5.1601,
+      "step": 952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027217642209398184,
+      "loss": 5.2512,
+      "step": 953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027211459192085733,
+      "loss": 4.6179,
+      "step": 954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002720527617477329,
+      "loss": 5.0048,
+      "step": 955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027199093157460836,
+      "loss": 4.8727,
+      "step": 956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002719291014014839,
+      "loss": 5.1898,
+      "step": 957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027186727122835945,
+      "loss": 5.0084,
+      "step": 958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027180544105523494,
+      "loss": 6.1726,
+      "step": 959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027174361088211043,
+      "loss": 5.191,
+      "step": 960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000271681780708986,
+      "loss": 4.7164,
+      "step": 961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027161995053586146,
+      "loss": 5.0394,
+      "step": 962
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000271558120362737,
+      "loss": 5.3735,
+      "step": 963
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002714962901896125,
+      "loss": 4.9791,
+      "step": 964
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027143446001648804,
+      "loss": 4.8134,
+      "step": 965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027137262984336353,
+      "loss": 4.8007,
+      "step": 966
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002713107996702391,
+      "loss": 5.2374,
+      "step": 967
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027124896949711456,
+      "loss": 4.8277,
+      "step": 968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027118713932399005,
+      "loss": 5.1394,
+      "step": 969
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002711253091508656,
+      "loss": 4.7522,
+      "step": 970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027106347897774114,
+      "loss": 4.8956,
+      "step": 971
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027100164880461663,
+      "loss": 5.0277,
+      "step": 972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002709398186314921,
+      "loss": 4.9744,
+      "step": 973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00027087798845836766,
+      "loss": 6.2504,
+      "step": 974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00027081615828524315,
+      "loss": 5.0479,
+      "step": 975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002707543281121187,
+      "loss": 5.2768,
+      "step": 976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002706924979389942,
+      "loss": 5.7139,
+      "step": 977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00027063066776586973,
+      "loss": 5.0248,
+      "step": 978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002705688375927452,
+      "loss": 5.6865,
+      "step": 979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00027050700741962076,
+      "loss": 4.9049,
+      "step": 980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002704451772464963,
+      "loss": 4.4972,
+      "step": 981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002703833470733718,
+      "loss": 4.5203,
+      "step": 982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002703215169002473,
+      "loss": 5.1983,
+      "step": 983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00027025968672712283,
+      "loss": 5.1435,
+      "step": 984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002701978565539983,
+      "loss": 4.5972,
+      "step": 985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002701360263808738,
+      "loss": 5.1541,
+      "step": 986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00027007419620774935,
+      "loss": 4.8833,
+      "step": 987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002700123660346249,
+      "loss": 4.773,
+      "step": 988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002699505358615004,
+      "loss": 5.0686,
+      "step": 989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00026988870568837593,
+      "loss": 4.696,
+      "step": 990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002698268755152514,
+      "loss": 5.2269,
+      "step": 991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002697650453421269,
+      "loss": 5.4459,
+      "step": 992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00026970321516900245,
+      "loss": 5.2011,
+      "step": 993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000269641384995878,
+      "loss": 5.7754,
+      "step": 994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002695795548227535,
+      "loss": 4.798,
+      "step": 995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000269517724649629,
+      "loss": 4.6332,
+      "step": 996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002694558944765045,
+      "loss": 4.7575,
+      "step": 997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00026939406430338,
+      "loss": 5.8504,
+      "step": 998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00026933223413025555,
+      "loss": 5.316,
+      "step": 999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00026927040395713104,
+      "loss": 4.808,
+      "step": 1000
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 5.316086292266846,
+      "eval_runtime": 962.3868,
+      "eval_samples_per_second": 2.745,
+      "eval_steps_per_second": 0.344,
+      "eval_wer": 1.9636850456168187,
+      "step": 1000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002692085737840066,
+      "loss": 5.3972,
+      "step": 1001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00026914674361088207,
+      "loss": 4.9263,
+      "step": 1002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002690849134377576,
+      "loss": 4.9292,
+      "step": 1003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002690230832646331,
+      "loss": 5.3926,
+      "step": 1004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002689612530915086,
+      "loss": 5.1111,
+      "step": 1005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00026889942291838414,
+      "loss": 4.9919,
+      "step": 1006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002688375927452597,
+      "loss": 6.09,
+      "step": 1007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00026877576257213517,
+      "loss": 5.5871,
+      "step": 1008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00026871393239901066,
+      "loss": 5.4539,
+      "step": 1009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002686521022258862,
+      "loss": 4.7883,
+      "step": 1010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00026859027205276175,
+      "loss": 5.1647,
+      "step": 1011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00026852844187963724,
+      "loss": 5.0851,
+      "step": 1012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002684666117065128,
+      "loss": 5.1146,
+      "step": 1013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00026840478153338827,
+      "loss": 5.3235,
+      "step": 1014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00026834295136026376,
+      "loss": 6.1531,
+      "step": 1015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002682811211871393,
+      "loss": 4.793,
+      "step": 1016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00026821929101401485,
+      "loss": 4.9152,
+      "step": 1017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00026815746084089034,
+      "loss": 4.8681,
+      "step": 1018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00026809563066776583,
+      "loss": 6.0258,
+      "step": 1019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00026803380049464137,
+      "loss": 5.5548,
+      "step": 1020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00026797197032151686,
+      "loss": 5.6167,
+      "step": 1021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002679101401483924,
+      "loss": 5.4006,
+      "step": 1022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002678483099752679,
+      "loss": 4.8463,
+      "step": 1023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00026778647980214344,
+      "loss": 5.2015,
+      "step": 1024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002677246496290189,
+      "loss": 5.5833,
+      "step": 1025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00026766281945589447,
+      "loss": 5.2936,
+      "step": 1026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00026760098928276996,
+      "loss": 5.0993,
+      "step": 1027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00026753915910964545,
+      "loss": 5.2281,
+      "step": 1028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000267477328936521,
+      "loss": 5.2282,
+      "step": 1029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00026741549876339654,
+      "loss": 5.2021,
+      "step": 1030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000267353668590272,
+      "loss": 5.2786,
+      "step": 1031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002672918384171475,
+      "loss": 5.6146,
+      "step": 1032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00026723000824402306,
+      "loss": 4.8739,
+      "step": 1033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00026716817807089855,
+      "loss": 4.6158,
+      "step": 1034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002671063478977741,
+      "loss": 4.947,
+      "step": 1035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00026704451772464964,
+      "loss": 5.6075,
+      "step": 1036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002669826875515251,
+      "loss": 5.0564,
+      "step": 1037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002669208573784006,
+      "loss": 4.9035,
+      "step": 1038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00026685902720527616,
+      "loss": 4.9699,
+      "step": 1039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002667971970321517,
+      "loss": 5.6175,
+      "step": 1040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002667353668590272,
+      "loss": 5.8511,
+      "step": 1041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002666735366859027,
+      "loss": 4.9707,
+      "step": 1042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002666117065127782,
+      "loss": 5.8618,
+      "step": 1043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002665498763396537,
+      "loss": 5.1241,
+      "step": 1044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00026648804616652926,
+      "loss": 5.6292,
+      "step": 1045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00026642621599340475,
+      "loss": 5.1288,
+      "step": 1046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002663643858202803,
+      "loss": 5.1893,
+      "step": 1047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002663025556471558,
+      "loss": 5.2206,
+      "step": 1048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002662407254740313,
+      "loss": 5.7287,
+      "step": 1049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002661788953009068,
+      "loss": 6.316,
+      "step": 1050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002661170651277823,
+      "loss": 5.9071,
+      "step": 1051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00026605523495465785,
+      "loss": 4.7441,
+      "step": 1052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002659934047815334,
+      "loss": 4.8661,
+      "step": 1053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002659315746084089,
+      "loss": 4.6875,
+      "step": 1054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00026586974443528437,
+      "loss": 5.2006,
+      "step": 1055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002658079142621599,
+      "loss": 4.8059,
+      "step": 1056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002657460840890354,
+      "loss": 5.4153,
+      "step": 1057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00026568425391591095,
+      "loss": 4.8661,
+      "step": 1058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002656224237427865,
+      "loss": 6.2058,
+      "step": 1059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000265560593569662,
+      "loss": 4.75,
+      "step": 1060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00026549876339653747,
+      "loss": 5.0393,
+      "step": 1061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000265436933223413,
+      "loss": 4.9394,
+      "step": 1062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002653751030502885,
+      "loss": 5.8232,
+      "step": 1063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000265313272877164,
+      "loss": 4.8857,
+      "step": 1064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026525144270403954,
+      "loss": 5.2487,
+      "step": 1065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002651896125309151,
+      "loss": 4.7135,
+      "step": 1066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026512778235779057,
+      "loss": 5.098,
+      "step": 1067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002650659521846661,
+      "loss": 5.0074,
+      "step": 1068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002650041220115416,
+      "loss": 5.1129,
+      "step": 1069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026494229183841715,
+      "loss": 4.9466,
+      "step": 1070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026488046166529264,
+      "loss": 4.8283,
+      "step": 1071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002648186314921682,
+      "loss": 4.5856,
+      "step": 1072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026475680131904367,
+      "loss": 4.8334,
+      "step": 1073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026469497114591916,
+      "loss": 5.6573,
+      "step": 1074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002646331409727947,
+      "loss": 5.2807,
+      "step": 1075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026457131079967025,
+      "loss": 5.8586,
+      "step": 1076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026450948062654574,
+      "loss": 5.105,
+      "step": 1077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002644476504534212,
+      "loss": 4.9834,
+      "step": 1078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026438582028029677,
+      "loss": 4.9477,
+      "step": 1079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026432399010717226,
+      "loss": 5.5048,
+      "step": 1080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002642621599340478,
+      "loss": 5.1965,
+      "step": 1081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026420032976092335,
+      "loss": 5.1853,
+      "step": 1082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026413849958779883,
+      "loss": 4.5478,
+      "step": 1083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002640766694146743,
+      "loss": 4.8392,
+      "step": 1084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026401483924154987,
+      "loss": 4.9108,
+      "step": 1085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026395300906842536,
+      "loss": 5.4057,
+      "step": 1086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026389117889530085,
+      "loss": 5.3023,
+      "step": 1087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002638293487221764,
+      "loss": 5.7772,
+      "step": 1088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026376751854905193,
+      "loss": 5.8115,
+      "step": 1089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002637056883759274,
+      "loss": 4.7914,
+      "step": 1090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026364385820280297,
+      "loss": 4.7547,
+      "step": 1091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026358202802967846,
+      "loss": 4.9224,
+      "step": 1092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026352019785655395,
+      "loss": 4.988,
+      "step": 1093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002634583676834295,
+      "loss": 5.6971,
+      "step": 1094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026339653751030503,
+      "loss": 4.9822,
+      "step": 1095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002633347073371805,
+      "loss": 5.8275,
+      "step": 1096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000263272877164056,
+      "loss": 5.8619,
+      "step": 1097
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00026321104699093156,
+      "loss": 7.224,
+      "step": 1098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002631492168178071,
+      "loss": 4.995,
+      "step": 1099
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002630873866446826,
+      "loss": 5.3585,
+      "step": 1100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002630255564715581,
+      "loss": 4.8913,
+      "step": 1101
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002629637262984336,
+      "loss": 5.0946,
+      "step": 1102
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002629018961253091,
+      "loss": 5.0516,
+      "step": 1103
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00026284006595218466,
+      "loss": 4.588,
+      "step": 1104
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00026277823577906015,
+      "loss": 4.79,
+      "step": 1105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002627164056059357,
+      "loss": 4.5724,
+      "step": 1106
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002626545754328112,
+      "loss": 4.8343,
+      "step": 1107
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002625927452596867,
+      "loss": 4.6241,
+      "step": 1108
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002625309150865622,
+      "loss": 4.8808,
+      "step": 1109
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002624690849134377,
+      "loss": 5.1657,
+      "step": 1110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00026240725474031325,
+      "loss": 4.9605,
+      "step": 1111
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002623454245671888,
+      "loss": 4.9816,
+      "step": 1112
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002622835943940643,
+      "loss": 5.2379,
+      "step": 1113
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002622217642209398,
+      "loss": 5.3536,
+      "step": 1114
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002621599340478153,
+      "loss": 5.4409,
+      "step": 1115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002620981038746908,
+      "loss": 4.8492,
+      "step": 1116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00026203627370156634,
+      "loss": 4.9415,
+      "step": 1117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002619744435284419,
+      "loss": 5.262,
+      "step": 1118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002619126133553174,
+      "loss": 4.8039,
+      "step": 1119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00026185078318219287,
+      "loss": 5.1719,
+      "step": 1120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002617889530090684,
+      "loss": 4.8792,
+      "step": 1121
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002617271228359439,
+      "loss": 4.9534,
+      "step": 1122
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00026166529266281944,
+      "loss": 4.8145,
+      "step": 1123
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00026160346248969493,
+      "loss": 4.6025,
+      "step": 1124
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002615416323165705,
+      "loss": 5.454,
+      "step": 1125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00026147980214344597,
+      "loss": 5.2296,
+      "step": 1126
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002614179719703215,
+      "loss": 4.6148,
+      "step": 1127
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000261356141797197,
+      "loss": 5.1548,
+      "step": 1128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002612943116240725,
+      "loss": 4.5442,
+      "step": 1129
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00026123248145094803,
+      "loss": 5.6391,
+      "step": 1130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002611706512778236,
+      "loss": 5.0214,
+      "step": 1131
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00026110882110469907,
+      "loss": 4.8459,
+      "step": 1132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00026104699093157456,
+      "loss": 5.431,
+      "step": 1133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002609851607584501,
+      "loss": 4.8109,
+      "step": 1134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00026092333058532564,
+      "loss": 5.1768,
+      "step": 1135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00026086150041220113,
+      "loss": 4.6713,
+      "step": 1136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002607996702390766,
+      "loss": 5.2599,
+      "step": 1137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00026073784006595217,
+      "loss": 4.9369,
+      "step": 1138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00026067600989282766,
+      "loss": 4.6364,
+      "step": 1139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002606141797197032,
+      "loss": 5.0594,
+      "step": 1140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00026055234954657874,
+      "loss": 5.0007,
+      "step": 1141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00026049051937345423,
+      "loss": 5.4478,
+      "step": 1142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002604286892003297,
+      "loss": 5.6537,
+      "step": 1143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00026036685902720527,
+      "loss": 5.2074,
+      "step": 1144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00026030502885408075,
+      "loss": 4.7121,
+      "step": 1145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002602431986809563,
+      "loss": 5.1745,
+      "step": 1146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002601813685078318,
+      "loss": 4.9628,
+      "step": 1147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00026011953833470733,
+      "loss": 5.3851,
+      "step": 1148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002600577081615828,
+      "loss": 7.8282,
+      "step": 1149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00025999587798845836,
+      "loss": 5.1095,
+      "step": 1150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025993404781533385,
+      "loss": 5.0597,
+      "step": 1151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025987221764220934,
+      "loss": 5.4338,
+      "step": 1152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002598103874690849,
+      "loss": 4.8596,
+      "step": 1153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025974855729596043,
+      "loss": 4.9756,
+      "step": 1154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002596867271228359,
+      "loss": 4.9233,
+      "step": 1155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002596248969497114,
+      "loss": 5.14,
+      "step": 1156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025956306677658695,
+      "loss": 4.6902,
+      "step": 1157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025950123660346244,
+      "loss": 4.9915,
+      "step": 1158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000259439406430338,
+      "loss": 4.6641,
+      "step": 1159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002593775762572135,
+      "loss": 5.2024,
+      "step": 1160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000259315746084089,
+      "loss": 5.2313,
+      "step": 1161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002592539159109645,
+      "loss": 4.8624,
+      "step": 1162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025919208573784005,
+      "loss": 5.1057,
+      "step": 1163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002591302555647156,
+      "loss": 4.6986,
+      "step": 1164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002590684253915911,
+      "loss": 4.7582,
+      "step": 1165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002590065952184666,
+      "loss": 5.2597,
+      "step": 1166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002589447650453421,
+      "loss": 4.9493,
+      "step": 1167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002588829348722176,
+      "loss": 4.5007,
+      "step": 1168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025882110469909315,
+      "loss": 4.9723,
+      "step": 1169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025875927452596864,
+      "loss": 5.3528,
+      "step": 1170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002586974443528442,
+      "loss": 5.7956,
+      "step": 1171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002586356141797197,
+      "loss": 4.5666,
+      "step": 1172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002585737840065952,
+      "loss": 5.2376,
+      "step": 1173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002585119538334707,
+      "loss": 4.6477,
+      "step": 1174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002584501236603462,
+      "loss": 5.1603,
+      "step": 1175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025838829348722174,
+      "loss": 5.5929,
+      "step": 1176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002583264633140973,
+      "loss": 5.2497,
+      "step": 1177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002582646331409728,
+      "loss": 5.3413,
+      "step": 1178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025820280296784826,
+      "loss": 4.707,
+      "step": 1179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002581409727947238,
+      "loss": 4.8615,
+      "step": 1180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002580791426215993,
+      "loss": 4.6586,
+      "step": 1181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025801731244847484,
+      "loss": 5.4502,
+      "step": 1182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025795548227535033,
+      "loss": 5.5671,
+      "step": 1183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002578936521022259,
+      "loss": 4.7177,
+      "step": 1184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025783182192910136,
+      "loss": 5.5442,
+      "step": 1185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002577699917559769,
+      "loss": 5.2643,
+      "step": 1186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002577081615828524,
+      "loss": 5.4082,
+      "step": 1187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002576463314097279,
+      "loss": 4.9577,
+      "step": 1188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025758450123660343,
+      "loss": 5.1235,
+      "step": 1189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000257522671063479,
+      "loss": 5.0458,
+      "step": 1190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025746084089035446,
+      "loss": 4.9827,
+      "step": 1191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025739901071722995,
+      "loss": 5.3889,
+      "step": 1192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002573371805441055,
+      "loss": 5.7737,
+      "step": 1193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025727535037098104,
+      "loss": 4.7976,
+      "step": 1194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025721352019785653,
+      "loss": 5.1833,
+      "step": 1195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002571516900247321,
+      "loss": 4.6816,
+      "step": 1196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025708985985160756,
+      "loss": 4.5395,
+      "step": 1197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025702802967848305,
+      "loss": 5.0498,
+      "step": 1198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002569661995053586,
+      "loss": 4.9895,
+      "step": 1199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025690436933223414,
+      "loss": 4.5757,
+      "step": 1200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025684253915910963,
+      "loss": 5.0415,
+      "step": 1201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002567807089859851,
+      "loss": 4.9144,
+      "step": 1202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025671887881286066,
+      "loss": 4.7278,
+      "step": 1203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025665704863973615,
+      "loss": 4.6963,
+      "step": 1204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002565952184666117,
+      "loss": 4.4681,
+      "step": 1205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002565333882934872,
+      "loss": 5.1195,
+      "step": 1206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025647155812036273,
+      "loss": 5.0423,
+      "step": 1207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002564097279472382,
+      "loss": 4.9226,
+      "step": 1208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025634789777411376,
+      "loss": 5.0989,
+      "step": 1209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025628606760098925,
+      "loss": 4.7473,
+      "step": 1210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025622423742786474,
+      "loss": 5.0641,
+      "step": 1211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002561624072547403,
+      "loss": 5.2754,
+      "step": 1212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025610057708161583,
+      "loss": 5.2081,
+      "step": 1213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002560387469084913,
+      "loss": 4.9792,
+      "step": 1214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002559769167353668,
+      "loss": 4.9652,
+      "step": 1215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025591508656224235,
+      "loss": 4.967,
+      "step": 1216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025585325638911784,
+      "loss": 5.1002,
+      "step": 1217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002557914262159934,
+      "loss": 4.4061,
+      "step": 1218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025572959604286893,
+      "loss": 4.7349,
+      "step": 1219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002556677658697444,
+      "loss": 5.0702,
+      "step": 1220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002556059356966199,
+      "loss": 4.8113,
+      "step": 1221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025554410552349545,
+      "loss": 5.4536,
+      "step": 1222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000255482275350371,
+      "loss": 4.9873,
+      "step": 1223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002554204451772465,
+      "loss": 5.2266,
+      "step": 1224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000255358615004122,
+      "loss": 5.8476,
+      "step": 1225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002552967848309975,
+      "loss": 4.8324,
+      "step": 1226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000255234954657873,
+      "loss": 4.6393,
+      "step": 1227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025517312448474855,
+      "loss": 4.7609,
+      "step": 1228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025511129431162404,
+      "loss": 4.6344,
+      "step": 1229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002550494641384996,
+      "loss": 5.5517,
+      "step": 1230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025498763396537507,
+      "loss": 4.9565,
+      "step": 1231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002549258037922506,
+      "loss": 5.4854,
+      "step": 1232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002548639736191261,
+      "loss": 4.9927,
+      "step": 1233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002548021434460016,
+      "loss": 4.6621,
+      "step": 1234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025474031327287714,
+      "loss": 5.2237,
+      "step": 1235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002546784830997527,
+      "loss": 4.8469,
+      "step": 1236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025461665292662817,
+      "loss": 5.0383,
+      "step": 1237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025455482275350366,
+      "loss": 4.8003,
+      "step": 1238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002544929925803792,
+      "loss": 5.0467,
+      "step": 1239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002544311624072547,
+      "loss": 5.167,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00025436933223413024,
+      "loss": 4.8573,
+      "step": 1241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002543075020610058,
+      "loss": 5.255,
+      "step": 1242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00025424567188788127,
+      "loss": 5.349,
+      "step": 1243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00025418384171475676,
+      "loss": 4.8066,
+      "step": 1244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002541220115416323,
+      "loss": 4.5208,
+      "step": 1245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002540601813685078,
+      "loss": 5.0155,
+      "step": 1246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002539983511953833,
+      "loss": 5.0189,
+      "step": 1247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00025393652102225883,
+      "loss": 4.9931,
+      "step": 1248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00025387469084913437,
+      "loss": 5.0475,
+      "step": 1249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00025381286067600986,
+      "loss": 5.3964,
+      "step": 1250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002537510305028854,
+      "loss": 5.232,
+      "step": 1251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002536892003297609,
+      "loss": 5.632,
+      "step": 1252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00025362737015663644,
+      "loss": 5.1756,
+      "step": 1253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00025356553998351193,
+      "loss": 4.8181,
+      "step": 1254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00025350370981038747,
+      "loss": 5.5256,
+      "step": 1255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00025344187963726296,
+      "loss": 4.9942,
+      "step": 1256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00025338004946413845,
+      "loss": 4.7655,
+      "step": 1257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000253318219291014,
+      "loss": 4.6677,
+      "step": 1258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00025325638911788954,
+      "loss": 4.8027,
+      "step": 1259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000253194558944765,
+      "loss": 5.3053,
+      "step": 1260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002531327287716405,
+      "loss": 4.9574,
+      "step": 1261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00025307089859851606,
+      "loss": 5.0599,
+      "step": 1262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00025300906842539155,
+      "loss": 4.8895,
+      "step": 1263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002529472382522671,
+      "loss": 4.8101,
+      "step": 1264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00025288540807914264,
+      "loss": 5.4785,
+      "step": 1265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002528235779060181,
+      "loss": 5.21,
+      "step": 1266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002527617477328936,
+      "loss": 5.3368,
+      "step": 1267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00025269991755976916,
+      "loss": 4.8664,
+      "step": 1268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00025263808738664465,
+      "loss": 4.8469,
+      "step": 1269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00025257625721352014,
+      "loss": 4.8819,
+      "step": 1270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002525144270403957,
+      "loss": 5.1459,
+      "step": 1271
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002524525968672712,
+      "loss": 4.9804,
+      "step": 1272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002523907666941467,
+      "loss": 5.121,
+      "step": 1273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00025232893652102226,
+      "loss": 4.6535,
+      "step": 1274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00025226710634789775,
+      "loss": 4.723,
+      "step": 1275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00025220527617477324,
+      "loss": 5.1922,
+      "step": 1276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002521434460016488,
+      "loss": 5.014,
+      "step": 1277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002520816158285243,
+      "loss": 5.2942,
+      "step": 1278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002520197856553998,
+      "loss": 5.0539,
+      "step": 1279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002519579554822753,
+      "loss": 4.9986,
+      "step": 1280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00025189612530915085,
+      "loss": 4.807,
+      "step": 1281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002518342951360264,
+      "loss": 5.4869,
+      "step": 1282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002517724649629019,
+      "loss": 4.8406,
+      "step": 1283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00025171063478977737,
+      "loss": 4.7268,
+      "step": 1284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002516488046166529,
+      "loss": 4.7229,
+      "step": 1285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002515869744435284,
+      "loss": 4.8856,
+      "step": 1286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00025152514427040395,
+      "loss": 4.8853,
+      "step": 1287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002514633140972795,
+      "loss": 4.738,
+      "step": 1288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000251401483924155,
+      "loss": 4.7853,
+      "step": 1289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00025133965375103047,
+      "loss": 4.8232,
+      "step": 1290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000251277823577906,
+      "loss": 5.1734,
+      "step": 1291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002512159934047815,
+      "loss": 6.4219,
+      "step": 1292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000251154163231657,
+      "loss": 6.9067,
+      "step": 1293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00025109233305853254,
+      "loss": 4.6326,
+      "step": 1294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002510305028854081,
+      "loss": 4.5534,
+      "step": 1295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00025096867271228357,
+      "loss": 4.528,
+      "step": 1296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002509068425391591,
+      "loss": 4.5179,
+      "step": 1297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002508450123660346,
+      "loss": 4.3005,
+      "step": 1298
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002507831821929101,
+      "loss": 5.5905,
+      "step": 1299
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00025072135201978564,
+      "loss": 4.5032,
+      "step": 1300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002506595218466612,
+      "loss": 5.4848,
+      "step": 1301
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00025059769167353667,
+      "loss": 5.0562,
+      "step": 1302
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00025053586150041216,
+      "loss": 4.9049,
+      "step": 1303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002504740313272877,
+      "loss": 4.9295,
+      "step": 1304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002504122011541632,
+      "loss": 4.7313,
+      "step": 1305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00025035037098103874,
+      "loss": 5.2901,
+      "step": 1306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002502885408079142,
+      "loss": 4.9584,
+      "step": 1307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00025022671063478977,
+      "loss": 4.5765,
+      "step": 1308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00025016488046166526,
+      "loss": 4.9816,
+      "step": 1309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002501030502885408,
+      "loss": 4.6976,
+      "step": 1310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002500412201154163,
+      "loss": 4.6605,
+      "step": 1311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002499793899422918,
+      "loss": 4.9343,
+      "step": 1312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002499175597691673,
+      "loss": 4.7023,
+      "step": 1313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00024985572959604287,
+      "loss": 4.5407,
+      "step": 1314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00024979389942291836,
+      "loss": 5.2092,
+      "step": 1315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00024973206924979385,
+      "loss": 5.3434,
+      "step": 1316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002496702390766694,
+      "loss": 4.5685,
+      "step": 1317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00024960840890354493,
+      "loss": 6.1873,
+      "step": 1318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002495465787304204,
+      "loss": 4.8428,
+      "step": 1319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00024948474855729597,
+      "loss": 5.1651,
+      "step": 1320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00024942291838417146,
+      "loss": 4.4094,
+      "step": 1321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00024936108821104695,
+      "loss": 4.6467,
+      "step": 1322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002492992580379225,
+      "loss": 4.7516,
+      "step": 1323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00024923742786479803,
+      "loss": 4.8888,
+      "step": 1324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002491755976916735,
+      "loss": 5.6543,
+      "step": 1325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000249113767518549,
+      "loss": 4.7281,
+      "step": 1326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00024905193734542456,
+      "loss": 5.2736,
+      "step": 1327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00024899010717230005,
+      "loss": 5.3608,
+      "step": 1328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002489282769991756,
+      "loss": 5.1933,
+      "step": 1329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002488664468260511,
+      "loss": 5.5551,
+      "step": 1330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002488046166529266,
+      "loss": 5.1322,
+      "step": 1331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002487427864798021,
+      "loss": 4.9818,
+      "step": 1332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00024868095630667766,
+      "loss": 4.7844,
+      "step": 1333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00024861912613355315,
+      "loss": 5.2142,
+      "step": 1334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00024855729596042864,
+      "loss": 4.5375,
+      "step": 1335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002484954657873042,
+      "loss": 4.9879,
+      "step": 1336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002484336356141797,
+      "loss": 4.9624,
+      "step": 1337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002483718054410552,
+      "loss": 5.0347,
+      "step": 1338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002483099752679307,
+      "loss": 5.0021,
+      "step": 1339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00024824814509480625,
+      "loss": 6.4249,
+      "step": 1340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00024818631492168173,
+      "loss": 5.0749,
+      "step": 1341
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002481244847485573,
+      "loss": 5.5644,
+      "step": 1342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00024806265457543277,
+      "loss": 4.941,
+      "step": 1343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002480008244023083,
+      "loss": 5.584,
+      "step": 1344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002479389942291838,
+      "loss": 5.2354,
+      "step": 1345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00024787716405605934,
+      "loss": 4.8508,
+      "step": 1346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002478153338829349,
+      "loss": 6.9728,
+      "step": 1347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002477535037098104,
+      "loss": 4.6134,
+      "step": 1348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00024769167353668587,
+      "loss": 4.9284,
+      "step": 1349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002476298433635614,
+      "loss": 4.8686,
+      "step": 1350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002475680131904369,
+      "loss": 5.0436,
+      "step": 1351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00024750618301731244,
+      "loss": 5.0376,
+      "step": 1352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00024744435284418793,
+      "loss": 4.9365,
+      "step": 1353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002473825226710635,
+      "loss": 5.796,
+      "step": 1354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00024732069249793897,
+      "loss": 4.671,
+      "step": 1355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002472588623248145,
+      "loss": 5.4599,
+      "step": 1356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00024719703215169,
+      "loss": 4.5669,
+      "step": 1357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002471352019785655,
+      "loss": 5.1039,
+      "step": 1358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00024707337180544103,
+      "loss": 5.589,
+      "step": 1359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002470115416323166,
+      "loss": 5.5365,
+      "step": 1360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00024694971145919207,
+      "loss": 4.7548,
+      "step": 1361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00024688788128606756,
+      "loss": 4.8798,
+      "step": 1362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002468260511129431,
+      "loss": 4.562,
+      "step": 1363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002467642209398186,
+      "loss": 5.4912,
+      "step": 1364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024670239076669413,
+      "loss": 5.1,
+      "step": 1365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002466405605935696,
+      "loss": 4.5145,
+      "step": 1366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024657873042044517,
+      "loss": 5.2164,
+      "step": 1367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024651690024732066,
+      "loss": 4.5857,
+      "step": 1368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002464550700741962,
+      "loss": 4.8257,
+      "step": 1369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002463932399010717,
+      "loss": 4.9395,
+      "step": 1370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002463314097279472,
+      "loss": 5.264,
+      "step": 1371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002462695795548227,
+      "loss": 4.5067,
+      "step": 1372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024620774938169827,
+      "loss": 5.2336,
+      "step": 1373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024614591920857375,
+      "loss": 4.9007,
+      "step": 1374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024608408903544924,
+      "loss": 4.721,
+      "step": 1375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002460222588623248,
+      "loss": 4.755,
+      "step": 1376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024596042868920033,
+      "loss": 5.3678,
+      "step": 1377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002458985985160758,
+      "loss": 5.1153,
+      "step": 1378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024583676834295137,
+      "loss": 4.7021,
+      "step": 1379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024577493816982685,
+      "loss": 4.6078,
+      "step": 1380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024571310799670234,
+      "loss": 4.356,
+      "step": 1381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002456512778235779,
+      "loss": 4.8965,
+      "step": 1382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024558944765045343,
+      "loss": 4.7533,
+      "step": 1383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002455276174773289,
+      "loss": 5.349,
+      "step": 1384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002454657873042044,
+      "loss": 5.5817,
+      "step": 1385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024540395713107995,
+      "loss": 4.7784,
+      "step": 1386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024534212695795544,
+      "loss": 5.2665,
+      "step": 1387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000245280296784831,
+      "loss": 4.9542,
+      "step": 1388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002452184666117065,
+      "loss": 4.688,
+      "step": 1389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000245156636438582,
+      "loss": 5.4312,
+      "step": 1390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002450948062654575,
+      "loss": 4.7893,
+      "step": 1391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024503297609233305,
+      "loss": 4.6456,
+      "step": 1392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024497114591920854,
+      "loss": 4.6909,
+      "step": 1393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024490931574608403,
+      "loss": 4.93,
+      "step": 1394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002448474855729596,
+      "loss": 7.829,
+      "step": 1395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002447856553998351,
+      "loss": 4.6632,
+      "step": 1396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002447238252267106,
+      "loss": 5.3533,
+      "step": 1397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002446619950535861,
+      "loss": 4.7502,
+      "step": 1398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024460016488046164,
+      "loss": 4.8455,
+      "step": 1399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024453833470733713,
+      "loss": 5.0882,
+      "step": 1400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002444765045342127,
+      "loss": 4.7681,
+      "step": 1401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002444146743610882,
+      "loss": 5.4333,
+      "step": 1402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002443528441879637,
+      "loss": 4.6288,
+      "step": 1403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002442910140148392,
+      "loss": 5.6259,
+      "step": 1404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024422918384171474,
+      "loss": 4.8853,
+      "step": 1405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002441673536685903,
+      "loss": 6.0669,
+      "step": 1406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024410552349546578,
+      "loss": 4.9232,
+      "step": 1407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024404369332234126,
+      "loss": 5.0711,
+      "step": 1408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024398186314921678,
+      "loss": 4.845,
+      "step": 1409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024392003297609232,
+      "loss": 4.8903,
+      "step": 1410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024385820280296784,
+      "loss": 4.5311,
+      "step": 1411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024379637262984333,
+      "loss": 4.6134,
+      "step": 1412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024373454245671885,
+      "loss": 5.0424,
+      "step": 1413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024367271228359436,
+      "loss": 5.3233,
+      "step": 1414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002436108821104699,
+      "loss": 4.8807,
+      "step": 1415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024354905193734542,
+      "loss": 5.0912,
+      "step": 1416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024348722176422091,
+      "loss": 5.0696,
+      "step": 1417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024342539159109643,
+      "loss": 6.027,
+      "step": 1418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024336356141797195,
+      "loss": 5.3156,
+      "step": 1419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024330173124484746,
+      "loss": 4.908,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024323990107172295,
+      "loss": 4.9124,
+      "step": 1421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002431780708985985,
+      "loss": 5.1188,
+      "step": 1422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000243116240725474,
+      "loss": 4.6602,
+      "step": 1423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024305441055234953,
+      "loss": 4.9584,
+      "step": 1424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024299258037922505,
+      "loss": 5.1636,
+      "step": 1425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024293075020610054,
+      "loss": 4.9277,
+      "step": 1426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024286892003297608,
+      "loss": 4.7002,
+      "step": 1427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002428070898598516,
+      "loss": 5.096,
+      "step": 1428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002427452596867271,
+      "loss": 5.2968,
+      "step": 1429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002426834295136026,
+      "loss": 5.618,
+      "step": 1430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024262159934047812,
+      "loss": 5.8179,
+      "step": 1431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024255976916735364,
+      "loss": 4.9093,
+      "step": 1432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024249793899422918,
+      "loss": 5.5573,
+      "step": 1433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002424361088211047,
+      "loss": 4.784,
+      "step": 1434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024237427864798019,
+      "loss": 4.7103,
+      "step": 1435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002423124484748557,
+      "loss": 4.8006,
+      "step": 1436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024225061830173122,
+      "loss": 4.9781,
+      "step": 1437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024218878812860674,
+      "loss": 4.7442,
+      "step": 1438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024212695795548228,
+      "loss": 4.9979,
+      "step": 1439
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024206512778235777,
+      "loss": 5.4074,
+      "step": 1440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024200329760923328,
+      "loss": 6.414,
+      "step": 1441
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002419414674361088,
+      "loss": 4.649,
+      "step": 1442
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024187963726298432,
+      "loss": 5.0094,
+      "step": 1443
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002418178070898598,
+      "loss": 4.9381,
+      "step": 1444
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024175597691673535,
+      "loss": 4.9883,
+      "step": 1445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024169414674361087,
+      "loss": 5.4105,
+      "step": 1446
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024163231657048638,
+      "loss": 4.9757,
+      "step": 1447
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002415704863973619,
+      "loss": 4.9049,
+      "step": 1448
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002415086562242374,
+      "loss": 5.671,
+      "step": 1449
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002414468260511129,
+      "loss": 4.7704,
+      "step": 1450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024138499587798845,
+      "loss": 4.8881,
+      "step": 1451
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024132316570486397,
+      "loss": 4.9539,
+      "step": 1452
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024126133553173946,
+      "loss": 4.8245,
+      "step": 1453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024119950535861497,
+      "loss": 4.5529,
+      "step": 1454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002411376751854905,
+      "loss": 5.2261,
+      "step": 1455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000241075845012366,
+      "loss": 5.1069,
+      "step": 1456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024101401483924155,
+      "loss": 5.2203,
+      "step": 1457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024095218466611704,
+      "loss": 5.0829,
+      "step": 1458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024089035449299256,
+      "loss": 5.0143,
+      "step": 1459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024082852431986807,
+      "loss": 5.5523,
+      "step": 1460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002407666941467436,
+      "loss": 4.9239,
+      "step": 1461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024070486397361908,
+      "loss": 4.8839,
+      "step": 1462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024064303380049462,
+      "loss": 4.7183,
+      "step": 1463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024058120362737014,
+      "loss": 4.7392,
+      "step": 1464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024051937345424566,
+      "loss": 5.2783,
+      "step": 1465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024045754328112117,
+      "loss": 4.6103,
+      "step": 1466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024039571310799666,
+      "loss": 5.1538,
+      "step": 1467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024033388293487218,
+      "loss": 5.0218,
+      "step": 1468
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024027205276174772,
+      "loss": 4.5367,
+      "step": 1469
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024021022258862324,
+      "loss": 4.4913,
+      "step": 1470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024014839241549876,
+      "loss": 4.8984,
+      "step": 1471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00024008656224237424,
+      "loss": 4.674,
+      "step": 1472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00024002473206924976,
+      "loss": 4.6341,
+      "step": 1473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002399629018961253,
+      "loss": 4.8155,
+      "step": 1474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00023990107172300082,
+      "loss": 5.0706,
+      "step": 1475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002398392415498763,
+      "loss": 5.1294,
+      "step": 1476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00023977741137675183,
+      "loss": 5.0394,
+      "step": 1477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00023971558120362734,
+      "loss": 4.421,
+      "step": 1478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00023965375103050286,
+      "loss": 4.8376,
+      "step": 1479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002395919208573784,
+      "loss": 4.4904,
+      "step": 1480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002395300906842539,
+      "loss": 4.942,
+      "step": 1481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002394682605111294,
+      "loss": 5.7279,
+      "step": 1482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00023940643033800493,
+      "loss": 5.0522,
+      "step": 1483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00023934460016488044,
+      "loss": 4.5591,
+      "step": 1484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00023928276999175593,
+      "loss": 4.7385,
+      "step": 1485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00023922093981863145,
+      "loss": 5.0969,
+      "step": 1486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000239159109645507,
+      "loss": 5.3959,
+      "step": 1487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002390972794723825,
+      "loss": 4.8702,
+      "step": 1488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00023903544929925803,
+      "loss": 5.241,
+      "step": 1489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00023897361912613352,
+      "loss": 4.5539,
+      "step": 1490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00023891178895300903,
+      "loss": 5.4145,
+      "step": 1491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00023884995877988458,
+      "loss": 4.8933,
+      "step": 1492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002387881286067601,
+      "loss": 4.2696,
+      "step": 1493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00023872629843363558,
+      "loss": 7.4759,
+      "step": 1494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002386644682605111,
+      "loss": 4.568,
+      "step": 1495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00023860263808738662,
+      "loss": 7.4993,
+      "step": 1496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00023854080791426213,
+      "loss": 5.1732,
+      "step": 1497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00023847897774113768,
+      "loss": 5.0448,
+      "step": 1498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00023841714756801317,
+      "loss": 4.4001,
+      "step": 1499
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00023835531739488868,
+      "loss": 4.8826,
+      "step": 1500
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 5.110193729400635,
+      "eval_runtime": 963.6936,
+      "eval_samples_per_second": 2.742,
+      "eval_steps_per_second": 0.343,
+      "eval_wer": 1.9787385957953194,
+      "step": 1500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002382934872217642,
+      "loss": 4.8312,
+      "step": 1501
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00023823165704863972,
+      "loss": 4.7393,
+      "step": 1502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00023816982687551526,
+      "loss": 5.1558,
+      "step": 1503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00023810799670239075,
+      "loss": 4.8301,
+      "step": 1504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00023804616652926627,
+      "loss": 4.9438,
+      "step": 1505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00023798433635614178,
+      "loss": 5.2592,
+      "step": 1506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002379225061830173,
+      "loss": 4.6101,
+      "step": 1507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002378606760098928,
+      "loss": 4.8245,
+      "step": 1508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002377988458367683,
+      "loss": 4.8601,
+      "step": 1509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023773701566364385,
+      "loss": 4.816,
+      "step": 1510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023767518549051936,
+      "loss": 4.4559,
+      "step": 1511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023761335531739488,
+      "loss": 4.7376,
+      "step": 1512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023755152514427037,
+      "loss": 4.4704,
+      "step": 1513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002374896949711459,
+      "loss": 4.6591,
+      "step": 1514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002374278647980214,
+      "loss": 4.8071,
+      "step": 1515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023736603462489695,
+      "loss": 5.7788,
+      "step": 1516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023730420445177244,
+      "loss": 4.589,
+      "step": 1517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023724237427864795,
+      "loss": 5.1485,
+      "step": 1518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023718054410552347,
+      "loss": 4.9786,
+      "step": 1519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000237118713932399,
+      "loss": 4.9236,
+      "step": 1520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023705688375927453,
+      "loss": 5.0527,
+      "step": 1521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023699505358615002,
+      "loss": 5.3935,
+      "step": 1522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023693322341302554,
+      "loss": 5.5351,
+      "step": 1523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023687139323990105,
+      "loss": 4.6051,
+      "step": 1524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023680956306677657,
+      "loss": 4.5697,
+      "step": 1525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023674773289365209,
+      "loss": 4.543,
+      "step": 1526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023668590272052758,
+      "loss": 4.789,
+      "step": 1527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023662407254740312,
+      "loss": 4.5968,
+      "step": 1528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023656224237427864,
+      "loss": 4.6441,
+      "step": 1529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023650041220115415,
+      "loss": 4.6027,
+      "step": 1530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023643858202802964,
+      "loss": 4.8215,
+      "step": 1531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023637675185490516,
+      "loss": 4.6532,
+      "step": 1532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002363149216817807,
+      "loss": 5.0906,
+      "step": 1533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023625309150865622,
+      "loss": 4.6962,
+      "step": 1534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023619126133553174,
+      "loss": 4.9397,
+      "step": 1535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023612943116240723,
+      "loss": 4.459,
+      "step": 1536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023606760098928274,
+      "loss": 5.5226,
+      "step": 1537
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023600577081615826,
+      "loss": 5.0244,
+      "step": 1538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002359439406430338,
+      "loss": 4.6483,
+      "step": 1539
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002358821104699093,
+      "loss": 4.9378,
+      "step": 1540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002358202802967848,
+      "loss": 5.1525,
+      "step": 1541
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023575845012366032,
+      "loss": 4.7636,
+      "step": 1542
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023569661995053584,
+      "loss": 5.0054,
+      "step": 1543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023563478977741136,
+      "loss": 5.531,
+      "step": 1544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023557295960428685,
+      "loss": 5.6721,
+      "step": 1545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002355111294311624,
+      "loss": 4.4233,
+      "step": 1546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002354492992580379,
+      "loss": 5.3384,
+      "step": 1547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023538746908491342,
+      "loss": 4.8532,
+      "step": 1548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023532563891178891,
+      "loss": 5.2541,
+      "step": 1549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023526380873866443,
+      "loss": 4.4401,
+      "step": 1550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023520197856553997,
+      "loss": 5.4976,
+      "step": 1551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002351401483924155,
+      "loss": 5.3269,
+      "step": 1552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000235078318219291,
+      "loss": 4.7526,
+      "step": 1553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002350164880461665,
+      "loss": 4.7226,
+      "step": 1554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000234954657873042,
+      "loss": 5.2841,
+      "step": 1555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023489282769991753,
+      "loss": 5.2499,
+      "step": 1556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023483099752679307,
+      "loss": 4.9941,
+      "step": 1557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002347691673536686,
+      "loss": 5.1372,
+      "step": 1558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023470733718054408,
+      "loss": 4.8369,
+      "step": 1559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002346455070074196,
+      "loss": 4.9582,
+      "step": 1560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002345836768342951,
+      "loss": 4.9188,
+      "step": 1561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023452184666117063,
+      "loss": 4.9009,
+      "step": 1562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023446001648804612,
+      "loss": 5.0273,
+      "step": 1563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023439818631492166,
+      "loss": 4.7182,
+      "step": 1564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023433635614179718,
+      "loss": 5.2341,
+      "step": 1565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002342745259686727,
+      "loss": 4.6632,
+      "step": 1566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002342126957955482,
+      "loss": 4.5125,
+      "step": 1567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002341508656224237,
+      "loss": 4.9911,
+      "step": 1568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023408903544929925,
+      "loss": 4.7598,
+      "step": 1569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023402720527617476,
+      "loss": 4.6793,
+      "step": 1570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023396537510305028,
+      "loss": 4.6901,
+      "step": 1571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023390354492992577,
+      "loss": 5.2197,
+      "step": 1572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023384171475680128,
+      "loss": 4.8699,
+      "step": 1573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002337798845836768,
+      "loss": 4.4644,
+      "step": 1574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023371805441055234,
+      "loss": 4.6069,
+      "step": 1575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023365622423742786,
+      "loss": 5.0408,
+      "step": 1576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023359439406430335,
+      "loss": 5.0249,
+      "step": 1577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023353256389117887,
+      "loss": 5.2805,
+      "step": 1578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023347073371805438,
+      "loss": 5.2833,
+      "step": 1579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023340890354492993,
+      "loss": 5.4227,
+      "step": 1580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023334707337180542,
+      "loss": 5.0136,
+      "step": 1581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023328524319868093,
+      "loss": 4.8805,
+      "step": 1582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023322341302555645,
+      "loss": 4.6897,
+      "step": 1583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023316158285243197,
+      "loss": 4.4772,
+      "step": 1584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023309975267930748,
+      "loss": 4.8691,
+      "step": 1585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023303792250618297,
+      "loss": 4.6971,
+      "step": 1586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023297609233305852,
+      "loss": 4.6697,
+      "step": 1587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023291426215993403,
+      "loss": 4.8135,
+      "step": 1588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023285243198680955,
+      "loss": 5.2713,
+      "step": 1589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023279060181368507,
+      "loss": 4.5673,
+      "step": 1590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023272877164056056,
+      "loss": 4.6444,
+      "step": 1591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023266694146743607,
+      "loss": 4.7181,
+      "step": 1592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023260511129431162,
+      "loss": 4.5059,
+      "step": 1593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023254328112118713,
+      "loss": 4.7838,
+      "step": 1594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023248145094806262,
+      "loss": 4.5176,
+      "step": 1595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023241962077493814,
+      "loss": 4.805,
+      "step": 1596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023235779060181366,
+      "loss": 5.0844,
+      "step": 1597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002322959604286892,
+      "loss": 5.0018,
+      "step": 1598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023223413025556472,
+      "loss": 5.8315,
+      "step": 1599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002321723000824402,
+      "loss": 5.0351,
+      "step": 1600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023211046990931572,
+      "loss": 4.9223,
+      "step": 1601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023204863973619124,
+      "loss": 4.5704,
+      "step": 1602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023198680956306676,
+      "loss": 4.8067,
+      "step": 1603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023192497938994224,
+      "loss": 4.9751,
+      "step": 1604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002318631492168178,
+      "loss": 4.4727,
+      "step": 1605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002318013190436933,
+      "loss": 4.8956,
+      "step": 1606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023173948887056882,
+      "loss": 4.9467,
+      "step": 1607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023167765869744434,
+      "loss": 5.1108,
+      "step": 1608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023161582852431983,
+      "loss": 4.9339,
+      "step": 1609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023155399835119537,
+      "loss": 4.7366,
+      "step": 1610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002314921681780709,
+      "loss": 4.9438,
+      "step": 1611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002314303380049464,
+      "loss": 4.791,
+      "step": 1612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023136850783182192,
+      "loss": 5.3491,
+      "step": 1613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002313066776586974,
+      "loss": 4.5997,
+      "step": 1614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023124484748557293,
+      "loss": 5.0787,
+      "step": 1615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023118301731244847,
+      "loss": 4.9463,
+      "step": 1616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000231121187139324,
+      "loss": 5.0195,
+      "step": 1617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023105935696619948,
+      "loss": 4.8316,
+      "step": 1618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000230997526793075,
+      "loss": 4.7047,
+      "step": 1619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002309356966199505,
+      "loss": 4.7155,
+      "step": 1620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023087386644682603,
+      "loss": 5.2896,
+      "step": 1621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023081203627370157,
+      "loss": 4.4555,
+      "step": 1622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023075020610057706,
+      "loss": 4.793,
+      "step": 1623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023068837592745258,
+      "loss": 4.8113,
+      "step": 1624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002306265457543281,
+      "loss": 4.6833,
+      "step": 1625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002305647155812036,
+      "loss": 5.122,
+      "step": 1626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002305028854080791,
+      "loss": 4.8652,
+      "step": 1627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023044105523495464,
+      "loss": 4.5311,
+      "step": 1628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023037922506183016,
+      "loss": 4.7945,
+      "step": 1629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023031739488870568,
+      "loss": 4.768,
+      "step": 1630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002302555647155812,
+      "loss": 4.7249,
+      "step": 1631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023019373454245668,
+      "loss": 4.9132,
+      "step": 1632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002301319043693322,
+      "loss": 5.35,
+      "step": 1633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00023007007419620774,
+      "loss": 4.6074,
+      "step": 1634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00023000824402308326,
+      "loss": 4.7533,
+      "step": 1635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00022994641384995875,
+      "loss": 4.7451,
+      "step": 1636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00022988458367683426,
+      "loss": 4.5461,
+      "step": 1637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00022982275350370978,
+      "loss": 5.5378,
+      "step": 1638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002297609233305853,
+      "loss": 4.4944,
+      "step": 1639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00022969909315746084,
+      "loss": 5.2866,
+      "step": 1640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00022963726298433633,
+      "loss": 4.8992,
+      "step": 1641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00022957543281121185,
+      "loss": 4.5945,
+      "step": 1642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00022951360263808736,
+      "loss": 5.1465,
+      "step": 1643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00022945177246496288,
+      "loss": 4.778,
+      "step": 1644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00022938994229183842,
+      "loss": 5.5904,
+      "step": 1645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00022932811211871391,
+      "loss": 5.1272,
+      "step": 1646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00022926628194558943,
+      "loss": 5.0272,
+      "step": 1647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00022920445177246495,
+      "loss": 6.3811,
+      "step": 1648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00022914262159934046,
+      "loss": 4.9819,
+      "step": 1649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00022908079142621595,
+      "loss": 5.0968,
+      "step": 1650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00022901896125309147,
+      "loss": 4.893,
+      "step": 1651
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00022895713107996701,
+      "loss": 4.7593,
+      "step": 1652
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00022889530090684253,
+      "loss": 4.9498,
+      "step": 1653
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00022883347073371805,
+      "loss": 4.7937,
+      "step": 1654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00022877164056059354,
+      "loss": 4.5211,
+      "step": 1655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00022870981038746905,
+      "loss": 4.7016,
+      "step": 1656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002286479802143446,
+      "loss": 5.2413,
+      "step": 1657
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002285861500412201,
+      "loss": 4.9513,
+      "step": 1658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002285243198680956,
+      "loss": 4.8728,
+      "step": 1659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00022846248969497112,
+      "loss": 4.7051,
+      "step": 1660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00022840065952184664,
+      "loss": 4.7876,
+      "step": 1661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00022833882934872215,
+      "loss": 4.722,
+      "step": 1662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002282769991755977,
+      "loss": 4.5457,
+      "step": 1663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00022821516900247319,
+      "loss": 4.8936,
+      "step": 1664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002281533388293487,
+      "loss": 4.4923,
+      "step": 1665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00022809150865622422,
+      "loss": 4.9468,
+      "step": 1666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00022802967848309974,
+      "loss": 4.982,
+      "step": 1667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00022796784830997522,
+      "loss": 5.1383,
+      "step": 1668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00022790601813685074,
+      "loss": 4.5065,
+      "step": 1669
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00022784418796372629,
+      "loss": 4.893,
+      "step": 1670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002277823577906018,
+      "loss": 5.3951,
+      "step": 1671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00022772052761747732,
+      "loss": 5.1552,
+      "step": 1672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002276586974443528,
+      "loss": 4.7542,
+      "step": 1673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00022759686727122832,
+      "loss": 4.769,
+      "step": 1674
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00022753503709810387,
+      "loss": 4.9744,
+      "step": 1675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00022747320692497938,
+      "loss": 4.5423,
+      "step": 1676
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002274113767518549,
+      "loss": 4.4329,
+      "step": 1677
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002273495465787304,
+      "loss": 4.6472,
+      "step": 1678
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002272877164056059,
+      "loss": 5.2094,
+      "step": 1679
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00022722588623248142,
+      "loss": 5.6329,
+      "step": 1680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00022716405605935697,
+      "loss": 4.3759,
+      "step": 1681
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00022710222588623246,
+      "loss": 4.7409,
+      "step": 1682
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00022704039571310797,
+      "loss": 4.9842,
+      "step": 1683
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002269785655399835,
+      "loss": 4.6975,
+      "step": 1684
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000226916735366859,
+      "loss": 4.8755,
+      "step": 1685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022685490519373455,
+      "loss": 4.7109,
+      "step": 1686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022679307502061004,
+      "loss": 4.6217,
+      "step": 1687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022673124484748556,
+      "loss": 4.9188,
+      "step": 1688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022666941467436107,
+      "loss": 4.6157,
+      "step": 1689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002266075845012366,
+      "loss": 4.4872,
+      "step": 1690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022654575432811208,
+      "loss": 4.9654,
+      "step": 1691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002264839241549876,
+      "loss": 5.0463,
+      "step": 1692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022642209398186314,
+      "loss": 4.9781,
+      "step": 1693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022636026380873866,
+      "loss": 5.0893,
+      "step": 1694
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022629843363561417,
+      "loss": 5.1739,
+      "step": 1695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022623660346248966,
+      "loss": 7.2466,
+      "step": 1696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022617477328936518,
+      "loss": 4.6718,
+      "step": 1697
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002261129431162407,
+      "loss": 5.1104,
+      "step": 1698
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022605111294311624,
+      "loss": 5.3072,
+      "step": 1699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022598928276999173,
+      "loss": 5.1041,
+      "step": 1700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022592745259686725,
+      "loss": 5.7561,
+      "step": 1701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022586562242374276,
+      "loss": 5.1465,
+      "step": 1702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022580379225061828,
+      "loss": 4.7081,
+      "step": 1703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022574196207749382,
+      "loss": 5.0748,
+      "step": 1704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002256801319043693,
+      "loss": 4.7419,
+      "step": 1705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022561830173124483,
+      "loss": 4.6796,
+      "step": 1706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022555647155812034,
+      "loss": 4.5807,
+      "step": 1707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022549464138499586,
+      "loss": 5.0832,
+      "step": 1708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022543281121187138,
+      "loss": 5.895,
+      "step": 1709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022537098103874687,
+      "loss": 5.0504,
+      "step": 1710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002253091508656224,
+      "loss": 4.5521,
+      "step": 1711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022524732069249793,
+      "loss": 4.8359,
+      "step": 1712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022518549051937344,
+      "loss": 5.3224,
+      "step": 1713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022512366034624893,
+      "loss": 4.6264,
+      "step": 1714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022506183017312445,
+      "loss": 4.7588,
+      "step": 1715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000225,
+      "loss": 5.4298,
+      "step": 1716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002249381698268755,
+      "loss": 5.3282,
+      "step": 1717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022487633965375103,
+      "loss": 5.7339,
+      "step": 1718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022481450948062652,
+      "loss": 5.0199,
+      "step": 1719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022475267930750203,
+      "loss": 4.7039,
+      "step": 1720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022469084913437755,
+      "loss": 5.0596,
+      "step": 1721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002246290189612531,
+      "loss": 5.1479,
+      "step": 1722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022456718878812858,
+      "loss": 5.2915,
+      "step": 1723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002245053586150041,
+      "loss": 4.6768,
+      "step": 1724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022444352844187962,
+      "loss": 4.9414,
+      "step": 1725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022438169826875513,
+      "loss": 4.8027,
+      "step": 1726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022431986809563065,
+      "loss": 5.0499,
+      "step": 1727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022425803792250614,
+      "loss": 4.8452,
+      "step": 1728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022419620774938168,
+      "loss": 5.1527,
+      "step": 1729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002241343775762572,
+      "loss": 5.0717,
+      "step": 1730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022407254740313272,
+      "loss": 4.323,
+      "step": 1731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022401071723000823,
+      "loss": 4.5814,
+      "step": 1732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022394888705688372,
+      "loss": 4.684,
+      "step": 1733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022388705688375927,
+      "loss": 4.8464,
+      "step": 1734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022382522671063478,
+      "loss": 5.2516,
+      "step": 1735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002237633965375103,
+      "loss": 6.3892,
+      "step": 1736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002237015663643858,
+      "loss": 5.2744,
+      "step": 1737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002236397361912613,
+      "loss": 4.6835,
+      "step": 1738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022357790601813682,
+      "loss": 4.7632,
+      "step": 1739
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022351607584501236,
+      "loss": 4.683,
+      "step": 1740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022345424567188788,
+      "loss": 5.1256,
+      "step": 1741
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022339241549876337,
+      "loss": 4.7506,
+      "step": 1742
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002233305853256389,
+      "loss": 4.7853,
+      "step": 1743
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002232687551525144,
+      "loss": 5.4279,
+      "step": 1744
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022320692497938992,
+      "loss": 4.9857,
+      "step": 1745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002231450948062654,
+      "loss": 4.8214,
+      "step": 1746
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022308326463314095,
+      "loss": 4.8111,
+      "step": 1747
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022302143446001647,
+      "loss": 4.5853,
+      "step": 1748
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000222959604286892,
+      "loss": 4.8762,
+      "step": 1749
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002228977741137675,
+      "loss": 5.439,
+      "step": 1750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000222835943940643,
+      "loss": 4.5937,
+      "step": 1751
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022277411376751854,
+      "loss": 4.8131,
+      "step": 1752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022271228359439405,
+      "loss": 4.9153,
+      "step": 1753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022265045342126957,
+      "loss": 5.4705,
+      "step": 1754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022258862324814506,
+      "loss": 5.0429,
+      "step": 1755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022252679307502058,
+      "loss": 4.7381,
+      "step": 1756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002224649629018961,
+      "loss": 4.778,
+      "step": 1757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022240313272877164,
+      "loss": 4.4948,
+      "step": 1758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022234130255564715,
+      "loss": 5.5305,
+      "step": 1759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022227947238252264,
+      "loss": 4.8582,
+      "step": 1760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022221764220939816,
+      "loss": 5.2286,
+      "step": 1761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022215581203627368,
+      "loss": 4.6681,
+      "step": 1762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022209398186314922,
+      "loss": 4.8589,
+      "step": 1763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022203215169002474,
+      "loss": 4.8424,
+      "step": 1764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022197032151690023,
+      "loss": 4.6714,
+      "step": 1765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022190849134377574,
+      "loss": 4.684,
+      "step": 1766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022184666117065126,
+      "loss": 4.9776,
+      "step": 1767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022178483099752678,
+      "loss": 4.8745,
+      "step": 1768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022172300082440226,
+      "loss": 4.6617,
+      "step": 1769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002216611706512778,
+      "loss": 4.6106,
+      "step": 1770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022159934047815332,
+      "loss": 4.8001,
+      "step": 1771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022153751030502884,
+      "loss": 4.5968,
+      "step": 1772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022147568013190436,
+      "loss": 4.7554,
+      "step": 1773
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022141384995877985,
+      "loss": 4.7098,
+      "step": 1774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022135201978565536,
+      "loss": 4.5689,
+      "step": 1775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002212901896125309,
+      "loss": 4.6485,
+      "step": 1776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022122835943940642,
+      "loss": 5.3558,
+      "step": 1777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022116652926628191,
+      "loss": 4.6398,
+      "step": 1778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022110469909315743,
+      "loss": 6.0466,
+      "step": 1779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022104286892003295,
+      "loss": 4.8022,
+      "step": 1780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002209810387469085,
+      "loss": 5.484,
+      "step": 1781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000220919208573784,
+      "loss": 4.9977,
+      "step": 1782
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002208573784006595,
+      "loss": 5.7127,
+      "step": 1783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000220795548227535,
+      "loss": 4.1483,
+      "step": 1784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022073371805441053,
+      "loss": 4.9275,
+      "step": 1785
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022067188788128605,
+      "loss": 4.783,
+      "step": 1786
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022061005770816154,
+      "loss": 4.2222,
+      "step": 1787
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022054822753503708,
+      "loss": 4.1721,
+      "step": 1788
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002204863973619126,
+      "loss": 4.708,
+      "step": 1789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002204245671887881,
+      "loss": 4.9501,
+      "step": 1790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022036273701566363,
+      "loss": 4.231,
+      "step": 1791
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022030090684253912,
+      "loss": 4.2647,
+      "step": 1792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00022023907666941466,
+      "loss": 4.8888,
+      "step": 1793
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00022017724649629018,
+      "loss": 4.239,
+      "step": 1794
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002201154163231657,
+      "loss": 4.6312,
+      "step": 1795
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002200535861500412,
+      "loss": 4.5236,
+      "step": 1796
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002199917559769167,
+      "loss": 4.0982,
+      "step": 1797
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00021992992580379222,
+      "loss": 4.5575,
+      "step": 1798
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00021986809563066776,
+      "loss": 4.2895,
+      "step": 1799
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00021980626545754328,
+      "loss": 4.4954,
+      "step": 1800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00021974443528441877,
+      "loss": 4.3467,
+      "step": 1801
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00021968260511129428,
+      "loss": 4.0511,
+      "step": 1802
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002196207749381698,
+      "loss": 4.3306,
+      "step": 1803
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00021955894476504532,
+      "loss": 4.8846,
+      "step": 1804
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00021949711459192086,
+      "loss": 4.0288,
+      "step": 1805
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00021943528441879635,
+      "loss": 4.6451,
+      "step": 1806
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00021937345424567187,
+      "loss": 3.9867,
+      "step": 1807
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00021931162407254738,
+      "loss": 4.2205,
+      "step": 1808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002192497938994229,
+      "loss": 4.1628,
+      "step": 1809
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002191879637262984,
+      "loss": 4.1805,
+      "step": 1810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00021912613355317393,
+      "loss": 4.3665,
+      "step": 1811
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00021906430338004945,
+      "loss": 4.5182,
+      "step": 1812
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00021900247320692497,
+      "loss": 4.3839,
+      "step": 1813
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00021894064303380048,
+      "loss": 3.8271,
+      "step": 1814
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00021887881286067597,
+      "loss": 4.3602,
+      "step": 1815
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002188169826875515,
+      "loss": 4.7941,
+      "step": 1816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00021875515251442703,
+      "loss": 4.6545,
+      "step": 1817
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00021869332234130255,
+      "loss": 4.0513,
+      "step": 1818
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00021863149216817804,
+      "loss": 4.5893,
+      "step": 1819
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00021856966199505356,
+      "loss": 4.3379,
+      "step": 1820
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00021850783182192907,
+      "loss": 4.6565,
+      "step": 1821
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002184460016488046,
+      "loss": 4.2378,
+      "step": 1822
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00021838417147568013,
+      "loss": 4.5566,
+      "step": 1823
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00021832234130255562,
+      "loss": 4.9048,
+      "step": 1824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00021826051112943114,
+      "loss": 4.1219,
+      "step": 1825
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00021819868095630666,
+      "loss": 4.9947,
+      "step": 1826
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00021813685078318217,
+      "loss": 4.0081,
+      "step": 1827
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00021807502061005772,
+      "loss": 4.614,
+      "step": 1828
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002180131904369332,
+      "loss": 4.6027,
+      "step": 1829
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00021795136026380872,
+      "loss": 5.2473,
+      "step": 1830
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00021788953009068424,
+      "loss": 3.6428,
+      "step": 1831
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00021782769991755976,
+      "loss": 4.9894,
+      "step": 1832
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00021776586974443524,
+      "loss": 6.4707,
+      "step": 1833
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00021770403957131076,
+      "loss": 4.2968,
+      "step": 1834
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002176422093981863,
+      "loss": 4.5087,
+      "step": 1835
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00021758037922506182,
+      "loss": 4.7255,
+      "step": 1836
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00021751854905193734,
+      "loss": 4.3344,
+      "step": 1837
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00021745671887881283,
+      "loss": 4.4766,
+      "step": 1838
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00021739488870568834,
+      "loss": 4.0547,
+      "step": 1839
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002173330585325639,
+      "loss": 4.4878,
+      "step": 1840
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002172712283594394,
+      "loss": 4.4728,
+      "step": 1841
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002172093981863149,
+      "loss": 4.2672,
+      "step": 1842
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002171475680131904,
+      "loss": 4.558,
+      "step": 1843
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00021708573784006593,
+      "loss": 4.9313,
+      "step": 1844
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00021702390766694144,
+      "loss": 4.5943,
+      "step": 1845
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000216962077493817,
+      "loss": 4.1062,
+      "step": 1846
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021690024732069248,
+      "loss": 4.6563,
+      "step": 1847
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000216838417147568,
+      "loss": 4.3016,
+      "step": 1848
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002167765869744435,
+      "loss": 4.3437,
+      "step": 1849
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021671475680131903,
+      "loss": 4.3573,
+      "step": 1850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021665292662819454,
+      "loss": 4.1072,
+      "step": 1851
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021659109645507003,
+      "loss": 4.7962,
+      "step": 1852
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021652926628194558,
+      "loss": 4.1764,
+      "step": 1853
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002164674361088211,
+      "loss": 3.9577,
+      "step": 1854
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002164056059356966,
+      "loss": 4.2439,
+      "step": 1855
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002163437757625721,
+      "loss": 4.2247,
+      "step": 1856
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021628194558944762,
+      "loss": 4.2786,
+      "step": 1857
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021622011541632316,
+      "loss": 4.2887,
+      "step": 1858
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021615828524319868,
+      "loss": 4.4225,
+      "step": 1859
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002160964550700742,
+      "loss": 4.202,
+      "step": 1860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021603462489694968,
+      "loss": 3.9829,
+      "step": 1861
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002159727947238252,
+      "loss": 4.2176,
+      "step": 1862
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021591096455070072,
+      "loss": 4.3083,
+      "step": 1863
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021584913437757626,
+      "loss": 4.3716,
+      "step": 1864
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021578730420445175,
+      "loss": 4.4742,
+      "step": 1865
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021572547403132726,
+      "loss": 3.8497,
+      "step": 1866
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021566364385820278,
+      "loss": 4.2625,
+      "step": 1867
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002156018136850783,
+      "loss": 4.1776,
+      "step": 1868
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021553998351195384,
+      "loss": 4.0808,
+      "step": 1869
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021547815333882933,
+      "loss": 4.664,
+      "step": 1870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021541632316570485,
+      "loss": 4.1419,
+      "step": 1871
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021535449299258036,
+      "loss": 4.4966,
+      "step": 1872
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021529266281945588,
+      "loss": 3.9461,
+      "step": 1873
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021523083264633137,
+      "loss": 4.1367,
+      "step": 1874
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002151690024732069,
+      "loss": 4.1444,
+      "step": 1875
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021510717230008243,
+      "loss": 4.2955,
+      "step": 1876
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021504534212695795,
+      "loss": 4.4176,
+      "step": 1877
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021498351195383346,
+      "loss": 4.0552,
+      "step": 1878
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021492168178070895,
+      "loss": 4.3888,
+      "step": 1879
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021485985160758447,
+      "loss": 3.7147,
+      "step": 1880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021479802143446,
+      "loss": 3.945,
+      "step": 1881
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021473619126133553,
+      "loss": 5.1081,
+      "step": 1882
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021467436108821105,
+      "loss": 3.7242,
+      "step": 1883
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021461253091508654,
+      "loss": 4.9485,
+      "step": 1884
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021455070074196205,
+      "loss": 4.1513,
+      "step": 1885
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021448887056883757,
+      "loss": 4.262,
+      "step": 1886
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002144270403957131,
+      "loss": 4.8252,
+      "step": 1887
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002143652102225886,
+      "loss": 4.3994,
+      "step": 1888
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021430338004946412,
+      "loss": 4.123,
+      "step": 1889
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021424154987633964,
+      "loss": 4.7568,
+      "step": 1890
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021417971970321515,
+      "loss": 4.2343,
+      "step": 1891
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021411788953009067,
+      "loss": 4.3639,
+      "step": 1892
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021405605935696616,
+      "loss": 4.2373,
+      "step": 1893
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002139942291838417,
+      "loss": 4.1435,
+      "step": 1894
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021393239901071722,
+      "loss": 4.5327,
+      "step": 1895
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021387056883759274,
+      "loss": 4.2353,
+      "step": 1896
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021380873866446822,
+      "loss": 4.2919,
+      "step": 1897
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021374690849134374,
+      "loss": 4.4488,
+      "step": 1898
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021368507831821926,
+      "loss": 4.2019,
+      "step": 1899
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002136232481450948,
+      "loss": 4.0895,
+      "step": 1900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021356141797197032,
+      "loss": 4.4734,
+      "step": 1901
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002134995877988458,
+      "loss": 4.2665,
+      "step": 1902
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021343775762572132,
+      "loss": 4.5362,
+      "step": 1903
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021337592745259684,
+      "loss": 4.555,
+      "step": 1904
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021331409727947238,
+      "loss": 4.4678,
+      "step": 1905
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021325226710634787,
+      "loss": 4.7331,
+      "step": 1906
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002131904369332234,
+      "loss": 4.7416,
+      "step": 1907
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002131286067600989,
+      "loss": 4.3882,
+      "step": 1908
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021306677658697442,
+      "loss": 4.302,
+      "step": 1909
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021300494641384994,
+      "loss": 4.7488,
+      "step": 1910
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021294311624072543,
+      "loss": 4.7296,
+      "step": 1911
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021288128606760097,
+      "loss": 4.2448,
+      "step": 1912
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002128194558944765,
+      "loss": 4.5813,
+      "step": 1913
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000212757625721352,
+      "loss": 4.8866,
+      "step": 1914
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021269579554822752,
+      "loss": 4.528,
+      "step": 1915
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000212633965375103,
+      "loss": 4.2495,
+      "step": 1916
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021257213520197856,
+      "loss": 4.2151,
+      "step": 1917
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021251030502885407,
+      "loss": 4.4869,
+      "step": 1918
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002124484748557296,
+      "loss": 4.2424,
+      "step": 1919
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021238664468260508,
+      "loss": 3.9412,
+      "step": 1920
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002123248145094806,
+      "loss": 4.0781,
+      "step": 1921
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002122629843363561,
+      "loss": 5.0096,
+      "step": 1922
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021220115416323166,
+      "loss": 4.5541,
+      "step": 1923
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021213932399010717,
+      "loss": 4.224,
+      "step": 1924
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021207749381698266,
+      "loss": 4.5065,
+      "step": 1925
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021201566364385818,
+      "loss": 4.2974,
+      "step": 1926
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002119538334707337,
+      "loss": 4.2464,
+      "step": 1927
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002118920032976092,
+      "loss": 4.4322,
+      "step": 1928
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002118301731244847,
+      "loss": 3.9618,
+      "step": 1929
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021176834295136025,
+      "loss": 3.8476,
+      "step": 1930
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021170651277823576,
+      "loss": 3.7236,
+      "step": 1931
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021164468260511128,
+      "loss": 5.355,
+      "step": 1932
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002115828524319868,
+      "loss": 3.9975,
+      "step": 1933
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021152102225886228,
+      "loss": 5.1416,
+      "step": 1934
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021145919208573783,
+      "loss": 4.4639,
+      "step": 1935
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021139736191261334,
+      "loss": 4.3295,
+      "step": 1936
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021133553173948886,
+      "loss": 4.3668,
+      "step": 1937
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021127370156636438,
+      "loss": 5.3182,
+      "step": 1938
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021121187139323987,
+      "loss": 4.3346,
+      "step": 1939
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021115004122011538,
+      "loss": 4.7043,
+      "step": 1940
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021108821104699093,
+      "loss": 4.0101,
+      "step": 1941
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021102638087386644,
+      "loss": 4.465,
+      "step": 1942
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021096455070074193,
+      "loss": 4.6536,
+      "step": 1943
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021090272052761745,
+      "loss": 4.5671,
+      "step": 1944
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021084089035449297,
+      "loss": 3.9908,
+      "step": 1945
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002107790601813685,
+      "loss": 4.1181,
+      "step": 1946
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021071723000824403,
+      "loss": 4.812,
+      "step": 1947
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021065539983511952,
+      "loss": 4.3629,
+      "step": 1948
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021059356966199503,
+      "loss": 4.2942,
+      "step": 1949
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021053173948887055,
+      "loss": 4.2031,
+      "step": 1950
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021046990931574607,
+      "loss": 4.855,
+      "step": 1951
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021040807914262156,
+      "loss": 3.8943,
+      "step": 1952
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002103462489694971,
+      "loss": 4.9096,
+      "step": 1953
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00021028441879637262,
+      "loss": 4.2225,
+      "step": 1954
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00021022258862324813,
+      "loss": 4.478,
+      "step": 1955
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00021016075845012365,
+      "loss": 4.3215,
+      "step": 1956
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00021009892827699914,
+      "loss": 4.2379,
+      "step": 1957
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00021003709810387466,
+      "loss": 4.4387,
+      "step": 1958
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002099752679307502,
+      "loss": 4.4451,
+      "step": 1959
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00020991343775762572,
+      "loss": 4.0919,
+      "step": 1960
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002098516075845012,
+      "loss": 4.1237,
+      "step": 1961
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00020978977741137672,
+      "loss": 3.9294,
+      "step": 1962
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00020972794723825224,
+      "loss": 4.3057,
+      "step": 1963
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00020966611706512778,
+      "loss": 4.7266,
+      "step": 1964
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002096042868920033,
+      "loss": 5.0877,
+      "step": 1965
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002095424567188788,
+      "loss": 4.0448,
+      "step": 1966
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002094806265457543,
+      "loss": 4.4972,
+      "step": 1967
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00020941879637262982,
+      "loss": 3.9749,
+      "step": 1968
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00020935696619950534,
+      "loss": 4.395,
+      "step": 1969
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00020929513602638088,
+      "loss": 4.2235,
+      "step": 1970
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00020923330585325637,
+      "loss": 4.8132,
+      "step": 1971
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002091714756801319,
+      "loss": 4.1994,
+      "step": 1972
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002091096455070074,
+      "loss": 4.3047,
+      "step": 1973
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00020904781533388292,
+      "loss": 4.5337,
+      "step": 1974
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002089859851607584,
+      "loss": 4.482,
+      "step": 1975
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00020892415498763395,
+      "loss": 4.0722,
+      "step": 1976
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00020886232481450947,
+      "loss": 4.4456,
+      "step": 1977
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000208800494641385,
+      "loss": 3.9115,
+      "step": 1978
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002087386644682605,
+      "loss": 3.6723,
+      "step": 1979
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000208676834295136,
+      "loss": 4.7251,
+      "step": 1980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002086150041220115,
+      "loss": 4.4887,
+      "step": 1981
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00020855317394888705,
+      "loss": 4.1576,
+      "step": 1982
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00020849134377576257,
+      "loss": 4.2002,
+      "step": 1983
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00020842951360263806,
+      "loss": 3.6878,
+      "step": 1984
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00020836768342951358,
+      "loss": 4.4889,
+      "step": 1985
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002083058532563891,
+      "loss": 4.2417,
+      "step": 1986
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002082440230832646,
+      "loss": 4.2297,
+      "step": 1987
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00020818219291014015,
+      "loss": 4.7065,
+      "step": 1988
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00020812036273701564,
+      "loss": 4.5499,
+      "step": 1989
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020805853256389116,
+      "loss": 4.7985,
+      "step": 1990
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020799670239076668,
+      "loss": 4.4748,
+      "step": 1991
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002079348722176422,
+      "loss": 4.1624,
+      "step": 1992
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020787304204451768,
+      "loss": 4.1525,
+      "step": 1993
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020781121187139323,
+      "loss": 4.3708,
+      "step": 1994
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020774938169826874,
+      "loss": 4.3425,
+      "step": 1995
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020768755152514426,
+      "loss": 4.8552,
+      "step": 1996
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020762572135201978,
+      "loss": 4.4022,
+      "step": 1997
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020756389117889526,
+      "loss": 4.4513,
+      "step": 1998
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020750206100577078,
+      "loss": 4.3355,
+      "step": 1999
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020744023083264632,
+      "loss": 4.1926,
+      "step": 2000
+    },
+    {
+      "epoch": 1.12,
+      "eval_loss": 5.016541004180908,
+      "eval_runtime": 964.9547,
+      "eval_samples_per_second": 2.738,
+      "eval_steps_per_second": 0.343,
+      "eval_wer": 1.9480563268544229,
+      "step": 2000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020737840065952184,
+      "loss": 4.5581,
+      "step": 2001
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020731657048639736,
+      "loss": 4.048,
+      "step": 2002
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020725474031327285,
+      "loss": 4.2169,
+      "step": 2003
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020719291014014836,
+      "loss": 4.3366,
+      "step": 2004
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020713107996702388,
+      "loss": 4.144,
+      "step": 2005
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020706924979389942,
+      "loss": 4.1249,
+      "step": 2006
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020700741962077491,
+      "loss": 4.1117,
+      "step": 2007
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020694558944765043,
+      "loss": 4.2399,
+      "step": 2008
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020688375927452595,
+      "loss": 4.2206,
+      "step": 2009
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020682192910140146,
+      "loss": 4.0255,
+      "step": 2010
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000206760098928277,
+      "loss": 3.9022,
+      "step": 2011
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002066982687551525,
+      "loss": 4.2656,
+      "step": 2012
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000206636438582028,
+      "loss": 5.009,
+      "step": 2013
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020657460840890353,
+      "loss": 4.9038,
+      "step": 2014
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020651277823577905,
+      "loss": 4.281,
+      "step": 2015
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020645094806265454,
+      "loss": 4.4258,
+      "step": 2016
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020638911788953005,
+      "loss": 4.5296,
+      "step": 2017
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002063272877164056,
+      "loss": 4.3366,
+      "step": 2018
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002062654575432811,
+      "loss": 4.2908,
+      "step": 2019
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020620362737015663,
+      "loss": 4.2209,
+      "step": 2020
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020614179719703212,
+      "loss": 4.6234,
+      "step": 2021
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020607996702390764,
+      "loss": 4.0819,
+      "step": 2022
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020601813685078318,
+      "loss": 4.0341,
+      "step": 2023
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002059563066776587,
+      "loss": 4.275,
+      "step": 2024
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020589447650453419,
+      "loss": 4.5047,
+      "step": 2025
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002058326463314097,
+      "loss": 4.3766,
+      "step": 2026
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020577081615828522,
+      "loss": 5.0109,
+      "step": 2027
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020570898598516074,
+      "loss": 4.0102,
+      "step": 2028
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020564715581203628,
+      "loss": 4.52,
+      "step": 2029
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020558532563891177,
+      "loss": 4.5957,
+      "step": 2030
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020552349546578728,
+      "loss": 4.9114,
+      "step": 2031
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002054616652926628,
+      "loss": 3.8494,
+      "step": 2032
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020539983511953832,
+      "loss": 4.4098,
+      "step": 2033
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020533800494641383,
+      "loss": 4.3071,
+      "step": 2034
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020527617477328932,
+      "loss": 4.2923,
+      "step": 2035
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020521434460016487,
+      "loss": 4.2879,
+      "step": 2036
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020515251442704038,
+      "loss": 4.5749,
+      "step": 2037
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002050906842539159,
+      "loss": 4.2714,
+      "step": 2038
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002050288540807914,
+      "loss": 4.4426,
+      "step": 2039
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002049670239076669,
+      "loss": 5.2494,
+      "step": 2040
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020490519373454245,
+      "loss": 4.0955,
+      "step": 2041
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020484336356141797,
+      "loss": 4.6205,
+      "step": 2042
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020478153338829348,
+      "loss": 4.2157,
+      "step": 2043
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020471970321516897,
+      "loss": 4.1929,
+      "step": 2044
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002046578730420445,
+      "loss": 4.8054,
+      "step": 2045
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020459604286892,
+      "loss": 4.4631,
+      "step": 2046
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020453421269579555,
+      "loss": 4.0652,
+      "step": 2047
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020447238252267104,
+      "loss": 5.0021,
+      "step": 2048
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020441055234954656,
+      "loss": 4.6101,
+      "step": 2049
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020434872217642207,
+      "loss": 4.7676,
+      "step": 2050
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002042868920032976,
+      "loss": 4.0998,
+      "step": 2051
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020422506183017313,
+      "loss": 4.2169,
+      "step": 2052
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020416323165704862,
+      "loss": 4.317,
+      "step": 2053
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020410140148392414,
+      "loss": 4.0591,
+      "step": 2054
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020403957131079966,
+      "loss": 4.353,
+      "step": 2055
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020397774113767517,
+      "loss": 4.0603,
+      "step": 2056
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002039159109645507,
+      "loss": 4.1688,
+      "step": 2057
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020385408079142618,
+      "loss": 4.0526,
+      "step": 2058
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020379225061830172,
+      "loss": 4.4744,
+      "step": 2059
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020373042044517724,
+      "loss": 4.2508,
+      "step": 2060
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020366859027205276,
+      "loss": 5.5821,
+      "step": 2061
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020360676009892824,
+      "loss": 4.6607,
+      "step": 2062
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020354492992580376,
+      "loss": 4.2335,
+      "step": 2063
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020348309975267928,
+      "loss": 4.0746,
+      "step": 2064
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020342126957955482,
+      "loss": 4.3043,
+      "step": 2065
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020335943940643034,
+      "loss": 4.5497,
+      "step": 2066
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020329760923330583,
+      "loss": 3.9801,
+      "step": 2067
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020323577906018134,
+      "loss": 4.3226,
+      "step": 2068
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020317394888705686,
+      "loss": 4.4148,
+      "step": 2069
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002031121187139324,
+      "loss": 4.4479,
+      "step": 2070
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002030502885408079,
+      "loss": 4.1789,
+      "step": 2071
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002029884583676834,
+      "loss": 4.3936,
+      "step": 2072
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020292662819455893,
+      "loss": 4.1474,
+      "step": 2073
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020286479802143444,
+      "loss": 4.1087,
+      "step": 2074
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020280296784830996,
+      "loss": 4.2359,
+      "step": 2075
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020274113767518545,
+      "loss": 4.2856,
+      "step": 2076
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000202679307502061,
+      "loss": 4.7534,
+      "step": 2077
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002026174773289365,
+      "loss": 4.5953,
+      "step": 2078
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020255564715581203,
+      "loss": 3.7334,
+      "step": 2079
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020249381698268752,
+      "loss": 5.126,
+      "step": 2080
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020243198680956303,
+      "loss": 4.5141,
+      "step": 2081
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020237015663643855,
+      "loss": 4.2326,
+      "step": 2082
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002023083264633141,
+      "loss": 6.0161,
+      "step": 2083
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002022464962901896,
+      "loss": 6.3244,
+      "step": 2084
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002021846661170651,
+      "loss": 4.4372,
+      "step": 2085
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020212283594394062,
+      "loss": 3.9703,
+      "step": 2086
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020206100577081613,
+      "loss": 4.3571,
+      "step": 2087
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020199917559769168,
+      "loss": 4.2938,
+      "step": 2088
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002019373454245672,
+      "loss": 4.1363,
+      "step": 2089
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020187551525144268,
+      "loss": 4.3989,
+      "step": 2090
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002018136850783182,
+      "loss": 4.2811,
+      "step": 2091
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020175185490519372,
+      "loss": 4.2564,
+      "step": 2092
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020169002473206923,
+      "loss": 4.3685,
+      "step": 2093
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020162819455894472,
+      "loss": 4.2374,
+      "step": 2094
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020156636438582027,
+      "loss": 4.4376,
+      "step": 2095
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020150453421269578,
+      "loss": 4.5132,
+      "step": 2096
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002014427040395713,
+      "loss": 4.4888,
+      "step": 2097
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020138087386644681,
+      "loss": 4.6262,
+      "step": 2098
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002013190436933223,
+      "loss": 5.0821,
+      "step": 2099
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020125721352019785,
+      "loss": 4.0241,
+      "step": 2100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020119538334707336,
+      "loss": 4.3985,
+      "step": 2101
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020113355317394888,
+      "loss": 4.1768,
+      "step": 2102
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020107172300082437,
+      "loss": 4.1513,
+      "step": 2103
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002010098928276999,
+      "loss": 4.8351,
+      "step": 2104
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002009480626545754,
+      "loss": 4.2565,
+      "step": 2105
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020088623248145095,
+      "loss": 4.4269,
+      "step": 2106
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020082440230832646,
+      "loss": 4.3464,
+      "step": 2107
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020076257213520195,
+      "loss": 4.6812,
+      "step": 2108
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020070074196207747,
+      "loss": 4.6333,
+      "step": 2109
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000200638911788953,
+      "loss": 4.3861,
+      "step": 2110
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002005770816158285,
+      "loss": 4.4086,
+      "step": 2111
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000200515251442704,
+      "loss": 4.6948,
+      "step": 2112
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020045342126957954,
+      "loss": 4.4821,
+      "step": 2113
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020039159109645505,
+      "loss": 4.2755,
+      "step": 2114
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00020032976092333057,
+      "loss": 4.3998,
+      "step": 2115
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00020026793075020609,
+      "loss": 4.1146,
+      "step": 2116
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00020020610057708158,
+      "loss": 3.8942,
+      "step": 2117
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00020014427040395712,
+      "loss": 4.2122,
+      "step": 2118
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00020008244023083264,
+      "loss": 4.0814,
+      "step": 2119
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00020002061005770815,
+      "loss": 4.3404,
+      "step": 2120
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00019995877988458367,
+      "loss": 4.4704,
+      "step": 2121
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00019989694971145916,
+      "loss": 4.3054,
+      "step": 2122
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00019983511953833468,
+      "loss": 4.2301,
+      "step": 2123
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00019977328936521022,
+      "loss": 4.104,
+      "step": 2124
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00019971145919208574,
+      "loss": 5.5095,
+      "step": 2125
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00019964962901896123,
+      "loss": 4.3428,
+      "step": 2126
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00019958779884583674,
+      "loss": 3.9532,
+      "step": 2127
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00019952596867271226,
+      "loss": 4.5419,
+      "step": 2128
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001994641384995878,
+      "loss": 4.1444,
+      "step": 2129
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00019940230832646332,
+      "loss": 4.6509,
+      "step": 2130
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001993404781533388,
+      "loss": 4.0253,
+      "step": 2131
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00019927864798021432,
+      "loss": 4.2822,
+      "step": 2132
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00019921681780708984,
+      "loss": 4.9368,
+      "step": 2133
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00019915498763396536,
+      "loss": 4.1643,
+      "step": 2134
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00019909315746084085,
+      "loss": 4.5168,
+      "step": 2135
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001990313272877164,
+      "loss": 4.5065,
+      "step": 2136
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001989694971145919,
+      "loss": 4.2376,
+      "step": 2137
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00019890766694146742,
+      "loss": 4.0584,
+      "step": 2138
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00019884583676834294,
+      "loss": 4.1995,
+      "step": 2139
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00019878400659521843,
+      "loss": 4.6928,
+      "step": 2140
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00019872217642209395,
+      "loss": 4.6509,
+      "step": 2141
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001986603462489695,
+      "loss": 4.1322,
+      "step": 2142
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000198598516075845,
+      "loss": 4.245,
+      "step": 2143
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001985366859027205,
+      "loss": 4.4636,
+      "step": 2144
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000198474855729596,
+      "loss": 4.6273,
+      "step": 2145
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00019841302555647153,
+      "loss": 4.6772,
+      "step": 2146
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00019835119538334707,
+      "loss": 4.2381,
+      "step": 2147
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001982893652102226,
+      "loss": 4.2548,
+      "step": 2148
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00019822753503709808,
+      "loss": 4.1426,
+      "step": 2149
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001981657048639736,
+      "loss": 4.5471,
+      "step": 2150
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001981038746908491,
+      "loss": 4.5037,
+      "step": 2151
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019804204451772463,
+      "loss": 4.8702,
+      "step": 2152
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019798021434460017,
+      "loss": 4.4314,
+      "step": 2153
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019791838417147566,
+      "loss": 4.614,
+      "step": 2154
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019785655399835118,
+      "loss": 4.4746,
+      "step": 2155
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001977947238252267,
+      "loss": 4.2257,
+      "step": 2156
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001977328936521022,
+      "loss": 4.6069,
+      "step": 2157
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001976710634789777,
+      "loss": 4.3038,
+      "step": 2158
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019760923330585325,
+      "loss": 4.3436,
+      "step": 2159
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019754740313272876,
+      "loss": 4.517,
+      "step": 2160
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019748557295960428,
+      "loss": 4.2689,
+      "step": 2161
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001974237427864798,
+      "loss": 4.077,
+      "step": 2162
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019736191261335528,
+      "loss": 4.4126,
+      "step": 2163
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001973000824402308,
+      "loss": 4.2489,
+      "step": 2164
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019723825226710634,
+      "loss": 4.6906,
+      "step": 2165
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019717642209398186,
+      "loss": 4.3048,
+      "step": 2166
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019711459192085735,
+      "loss": 4.4959,
+      "step": 2167
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019705276174773287,
+      "loss": 4.1282,
+      "step": 2168
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019699093157460838,
+      "loss": 4.4104,
+      "step": 2169
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001969291014014839,
+      "loss": 4.4102,
+      "step": 2170
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019686727122835944,
+      "loss": 4.0834,
+      "step": 2171
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019680544105523493,
+      "loss": 4.7637,
+      "step": 2172
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019674361088211045,
+      "loss": 4.2248,
+      "step": 2173
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019668178070898597,
+      "loss": 4.0649,
+      "step": 2174
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019661995053586148,
+      "loss": 4.1947,
+      "step": 2175
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019655812036273703,
+      "loss": 3.949,
+      "step": 2176
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019649629018961252,
+      "loss": 3.751,
+      "step": 2177
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019643446001648803,
+      "loss": 4.6825,
+      "step": 2178
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019637262984336355,
+      "loss": 3.6498,
+      "step": 2179
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019631079967023907,
+      "loss": 4.3444,
+      "step": 2180
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019624896949711456,
+      "loss": 3.7872,
+      "step": 2181
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019618713932399007,
+      "loss": 4.2444,
+      "step": 2182
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019612530915086562,
+      "loss": 6.8173,
+      "step": 2183
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019606347897774113,
+      "loss": 3.8492,
+      "step": 2184
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019600164880461665,
+      "loss": 4.5002,
+      "step": 2185
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019593981863149214,
+      "loss": 4.4665,
+      "step": 2186
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019587798845836766,
+      "loss": 4.1492,
+      "step": 2187
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019581615828524317,
+      "loss": 4.4674,
+      "step": 2188
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019575432811211872,
+      "loss": 4.5616,
+      "step": 2189
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001956924979389942,
+      "loss": 4.2153,
+      "step": 2190
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019563066776586972,
+      "loss": 4.2296,
+      "step": 2191
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019556883759274524,
+      "loss": 4.5537,
+      "step": 2192
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019550700741962076,
+      "loss": 5.3998,
+      "step": 2193
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001954451772464963,
+      "loss": 4.4875,
+      "step": 2194
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001953833470733718,
+      "loss": 4.1127,
+      "step": 2195
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001953215169002473,
+      "loss": 4.411,
+      "step": 2196
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019525968672712282,
+      "loss": 4.4252,
+      "step": 2197
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019519785655399834,
+      "loss": 4.7354,
+      "step": 2198
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019513602638087383,
+      "loss": 4.1897,
+      "step": 2199
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019507419620774934,
+      "loss": 4.7718,
+      "step": 2200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001950123660346249,
+      "loss": 4.4185,
+      "step": 2201
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001949505358615004,
+      "loss": 4.4487,
+      "step": 2202
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019488870568837592,
+      "loss": 4.1434,
+      "step": 2203
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001948268755152514,
+      "loss": 4.401,
+      "step": 2204
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019476504534212693,
+      "loss": 4.4078,
+      "step": 2205
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019470321516900247,
+      "loss": 4.3002,
+      "step": 2206
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000194641384995878,
+      "loss": 4.6281,
+      "step": 2207
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001945795548227535,
+      "loss": 4.1061,
+      "step": 2208
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000194517724649629,
+      "loss": 4.2288,
+      "step": 2209
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001944558944765045,
+      "loss": 4.4523,
+      "step": 2210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019439406430338003,
+      "loss": 4.2977,
+      "step": 2211
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019433223413025557,
+      "loss": 4.0619,
+      "step": 2212
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019427040395713106,
+      "loss": 4.6526,
+      "step": 2213
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019420857378400658,
+      "loss": 4.3712,
+      "step": 2214
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001941467436108821,
+      "loss": 4.1264,
+      "step": 2215
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001940849134377576,
+      "loss": 4.2969,
+      "step": 2216
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019402308326463313,
+      "loss": 4.655,
+      "step": 2217
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019396125309150862,
+      "loss": 4.536,
+      "step": 2218
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019389942291838416,
+      "loss": 4.3913,
+      "step": 2219
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019383759274525968,
+      "loss": 4.3478,
+      "step": 2220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001937757625721352,
+      "loss": 4.5299,
+      "step": 2221
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019371393239901068,
+      "loss": 4.0866,
+      "step": 2222
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001936521022258862,
+      "loss": 4.0634,
+      "step": 2223
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019359027205276174,
+      "loss": 4.5467,
+      "step": 2224
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019352844187963726,
+      "loss": 4.4253,
+      "step": 2225
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019346661170651278,
+      "loss": 4.2333,
+      "step": 2226
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019340478153338826,
+      "loss": 4.1808,
+      "step": 2227
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019334295136026378,
+      "loss": 4.3517,
+      "step": 2228
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001932811211871393,
+      "loss": 4.3603,
+      "step": 2229
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019321929101401484,
+      "loss": 5.0476,
+      "step": 2230
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019315746084089033,
+      "loss": 4.0974,
+      "step": 2231
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019309563066776585,
+      "loss": 4.0172,
+      "step": 2232
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019303380049464136,
+      "loss": 4.6374,
+      "step": 2233
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019297197032151688,
+      "loss": 4.1264,
+      "step": 2234
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019291014014839242,
+      "loss": 4.6354,
+      "step": 2235
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019284830997526791,
+      "loss": 4.0226,
+      "step": 2236
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019278647980214343,
+      "loss": 4.0391,
+      "step": 2237
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019272464962901895,
+      "loss": 4.4654,
+      "step": 2238
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019266281945589446,
+      "loss": 4.315,
+      "step": 2239
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019260098928276998,
+      "loss": 5.3824,
+      "step": 2240
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019253915910964547,
+      "loss": 4.4389,
+      "step": 2241
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019247732893652101,
+      "loss": 4.9573,
+      "step": 2242
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019241549876339653,
+      "loss": 4.2102,
+      "step": 2243
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019235366859027205,
+      "loss": 4.1575,
+      "step": 2244
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019229183841714754,
+      "loss": 4.1612,
+      "step": 2245
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019223000824402305,
+      "loss": 4.1298,
+      "step": 2246
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019216817807089857,
+      "loss": 4.0527,
+      "step": 2247
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001921063478977741,
+      "loss": 4.5568,
+      "step": 2248
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019204451772464963,
+      "loss": 4.0292,
+      "step": 2249
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019198268755152512,
+      "loss": 4.0544,
+      "step": 2250
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019192085737840064,
+      "loss": 4.6097,
+      "step": 2251
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019185902720527615,
+      "loss": 3.9369,
+      "step": 2252
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001917971970321517,
+      "loss": 4.3381,
+      "step": 2253
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019173536685902719,
+      "loss": 4.4533,
+      "step": 2254
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001916735366859027,
+      "loss": 4.3471,
+      "step": 2255
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019161170651277822,
+      "loss": 4.4169,
+      "step": 2256
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019154987633965374,
+      "loss": 3.9839,
+      "step": 2257
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019148804616652925,
+      "loss": 4.0178,
+      "step": 2258
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019142621599340474,
+      "loss": 4.0981,
+      "step": 2259
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019136438582028029,
+      "loss": 4.8785,
+      "step": 2260
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001913025556471558,
+      "loss": 4.3341,
+      "step": 2261
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019124072547403132,
+      "loss": 4.0017,
+      "step": 2262
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001911788953009068,
+      "loss": 4.3286,
+      "step": 2263
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019111706512778232,
+      "loss": 4.2555,
+      "step": 2264
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019105523495465784,
+      "loss": 4.642,
+      "step": 2265
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019099340478153338,
+      "loss": 4.5935,
+      "step": 2266
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001909315746084089,
+      "loss": 5.2566,
+      "step": 2267
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001908697444352844,
+      "loss": 4.0273,
+      "step": 2268
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001908079142621599,
+      "loss": 4.2581,
+      "step": 2269
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019074608408903542,
+      "loss": 4.0552,
+      "step": 2270
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019068425391591097,
+      "loss": 4.5656,
+      "step": 2271
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019062242374278648,
+      "loss": 4.5885,
+      "step": 2272
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019056059356966197,
+      "loss": 4.2681,
+      "step": 2273
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001904987633965375,
+      "loss": 3.9055,
+      "step": 2274
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000190436933223413,
+      "loss": 4.6501,
+      "step": 2275
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00019037510305028852,
+      "loss": 4.4079,
+      "step": 2276
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000190313272877164,
+      "loss": 3.889,
+      "step": 2277
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00019025144270403956,
+      "loss": 3.8225,
+      "step": 2278
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00019018961253091507,
+      "loss": 4.1693,
+      "step": 2279
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001901277823577906,
+      "loss": 4.6065,
+      "step": 2280
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001900659521846661,
+      "loss": 4.2735,
+      "step": 2281
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001900041220115416,
+      "loss": 4.5255,
+      "step": 2282
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018994229183841714,
+      "loss": 4.3245,
+      "step": 2283
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018988046166529266,
+      "loss": 4.0601,
+      "step": 2284
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018981863149216817,
+      "loss": 4.5977,
+      "step": 2285
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018975680131904366,
+      "loss": 4.4614,
+      "step": 2286
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018969497114591918,
+      "loss": 4.3322,
+      "step": 2287
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001896331409727947,
+      "loss": 4.5522,
+      "step": 2288
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018957131079967024,
+      "loss": 4.3444,
+      "step": 2289
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018950948062654576,
+      "loss": 4.2223,
+      "step": 2290
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018944765045342124,
+      "loss": 4.6983,
+      "step": 2291
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018938582028029676,
+      "loss": 4.4067,
+      "step": 2292
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018932399010717228,
+      "loss": 4.2571,
+      "step": 2293
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001892621599340478,
+      "loss": 4.697,
+      "step": 2294
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018920032976092334,
+      "loss": 4.3857,
+      "step": 2295
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018913849958779883,
+      "loss": 4.3765,
+      "step": 2296
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018907666941467434,
+      "loss": 4.1534,
+      "step": 2297
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018901483924154986,
+      "loss": 4.5824,
+      "step": 2298
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018895300906842538,
+      "loss": 4.9713,
+      "step": 2299
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018889117889530087,
+      "loss": 4.1792,
+      "step": 2300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001888293487221764,
+      "loss": 4.1049,
+      "step": 2301
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018876751854905193,
+      "loss": 4.2729,
+      "step": 2302
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018870568837592744,
+      "loss": 4.3244,
+      "step": 2303
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018864385820280296,
+      "loss": 4.1311,
+      "step": 2304
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018858202802967845,
+      "loss": 4.0728,
+      "step": 2305
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018852019785655397,
+      "loss": 4.7833,
+      "step": 2306
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001884583676834295,
+      "loss": 4.6293,
+      "step": 2307
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018839653751030503,
+      "loss": 4.4349,
+      "step": 2308
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018833470733718052,
+      "loss": 4.2059,
+      "step": 2309
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018827287716405603,
+      "loss": 4.152,
+      "step": 2310
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018821104699093155,
+      "loss": 5.1145,
+      "step": 2311
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001881492168178071,
+      "loss": 4.2765,
+      "step": 2312
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001880873866446826,
+      "loss": 4.4382,
+      "step": 2313
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001880255564715581,
+      "loss": 4.6155,
+      "step": 2314
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018796372629843362,
+      "loss": 4.2142,
+      "step": 2315
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018790189612530913,
+      "loss": 4.4822,
+      "step": 2316
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018784006595218465,
+      "loss": 4.3556,
+      "step": 2317
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018777823577906014,
+      "loss": 4.571,
+      "step": 2318
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018771640560593568,
+      "loss": 4.3069,
+      "step": 2319
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001876545754328112,
+      "loss": 4.3641,
+      "step": 2320
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018759274525968672,
+      "loss": 4.2489,
+      "step": 2321
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018753091508656223,
+      "loss": 3.9271,
+      "step": 2322
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018746908491343772,
+      "loss": 4.3267,
+      "step": 2323
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018740725474031324,
+      "loss": 4.1141,
+      "step": 2324
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018734542456718878,
+      "loss": 4.5104,
+      "step": 2325
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001872835943940643,
+      "loss": 3.9583,
+      "step": 2326
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018722176422093981,
+      "loss": 3.9337,
+      "step": 2327
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001871599340478153,
+      "loss": 3.8943,
+      "step": 2328
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018709810387469082,
+      "loss": 4.2283,
+      "step": 2329
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018703627370156636,
+      "loss": 4.1975,
+      "step": 2330
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018697444352844188,
+      "loss": 3.7907,
+      "step": 2331
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018691261335531737,
+      "loss": 4.5986,
+      "step": 2332
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001868507831821929,
+      "loss": 4.1209,
+      "step": 2333
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001867889530090684,
+      "loss": 4.565,
+      "step": 2334
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018672712283594392,
+      "loss": 4.6837,
+      "step": 2335
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018666529266281946,
+      "loss": 4.0879,
+      "step": 2336
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018660346248969495,
+      "loss": 4.2608,
+      "step": 2337
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018654163231657047,
+      "loss": 4.3112,
+      "step": 2338
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000186479802143446,
+      "loss": 4.2887,
+      "step": 2339
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001864179719703215,
+      "loss": 4.3246,
+      "step": 2340
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000186356141797197,
+      "loss": 4.7412,
+      "step": 2341
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001862943116240725,
+      "loss": 4.3862,
+      "step": 2342
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018623248145094805,
+      "loss": 4.8469,
+      "step": 2343
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018617065127782357,
+      "loss": 4.2282,
+      "step": 2344
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018610882110469909,
+      "loss": 4.3703,
+      "step": 2345
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018604699093157458,
+      "loss": 4.6155,
+      "step": 2346
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001859851607584501,
+      "loss": 4.6382,
+      "step": 2347
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018592333058532564,
+      "loss": 4.212,
+      "step": 2348
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018586150041220115,
+      "loss": 4.1224,
+      "step": 2349
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018579967023907664,
+      "loss": 4.6669,
+      "step": 2350
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018573784006595216,
+      "loss": 4.026,
+      "step": 2351
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018567600989282768,
+      "loss": 3.9188,
+      "step": 2352
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001856141797197032,
+      "loss": 4.0524,
+      "step": 2353
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018555234954657874,
+      "loss": 4.0541,
+      "step": 2354
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018549051937345423,
+      "loss": 4.1762,
+      "step": 2355
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018542868920032974,
+      "loss": 3.9575,
+      "step": 2356
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018536685902720526,
+      "loss": 4.8002,
+      "step": 2357
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018530502885408077,
+      "loss": 4.394,
+      "step": 2358
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018524319868095632,
+      "loss": 4.176,
+      "step": 2359
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001851813685078318,
+      "loss": 4.4878,
+      "step": 2360
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018511953833470732,
+      "loss": 4.3342,
+      "step": 2361
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018505770816158284,
+      "loss": 4.2079,
+      "step": 2362
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018499587798845836,
+      "loss": 4.6162,
+      "step": 2363
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018493404781533385,
+      "loss": 4.6448,
+      "step": 2364
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018487221764220936,
+      "loss": 4.312,
+      "step": 2365
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001848103874690849,
+      "loss": 4.1187,
+      "step": 2366
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018474855729596042,
+      "loss": 4.5255,
+      "step": 2367
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018468672712283594,
+      "loss": 4.3503,
+      "step": 2368
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018462489694971143,
+      "loss": 4.4079,
+      "step": 2369
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018456306677658695,
+      "loss": 4.2807,
+      "step": 2370
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018450123660346246,
+      "loss": 4.2424,
+      "step": 2371
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000184439406430338,
+      "loss": 4.1302,
+      "step": 2372
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001843775762572135,
+      "loss": 4.0609,
+      "step": 2373
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000184315746084089,
+      "loss": 4.1346,
+      "step": 2374
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018425391591096453,
+      "loss": 4.4258,
+      "step": 2375
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018419208573784005,
+      "loss": 4.3267,
+      "step": 2376
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001841302555647156,
+      "loss": 4.1764,
+      "step": 2377
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018406842539159108,
+      "loss": 4.2069,
+      "step": 2378
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001840065952184666,
+      "loss": 3.9963,
+      "step": 2379
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001839447650453421,
+      "loss": 4.278,
+      "step": 2380
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018388293487221763,
+      "loss": 4.055,
+      "step": 2381
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018382110469909315,
+      "loss": 4.2495,
+      "step": 2382
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018375927452596864,
+      "loss": 4.0001,
+      "step": 2383
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018369744435284418,
+      "loss": 4.3167,
+      "step": 2384
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001836356141797197,
+      "loss": 4.7528,
+      "step": 2385
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001835737840065952,
+      "loss": 4.2825,
+      "step": 2386
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001835119538334707,
+      "loss": 4.3875,
+      "step": 2387
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018345012366034622,
+      "loss": 4.0634,
+      "step": 2388
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018338829348722176,
+      "loss": 4.2285,
+      "step": 2389
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018332646331409728,
+      "loss": 4.5073,
+      "step": 2390
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001832646331409728,
+      "loss": 4.3015,
+      "step": 2391
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018320280296784828,
+      "loss": 4.1484,
+      "step": 2392
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001831409727947238,
+      "loss": 4.4696,
+      "step": 2393
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018307914262159932,
+      "loss": 4.204,
+      "step": 2394
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018301731244847486,
+      "loss": 3.8424,
+      "step": 2395
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018295548227535035,
+      "loss": 4.0342,
+      "step": 2396
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018289365210222587,
+      "loss": 4.2756,
+      "step": 2397
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018283182192910138,
+      "loss": 4.0748,
+      "step": 2398
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001827699917559769,
+      "loss": 5.9426,
+      "step": 2399
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018270816158285242,
+      "loss": 4.2566,
+      "step": 2400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001826463314097279,
+      "loss": 4.7347,
+      "step": 2401
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018258450123660345,
+      "loss": 4.1144,
+      "step": 2402
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018252267106347897,
+      "loss": 4.7693,
+      "step": 2403
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018246084089035448,
+      "loss": 4.6002,
+      "step": 2404
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018239901071722997,
+      "loss": 4.2319,
+      "step": 2405
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001823371805441055,
+      "loss": 4.0796,
+      "step": 2406
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018227535037098103,
+      "loss": 4.2613,
+      "step": 2407
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018221352019785655,
+      "loss": 4.1197,
+      "step": 2408
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018215169002473207,
+      "loss": 4.3857,
+      "step": 2409
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018208985985160756,
+      "loss": 4.1343,
+      "step": 2410
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018202802967848307,
+      "loss": 4.1643,
+      "step": 2411
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001819661995053586,
+      "loss": 4.0935,
+      "step": 2412
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018190436933223413,
+      "loss": 5.1482,
+      "step": 2413
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018184253915910965,
+      "loss": 4.0839,
+      "step": 2414
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018178070898598514,
+      "loss": 4.2921,
+      "step": 2415
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018171887881286066,
+      "loss": 5.2294,
+      "step": 2416
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018165704863973617,
+      "loss": 4.2379,
+      "step": 2417
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018159521846661172,
+      "loss": 4.2379,
+      "step": 2418
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001815333882934872,
+      "loss": 4.4058,
+      "step": 2419
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018147155812036272,
+      "loss": 3.9027,
+      "step": 2420
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018140972794723824,
+      "loss": 4.3838,
+      "step": 2421
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018134789777411376,
+      "loss": 4.7827,
+      "step": 2422
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018128606760098927,
+      "loss": 4.006,
+      "step": 2423
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018122423742786476,
+      "loss": 4.0903,
+      "step": 2424
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001811624072547403,
+      "loss": 4.5285,
+      "step": 2425
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018110057708161582,
+      "loss": 4.2161,
+      "step": 2426
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018103874690849134,
+      "loss": 4.5388,
+      "step": 2427
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018097691673536683,
+      "loss": 4.1773,
+      "step": 2428
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018091508656224234,
+      "loss": 4.0,
+      "step": 2429
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018085325638911786,
+      "loss": 4.4492,
+      "step": 2430
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001807914262159934,
+      "loss": 4.4096,
+      "step": 2431
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018072959604286892,
+      "loss": 3.7195,
+      "step": 2432
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001806677658697444,
+      "loss": 6.0865,
+      "step": 2433
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018060593569661993,
+      "loss": 4.4198,
+      "step": 2434
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018054410552349544,
+      "loss": 4.2215,
+      "step": 2435
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000180482275350371,
+      "loss": 5.8872,
+      "step": 2436
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018042044517724648,
+      "loss": 4.5043,
+      "step": 2437
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000180358615004122,
+      "loss": 4.8598,
+      "step": 2438
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001802967848309975,
+      "loss": 4.9862,
+      "step": 2439
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018023495465787303,
+      "loss": 4.286,
+      "step": 2440
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018017312448474854,
+      "loss": 4.7215,
+      "step": 2441
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018011129431162403,
+      "loss": 4.4592,
+      "step": 2442
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018004946413849958,
+      "loss": 4.1455,
+      "step": 2443
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001799876339653751,
+      "loss": 4.1635,
+      "step": 2444
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001799258037922506,
+      "loss": 4.4163,
+      "step": 2445
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017986397361912613,
+      "loss": 4.3235,
+      "step": 2446
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017980214344600162,
+      "loss": 4.6524,
+      "step": 2447
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017974031327287713,
+      "loss": 4.2119,
+      "step": 2448
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017967848309975268,
+      "loss": 4.6738,
+      "step": 2449
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001796166529266282,
+      "loss": 4.1305,
+      "step": 2450
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017955482275350368,
+      "loss": 4.315,
+      "step": 2451
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001794929925803792,
+      "loss": 4.5648,
+      "step": 2452
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017943116240725472,
+      "loss": 4.6087,
+      "step": 2453
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017936933223413026,
+      "loss": 4.0065,
+      "step": 2454
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017930750206100578,
+      "loss": 4.3279,
+      "step": 2455
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017924567188788126,
+      "loss": 4.1729,
+      "step": 2456
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017918384171475678,
+      "loss": 4.8546,
+      "step": 2457
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001791220115416323,
+      "loss": 4.0476,
+      "step": 2458
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017906018136850781,
+      "loss": 4.3535,
+      "step": 2459
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001789983511953833,
+      "loss": 4.0868,
+      "step": 2460
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017893652102225885,
+      "loss": 4.1507,
+      "step": 2461
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017887469084913436,
+      "loss": 4.2389,
+      "step": 2462
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017881286067600988,
+      "loss": 4.3865,
+      "step": 2463
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001787510305028854,
+      "loss": 4.3016,
+      "step": 2464
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001786892003297609,
+      "loss": 4.4627,
+      "step": 2465
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017862737015663643,
+      "loss": 3.9502,
+      "step": 2466
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017856553998351195,
+      "loss": 4.1461,
+      "step": 2467
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017850370981038746,
+      "loss": 4.4565,
+      "step": 2468
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017844187963726295,
+      "loss": 4.6421,
+      "step": 2469
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017838004946413847,
+      "loss": 4.25,
+      "step": 2470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000178318219291014,
+      "loss": 4.1975,
+      "step": 2471
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017825638911788953,
+      "loss": 4.0848,
+      "step": 2472
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017819455894476505,
+      "loss": 4.0462,
+      "step": 2473
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017813272877164054,
+      "loss": 4.0139,
+      "step": 2474
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017807089859851605,
+      "loss": 4.3596,
+      "step": 2475
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017800906842539157,
+      "loss": 5.0695,
+      "step": 2476
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017794723825226709,
+      "loss": 4.3106,
+      "step": 2477
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017788540807914263,
+      "loss": 4.431,
+      "step": 2478
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017782357790601812,
+      "loss": 3.9662,
+      "step": 2479
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017776174773289364,
+      "loss": 4.5709,
+      "step": 2480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017769991755976915,
+      "loss": 3.8524,
+      "step": 2481
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017763808738664467,
+      "loss": 3.9325,
+      "step": 2482
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017757625721352016,
+      "loss": 3.9439,
+      "step": 2483
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001775144270403957,
+      "loss": 4.1566,
+      "step": 2484
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017745259686727122,
+      "loss": 4.4481,
+      "step": 2485
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017739076669414674,
+      "loss": 4.5523,
+      "step": 2486
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017732893652102225,
+      "loss": 4.4619,
+      "step": 2487
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017726710634789774,
+      "loss": 4.8993,
+      "step": 2488
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017720527617477326,
+      "loss": 4.4693,
+      "step": 2489
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001771434460016488,
+      "loss": 4.5871,
+      "step": 2490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017708161582852432,
+      "loss": 4.4664,
+      "step": 2491
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001770197856553998,
+      "loss": 4.6022,
+      "step": 2492
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017695795548227532,
+      "loss": 4.2218,
+      "step": 2493
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017689612530915084,
+      "loss": 4.2657,
+      "step": 2494
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017683429513602638,
+      "loss": 4.2763,
+      "step": 2495
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001767724649629019,
+      "loss": 4.1854,
+      "step": 2496
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001767106347897774,
+      "loss": 4.282,
+      "step": 2497
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001766488046166529,
+      "loss": 4.5275,
+      "step": 2498
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017658697444352842,
+      "loss": 4.7273,
+      "step": 2499
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017652514427040394,
+      "loss": 4.3364,
+      "step": 2500
+    },
+    {
+      "epoch": 1.4,
+      "eval_loss": 4.967714786529541,
+      "eval_runtime": 947.3798,
+      "eval_samples_per_second": 2.789,
+      "eval_steps_per_second": 0.349,
+      "eval_wer": 1.7882586275287584,
+      "step": 2500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017646331409727948,
+      "loss": 4.08,
+      "step": 2501
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017640148392415497,
+      "loss": 4.2011,
+      "step": 2502
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001763396537510305,
+      "loss": 4.2336,
+      "step": 2503
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000176277823577906,
+      "loss": 4.198,
+      "step": 2504
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017621599340478152,
+      "loss": 4.0383,
+      "step": 2505
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000176154163231657,
+      "loss": 4.2402,
+      "step": 2506
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017609233305853253,
+      "loss": 3.9221,
+      "step": 2507
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017603050288540807,
+      "loss": 4.0672,
+      "step": 2508
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001759686727122836,
+      "loss": 4.0355,
+      "step": 2509
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001759068425391591,
+      "loss": 4.7416,
+      "step": 2510
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001758450123660346,
+      "loss": 4.3021,
+      "step": 2511
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001757831821929101,
+      "loss": 4.0322,
+      "step": 2512
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017572135201978566,
+      "loss": 4.119,
+      "step": 2513
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017565952184666117,
+      "loss": 4.274,
+      "step": 2514
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017559769167353666,
+      "loss": 4.7404,
+      "step": 2515
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017553586150041218,
+      "loss": 4.1956,
+      "step": 2516
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001754740313272877,
+      "loss": 4.3372,
+      "step": 2517
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001754122011541632,
+      "loss": 5.0094,
+      "step": 2518
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017535037098103876,
+      "loss": 4.4075,
+      "step": 2519
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017528854080791425,
+      "loss": 4.7933,
+      "step": 2520
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017522671063478976,
+      "loss": 4.3391,
+      "step": 2521
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017516488046166528,
+      "loss": 3.9804,
+      "step": 2522
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001751030502885408,
+      "loss": 3.9538,
+      "step": 2523
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017504122011541628,
+      "loss": 4.04,
+      "step": 2524
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001749793899422918,
+      "loss": 3.9676,
+      "step": 2525
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017491755976916734,
+      "loss": 4.4067,
+      "step": 2526
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017485572959604286,
+      "loss": 4.5938,
+      "step": 2527
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017479389942291838,
+      "loss": 4.0748,
+      "step": 2528
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017473206924979387,
+      "loss": 3.9802,
+      "step": 2529
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017467023907666938,
+      "loss": 4.3661,
+      "step": 2530
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017460840890354493,
+      "loss": 4.3319,
+      "step": 2531
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017454657873042044,
+      "loss": 4.5113,
+      "step": 2532
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017448474855729596,
+      "loss": 4.412,
+      "step": 2533
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017442291838417145,
+      "loss": 4.6128,
+      "step": 2534
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017436108821104697,
+      "loss": 4.1836,
+      "step": 2535
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017429925803792248,
+      "loss": 4.6686,
+      "step": 2536
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017423742786479803,
+      "loss": 4.6149,
+      "step": 2537
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017417559769167352,
+      "loss": 5.2281,
+      "step": 2538
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017411376751854903,
+      "loss": 4.2074,
+      "step": 2539
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017405193734542455,
+      "loss": 4.3956,
+      "step": 2540
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017399010717230007,
+      "loss": 4.1584,
+      "step": 2541
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001739282769991756,
+      "loss": 4.2826,
+      "step": 2542
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001738664468260511,
+      "loss": 4.2688,
+      "step": 2543
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017380461665292662,
+      "loss": 4.1466,
+      "step": 2544
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017374278647980213,
+      "loss": 4.2962,
+      "step": 2545
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017368095630667765,
+      "loss": 4.257,
+      "step": 2546
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017361912613355314,
+      "loss": 4.4328,
+      "step": 2547
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017355729596042866,
+      "loss": 4.2988,
+      "step": 2548
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001734954657873042,
+      "loss": 4.0333,
+      "step": 2549
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017343363561417972,
+      "loss": 4.1649,
+      "step": 2550
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017337180544105523,
+      "loss": 4.1032,
+      "step": 2551
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017330997526793072,
+      "loss": 4.2966,
+      "step": 2552
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017324814509480624,
+      "loss": 4.5798,
+      "step": 2553
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017318631492168175,
+      "loss": 4.3704,
+      "step": 2554
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001731244847485573,
+      "loss": 4.4276,
+      "step": 2555
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001730626545754328,
+      "loss": 3.9958,
+      "step": 2556
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001730008244023083,
+      "loss": 4.4693,
+      "step": 2557
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017293899422918382,
+      "loss": 4.0444,
+      "step": 2558
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017287716405605934,
+      "loss": 4.5717,
+      "step": 2559
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017281533388293488,
+      "loss": 4.375,
+      "step": 2560
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017275350370981037,
+      "loss": 4.106,
+      "step": 2561
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001726916735366859,
+      "loss": 4.157,
+      "step": 2562
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001726298433635614,
+      "loss": 4.2146,
+      "step": 2563
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017256801319043692,
+      "loss": 4.1748,
+      "step": 2564
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017250618301731244,
+      "loss": 4.1688,
+      "step": 2565
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017244435284418793,
+      "loss": 4.2318,
+      "step": 2566
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017238252267106347,
+      "loss": 4.3995,
+      "step": 2567
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000172320692497939,
+      "loss": 4.1985,
+      "step": 2568
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001722588623248145,
+      "loss": 4.487,
+      "step": 2569
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017219703215169,
+      "loss": 4.1984,
+      "step": 2570
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001721352019785655,
+      "loss": 4.0619,
+      "step": 2571
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017207337180544105,
+      "loss": 4.593,
+      "step": 2572
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017201154163231657,
+      "loss": 4.4714,
+      "step": 2573
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001719497114591921,
+      "loss": 4.4901,
+      "step": 2574
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017188788128606758,
+      "loss": 3.875,
+      "step": 2575
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001718260511129431,
+      "loss": 3.953,
+      "step": 2576
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001717642209398186,
+      "loss": 4.6639,
+      "step": 2577
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017170239076669415,
+      "loss": 4.3526,
+      "step": 2578
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017164056059356964,
+      "loss": 4.1232,
+      "step": 2579
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017157873042044516,
+      "loss": 4.1924,
+      "step": 2580
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017151690024732068,
+      "loss": 4.4207,
+      "step": 2581
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001714550700741962,
+      "loss": 4.8917,
+      "step": 2582
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001713932399010717,
+      "loss": 4.4101,
+      "step": 2583
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001713314097279472,
+      "loss": 4.3187,
+      "step": 2584
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017126957955482274,
+      "loss": 4.2082,
+      "step": 2585
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017120774938169826,
+      "loss": 4.5202,
+      "step": 2586
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017114591920857378,
+      "loss": 4.4483,
+      "step": 2587
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017108408903544926,
+      "loss": 4.1583,
+      "step": 2588
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017102225886232478,
+      "loss": 4.3345,
+      "step": 2589
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017096042868920032,
+      "loss": 4.2461,
+      "step": 2590
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017089859851607584,
+      "loss": 4.0939,
+      "step": 2591
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017083676834295136,
+      "loss": 4.4848,
+      "step": 2592
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017077493816982685,
+      "loss": 4.3236,
+      "step": 2593
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017071310799670236,
+      "loss": 4.3639,
+      "step": 2594
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017065127782357788,
+      "loss": 4.4683,
+      "step": 2595
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017058944765045342,
+      "loss": 4.1614,
+      "step": 2596
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017052761747732894,
+      "loss": 4.3061,
+      "step": 2597
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017046578730420443,
+      "loss": 4.3981,
+      "step": 2598
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017040395713107995,
+      "loss": 4.0241,
+      "step": 2599
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017034212695795546,
+      "loss": 4.3257,
+      "step": 2600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000170280296784831,
+      "loss": 4.2182,
+      "step": 2601
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001702184666117065,
+      "loss": 4.3287,
+      "step": 2602
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000170156636438582,
+      "loss": 4.6554,
+      "step": 2603
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017009480626545753,
+      "loss": 4.4016,
+      "step": 2604
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017003297609233305,
+      "loss": 4.2556,
+      "step": 2605
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016997114591920856,
+      "loss": 3.9328,
+      "step": 2606
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016990931574608405,
+      "loss": 4.0076,
+      "step": 2607
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001698474855729596,
+      "loss": 4.1666,
+      "step": 2608
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001697856553998351,
+      "loss": 3.99,
+      "step": 2609
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016972382522671063,
+      "loss": 4.4872,
+      "step": 2610
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016966199505358612,
+      "loss": 4.3516,
+      "step": 2611
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016960016488046164,
+      "loss": 3.9909,
+      "step": 2612
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016953833470733715,
+      "loss": 4.2851,
+      "step": 2613
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001694765045342127,
+      "loss": 4.2985,
+      "step": 2614
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001694146743610882,
+      "loss": 4.5875,
+      "step": 2615
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001693528441879637,
+      "loss": 4.2754,
+      "step": 2616
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016929101401483922,
+      "loss": 4.1285,
+      "step": 2617
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016922918384171474,
+      "loss": 3.8739,
+      "step": 2618
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016916735366859028,
+      "loss": 4.4014,
+      "step": 2619
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001691055234954658,
+      "loss": 4.1212,
+      "step": 2620
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016904369332234128,
+      "loss": 4.5203,
+      "step": 2621
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001689818631492168,
+      "loss": 4.2614,
+      "step": 2622
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016892003297609232,
+      "loss": 4.1899,
+      "step": 2623
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016885820280296783,
+      "loss": 4.0708,
+      "step": 2624
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016879637262984332,
+      "loss": 4.316,
+      "step": 2625
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016873454245671887,
+      "loss": 4.0306,
+      "step": 2626
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016867271228359438,
+      "loss": 4.036,
+      "step": 2627
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001686108821104699,
+      "loss": 4.6019,
+      "step": 2628
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016854905193734542,
+      "loss": 3.8785,
+      "step": 2629
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001684872217642209,
+      "loss": 4.3177,
+      "step": 2630
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016842539159109642,
+      "loss": 4.2544,
+      "step": 2631
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016836356141797197,
+      "loss": 4.2142,
+      "step": 2632
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016830173124484748,
+      "loss": 4.1103,
+      "step": 2633
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016823990107172297,
+      "loss": 4.0658,
+      "step": 2634
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001681780708985985,
+      "loss": 4.2792,
+      "step": 2635
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000168116240725474,
+      "loss": 4.3945,
+      "step": 2636
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016805441055234955,
+      "loss": 4.2559,
+      "step": 2637
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016799258037922507,
+      "loss": 4.1571,
+      "step": 2638
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016793075020610056,
+      "loss": 4.1757,
+      "step": 2639
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016786892003297607,
+      "loss": 4.3393,
+      "step": 2640
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001678070898598516,
+      "loss": 4.1822,
+      "step": 2641
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001677452596867271,
+      "loss": 4.2338,
+      "step": 2642
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001676834295136026,
+      "loss": 4.1816,
+      "step": 2643
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016762159934047814,
+      "loss": 4.2847,
+      "step": 2644
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016755976916735366,
+      "loss": 4.4194,
+      "step": 2645
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016749793899422917,
+      "loss": 4.2069,
+      "step": 2646
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001674361088211047,
+      "loss": 4.925,
+      "step": 2647
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016737427864798018,
+      "loss": 3.9961,
+      "step": 2648
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016731244847485572,
+      "loss": 4.1261,
+      "step": 2649
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016725061830173124,
+      "loss": 4.158,
+      "step": 2650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016718878812860676,
+      "loss": 4.4855,
+      "step": 2651
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016712695795548227,
+      "loss": 4.1253,
+      "step": 2652
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016706512778235776,
+      "loss": 4.07,
+      "step": 2653
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016700329760923328,
+      "loss": 4.4006,
+      "step": 2654
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016694146743610882,
+      "loss": 3.9602,
+      "step": 2655
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016687963726298434,
+      "loss": 4.5092,
+      "step": 2656
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016681780708985983,
+      "loss": 4.3365,
+      "step": 2657
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016675597691673534,
+      "loss": 4.3174,
+      "step": 2658
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016669414674361086,
+      "loss": 4.7018,
+      "step": 2659
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016663231657048638,
+      "loss": 4.0241,
+      "step": 2660
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016657048639736192,
+      "loss": 4.2854,
+      "step": 2661
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001665086562242374,
+      "loss": 4.2149,
+      "step": 2662
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016644682605111293,
+      "loss": 4.9532,
+      "step": 2663
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016638499587798844,
+      "loss": 4.1503,
+      "step": 2664
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016632316570486396,
+      "loss": 4.3993,
+      "step": 2665
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016626133553173945,
+      "loss": 4.2617,
+      "step": 2666
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000166199505358615,
+      "loss": 4.7419,
+      "step": 2667
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001661376751854905,
+      "loss": 4.2467,
+      "step": 2668
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016607584501236603,
+      "loss": 4.4842,
+      "step": 2669
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016601401483924154,
+      "loss": 4.7518,
+      "step": 2670
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016595218466611703,
+      "loss": 4.4537,
+      "step": 2671
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016589035449299255,
+      "loss": 4.4919,
+      "step": 2672
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001658285243198681,
+      "loss": 4.2137,
+      "step": 2673
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001657666941467436,
+      "loss": 4.123,
+      "step": 2674
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001657048639736191,
+      "loss": 4.3881,
+      "step": 2675
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016564303380049462,
+      "loss": 4.8815,
+      "step": 2676
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016558120362737013,
+      "loss": 4.254,
+      "step": 2677
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016551937345424568,
+      "loss": 4.4383,
+      "step": 2678
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001654575432811212,
+      "loss": 4.2515,
+      "step": 2679
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016539571310799668,
+      "loss": 4.0853,
+      "step": 2680
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001653338829348722,
+      "loss": 4.0884,
+      "step": 2681
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016527205276174772,
+      "loss": 4.1187,
+      "step": 2682
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016521022258862323,
+      "loss": 4.1804,
+      "step": 2683
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016514839241549878,
+      "loss": 3.5325,
+      "step": 2684
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016508656224237427,
+      "loss": 4.4883,
+      "step": 2685
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016502473206924978,
+      "loss": 4.0911,
+      "step": 2686
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001649629018961253,
+      "loss": 4.1476,
+      "step": 2687
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016490107172300081,
+      "loss": 4.4648,
+      "step": 2688
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001648392415498763,
+      "loss": 4.2674,
+      "step": 2689
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016477741137675182,
+      "loss": 4.2404,
+      "step": 2690
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016471558120362736,
+      "loss": 4.1419,
+      "step": 2691
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016465375103050288,
+      "loss": 4.2329,
+      "step": 2692
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001645919208573784,
+      "loss": 4.1333,
+      "step": 2693
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001645300906842539,
+      "loss": 4.3508,
+      "step": 2694
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001644682605111294,
+      "loss": 4.5458,
+      "step": 2695
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016440643033800495,
+      "loss": 4.0436,
+      "step": 2696
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016434460016488046,
+      "loss": 4.1459,
+      "step": 2697
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016428276999175595,
+      "loss": 4.1083,
+      "step": 2698
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016422093981863147,
+      "loss": 4.4987,
+      "step": 2699
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000164159109645507,
+      "loss": 4.9806,
+      "step": 2700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001640972794723825,
+      "loss": 4.0605,
+      "step": 2701
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016403544929925805,
+      "loss": 3.9136,
+      "step": 2702
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016397361912613354,
+      "loss": 4.1206,
+      "step": 2703
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016391178895300905,
+      "loss": 4.4342,
+      "step": 2704
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016384995877988457,
+      "loss": 4.5537,
+      "step": 2705
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016378812860676009,
+      "loss": 4.4167,
+      "step": 2706
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016372629843363558,
+      "loss": 4.1507,
+      "step": 2707
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001636644682605111,
+      "loss": 4.2838,
+      "step": 2708
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016360263808738664,
+      "loss": 4.3561,
+      "step": 2709
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016354080791426215,
+      "loss": 4.1237,
+      "step": 2710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016347897774113767,
+      "loss": 3.9667,
+      "step": 2711
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016341714756801316,
+      "loss": 4.2777,
+      "step": 2712
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016335531739488868,
+      "loss": 4.0954,
+      "step": 2713
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016329348722176422,
+      "loss": 4.4786,
+      "step": 2714
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016323165704863974,
+      "loss": 4.3266,
+      "step": 2715
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016316982687551525,
+      "loss": 4.3637,
+      "step": 2716
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016310799670239074,
+      "loss": 4.3237,
+      "step": 2717
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016304616652926626,
+      "loss": 4.0447,
+      "step": 2718
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016298433635614177,
+      "loss": 4.1503,
+      "step": 2719
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016292250618301732,
+      "loss": 4.1161,
+      "step": 2720
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001628606760098928,
+      "loss": 4.1645,
+      "step": 2721
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016279884583676832,
+      "loss": 4.2982,
+      "step": 2722
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016273701566364384,
+      "loss": 4.2018,
+      "step": 2723
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016267518549051936,
+      "loss": 4.0788,
+      "step": 2724
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001626133553173949,
+      "loss": 4.3285,
+      "step": 2725
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001625515251442704,
+      "loss": 4.1045,
+      "step": 2726
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001624896949711459,
+      "loss": 4.3126,
+      "step": 2727
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016242786479802142,
+      "loss": 4.2208,
+      "step": 2728
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016236603462489694,
+      "loss": 4.6796,
+      "step": 2729
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016230420445177243,
+      "loss": 4.8786,
+      "step": 2730
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016224237427864795,
+      "loss": 3.8599,
+      "step": 2731
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001621805441055235,
+      "loss": 3.9384,
+      "step": 2732
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000162118713932399,
+      "loss": 5.0368,
+      "step": 2733
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016205688375927452,
+      "loss": 3.7891,
+      "step": 2734
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016199505358615,
+      "loss": 4.9868,
+      "step": 2735
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016193322341302553,
+      "loss": 4.4479,
+      "step": 2736
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016187139323990105,
+      "loss": 4.5552,
+      "step": 2737
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001618095630667766,
+      "loss": 4.6102,
+      "step": 2738
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001617477328936521,
+      "loss": 4.0834,
+      "step": 2739
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001616859027205276,
+      "loss": 4.1127,
+      "step": 2740
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001616240725474031,
+      "loss": 4.7225,
+      "step": 2741
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016156224237427863,
+      "loss": 4.1624,
+      "step": 2742
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016150041220115417,
+      "loss": 4.2638,
+      "step": 2743
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016143858202802966,
+      "loss": 4.3138,
+      "step": 2744
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016137675185490518,
+      "loss": 4.5146,
+      "step": 2745
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001613149216817807,
+      "loss": 4.1827,
+      "step": 2746
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001612530915086562,
+      "loss": 4.1394,
+      "step": 2747
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016119126133553173,
+      "loss": 4.7112,
+      "step": 2748
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016112943116240722,
+      "loss": 4.891,
+      "step": 2749
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016106760098928276,
+      "loss": 4.7895,
+      "step": 2750
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016100577081615828,
+      "loss": 4.2906,
+      "step": 2751
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001609439406430338,
+      "loss": 4.022,
+      "step": 2752
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016088211046990928,
+      "loss": 4.2226,
+      "step": 2753
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001608202802967848,
+      "loss": 4.3312,
+      "step": 2754
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016075845012366034,
+      "loss": 4.3469,
+      "step": 2755
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016069661995053586,
+      "loss": 4.5174,
+      "step": 2756
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016063478977741138,
+      "loss": 4.3703,
+      "step": 2757
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016057295960428687,
+      "loss": 4.4953,
+      "step": 2758
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016051112943116238,
+      "loss": 4.3634,
+      "step": 2759
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001604492992580379,
+      "loss": 4.3892,
+      "step": 2760
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016038746908491344,
+      "loss": 4.2824,
+      "step": 2761
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016032563891178893,
+      "loss": 4.3466,
+      "step": 2762
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016026380873866445,
+      "loss": 4.3565,
+      "step": 2763
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016020197856553997,
+      "loss": 4.4975,
+      "step": 2764
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016014014839241548,
+      "loss": 3.9054,
+      "step": 2765
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000160078318219291,
+      "loss": 3.9464,
+      "step": 2766
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001600164880461665,
+      "loss": 4.2423,
+      "step": 2767
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00015995465787304203,
+      "loss": 4.2861,
+      "step": 2768
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00015989282769991755,
+      "loss": 3.7544,
+      "step": 2769
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00015983099752679307,
+      "loss": 4.2155,
+      "step": 2770
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00015976916735366858,
+      "loss": 4.2858,
+      "step": 2771
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00015970733718054407,
+      "loss": 4.0404,
+      "step": 2772
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00015964550700741962,
+      "loss": 3.9748,
+      "step": 2773
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00015958367683429513,
+      "loss": 4.124,
+      "step": 2774
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00015952184666117065,
+      "loss": 3.8862,
+      "step": 2775
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00015946001648804614,
+      "loss": 4.1761,
+      "step": 2776
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00015939818631492166,
+      "loss": 4.6388,
+      "step": 2777
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00015933635614179717,
+      "loss": 4.2182,
+      "step": 2778
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00015927452596867272,
+      "loss": 4.06,
+      "step": 2779
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00015921269579554823,
+      "loss": 4.1836,
+      "step": 2780
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00015915086562242372,
+      "loss": 3.5874,
+      "step": 2781
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00015908903544929924,
+      "loss": 3.8788,
+      "step": 2782
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00015902720527617475,
+      "loss": 3.929,
+      "step": 2783
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001589653751030503,
+      "loss": 3.5994,
+      "step": 2784
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00015890354492992576,
+      "loss": 4.6409,
+      "step": 2785
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001588417147568013,
+      "loss": 4.3906,
+      "step": 2786
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00015877988458367682,
+      "loss": 4.2538,
+      "step": 2787
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00015871805441055234,
+      "loss": 4.3747,
+      "step": 2788
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00015865622423742785,
+      "loss": 3.9465,
+      "step": 2789
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00015859439406430334,
+      "loss": 4.1066,
+      "step": 2790
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001585325638911789,
+      "loss": 3.9572,
+      "step": 2791
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001584707337180544,
+      "loss": 4.1621,
+      "step": 2792
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015840890354492992,
+      "loss": 4.2937,
+      "step": 2793
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001583470733718054,
+      "loss": 4.3335,
+      "step": 2794
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015828524319868093,
+      "loss": 4.1616,
+      "step": 2795
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015822341302555644,
+      "loss": 4.4244,
+      "step": 2796
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000158161582852432,
+      "loss": 4.177,
+      "step": 2797
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001580997526793075,
+      "loss": 4.4162,
+      "step": 2798
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000158037922506183,
+      "loss": 4.3156,
+      "step": 2799
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001579760923330585,
+      "loss": 4.3526,
+      "step": 2800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015791426215993403,
+      "loss": 4.0596,
+      "step": 2801
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015785243198680957,
+      "loss": 3.936,
+      "step": 2802
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001577906018136851,
+      "loss": 4.057,
+      "step": 2803
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015772877164056058,
+      "loss": 4.2211,
+      "step": 2804
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001576669414674361,
+      "loss": 4.1133,
+      "step": 2805
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001576051112943116,
+      "loss": 4.4324,
+      "step": 2806
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015754328112118713,
+      "loss": 4.615,
+      "step": 2807
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015748145094806262,
+      "loss": 4.0555,
+      "step": 2808
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015741962077493816,
+      "loss": 3.9415,
+      "step": 2809
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015735779060181368,
+      "loss": 3.7546,
+      "step": 2810
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001572959604286892,
+      "loss": 3.9908,
+      "step": 2811
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001572341302555647,
+      "loss": 4.2729,
+      "step": 2812
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001571723000824402,
+      "loss": 4.2774,
+      "step": 2813
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015711046990931571,
+      "loss": 4.1826,
+      "step": 2814
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015704863973619126,
+      "loss": 4.0917,
+      "step": 2815
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015698680956306678,
+      "loss": 3.9492,
+      "step": 2816
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015692497938994226,
+      "loss": 4.0602,
+      "step": 2817
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015686314921681778,
+      "loss": 3.9364,
+      "step": 2818
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001568013190436933,
+      "loss": 4.2055,
+      "step": 2819
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015673948887056884,
+      "loss": 4.0028,
+      "step": 2820
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015667765869744436,
+      "loss": 3.9101,
+      "step": 2821
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015661582852431985,
+      "loss": 4.3025,
+      "step": 2822
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015655399835119536,
+      "loss": 4.2426,
+      "step": 2823
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015649216817807088,
+      "loss": 4.2346,
+      "step": 2824
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001564303380049464,
+      "loss": 4.6075,
+      "step": 2825
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015636850783182194,
+      "loss": 4.0716,
+      "step": 2826
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015630667765869743,
+      "loss": 4.3965,
+      "step": 2827
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015624484748557295,
+      "loss": 4.7774,
+      "step": 2828
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015618301731244846,
+      "loss": 4.0364,
+      "step": 2829
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015612118713932398,
+      "loss": 4.6578,
+      "step": 2830
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015605935696619947,
+      "loss": 3.9939,
+      "step": 2831
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000155997526793075,
+      "loss": 4.3993,
+      "step": 2832
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015593569661995053,
+      "loss": 3.8956,
+      "step": 2833
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015587386644682605,
+      "loss": 4.2674,
+      "step": 2834
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015581203627370156,
+      "loss": 4.2169,
+      "step": 2835
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015575020610057705,
+      "loss": 4.068,
+      "step": 2836
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015568837592745257,
+      "loss": 3.802,
+      "step": 2837
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001556265457543281,
+      "loss": 4.0315,
+      "step": 2838
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015556471558120363,
+      "loss": 4.2165,
+      "step": 2839
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015550288540807912,
+      "loss": 3.9964,
+      "step": 2840
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015544105523495464,
+      "loss": 4.6568,
+      "step": 2841
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015537922506183015,
+      "loss": 4.3444,
+      "step": 2842
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015531739488870567,
+      "loss": 4.1593,
+      "step": 2843
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001552555647155812,
+      "loss": 4.783,
+      "step": 2844
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001551937345424567,
+      "loss": 4.4191,
+      "step": 2845
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015513190436933222,
+      "loss": 4.3172,
+      "step": 2846
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015507007419620774,
+      "loss": 4.5343,
+      "step": 2847
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015500824402308325,
+      "loss": 4.4063,
+      "step": 2848
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015494641384995874,
+      "loss": 4.1407,
+      "step": 2849
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015488458367683428,
+      "loss": 4.1587,
+      "step": 2850
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001548227535037098,
+      "loss": 4.2101,
+      "step": 2851
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015476092333058532,
+      "loss": 4.2917,
+      "step": 2852
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015469909315746083,
+      "loss": 4.025,
+      "step": 2853
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015463726298433632,
+      "loss": 3.9603,
+      "step": 2854
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015457543281121184,
+      "loss": 3.8577,
+      "step": 2855
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015451360263808738,
+      "loss": 4.2034,
+      "step": 2856
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001544517724649629,
+      "loss": 4.1188,
+      "step": 2857
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015438994229183842,
+      "loss": 4.2036,
+      "step": 2858
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001543281121187139,
+      "loss": 4.0516,
+      "step": 2859
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015426628194558942,
+      "loss": 4.363,
+      "step": 2860
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015420445177246497,
+      "loss": 4.3321,
+      "step": 2861
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015414262159934048,
+      "loss": 4.2317,
+      "step": 2862
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015408079142621597,
+      "loss": 4.2712,
+      "step": 2863
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001540189612530915,
+      "loss": 4.2909,
+      "step": 2864
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000153957131079967,
+      "loss": 3.9119,
+      "step": 2865
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015389530090684252,
+      "loss": 4.2593,
+      "step": 2866
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015383347073371807,
+      "loss": 4.3621,
+      "step": 2867
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015377164056059356,
+      "loss": 3.9224,
+      "step": 2868
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015370981038746907,
+      "loss": 4.4706,
+      "step": 2869
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001536479802143446,
+      "loss": 4.0268,
+      "step": 2870
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001535861500412201,
+      "loss": 4.0725,
+      "step": 2871
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001535243198680956,
+      "loss": 3.9515,
+      "step": 2872
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001534624896949711,
+      "loss": 4.483,
+      "step": 2873
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015340065952184666,
+      "loss": 4.4361,
+      "step": 2874
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015333882934872217,
+      "loss": 4.5398,
+      "step": 2875
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001532769991755977,
+      "loss": 4.0459,
+      "step": 2876
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015321516900247318,
+      "loss": 4.131,
+      "step": 2877
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001531533388293487,
+      "loss": 4.3882,
+      "step": 2878
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015309150865622424,
+      "loss": 3.86,
+      "step": 2879
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015302967848309976,
+      "loss": 3.6529,
+      "step": 2880
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015296784830997524,
+      "loss": 4.0798,
+      "step": 2881
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015290601813685076,
+      "loss": 5.3704,
+      "step": 2882
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015284418796372628,
+      "loss": 4.3631,
+      "step": 2883
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001527823577906018,
+      "loss": 3.7283,
+      "step": 2884
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015272052761747734,
+      "loss": 4.3869,
+      "step": 2885
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015265869744435283,
+      "loss": 4.2719,
+      "step": 2886
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015259686727122834,
+      "loss": 4.3036,
+      "step": 2887
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015253503709810386,
+      "loss": 4.0172,
+      "step": 2888
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015247320692497938,
+      "loss": 4.0192,
+      "step": 2889
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015241137675185492,
+      "loss": 4.1507,
+      "step": 2890
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015234954657873038,
+      "loss": 4.0966,
+      "step": 2891
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015228771640560593,
+      "loss": 4.3967,
+      "step": 2892
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015222588623248144,
+      "loss": 4.5019,
+      "step": 2893
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015216405605935696,
+      "loss": 4.0691,
+      "step": 2894
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015210222588623245,
+      "loss": 4.2817,
+      "step": 2895
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015204039571310797,
+      "loss": 4.094,
+      "step": 2896
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001519785655399835,
+      "loss": 4.3614,
+      "step": 2897
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015191673536685903,
+      "loss": 4.1314,
+      "step": 2898
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015185490519373454,
+      "loss": 4.0227,
+      "step": 2899
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015179307502061003,
+      "loss": 4.1131,
+      "step": 2900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015173124484748555,
+      "loss": 4.1814,
+      "step": 2901
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015166941467436107,
+      "loss": 4.0451,
+      "step": 2902
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001516075845012366,
+      "loss": 4.0912,
+      "step": 2903
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001515457543281121,
+      "loss": 4.4227,
+      "step": 2904
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015148392415498762,
+      "loss": 4.3325,
+      "step": 2905
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015142209398186313,
+      "loss": 4.1875,
+      "step": 2906
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015136026380873865,
+      "loss": 4.3884,
+      "step": 2907
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001512984336356142,
+      "loss": 4.3069,
+      "step": 2908
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015123660346248968,
+      "loss": 4.2318,
+      "step": 2909
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001511747732893652,
+      "loss": 4.2726,
+      "step": 2910
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015111294311624072,
+      "loss": 4.1201,
+      "step": 2911
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015105111294311623,
+      "loss": 3.9619,
+      "step": 2912
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015098928276999172,
+      "loss": 4.5153,
+      "step": 2913
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015092745259686724,
+      "loss": 4.5674,
+      "step": 2914
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015086562242374278,
+      "loss": 4.2549,
+      "step": 2915
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001508037922506183,
+      "loss": 3.834,
+      "step": 2916
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015074196207749381,
+      "loss": 3.8257,
+      "step": 2917
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001506801319043693,
+      "loss": 4.3548,
+      "step": 2918
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015061830173124482,
+      "loss": 4.0105,
+      "step": 2919
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015055647155812034,
+      "loss": 4.1607,
+      "step": 2920
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015049464138499588,
+      "loss": 3.9163,
+      "step": 2921
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001504328112118714,
+      "loss": 3.8818,
+      "step": 2922
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001503709810387469,
+      "loss": 4.1936,
+      "step": 2923
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001503091508656224,
+      "loss": 4.1474,
+      "step": 2924
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015024732069249792,
+      "loss": 4.149,
+      "step": 2925
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015018549051937346,
+      "loss": 4.2869,
+      "step": 2926
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015012366034624895,
+      "loss": 4.2542,
+      "step": 2927
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015006183017312447,
+      "loss": 4.338,
+      "step": 2928
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015,
+      "loss": 5.0285,
+      "step": 2929
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001499381698268755,
+      "loss": 4.4482,
+      "step": 2930
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00014987633965375102,
+      "loss": 3.8525,
+      "step": 2931
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00014981450948062654,
+      "loss": 4.2151,
+      "step": 2932
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00014975267930750205,
+      "loss": 4.6118,
+      "step": 2933
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00014969084913437757,
+      "loss": 3.4412,
+      "step": 2934
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00014962901896125306,
+      "loss": 3.8315,
+      "step": 2935
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001495671887881286,
+      "loss": 5.0155,
+      "step": 2936
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00014950535861500412,
+      "loss": 4.2408,
+      "step": 2937
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00014944352844187964,
+      "loss": 3.9003,
+      "step": 2938
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00014938169826875515,
+      "loss": 4.1961,
+      "step": 2939
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00014931986809563064,
+      "loss": 4.1004,
+      "step": 2940
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00014925803792250619,
+      "loss": 4.1489,
+      "step": 2941
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00014919620774938168,
+      "loss": 4.0844,
+      "step": 2942
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001491343775762572,
+      "loss": 4.4832,
+      "step": 2943
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001490725474031327,
+      "loss": 4.6278,
+      "step": 2944
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00014901071723000823,
+      "loss": 4.5664,
+      "step": 2945
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00014894888705688374,
+      "loss": 4.0889,
+      "step": 2946
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00014888705688375926,
+      "loss": 4.2104,
+      "step": 2947
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00014882522671063477,
+      "loss": 4.0328,
+      "step": 2948
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001487633965375103,
+      "loss": 4.0304,
+      "step": 2949
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001487015663643858,
+      "loss": 3.8821,
+      "step": 2950
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00014863973619126132,
+      "loss": 4.072,
+      "step": 2951
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00014857790601813684,
+      "loss": 4.0504,
+      "step": 2952
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014851607584501236,
+      "loss": 3.8243,
+      "step": 2953
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014845424567188787,
+      "loss": 4.2273,
+      "step": 2954
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001483924154987634,
+      "loss": 3.8885,
+      "step": 2955
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001483305853256389,
+      "loss": 4.6433,
+      "step": 2956
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014826875515251442,
+      "loss": 3.7781,
+      "step": 2957
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014820692497938991,
+      "loss": 4.4117,
+      "step": 2958
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014814509480626546,
+      "loss": 4.1236,
+      "step": 2959
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014808326463314095,
+      "loss": 4.099,
+      "step": 2960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014802143446001646,
+      "loss": 3.9118,
+      "step": 2961
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.000147959604286892,
+      "loss": 3.9597,
+      "step": 2962
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001478977741137675,
+      "loss": 4.6664,
+      "step": 2963
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.000147835943940643,
+      "loss": 4.9245,
+      "step": 2964
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014777411376751853,
+      "loss": 4.0001,
+      "step": 2965
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014771228359439405,
+      "loss": 3.8875,
+      "step": 2966
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014765045342126956,
+      "loss": 4.1171,
+      "step": 2967
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014758862324814508,
+      "loss": 4.1229,
+      "step": 2968
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001475267930750206,
+      "loss": 3.9181,
+      "step": 2969
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001474649629018961,
+      "loss": 4.0492,
+      "step": 2970
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014740313272877163,
+      "loss": 4.0068,
+      "step": 2971
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014734130255564715,
+      "loss": 4.2876,
+      "step": 2972
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014727947238252266,
+      "loss": 4.1921,
+      "step": 2973
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014721764220939818,
+      "loss": 4.358,
+      "step": 2974
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001471558120362737,
+      "loss": 4.1283,
+      "step": 2975
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014709398186314919,
+      "loss": 4.1907,
+      "step": 2976
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014703215169002473,
+      "loss": 4.3004,
+      "step": 2977
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014697032151690025,
+      "loss": 4.5539,
+      "step": 2978
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014690849134377573,
+      "loss": 4.0817,
+      "step": 2979
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014684666117065128,
+      "loss": 4.1215,
+      "step": 2980
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014678483099752677,
+      "loss": 4.1216,
+      "step": 2981
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014672300082440228,
+      "loss": 4.134,
+      "step": 2982
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001466611706512778,
+      "loss": 4.2976,
+      "step": 2983
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014659934047815332,
+      "loss": 4.3401,
+      "step": 2984
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014653751030502886,
+      "loss": 4.3141,
+      "step": 2985
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014647568013190435,
+      "loss": 4.4611,
+      "step": 2986
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014641384995877987,
+      "loss": 4.1446,
+      "step": 2987
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014635201978565538,
+      "loss": 4.084,
+      "step": 2988
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001462901896125309,
+      "loss": 4.3433,
+      "step": 2989
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014622835943940642,
+      "loss": 4.0296,
+      "step": 2990
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014616652926628193,
+      "loss": 4.2582,
+      "step": 2991
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014610469909315745,
+      "loss": 4.0768,
+      "step": 2992
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014604286892003297,
+      "loss": 4.6102,
+      "step": 2993
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014598103874690848,
+      "loss": 4.0457,
+      "step": 2994
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.000145919208573784,
+      "loss": 4.2412,
+      "step": 2995
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014585737840065952,
+      "loss": 4.1188,
+      "step": 2996
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.000145795548227535,
+      "loss": 4.4904,
+      "step": 2997
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014573371805441055,
+      "loss": 3.9423,
+      "step": 2998
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014567188788128604,
+      "loss": 4.0599,
+      "step": 2999
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014561005770816158,
+      "loss": 4.1573,
+      "step": 3000
+    },
+    {
+      "epoch": 1.68,
+      "eval_loss": 4.77024507522583,
+      "eval_runtime": 960.6414,
+      "eval_samples_per_second": 2.75,
+      "eval_steps_per_second": 0.345,
+      "eval_wer": 1.9015866719555732,
+      "step": 3000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001455482275350371,
+      "loss": 4.0447,
+      "step": 3001
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001454863973619126,
+      "loss": 4.6959,
+      "step": 3002
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014542456718878813,
+      "loss": 3.918,
+      "step": 3003
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014536273701566362,
+      "loss": 4.061,
+      "step": 3004
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014530090684253914,
+      "loss": 4.0473,
+      "step": 3005
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014523907666941466,
+      "loss": 4.3885,
+      "step": 3006
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014517724649629017,
+      "loss": 4.9364,
+      "step": 3007
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001451154163231657,
+      "loss": 4.1261,
+      "step": 3008
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001450535861500412,
+      "loss": 4.1401,
+      "step": 3009
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014499175597691672,
+      "loss": 4.1314,
+      "step": 3010
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014492992580379224,
+      "loss": 4.2334,
+      "step": 3011
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014486809563066776,
+      "loss": 3.7447,
+      "step": 3012
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014480626545754327,
+      "loss": 4.4207,
+      "step": 3013
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001447444352844188,
+      "loss": 4.2387,
+      "step": 3014
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001446826051112943,
+      "loss": 3.9491,
+      "step": 3015
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014462077493816982,
+      "loss": 4.1736,
+      "step": 3016
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014455894476504534,
+      "loss": 4.9579,
+      "step": 3017
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014449711459192085,
+      "loss": 3.9371,
+      "step": 3018
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014443528441879637,
+      "loss": 4.2165,
+      "step": 3019
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014437345424567186,
+      "loss": 4.0408,
+      "step": 3020
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001443116240725474,
+      "loss": 4.2062,
+      "step": 3021
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001442497938994229,
+      "loss": 4.0505,
+      "step": 3022
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001441879637262984,
+      "loss": 4.5659,
+      "step": 3023
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014412613355317393,
+      "loss": 4.0201,
+      "step": 3024
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014406430338004944,
+      "loss": 4.2657,
+      "step": 3025
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014400247320692496,
+      "loss": 4.1482,
+      "step": 3026
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014394064303380048,
+      "loss": 3.9601,
+      "step": 3027
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000143878812860676,
+      "loss": 4.4726,
+      "step": 3028
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001438169826875515,
+      "loss": 4.3444,
+      "step": 3029
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014375515251442703,
+      "loss": 5.8663,
+      "step": 3030
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014369332234130254,
+      "loss": 4.0892,
+      "step": 3031
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014363149216817806,
+      "loss": 4.0897,
+      "step": 3032
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014356966199505358,
+      "loss": 4.0522,
+      "step": 3033
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001435078318219291,
+      "loss": 3.9255,
+      "step": 3034
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001434460016488046,
+      "loss": 4.1576,
+      "step": 3035
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014338417147568013,
+      "loss": 4.0687,
+      "step": 3036
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014332234130255564,
+      "loss": 4.1643,
+      "step": 3037
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014326051112943113,
+      "loss": 4.1673,
+      "step": 3038
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014319868095630668,
+      "loss": 4.0722,
+      "step": 3039
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001431368507831822,
+      "loss": 4.186,
+      "step": 3040
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014307502061005768,
+      "loss": 4.1703,
+      "step": 3041
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014301319043693323,
+      "loss": 3.9305,
+      "step": 3042
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014295136026380872,
+      "loss": 4.1079,
+      "step": 3043
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014288953009068426,
+      "loss": 4.0519,
+      "step": 3044
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014282769991755975,
+      "loss": 4.1662,
+      "step": 3045
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014276586974443526,
+      "loss": 3.993,
+      "step": 3046
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014270403957131078,
+      "loss": 4.0605,
+      "step": 3047
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001426422093981863,
+      "loss": 4.1876,
+      "step": 3048
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014258037922506181,
+      "loss": 4.1476,
+      "step": 3049
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014251854905193733,
+      "loss": 3.9316,
+      "step": 3050
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014245671887881285,
+      "loss": 4.0934,
+      "step": 3051
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014239488870568836,
+      "loss": 3.9985,
+      "step": 3052
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014233305853256388,
+      "loss": 4.1816,
+      "step": 3053
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001422712283594394,
+      "loss": 4.0236,
+      "step": 3054
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014220939818631491,
+      "loss": 4.2479,
+      "step": 3055
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014214756801319043,
+      "loss": 4.1503,
+      "step": 3056
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014208573784006595,
+      "loss": 4.1864,
+      "step": 3057
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014202390766694146,
+      "loss": 4.6283,
+      "step": 3058
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014196207749381698,
+      "loss": 3.9792,
+      "step": 3059
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001419002473206925,
+      "loss": 3.9563,
+      "step": 3060
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.000141838417147568,
+      "loss": 3.91,
+      "step": 3061
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014177658697444353,
+      "loss": 4.0613,
+      "step": 3062
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014171475680131902,
+      "loss": 4.1274,
+      "step": 3063
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014165292662819454,
+      "loss": 4.3298,
+      "step": 3064
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014159109645507008,
+      "loss": 3.9149,
+      "step": 3065
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014152926628194557,
+      "loss": 4.1324,
+      "step": 3066
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014146743610882109,
+      "loss": 3.9867,
+      "step": 3067
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001414056059356966,
+      "loss": 4.2437,
+      "step": 3068
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014134377576257212,
+      "loss": 4.9571,
+      "step": 3069
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014128194558944764,
+      "loss": 4.2591,
+      "step": 3070
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014122011541632315,
+      "loss": 4.1704,
+      "step": 3071
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014115828524319867,
+      "loss": 4.0502,
+      "step": 3072
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014109645507007419,
+      "loss": 4.3646,
+      "step": 3073
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001410346248969497,
+      "loss": 4.128,
+      "step": 3074
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014097279472382522,
+      "loss": 4.1707,
+      "step": 3075
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014091096455070074,
+      "loss": 5.3233,
+      "step": 3076
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014084913437757625,
+      "loss": 4.0337,
+      "step": 3077
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014078730420445177,
+      "loss": 4.0906,
+      "step": 3078
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014072547403132726,
+      "loss": 4.5644,
+      "step": 3079
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001406636438582028,
+      "loss": 5.7245,
+      "step": 3080
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014060181368507832,
+      "loss": 3.8425,
+      "step": 3081
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001405399835119538,
+      "loss": 4.1779,
+      "step": 3082
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014047815333882935,
+      "loss": 4.1086,
+      "step": 3083
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014041632316570484,
+      "loss": 5.0802,
+      "step": 3084
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014035449299258036,
+      "loss": 4.76,
+      "step": 3085
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014029266281945587,
+      "loss": 4.5891,
+      "step": 3086
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001402308326463314,
+      "loss": 4.231,
+      "step": 3087
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001401690024732069,
+      "loss": 4.2571,
+      "step": 3088
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014010717230008242,
+      "loss": 4.155,
+      "step": 3089
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014004534212695794,
+      "loss": 3.8107,
+      "step": 3090
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00013998351195383346,
+      "loss": 4.3086,
+      "step": 3091
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00013992168178070897,
+      "loss": 4.1963,
+      "step": 3092
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001398598516075845,
+      "loss": 3.8885,
+      "step": 3093
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00013979802143446,
+      "loss": 4.3713,
+      "step": 3094
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00013973619126133552,
+      "loss": 4.3468,
+      "step": 3095
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013967436108821104,
+      "loss": 4.2309,
+      "step": 3096
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013961253091508656,
+      "loss": 4.4189,
+      "step": 3097
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013955070074196207,
+      "loss": 3.9388,
+      "step": 3098
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001394888705688376,
+      "loss": 4.1678,
+      "step": 3099
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013942704039571308,
+      "loss": 4.2535,
+      "step": 3100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013936521022258862,
+      "loss": 4.4182,
+      "step": 3101
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001393033800494641,
+      "loss": 4.0957,
+      "step": 3102
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013924154987633963,
+      "loss": 4.4586,
+      "step": 3103
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013917971970321517,
+      "loss": 4.0805,
+      "step": 3104
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013911788953009066,
+      "loss": 4.38,
+      "step": 3105
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001390560593569662,
+      "loss": 4.3997,
+      "step": 3106
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001389942291838417,
+      "loss": 4.1918,
+      "step": 3107
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001389323990107172,
+      "loss": 4.2121,
+      "step": 3108
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013887056883759273,
+      "loss": 4.0977,
+      "step": 3109
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013880873866446825,
+      "loss": 4.3208,
+      "step": 3110
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013874690849134376,
+      "loss": 4.0694,
+      "step": 3111
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013868507831821928,
+      "loss": 4.698,
+      "step": 3112
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001386232481450948,
+      "loss": 3.941,
+      "step": 3113
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001385614179719703,
+      "loss": 4.0718,
+      "step": 3114
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013849958779884583,
+      "loss": 4.282,
+      "step": 3115
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013843775762572134,
+      "loss": 4.0858,
+      "step": 3116
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013837592745259686,
+      "loss": 3.8843,
+      "step": 3117
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013831409727947235,
+      "loss": 3.9765,
+      "step": 3118
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001382522671063479,
+      "loss": 4.0493,
+      "step": 3119
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001381904369332234,
+      "loss": 4.0281,
+      "step": 3120
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013812860676009893,
+      "loss": 4.3233,
+      "step": 3121
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013806677658697444,
+      "loss": 3.9439,
+      "step": 3122
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013800494641384993,
+      "loss": 4.1182,
+      "step": 3123
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013794311624072548,
+      "loss": 4.1726,
+      "step": 3124
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013788128606760097,
+      "loss": 4.0285,
+      "step": 3125
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013781945589447648,
+      "loss": 3.9501,
+      "step": 3126
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000137757625721352,
+      "loss": 4.1901,
+      "step": 3127
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013769579554822752,
+      "loss": 3.9523,
+      "step": 3128
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013763396537510303,
+      "loss": 4.0084,
+      "step": 3129
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013757213520197855,
+      "loss": 3.8232,
+      "step": 3130
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013751030502885407,
+      "loss": 4.1239,
+      "step": 3131
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013744847485572958,
+      "loss": 5.4093,
+      "step": 3132
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001373866446826051,
+      "loss": 3.755,
+      "step": 3133
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013732481450948062,
+      "loss": 3.5529,
+      "step": 3134
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013726298433635613,
+      "loss": 4.9972,
+      "step": 3135
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013720115416323165,
+      "loss": 4.4023,
+      "step": 3136
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013713932399010717,
+      "loss": 4.4447,
+      "step": 3137
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013707749381698268,
+      "loss": 4.1634,
+      "step": 3138
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001370156636438582,
+      "loss": 4.3384,
+      "step": 3139
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013695383347073372,
+      "loss": 4.2049,
+      "step": 3140
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001368920032976092,
+      "loss": 4.569,
+      "step": 3141
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013683017312448475,
+      "loss": 4.7465,
+      "step": 3142
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013676834295136024,
+      "loss": 4.2833,
+      "step": 3143
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013670651277823575,
+      "loss": 4.3327,
+      "step": 3144
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001366446826051113,
+      "loss": 4.0117,
+      "step": 3145
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001365828524319868,
+      "loss": 3.8444,
+      "step": 3146
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001365210222588623,
+      "loss": 4.1323,
+      "step": 3147
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013645919208573782,
+      "loss": 3.9648,
+      "step": 3148
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013639736191261334,
+      "loss": 4.1223,
+      "step": 3149
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013633553173948885,
+      "loss": 4.3011,
+      "step": 3150
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013627370156636437,
+      "loss": 4.1134,
+      "step": 3151
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001362118713932399,
+      "loss": 4.5586,
+      "step": 3152
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001361500412201154,
+      "loss": 4.1857,
+      "step": 3153
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013608821104699092,
+      "loss": 4.4046,
+      "step": 3154
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013602638087386644,
+      "loss": 4.0756,
+      "step": 3155
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013596455070074195,
+      "loss": 3.7835,
+      "step": 3156
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013590272052761747,
+      "loss": 4.3052,
+      "step": 3157
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.000135840890354493,
+      "loss": 4.0762,
+      "step": 3158
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001357790601813685,
+      "loss": 3.9861,
+      "step": 3159
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013571723000824402,
+      "loss": 4.3756,
+      "step": 3160
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013565539983511954,
+      "loss": 4.2391,
+      "step": 3161
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013559356966199503,
+      "loss": 3.8784,
+      "step": 3162
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013553173948887057,
+      "loss": 3.9703,
+      "step": 3163
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013546990931574606,
+      "loss": 4.6538,
+      "step": 3164
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013540807914262158,
+      "loss": 4.2597,
+      "step": 3165
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001353462489694971,
+      "loss": 4.0625,
+      "step": 3166
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001352844187963726,
+      "loss": 4.3829,
+      "step": 3167
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013522258862324815,
+      "loss": 3.9121,
+      "step": 3168
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013516075845012364,
+      "loss": 3.915,
+      "step": 3169
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013509892827699916,
+      "loss": 3.9107,
+      "step": 3170
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013503709810387468,
+      "loss": 4.3006,
+      "step": 3171
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001349752679307502,
+      "loss": 3.8242,
+      "step": 3172
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001349134377576257,
+      "loss": 4.0845,
+      "step": 3173
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013485160758450123,
+      "loss": 4.0501,
+      "step": 3174
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013478977741137674,
+      "loss": 3.9311,
+      "step": 3175
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013472794723825226,
+      "loss": 4.29,
+      "step": 3176
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013466611706512778,
+      "loss": 4.2492,
+      "step": 3177
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001346042868920033,
+      "loss": 3.9831,
+      "step": 3178
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001345424567188788,
+      "loss": 4.444,
+      "step": 3179
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001344806265457543,
+      "loss": 4.2075,
+      "step": 3180
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013441879637262984,
+      "loss": 4.5413,
+      "step": 3181
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013435696619950533,
+      "loss": 3.3224,
+      "step": 3182
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013429513602638087,
+      "loss": 4.3161,
+      "step": 3183
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001342333058532564,
+      "loss": 4.0052,
+      "step": 3184
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013417147568013188,
+      "loss": 4.208,
+      "step": 3185
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013410964550700742,
+      "loss": 4.1778,
+      "step": 3186
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013404781533388291,
+      "loss": 4.1398,
+      "step": 3187
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013398598516075843,
+      "loss": 4.127,
+      "step": 3188
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013392415498763395,
+      "loss": 4.3973,
+      "step": 3189
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013386232481450946,
+      "loss": 4.355,
+      "step": 3190
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013380049464138498,
+      "loss": 4.2525,
+      "step": 3191
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001337386644682605,
+      "loss": 4.3409,
+      "step": 3192
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.000133676834295136,
+      "loss": 4.1669,
+      "step": 3193
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013361500412201153,
+      "loss": 4.0692,
+      "step": 3194
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013355317394888705,
+      "loss": 4.272,
+      "step": 3195
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013349134377576256,
+      "loss": 4.9176,
+      "step": 3196
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013342951360263808,
+      "loss": 3.9702,
+      "step": 3197
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001333676834295136,
+      "loss": 4.4383,
+      "step": 3198
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001333058532563891,
+      "loss": 4.4113,
+      "step": 3199
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013324402308326463,
+      "loss": 4.8919,
+      "step": 3200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013318219291014015,
+      "loss": 4.0214,
+      "step": 3201
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013312036273701566,
+      "loss": 4.04,
+      "step": 3202
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013305853256389115,
+      "loss": 4.1676,
+      "step": 3203
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001329967023907667,
+      "loss": 4.1267,
+      "step": 3204
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013293487221764219,
+      "loss": 4.211,
+      "step": 3205
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001328730420445177,
+      "loss": 4.1984,
+      "step": 3206
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013281121187139325,
+      "loss": 4.2898,
+      "step": 3207
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013274938169826873,
+      "loss": 4.1784,
+      "step": 3208
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013268755152514425,
+      "loss": 4.1755,
+      "step": 3209
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013262572135201977,
+      "loss": 4.8438,
+      "step": 3210
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013256389117889528,
+      "loss": 4.1534,
+      "step": 3211
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001325020610057708,
+      "loss": 3.9643,
+      "step": 3212
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013244023083264632,
+      "loss": 4.0056,
+      "step": 3213
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013237840065952183,
+      "loss": 4.5865,
+      "step": 3214
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013231657048639735,
+      "loss": 4.5989,
+      "step": 3215
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013225474031327287,
+      "loss": 4.2457,
+      "step": 3216
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013219291014014838,
+      "loss": 4.0354,
+      "step": 3217
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001321310799670239,
+      "loss": 4.0027,
+      "step": 3218
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013206924979389942,
+      "loss": 3.9559,
+      "step": 3219
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013200741962077493,
+      "loss": 3.951,
+      "step": 3220
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013194558944765042,
+      "loss": 4.0211,
+      "step": 3221
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013188375927452597,
+      "loss": 4.0364,
+      "step": 3222
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013182192910140148,
+      "loss": 3.8459,
+      "step": 3223
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013176009892827697,
+      "loss": 3.9646,
+      "step": 3224
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013169826875515252,
+      "loss": 4.1478,
+      "step": 3225
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.000131636438582028,
+      "loss": 4.3025,
+      "step": 3226
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013157460840890355,
+      "loss": 4.2187,
+      "step": 3227
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013151277823577904,
+      "loss": 3.8917,
+      "step": 3228
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013145094806265456,
+      "loss": 3.8391,
+      "step": 3229
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013138911788953007,
+      "loss": 3.8141,
+      "step": 3230
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001313272877164056,
+      "loss": 3.883,
+      "step": 3231
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001312654575432811,
+      "loss": 4.3223,
+      "step": 3232
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013120362737015662,
+      "loss": 4.4083,
+      "step": 3233
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013114179719703214,
+      "loss": 3.6711,
+      "step": 3234
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013107996702390766,
+      "loss": 4.7249,
+      "step": 3235
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013101813685078317,
+      "loss": 4.112,
+      "step": 3236
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001309563066776587,
+      "loss": 4.1084,
+      "step": 3237
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001308944765045342,
+      "loss": 4.2124,
+      "step": 3238
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013083264633140972,
+      "loss": 4.4603,
+      "step": 3239
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013077081615828524,
+      "loss": 4.0253,
+      "step": 3240
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013070898598516076,
+      "loss": 3.9617,
+      "step": 3241
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013064715581203624,
+      "loss": 4.3094,
+      "step": 3242
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001305853256389118,
+      "loss": 4.3093,
+      "step": 3243
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013052349546578728,
+      "loss": 4.5295,
+      "step": 3244
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013046166529266282,
+      "loss": 4.2364,
+      "step": 3245
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001303998351195383,
+      "loss": 4.1691,
+      "step": 3246
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013033800494641383,
+      "loss": 4.1204,
+      "step": 3247
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013027617477328937,
+      "loss": 4.0572,
+      "step": 3248
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013021434460016486,
+      "loss": 3.8138,
+      "step": 3249
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013015251442704038,
+      "loss": 4.0179,
+      "step": 3250
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001300906842539159,
+      "loss": 3.9825,
+      "step": 3251
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001300288540807914,
+      "loss": 4.1224,
+      "step": 3252
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00012996702390766693,
+      "loss": 4.2967,
+      "step": 3253
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00012990519373454244,
+      "loss": 4.3775,
+      "step": 3254
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00012984336356141796,
+      "loss": 4.2513,
+      "step": 3255
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012978153338829348,
+      "loss": 3.9416,
+      "step": 3256
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000129719703215169,
+      "loss": 3.9286,
+      "step": 3257
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001296578730420445,
+      "loss": 4.3107,
+      "step": 3258
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012959604286892003,
+      "loss": 4.2035,
+      "step": 3259
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012953421269579554,
+      "loss": 4.4403,
+      "step": 3260
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012947238252267106,
+      "loss": 4.301,
+      "step": 3261
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012941055234954658,
+      "loss": 4.6273,
+      "step": 3262
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001293487221764221,
+      "loss": 4.4354,
+      "step": 3263
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001292868920032976,
+      "loss": 3.9152,
+      "step": 3264
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001292250618301731,
+      "loss": 4.1847,
+      "step": 3265
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012916323165704864,
+      "loss": 3.8442,
+      "step": 3266
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012910140148392413,
+      "loss": 3.8223,
+      "step": 3267
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012903957131079965,
+      "loss": 3.9249,
+      "step": 3268
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012897774113767517,
+      "loss": 4.329,
+      "step": 3269
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012891591096455068,
+      "loss": 4.3822,
+      "step": 3270
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001288540807914262,
+      "loss": 4.1625,
+      "step": 3271
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012879225061830172,
+      "loss": 4.1492,
+      "step": 3272
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012873042044517723,
+      "loss": 4.0574,
+      "step": 3273
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012866859027205275,
+      "loss": 4.4321,
+      "step": 3274
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012860676009892826,
+      "loss": 4.2703,
+      "step": 3275
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012854492992580378,
+      "loss": 4.0525,
+      "step": 3276
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001284830997526793,
+      "loss": 3.6851,
+      "step": 3277
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012842126957955481,
+      "loss": 3.946,
+      "step": 3278
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012835943940643033,
+      "loss": 4.1382,
+      "step": 3279
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012829760923330585,
+      "loss": 3.6822,
+      "step": 3280
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012823577906018136,
+      "loss": 4.1678,
+      "step": 3281
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012817394888705688,
+      "loss": 3.5481,
+      "step": 3282
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012811211871393237,
+      "loss": 4.7968,
+      "step": 3283
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012805028854080791,
+      "loss": 3.3529,
+      "step": 3284
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001279884583676834,
+      "loss": 4.0186,
+      "step": 3285
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012792662819455892,
+      "loss": 4.2636,
+      "step": 3286
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012786479802143446,
+      "loss": 3.9528,
+      "step": 3287
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012780296784830995,
+      "loss": 3.6736,
+      "step": 3288
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001277411376751855,
+      "loss": 4.2806,
+      "step": 3289
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.000127679307502061,
+      "loss": 4.2875,
+      "step": 3290
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001276174773289365,
+      "loss": 4.8291,
+      "step": 3291
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012755564715581202,
+      "loss": 4.0898,
+      "step": 3292
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012749381698268754,
+      "loss": 4.2451,
+      "step": 3293
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012743198680956305,
+      "loss": 3.9998,
+      "step": 3294
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012737015663643857,
+      "loss": 4.1468,
+      "step": 3295
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012730832646331409,
+      "loss": 4.0707,
+      "step": 3296
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001272464962901896,
+      "loss": 4.1541,
+      "step": 3297
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012718466611706512,
+      "loss": 4.1158,
+      "step": 3298
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012712283594394064,
+      "loss": 4.3956,
+      "step": 3299
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012706100577081615,
+      "loss": 4.0425,
+      "step": 3300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012699917559769164,
+      "loss": 4.2032,
+      "step": 3301
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012693734542456719,
+      "loss": 3.9096,
+      "step": 3302
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001268755152514427,
+      "loss": 4.1699,
+      "step": 3303
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012681368507831822,
+      "loss": 4.1184,
+      "step": 3304
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012675185490519374,
+      "loss": 4.3026,
+      "step": 3305
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012669002473206922,
+      "loss": 4.0122,
+      "step": 3306
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012662819455894477,
+      "loss": 4.0079,
+      "step": 3307
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012656636438582026,
+      "loss": 4.0772,
+      "step": 3308
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012650453421269577,
+      "loss": 4.0205,
+      "step": 3309
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012644270403957132,
+      "loss": 4.297,
+      "step": 3310
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001263808738664468,
+      "loss": 4.0818,
+      "step": 3311
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012631904369332232,
+      "loss": 3.8514,
+      "step": 3312
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012625721352019784,
+      "loss": 3.9766,
+      "step": 3313
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012619538334707336,
+      "loss": 4.3845,
+      "step": 3314
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012613355317394887,
+      "loss": 4.0625,
+      "step": 3315
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001260717230008244,
+      "loss": 4.1819,
+      "step": 3316
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001260098928276999,
+      "loss": 3.9087,
+      "step": 3317
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012594806265457542,
+      "loss": 3.9908,
+      "step": 3318
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012588623248145094,
+      "loss": 3.895,
+      "step": 3319
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012582440230832646,
+      "loss": 3.8927,
+      "step": 3320
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012576257213520197,
+      "loss": 3.8207,
+      "step": 3321
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001257007419620775,
+      "loss": 4.0068,
+      "step": 3322
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.000125638911788953,
+      "loss": 4.6417,
+      "step": 3323
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001255770816158285,
+      "loss": 3.8094,
+      "step": 3324
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012551525144270404,
+      "loss": 3.9999,
+      "step": 3325
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012545342126957956,
+      "loss": 4.3992,
+      "step": 3326
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012539159109645505,
+      "loss": 4.2781,
+      "step": 3327
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001253297609233306,
+      "loss": 3.73,
+      "step": 3328
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012526793075020608,
+      "loss": 4.0383,
+      "step": 3329
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001252061005770816,
+      "loss": 3.9762,
+      "step": 3330
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001251442704039571,
+      "loss": 4.1993,
+      "step": 3331
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012508244023083263,
+      "loss": 4.0854,
+      "step": 3332
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012502061005770815,
+      "loss": 4.4156,
+      "step": 3333
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012495877988458366,
+      "loss": 3.7714,
+      "step": 3334
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012489694971145918,
+      "loss": 4.3203,
+      "step": 3335
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001248351195383347,
+      "loss": 4.3759,
+      "step": 3336
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001247732893652102,
+      "loss": 4.6485,
+      "step": 3337
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012471145919208573,
+      "loss": 4.1672,
+      "step": 3338
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012464962901896125,
+      "loss": 3.9131,
+      "step": 3339
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012458779884583676,
+      "loss": 4.1427,
+      "step": 3340
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012452596867271228,
+      "loss": 4.1806,
+      "step": 3341
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001244641384995878,
+      "loss": 4.0864,
+      "step": 3342
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001244023083264633,
+      "loss": 4.4541,
+      "step": 3343
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012434047815333883,
+      "loss": 3.8479,
+      "step": 3344
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012427864798021432,
+      "loss": 4.4354,
+      "step": 3345
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012421681780708986,
+      "loss": 3.8934,
+      "step": 3346
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012415498763396535,
+      "loss": 3.8975,
+      "step": 3347
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012409315746084087,
+      "loss": 4.2186,
+      "step": 3348
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012403132728771638,
+      "loss": 3.7763,
+      "step": 3349
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001239694971145919,
+      "loss": 4.1023,
+      "step": 3350
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012390766694146744,
+      "loss": 4.2418,
+      "step": 3351
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012384583676834293,
+      "loss": 4.122,
+      "step": 3352
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012378400659521845,
+      "loss": 3.919,
+      "step": 3353
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012372217642209397,
+      "loss": 4.0952,
+      "step": 3354
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012366034624896948,
+      "loss": 4.0951,
+      "step": 3355
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000123598516075845,
+      "loss": 3.8034,
+      "step": 3356
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012353668590272052,
+      "loss": 4.3046,
+      "step": 3357
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012347485572959603,
+      "loss": 3.7228,
+      "step": 3358
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012341302555647155,
+      "loss": 4.4463,
+      "step": 3359
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012335119538334707,
+      "loss": 4.1327,
+      "step": 3360
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012328936521022258,
+      "loss": 4.4574,
+      "step": 3361
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001232275350370981,
+      "loss": 4.2259,
+      "step": 3362
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001231657048639736,
+      "loss": 3.836,
+      "step": 3363
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012310387469084913,
+      "loss": 3.8181,
+      "step": 3364
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012304204451772462,
+      "loss": 4.5002,
+      "step": 3365
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012298021434460017,
+      "loss": 4.2206,
+      "step": 3366
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012291838417147568,
+      "loss": 4.1996,
+      "step": 3367
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012285655399835117,
+      "loss": 4.2524,
+      "step": 3368
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012279472382522672,
+      "loss": 4.2056,
+      "step": 3369
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001227328936521022,
+      "loss": 4.0508,
+      "step": 3370
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012267106347897772,
+      "loss": 4.1065,
+      "step": 3371
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012260923330585324,
+      "loss": 4.3521,
+      "step": 3372
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012254740313272875,
+      "loss": 4.0549,
+      "step": 3373
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012248557295960427,
+      "loss": 3.9454,
+      "step": 3374
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001224237427864798,
+      "loss": 4.5076,
+      "step": 3375
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001223619126133553,
+      "loss": 3.81,
+      "step": 3376
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012230008244023082,
+      "loss": 3.8516,
+      "step": 3377
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012223825226710634,
+      "loss": 4.0194,
+      "step": 3378
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012217642209398185,
+      "loss": 3.9065,
+      "step": 3379
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012211459192085737,
+      "loss": 4.0964,
+      "step": 3380
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012205276174773289,
+      "loss": 4.7263,
+      "step": 3381
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012199093157460839,
+      "loss": 4.0225,
+      "step": 3382
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012192910140148392,
+      "loss": 4.0479,
+      "step": 3383
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012186727122835942,
+      "loss": 3.7784,
+      "step": 3384
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012180544105523495,
+      "loss": 4.1097,
+      "step": 3385
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012174361088211046,
+      "loss": 4.1736,
+      "step": 3386
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012168178070898597,
+      "loss": 4.2292,
+      "step": 3387
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012161995053586148,
+      "loss": 4.2494,
+      "step": 3388
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.000121558120362737,
+      "loss": 3.9602,
+      "step": 3389
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012149629018961252,
+      "loss": 4.1911,
+      "step": 3390
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012143446001648804,
+      "loss": 4.165,
+      "step": 3391
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012137262984336356,
+      "loss": 4.1282,
+      "step": 3392
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012131079967023906,
+      "loss": 4.0757,
+      "step": 3393
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012124896949711459,
+      "loss": 4.3104,
+      "step": 3394
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012118713932399009,
+      "loss": 3.868,
+      "step": 3395
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012112530915086561,
+      "loss": 4.3142,
+      "step": 3396
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012106347897774114,
+      "loss": 4.0672,
+      "step": 3397
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012100164880461664,
+      "loss": 3.866,
+      "step": 3398
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012093981863149216,
+      "loss": 3.9813,
+      "step": 3399
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012087798845836768,
+      "loss": 4.1772,
+      "step": 3400
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012081615828524319,
+      "loss": 3.9875,
+      "step": 3401
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001207543281121187,
+      "loss": 3.7547,
+      "step": 3402
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012069249793899423,
+      "loss": 4.0252,
+      "step": 3403
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012063066776586973,
+      "loss": 3.9242,
+      "step": 3404
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012056883759274525,
+      "loss": 3.9934,
+      "step": 3405
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012050700741962078,
+      "loss": 4.0041,
+      "step": 3406
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012044517724649628,
+      "loss": 3.9213,
+      "step": 3407
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001203833470733718,
+      "loss": 4.6098,
+      "step": 3408
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012032151690024731,
+      "loss": 4.3141,
+      "step": 3409
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012025968672712283,
+      "loss": 4.1553,
+      "step": 3410
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012019785655399833,
+      "loss": 4.1137,
+      "step": 3411
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012013602638087386,
+      "loss": 4.0573,
+      "step": 3412
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012007419620774938,
+      "loss": 3.9277,
+      "step": 3413
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012001236603462488,
+      "loss": 4.0252,
+      "step": 3414
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00011995053586150041,
+      "loss": 4.2902,
+      "step": 3415
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00011988870568837591,
+      "loss": 3.9681,
+      "step": 3416
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011982687551525143,
+      "loss": 3.9802,
+      "step": 3417
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011976504534212695,
+      "loss": 4.0345,
+      "step": 3418
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011970321516900246,
+      "loss": 3.8485,
+      "step": 3419
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011964138499587797,
+      "loss": 3.9579,
+      "step": 3420
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001195795548227535,
+      "loss": 4.3862,
+      "step": 3421
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011951772464962901,
+      "loss": 3.9007,
+      "step": 3422
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011945589447650452,
+      "loss": 3.9484,
+      "step": 3423
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011939406430338005,
+      "loss": 3.9744,
+      "step": 3424
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011933223413025555,
+      "loss": 3.9154,
+      "step": 3425
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011927040395713107,
+      "loss": 3.7607,
+      "step": 3426
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011920857378400658,
+      "loss": 3.8091,
+      "step": 3427
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001191467436108821,
+      "loss": 3.8927,
+      "step": 3428
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011908491343775763,
+      "loss": 3.6307,
+      "step": 3429
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011902308326463313,
+      "loss": 4.1059,
+      "step": 3430
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011896125309150865,
+      "loss": 3.7422,
+      "step": 3431
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011889942291838415,
+      "loss": 3.9156,
+      "step": 3432
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011883759274525968,
+      "loss": 3.4254,
+      "step": 3433
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00011877576257213519,
+      "loss": 4.0777,
+      "step": 3434
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001187139323990107,
+      "loss": 4.0648,
+      "step": 3435
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011865210222588622,
+      "loss": 4.3676,
+      "step": 3436
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011859027205276174,
+      "loss": 4.015,
+      "step": 3437
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011852844187963727,
+      "loss": 4.0254,
+      "step": 3438
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011846661170651277,
+      "loss": 3.8896,
+      "step": 3439
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011840478153338828,
+      "loss": 4.3646,
+      "step": 3440
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011834295136026379,
+      "loss": 4.6024,
+      "step": 3441
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011828112118713932,
+      "loss": 4.1862,
+      "step": 3442
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011821929101401482,
+      "loss": 4.2096,
+      "step": 3443
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011815746084089035,
+      "loss": 4.0984,
+      "step": 3444
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011809563066776587,
+      "loss": 4.2078,
+      "step": 3445
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011803380049464137,
+      "loss": 4.0571,
+      "step": 3446
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001179719703215169,
+      "loss": 4.445,
+      "step": 3447
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001179101401483924,
+      "loss": 3.9143,
+      "step": 3448
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011784830997526792,
+      "loss": 3.9332,
+      "step": 3449
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011778647980214342,
+      "loss": 4.2979,
+      "step": 3450
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011772464962901895,
+      "loss": 4.0441,
+      "step": 3451
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011766281945589446,
+      "loss": 4.162,
+      "step": 3452
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011760098928276999,
+      "loss": 3.6389,
+      "step": 3453
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001175391591096455,
+      "loss": 4.0046,
+      "step": 3454
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.000117477328936521,
+      "loss": 3.8239,
+      "step": 3455
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011741549876339654,
+      "loss": 3.7678,
+      "step": 3456
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011735366859027204,
+      "loss": 4.0622,
+      "step": 3457
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011729183841714756,
+      "loss": 4.1194,
+      "step": 3458
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011723000824402306,
+      "loss": 3.9353,
+      "step": 3459
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011716817807089859,
+      "loss": 3.9988,
+      "step": 3460
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001171063478977741,
+      "loss": 3.924,
+      "step": 3461
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011704451772464962,
+      "loss": 3.7466,
+      "step": 3462
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011698268755152514,
+      "loss": 4.1753,
+      "step": 3463
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011692085737840064,
+      "loss": 3.8821,
+      "step": 3464
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011685902720527617,
+      "loss": 4.0039,
+      "step": 3465
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011679719703215168,
+      "loss": 4.1444,
+      "step": 3466
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011673536685902719,
+      "loss": 3.9271,
+      "step": 3467
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011667353668590271,
+      "loss": 3.5516,
+      "step": 3468
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011661170651277823,
+      "loss": 3.7136,
+      "step": 3469
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011654987633965374,
+      "loss": 4.5676,
+      "step": 3470
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011648804616652926,
+      "loss": 3.8254,
+      "step": 3471
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011642621599340477,
+      "loss": 3.764,
+      "step": 3472
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011636438582028028,
+      "loss": 4.512,
+      "step": 3473
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011630255564715581,
+      "loss": 4.3398,
+      "step": 3474
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011624072547403131,
+      "loss": 3.824,
+      "step": 3475
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011617889530090683,
+      "loss": 3.6856,
+      "step": 3476
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011611706512778236,
+      "loss": 4.1883,
+      "step": 3477
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011605523495465786,
+      "loss": 3.857,
+      "step": 3478
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011599340478153338,
+      "loss": 3.925,
+      "step": 3479
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001159315746084089,
+      "loss": 3.9467,
+      "step": 3480
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011586974443528441,
+      "loss": 4.2039,
+      "step": 3481
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011580791426215991,
+      "loss": 3.8095,
+      "step": 3482
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011574608408903544,
+      "loss": 4.0144,
+      "step": 3483
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011568425391591096,
+      "loss": 3.8046,
+      "step": 3484
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011562242374278646,
+      "loss": 4.1717,
+      "step": 3485
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.000115560593569662,
+      "loss": 3.9725,
+      "step": 3486
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001154987633965375,
+      "loss": 4.0654,
+      "step": 3487
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011543693322341301,
+      "loss": 4.0673,
+      "step": 3488
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011537510305028853,
+      "loss": 4.025,
+      "step": 3489
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011531327287716405,
+      "loss": 4.0851,
+      "step": 3490
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011525144270403955,
+      "loss": 4.056,
+      "step": 3491
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011518961253091508,
+      "loss": 4.3397,
+      "step": 3492
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001151277823577906,
+      "loss": 4.0819,
+      "step": 3493
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001150659521846661,
+      "loss": 4.1409,
+      "step": 3494
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011500412201154163,
+      "loss": 4.1023,
+      "step": 3495
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011494229183841713,
+      "loss": 4.318,
+      "step": 3496
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011488046166529265,
+      "loss": 4.0609,
+      "step": 3497
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011481863149216817,
+      "loss": 3.8442,
+      "step": 3498
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011475680131904368,
+      "loss": 3.9564,
+      "step": 3499
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011469497114591921,
+      "loss": 3.9193,
+      "step": 3500
+    },
+    {
+      "epoch": 1.96,
+      "eval_loss": 4.593017578125,
+      "eval_runtime": 951.667,
+      "eval_samples_per_second": 2.776,
+      "eval_steps_per_second": 0.348,
+      "eval_wer": 1.8681277270924237,
+      "step": 3500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011463314097279472,
+      "loss": 4.1359,
+      "step": 3501
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011457131079967023,
+      "loss": 4.7496,
+      "step": 3502
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011450948062654573,
+      "loss": 3.8741,
+      "step": 3503
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011444765045342127,
+      "loss": 3.8178,
+      "step": 3504
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011438582028029677,
+      "loss": 3.6953,
+      "step": 3505
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001143239901071723,
+      "loss": 4.0625,
+      "step": 3506
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001142621599340478,
+      "loss": 3.855,
+      "step": 3507
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011420032976092332,
+      "loss": 4.2252,
+      "step": 3508
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011413849958779885,
+      "loss": 4.2059,
+      "step": 3509
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011407666941467435,
+      "loss": 4.1987,
+      "step": 3510
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011401483924154987,
+      "loss": 3.6897,
+      "step": 3511
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011395300906842537,
+      "loss": 4.0671,
+      "step": 3512
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001138911788953009,
+      "loss": 3.9004,
+      "step": 3513
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001138293487221764,
+      "loss": 3.7086,
+      "step": 3514
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011376751854905193,
+      "loss": 4.1116,
+      "step": 3515
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011370568837592745,
+      "loss": 3.7376,
+      "step": 3516
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011364385820280295,
+      "loss": 3.7459,
+      "step": 3517
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011358202802967848,
+      "loss": 4.11,
+      "step": 3518
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011352019785655399,
+      "loss": 3.7787,
+      "step": 3519
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001134583676834295,
+      "loss": 3.8327,
+      "step": 3520
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011339653751030502,
+      "loss": 3.7858,
+      "step": 3521
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011333470733718054,
+      "loss": 3.9077,
+      "step": 3522
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011327287716405604,
+      "loss": 4.1213,
+      "step": 3523
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011321104699093157,
+      "loss": 4.0555,
+      "step": 3524
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011314921681780709,
+      "loss": 3.8582,
+      "step": 3525
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011308738664468259,
+      "loss": 4.1258,
+      "step": 3526
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011302555647155812,
+      "loss": 4.1445,
+      "step": 3527
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011296372629843362,
+      "loss": 4.2387,
+      "step": 3528
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011290189612530914,
+      "loss": 3.7353,
+      "step": 3529
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011284006595218466,
+      "loss": 4.0043,
+      "step": 3530
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011277823577906017,
+      "loss": 3.8491,
+      "step": 3531
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011271640560593569,
+      "loss": 3.8242,
+      "step": 3532
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001126545754328112,
+      "loss": 4.1889,
+      "step": 3533
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011259274525968672,
+      "loss": 3.373,
+      "step": 3534
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011253091508656223,
+      "loss": 3.9126,
+      "step": 3535
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011246908491343776,
+      "loss": 4.0806,
+      "step": 3536
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011240725474031326,
+      "loss": 4.0515,
+      "step": 3537
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011234542456718877,
+      "loss": 3.9649,
+      "step": 3538
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011228359439406429,
+      "loss": 3.8851,
+      "step": 3539
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011222176422093981,
+      "loss": 4.1861,
+      "step": 3540
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011215993404781532,
+      "loss": 4.2722,
+      "step": 3541
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011209810387469084,
+      "loss": 3.9182,
+      "step": 3542
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011203627370156636,
+      "loss": 4.2494,
+      "step": 3543
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011197444352844186,
+      "loss": 3.8839,
+      "step": 3544
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011191261335531739,
+      "loss": 3.989,
+      "step": 3545
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001118507831821929,
+      "loss": 4.0281,
+      "step": 3546
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011178895300906841,
+      "loss": 4.2588,
+      "step": 3547
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011172712283594394,
+      "loss": 4.1666,
+      "step": 3548
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011166529266281944,
+      "loss": 3.8474,
+      "step": 3549
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011160346248969496,
+      "loss": 3.8585,
+      "step": 3550
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011154163231657048,
+      "loss": 4.2128,
+      "step": 3551
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.000111479802143446,
+      "loss": 3.7825,
+      "step": 3552
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001114179719703215,
+      "loss": 4.0503,
+      "step": 3553
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011135614179719703,
+      "loss": 3.9775,
+      "step": 3554
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011129431162407253,
+      "loss": 4.3165,
+      "step": 3555
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011123248145094805,
+      "loss": 4.3706,
+      "step": 3556
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011117065127782358,
+      "loss": 3.9327,
+      "step": 3557
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011110882110469908,
+      "loss": 4.075,
+      "step": 3558
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011104699093157461,
+      "loss": 3.9738,
+      "step": 3559
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011098516075845011,
+      "loss": 3.5306,
+      "step": 3560
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011092333058532563,
+      "loss": 3.7208,
+      "step": 3561
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011086150041220113,
+      "loss": 3.9235,
+      "step": 3562
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011079967023907666,
+      "loss": 3.861,
+      "step": 3563
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011073784006595218,
+      "loss": 4.2272,
+      "step": 3564
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011067600989282768,
+      "loss": 3.9962,
+      "step": 3565
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011061417971970321,
+      "loss": 4.0947,
+      "step": 3566
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011055234954657872,
+      "loss": 4.0791,
+      "step": 3567
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011049051937345425,
+      "loss": 2.7234,
+      "step": 3568
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011042868920032975,
+      "loss": 3.5221,
+      "step": 3569
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011036685902720526,
+      "loss": 3.514,
+      "step": 3570
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011030502885408077,
+      "loss": 3.6217,
+      "step": 3571
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001102431986809563,
+      "loss": 3.4511,
+      "step": 3572
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011018136850783181,
+      "loss": 3.5199,
+      "step": 3573
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011011953833470733,
+      "loss": 3.335,
+      "step": 3574
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011005770816158285,
+      "loss": 3.5533,
+      "step": 3575
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010999587798845835,
+      "loss": 3.4697,
+      "step": 3576
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010993404781533388,
+      "loss": 3.6652,
+      "step": 3577
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010987221764220938,
+      "loss": 3.4757,
+      "step": 3578
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001098103874690849,
+      "loss": 3.5391,
+      "step": 3579
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010974855729596043,
+      "loss": 3.3846,
+      "step": 3580
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010968672712283593,
+      "loss": 3.6624,
+      "step": 3581
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010962489694971145,
+      "loss": 3.5137,
+      "step": 3582
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010956306677658697,
+      "loss": 3.4183,
+      "step": 3583
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010950123660346248,
+      "loss": 3.5575,
+      "step": 3584
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010943940643033799,
+      "loss": 3.631,
+      "step": 3585
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010937757625721352,
+      "loss": 3.8031,
+      "step": 3586
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010931574608408902,
+      "loss": 3.4168,
+      "step": 3587
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010925391591096454,
+      "loss": 3.4631,
+      "step": 3588
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010919208573784007,
+      "loss": 3.4396,
+      "step": 3589
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010913025556471557,
+      "loss": 3.2865,
+      "step": 3590
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010906842539159109,
+      "loss": 3.5926,
+      "step": 3591
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001090065952184666,
+      "loss": 3.7331,
+      "step": 3592
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010894476504534212,
+      "loss": 3.3497,
+      "step": 3593
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00010888293487221762,
+      "loss": 3.418,
+      "step": 3594
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010882110469909315,
+      "loss": 3.4401,
+      "step": 3595
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010875927452596867,
+      "loss": 3.3802,
+      "step": 3596
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010869744435284417,
+      "loss": 3.4148,
+      "step": 3597
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001086356141797197,
+      "loss": 3.4728,
+      "step": 3598
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001085737840065952,
+      "loss": 3.6015,
+      "step": 3599
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010851195383347072,
+      "loss": 3.4443,
+      "step": 3600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010845012366034624,
+      "loss": 3.3035,
+      "step": 3601
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010838829348722176,
+      "loss": 3.1783,
+      "step": 3602
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010832646331409727,
+      "loss": 3.0683,
+      "step": 3603
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010826463314097279,
+      "loss": 3.3679,
+      "step": 3604
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001082028029678483,
+      "loss": 3.523,
+      "step": 3605
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010814097279472381,
+      "loss": 3.2964,
+      "step": 3606
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010807914262159934,
+      "loss": 2.983,
+      "step": 3607
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010801731244847484,
+      "loss": 3.6521,
+      "step": 3608
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010795548227535036,
+      "loss": 3.1603,
+      "step": 3609
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010789365210222587,
+      "loss": 3.5163,
+      "step": 3610
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010783182192910139,
+      "loss": 2.9512,
+      "step": 3611
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010776999175597692,
+      "loss": 2.7894,
+      "step": 3612
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010770816158285242,
+      "loss": 2.9299,
+      "step": 3613
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010764633140972794,
+      "loss": 3.0387,
+      "step": 3614
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010758450123660344,
+      "loss": 2.5547,
+      "step": 3615
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010752267106347897,
+      "loss": 2.8206,
+      "step": 3616
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010746084089035448,
+      "loss": 2.8346,
+      "step": 3617
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010739901071723,
+      "loss": 3.0571,
+      "step": 3618
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010733718054410552,
+      "loss": 3.6649,
+      "step": 3619
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010727535037098103,
+      "loss": 3.2561,
+      "step": 3620
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010721352019785656,
+      "loss": 3.7564,
+      "step": 3621
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010715169002473206,
+      "loss": 3.6944,
+      "step": 3622
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010708985985160758,
+      "loss": 3.337,
+      "step": 3623
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010702802967848308,
+      "loss": 3.5699,
+      "step": 3624
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010696619950535861,
+      "loss": 3.8268,
+      "step": 3625
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010690436933223411,
+      "loss": 3.6919,
+      "step": 3626
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010684253915910963,
+      "loss": 3.4895,
+      "step": 3627
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010678070898598516,
+      "loss": 3.8211,
+      "step": 3628
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010671887881286066,
+      "loss": 3.633,
+      "step": 3629
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010665704863973619,
+      "loss": 3.4408,
+      "step": 3630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001065952184666117,
+      "loss": 3.3313,
+      "step": 3631
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010653338829348721,
+      "loss": 3.4673,
+      "step": 3632
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010647155812036272,
+      "loss": 3.9889,
+      "step": 3633
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010640972794723825,
+      "loss": 3.4183,
+      "step": 3634
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010634789777411376,
+      "loss": 3.4539,
+      "step": 3635
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010628606760098928,
+      "loss": 3.568,
+      "step": 3636
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001062242374278648,
+      "loss": 3.6576,
+      "step": 3637
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001061624072547403,
+      "loss": 3.4166,
+      "step": 3638
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010610057708161583,
+      "loss": 3.3949,
+      "step": 3639
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010603874690849133,
+      "loss": 3.238,
+      "step": 3640
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010597691673536685,
+      "loss": 3.1771,
+      "step": 3641
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010591508656224235,
+      "loss": 3.2838,
+      "step": 3642
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010585325638911788,
+      "loss": 3.4839,
+      "step": 3643
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001057914262159934,
+      "loss": 3.5221,
+      "step": 3644
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010572959604286891,
+      "loss": 3.0769,
+      "step": 3645
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010566776586974443,
+      "loss": 3.362,
+      "step": 3646
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010560593569661993,
+      "loss": 3.5061,
+      "step": 3647
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010554410552349546,
+      "loss": 3.2249,
+      "step": 3648
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010548227535037097,
+      "loss": 3.4412,
+      "step": 3649
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010542044517724648,
+      "loss": 3.1657,
+      "step": 3650
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010535861500412201,
+      "loss": 3.7311,
+      "step": 3651
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010529678483099752,
+      "loss": 3.4843,
+      "step": 3652
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010523495465787303,
+      "loss": 3.4826,
+      "step": 3653
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010517312448474855,
+      "loss": 3.0393,
+      "step": 3654
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010511129431162407,
+      "loss": 3.1854,
+      "step": 3655
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010504946413849957,
+      "loss": 3.2813,
+      "step": 3656
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001049876339653751,
+      "loss": 3.4058,
+      "step": 3657
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001049258037922506,
+      "loss": 3.0214,
+      "step": 3658
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010486397361912612,
+      "loss": 3.2857,
+      "step": 3659
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010480214344600165,
+      "loss": 3.4142,
+      "step": 3660
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010474031327287715,
+      "loss": 3.3393,
+      "step": 3661
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010467848309975267,
+      "loss": 3.1048,
+      "step": 3662
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010461665292662819,
+      "loss": 2.969,
+      "step": 3663
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001045548227535037,
+      "loss": 2.9614,
+      "step": 3664
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001044929925803792,
+      "loss": 3.3276,
+      "step": 3665
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010443116240725474,
+      "loss": 2.9292,
+      "step": 3666
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010436933223413025,
+      "loss": 2.9673,
+      "step": 3667
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010430750206100575,
+      "loss": 2.6155,
+      "step": 3668
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010424567188788129,
+      "loss": 3.6671,
+      "step": 3669
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010418384171475679,
+      "loss": 3.8733,
+      "step": 3670
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001041220115416323,
+      "loss": 3.2106,
+      "step": 3671
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010406018136850782,
+      "loss": 3.3048,
+      "step": 3672
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010399835119538334,
+      "loss": 3.486,
+      "step": 3673
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010393652102225884,
+      "loss": 3.3382,
+      "step": 3674
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010387469084913437,
+      "loss": 3.604,
+      "step": 3675
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010381286067600989,
+      "loss": 3.4762,
+      "step": 3676
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010375103050288539,
+      "loss": 3.4595,
+      "step": 3677
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010368920032976092,
+      "loss": 3.383,
+      "step": 3678
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010362737015663642,
+      "loss": 3.4247,
+      "step": 3679
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010356553998351194,
+      "loss": 3.6079,
+      "step": 3680
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010350370981038746,
+      "loss": 3.599,
+      "step": 3681
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010344187963726297,
+      "loss": 3.4048,
+      "step": 3682
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001033800494641385,
+      "loss": 3.4664,
+      "step": 3683
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.000103318219291014,
+      "loss": 3.5123,
+      "step": 3684
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010325638911788952,
+      "loss": 3.5518,
+      "step": 3685
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010319455894476503,
+      "loss": 3.6537,
+      "step": 3686
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010313272877164056,
+      "loss": 3.5728,
+      "step": 3687
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010307089859851606,
+      "loss": 3.8393,
+      "step": 3688
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010300906842539159,
+      "loss": 3.3941,
+      "step": 3689
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010294723825226709,
+      "loss": 3.4599,
+      "step": 3690
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010288540807914261,
+      "loss": 3.188,
+      "step": 3691
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010282357790601814,
+      "loss": 3.4735,
+      "step": 3692
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010276174773289364,
+      "loss": 3.1597,
+      "step": 3693
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010269991755976916,
+      "loss": 3.3161,
+      "step": 3694
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010263808738664466,
+      "loss": 3.5351,
+      "step": 3695
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010257625721352019,
+      "loss": 3.1906,
+      "step": 3696
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001025144270403957,
+      "loss": 3.5517,
+      "step": 3697
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010245259686727123,
+      "loss": 3.3077,
+      "step": 3698
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010239076669414674,
+      "loss": 3.3774,
+      "step": 3699
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010232893652102224,
+      "loss": 3.215,
+      "step": 3700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010226710634789778,
+      "loss": 3.435,
+      "step": 3701
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010220527617477328,
+      "loss": 3.3308,
+      "step": 3702
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001021434460016488,
+      "loss": 3.3117,
+      "step": 3703
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010208161582852431,
+      "loss": 3.3504,
+      "step": 3704
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010201978565539983,
+      "loss": 3.2041,
+      "step": 3705
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010195795548227534,
+      "loss": 3.6074,
+      "step": 3706
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010189612530915086,
+      "loss": 3.2592,
+      "step": 3707
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010183429513602638,
+      "loss": 3.2945,
+      "step": 3708
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010177246496290188,
+      "loss": 3.1592,
+      "step": 3709
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010171063478977741,
+      "loss": 3.3318,
+      "step": 3710
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010164880461665291,
+      "loss": 3.3356,
+      "step": 3711
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010158697444352843,
+      "loss": 3.0476,
+      "step": 3712
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010152514427040395,
+      "loss": 3.2715,
+      "step": 3713
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010146331409727946,
+      "loss": 3.2414,
+      "step": 3714
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010140148392415498,
+      "loss": 3.0203,
+      "step": 3715
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001013396537510305,
+      "loss": 2.7583,
+      "step": 3716
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010127782357790601,
+      "loss": 3.2491,
+      "step": 3717
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010121599340478152,
+      "loss": 2.6253,
+      "step": 3718
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010115416323165705,
+      "loss": 3.7084,
+      "step": 3719
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010109233305853255,
+      "loss": 3.6996,
+      "step": 3720
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010103050288540807,
+      "loss": 3.3502,
+      "step": 3721
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001009686727122836,
+      "loss": 3.5292,
+      "step": 3722
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001009068425391591,
+      "loss": 3.9092,
+      "step": 3723
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010084501236603462,
+      "loss": 3.4511,
+      "step": 3724
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010078318219291013,
+      "loss": 3.1355,
+      "step": 3725
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010072135201978565,
+      "loss": 3.316,
+      "step": 3726
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010065952184666115,
+      "loss": 3.185,
+      "step": 3727
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010059769167353668,
+      "loss": 3.4364,
+      "step": 3728
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010053586150041219,
+      "loss": 3.4842,
+      "step": 3729
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001004740313272877,
+      "loss": 3.3944,
+      "step": 3730
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010041220115416323,
+      "loss": 3.1389,
+      "step": 3731
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010035037098103874,
+      "loss": 3.3865,
+      "step": 3732
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010028854080791425,
+      "loss": 3.4244,
+      "step": 3733
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010022671063478977,
+      "loss": 3.8214,
+      "step": 3734
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010016488046166528,
+      "loss": 3.4013,
+      "step": 3735
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010010305028854079,
+      "loss": 3.4034,
+      "step": 3736
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010004122011541632,
+      "loss": 3.3569,
+      "step": 3737
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.997938994229183e-05,
+      "loss": 3.402,
+      "step": 3738
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.991755976916734e-05,
+      "loss": 3.5616,
+      "step": 3739
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.985572959604287e-05,
+      "loss": 3.5132,
+      "step": 3740
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.979389942291837e-05,
+      "loss": 3.2356,
+      "step": 3741
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.97320692497939e-05,
+      "loss": 3.3188,
+      "step": 3742
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.96702390766694e-05,
+      "loss": 3.367,
+      "step": 3743
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.960840890354492e-05,
+      "loss": 3.4863,
+      "step": 3744
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.954657873042042e-05,
+      "loss": 3.3745,
+      "step": 3745
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.948474855729595e-05,
+      "loss": 3.3875,
+      "step": 3746
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.942291838417147e-05,
+      "loss": 3.1298,
+      "step": 3747
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.936108821104697e-05,
+      "loss": 3.3834,
+      "step": 3748
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.92992580379225e-05,
+      "loss": 3.3731,
+      "step": 3749
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.9237427864798e-05,
+      "loss": 3.6494,
+      "step": 3750
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.917559769167354e-05,
+      "loss": 3.3987,
+      "step": 3751
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.911376751854904e-05,
+      "loss": 3.4053,
+      "step": 3752
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.905193734542456e-05,
+      "loss": 3.1273,
+      "step": 3753
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.899010717230009e-05,
+      "loss": 3.1355,
+      "step": 3754
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.892827699917559e-05,
+      "loss": 3.4569,
+      "step": 3755
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.88664468260511e-05,
+      "loss": 3.2717,
+      "step": 3756
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.880461665292662e-05,
+      "loss": 3.2553,
+      "step": 3757
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.874278647980214e-05,
+      "loss": 3.3744,
+      "step": 3758
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.868095630667764e-05,
+      "loss": 3.3911,
+      "step": 3759
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.861912613355317e-05,
+      "loss": 3.406,
+      "step": 3760
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.855729596042868e-05,
+      "loss": 3.1308,
+      "step": 3761
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.849546578730419e-05,
+      "loss": 3.1935,
+      "step": 3762
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.843363561417972e-05,
+      "loss": 3.1631,
+      "step": 3763
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.837180544105523e-05,
+      "loss": 3.1514,
+      "step": 3764
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.830997526793074e-05,
+      "loss": 2.8838,
+      "step": 3765
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.824814509480626e-05,
+      "loss": 2.9769,
+      "step": 3766
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.818631492168177e-05,
+      "loss": 3.1268,
+      "step": 3767
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.812448474855728e-05,
+      "loss": 2.7615,
+      "step": 3768
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.806265457543281e-05,
+      "loss": 3.6558,
+      "step": 3769
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.800082440230832e-05,
+      "loss": 3.3842,
+      "step": 3770
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.793899422918383e-05,
+      "loss": 3.4939,
+      "step": 3771
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.787716405605936e-05,
+      "loss": 3.6001,
+      "step": 3772
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.781533388293486e-05,
+      "loss": 3.4167,
+      "step": 3773
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.775350370981038e-05,
+      "loss": 3.1897,
+      "step": 3774
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.76916735366859e-05,
+      "loss": 3.3449,
+      "step": 3775
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.762984336356141e-05,
+      "loss": 3.5358,
+      "step": 3776
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.756801319043691e-05,
+      "loss": 3.6378,
+      "step": 3777
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.750618301731244e-05,
+      "loss": 3.5418,
+      "step": 3778
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.744435284418796e-05,
+      "loss": 3.1609,
+      "step": 3779
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.738252267106346e-05,
+      "loss": 3.2575,
+      "step": 3780
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.7320692497939e-05,
+      "loss": 3.4147,
+      "step": 3781
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.72588623248145e-05,
+      "loss": 3.9231,
+      "step": 3782
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.719703215169001e-05,
+      "loss": 3.1309,
+      "step": 3783
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.713520197856553e-05,
+      "loss": 3.3563,
+      "step": 3784
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.707337180544105e-05,
+      "loss": 3.4876,
+      "step": 3785
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.701154163231656e-05,
+      "loss": 3.2247,
+      "step": 3786
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.694971145919208e-05,
+      "loss": 3.218,
+      "step": 3787
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.68878812860676e-05,
+      "loss": 3.4807,
+      "step": 3788
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.68260511129431e-05,
+      "loss": 3.1692,
+      "step": 3789
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.676422093981863e-05,
+      "loss": 3.7293,
+      "step": 3790
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.670239076669413e-05,
+      "loss": 3.5361,
+      "step": 3791
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.664056059356965e-05,
+      "loss": 3.7783,
+      "step": 3792
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.657873042044517e-05,
+      "loss": 3.3568,
+      "step": 3793
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.651690024732068e-05,
+      "loss": 3.3104,
+      "step": 3794
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.645507007419621e-05,
+      "loss": 3.134,
+      "step": 3795
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.639323990107172e-05,
+      "loss": 3.6999,
+      "step": 3796
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.633140972794723e-05,
+      "loss": 3.3946,
+      "step": 3797
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.626957955482273e-05,
+      "loss": 3.5055,
+      "step": 3798
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.620774938169827e-05,
+      "loss": 3.4309,
+      "step": 3799
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.614591920857377e-05,
+      "loss": 3.4388,
+      "step": 3800
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.608408903544928e-05,
+      "loss": 3.1032,
+      "step": 3801
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.602225886232481e-05,
+      "loss": 3.5028,
+      "step": 3802
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.596042868920032e-05,
+      "loss": 3.1163,
+      "step": 3803
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.589859851607585e-05,
+      "loss": 3.3489,
+      "step": 3804
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.583676834295135e-05,
+      "loss": 3.0817,
+      "step": 3805
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.577493816982687e-05,
+      "loss": 3.2198,
+      "step": 3806
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.571310799670237e-05,
+      "loss": 2.8743,
+      "step": 3807
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.56512778235779e-05,
+      "loss": 2.9923,
+      "step": 3808
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.55894476504534e-05,
+      "loss": 3.1665,
+      "step": 3809
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.552761747732892e-05,
+      "loss": 3.1293,
+      "step": 3810
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.546578730420445e-05,
+      "loss": 3.1513,
+      "step": 3811
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.540395713107995e-05,
+      "loss": 2.6133,
+      "step": 3812
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.534212695795548e-05,
+      "loss": 2.9323,
+      "step": 3813
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.528029678483099e-05,
+      "loss": 2.7997,
+      "step": 3814
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.52184666117065e-05,
+      "loss": 2.8064,
+      "step": 3815
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.5156636438582e-05,
+      "loss": 2.5251,
+      "step": 3816
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.509480626545754e-05,
+      "loss": 2.4036,
+      "step": 3817
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.503297609233305e-05,
+      "loss": 2.9981,
+      "step": 3818
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.497114591920857e-05,
+      "loss": 3.5825,
+      "step": 3819
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.490931574608409e-05,
+      "loss": 3.5084,
+      "step": 3820
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.484748557295959e-05,
+      "loss": 3.6092,
+      "step": 3821
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.478565539983512e-05,
+      "loss": 3.4262,
+      "step": 3822
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.472382522671062e-05,
+      "loss": 3.6751,
+      "step": 3823
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.466199505358614e-05,
+      "loss": 3.4917,
+      "step": 3824
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.460016488046167e-05,
+      "loss": 3.5671,
+      "step": 3825
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.453833470733717e-05,
+      "loss": 3.3853,
+      "step": 3826
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.447650453421269e-05,
+      "loss": 3.8024,
+      "step": 3827
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.44146743610882e-05,
+      "loss": 3.3531,
+      "step": 3828
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.435284418796372e-05,
+      "loss": 3.4505,
+      "step": 3829
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.429101401483923e-05,
+      "loss": 3.4965,
+      "step": 3830
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.422918384171476e-05,
+      "loss": 3.5829,
+      "step": 3831
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.416735366859026e-05,
+      "loss": 3.2413,
+      "step": 3832
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.410552349546577e-05,
+      "loss": 3.3941,
+      "step": 3833
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.40436933223413e-05,
+      "loss": 3.307,
+      "step": 3834
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.398186314921681e-05,
+      "loss": 3.384,
+      "step": 3835
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.392003297609232e-05,
+      "loss": 3.2752,
+      "step": 3836
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.385820280296784e-05,
+      "loss": 3.2444,
+      "step": 3837
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.379637262984336e-05,
+      "loss": 3.3326,
+      "step": 3838
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.373454245671886e-05,
+      "loss": 3.1552,
+      "step": 3839
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.367271228359439e-05,
+      "loss": 3.4749,
+      "step": 3840
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.361088211046991e-05,
+      "loss": 3.2638,
+      "step": 3841
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.354905193734541e-05,
+      "loss": 3.103,
+      "step": 3842
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.348722176422094e-05,
+      "loss": 3.3682,
+      "step": 3843
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.342539159109644e-05,
+      "loss": 3.1935,
+      "step": 3844
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.336356141797196e-05,
+      "loss": 3.1564,
+      "step": 3845
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.330173124484748e-05,
+      "loss": 3.0756,
+      "step": 3846
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.3239901071723e-05,
+      "loss": 3.4468,
+      "step": 3847
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.31780708985985e-05,
+      "loss": 3.3947,
+      "step": 3848
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.311624072547403e-05,
+      "loss": 3.1995,
+      "step": 3849
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.305441055234954e-05,
+      "loss": 3.3325,
+      "step": 3850
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.299258037922505e-05,
+      "loss": 3.1295,
+      "step": 3851
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.293075020610058e-05,
+      "loss": 3.0761,
+      "step": 3852
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.286892003297608e-05,
+      "loss": 3.0754,
+      "step": 3853
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.28070898598516e-05,
+      "loss": 3.015,
+      "step": 3854
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.274525968672711e-05,
+      "loss": 3.3124,
+      "step": 3855
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.268342951360263e-05,
+      "loss": 3.1814,
+      "step": 3856
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.262159934047816e-05,
+      "loss": 3.1226,
+      "step": 3857
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.255976916735366e-05,
+      "loss": 3.0389,
+      "step": 3858
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.249793899422918e-05,
+      "loss": 3.3548,
+      "step": 3859
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.243610882110468e-05,
+      "loss": 3.0413,
+      "step": 3860
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.237427864798021e-05,
+      "loss": 2.8998,
+      "step": 3861
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.231244847485572e-05,
+      "loss": 2.6159,
+      "step": 3862
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.225061830173123e-05,
+      "loss": 2.6441,
+      "step": 3863
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.218878812860675e-05,
+      "loss": 2.8813,
+      "step": 3864
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.212695795548226e-05,
+      "loss": 2.8049,
+      "step": 3865
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.20651277823578e-05,
+      "loss": 2.7175,
+      "step": 3866
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.20032976092333e-05,
+      "loss": 3.2866,
+      "step": 3867
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.194146743610881e-05,
+      "loss": 2.428,
+      "step": 3868
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.187963726298432e-05,
+      "loss": 3.5377,
+      "step": 3869
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.181780708985985e-05,
+      "loss": 3.4922,
+      "step": 3870
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.175597691673535e-05,
+      "loss": 3.3141,
+      "step": 3871
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.169414674361088e-05,
+      "loss": 3.4566,
+      "step": 3872
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.16323165704864e-05,
+      "loss": 3.667,
+      "step": 3873
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.15704863973619e-05,
+      "loss": 3.5822,
+      "step": 3874
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.150865622423743e-05,
+      "loss": 3.4397,
+      "step": 3875
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.144682605111293e-05,
+      "loss": 3.5776,
+      "step": 3876
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.138499587798845e-05,
+      "loss": 3.4904,
+      "step": 3877
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.132316570486395e-05,
+      "loss": 3.1586,
+      "step": 3878
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.126133553173948e-05,
+      "loss": 3.4187,
+      "step": 3879
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.119950535861499e-05,
+      "loss": 3.7991,
+      "step": 3880
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.113767518549052e-05,
+      "loss": 3.1447,
+      "step": 3881
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.107584501236603e-05,
+      "loss": 3.3763,
+      "step": 3882
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.101401483924154e-05,
+      "loss": 3.4051,
+      "step": 3883
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.095218466611707e-05,
+      "loss": 3.6021,
+      "step": 3884
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.089035449299257e-05,
+      "loss": 3.1671,
+      "step": 3885
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.082852431986809e-05,
+      "loss": 3.1442,
+      "step": 3886
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.07666941467436e-05,
+      "loss": 3.1871,
+      "step": 3887
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.070486397361912e-05,
+      "loss": 3.3936,
+      "step": 3888
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.064303380049464e-05,
+      "loss": 3.1394,
+      "step": 3889
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.058120362737015e-05,
+      "loss": 3.4759,
+      "step": 3890
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.051937345424567e-05,
+      "loss": 3.2885,
+      "step": 3891
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.045754328112117e-05,
+      "loss": 3.1844,
+      "step": 3892
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.03957131079967e-05,
+      "loss": 3.2625,
+      "step": 3893
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.03338829348722e-05,
+      "loss": 3.1169,
+      "step": 3894
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.027205276174772e-05,
+      "loss": 3.1245,
+      "step": 3895
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.021022258862324e-05,
+      "loss": 3.2352,
+      "step": 3896
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.014839241549876e-05,
+      "loss": 3.4486,
+      "step": 3897
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.008656224237427e-05,
+      "loss": 3.267,
+      "step": 3898
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.002473206924979e-05,
+      "loss": 2.9787,
+      "step": 3899
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.99629018961253e-05,
+      "loss": 3.1578,
+      "step": 3900
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.990107172300081e-05,
+      "loss": 3.1365,
+      "step": 3901
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.983924154987634e-05,
+      "loss": 3.2998,
+      "step": 3902
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.977741137675184e-05,
+      "loss": 3.1836,
+      "step": 3903
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.971558120362736e-05,
+      "loss": 3.4544,
+      "step": 3904
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.965375103050289e-05,
+      "loss": 3.311,
+      "step": 3905
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.959192085737839e-05,
+      "loss": 3.2058,
+      "step": 3906
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.953009068425391e-05,
+      "loss": 3.396,
+      "step": 3907
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.946826051112942e-05,
+      "loss": 3.2184,
+      "step": 3908
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.940643033800494e-05,
+      "loss": 3.294,
+      "step": 3909
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.934460016488044e-05,
+      "loss": 4.0187,
+      "step": 3910
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.928276999175597e-05,
+      "loss": 3.1191,
+      "step": 3911
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.922093981863148e-05,
+      "loss": 2.9877,
+      "step": 3912
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.9159109645507e-05,
+      "loss": 2.8315,
+      "step": 3913
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.909727947238252e-05,
+      "loss": 2.7122,
+      "step": 3914
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.903544929925803e-05,
+      "loss": 3.0803,
+      "step": 3915
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.897361912613354e-05,
+      "loss": 2.4116,
+      "step": 3916
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.891178895300906e-05,
+      "loss": 2.7027,
+      "step": 3917
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.884995877988458e-05,
+      "loss": 2.2361,
+      "step": 3918
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.878812860676008e-05,
+      "loss": 3.8784,
+      "step": 3919
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.872629843363561e-05,
+      "loss": 3.4351,
+      "step": 3920
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.866446826051113e-05,
+      "loss": 3.2226,
+      "step": 3921
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.860263808738663e-05,
+      "loss": 3.5064,
+      "step": 3922
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.854080791426216e-05,
+      "loss": 3.4463,
+      "step": 3923
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.847897774113766e-05,
+      "loss": 3.4816,
+      "step": 3924
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.841714756801319e-05,
+      "loss": 3.2617,
+      "step": 3925
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.83553173948887e-05,
+      "loss": 3.2537,
+      "step": 3926
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.829348722176421e-05,
+      "loss": 3.2816,
+      "step": 3927
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.823165704863974e-05,
+      "loss": 3.2902,
+      "step": 3928
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.816982687551525e-05,
+      "loss": 3.3741,
+      "step": 3929
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.810799670239076e-05,
+      "loss": 3.3207,
+      "step": 3930
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.804616652926626e-05,
+      "loss": 3.3226,
+      "step": 3931
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.79843363561418e-05,
+      "loss": 3.2885,
+      "step": 3932
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.79225061830173e-05,
+      "loss": 3.3305,
+      "step": 3933
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.786067600989283e-05,
+      "loss": 3.2061,
+      "step": 3934
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.779884583676833e-05,
+      "loss": 3.3106,
+      "step": 3935
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.773701566364385e-05,
+      "loss": 3.1339,
+      "step": 3936
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.767518549051938e-05,
+      "loss": 3.4223,
+      "step": 3937
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.761335531739488e-05,
+      "loss": 3.3567,
+      "step": 3938
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.75515251442704e-05,
+      "loss": 3.3241,
+      "step": 3939
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.74896949711459e-05,
+      "loss": 3.1791,
+      "step": 3940
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.742786479802143e-05,
+      "loss": 3.2286,
+      "step": 3941
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.736603462489693e-05,
+      "loss": 3.0643,
+      "step": 3942
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.730420445177246e-05,
+      "loss": 3.3792,
+      "step": 3943
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.724237427864798e-05,
+      "loss": 3.2359,
+      "step": 3944
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.718054410552348e-05,
+      "loss": 3.2756,
+      "step": 3945
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.711871393239901e-05,
+      "loss": 3.4405,
+      "step": 3946
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.705688375927452e-05,
+      "loss": 3.1752,
+      "step": 3947
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.699505358615003e-05,
+      "loss": 3.3047,
+      "step": 3948
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.693322341302555e-05,
+      "loss": 3.103,
+      "step": 3949
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.687139323990107e-05,
+      "loss": 2.8688,
+      "step": 3950
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.680956306677657e-05,
+      "loss": 3.1102,
+      "step": 3951
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.67477328936521e-05,
+      "loss": 3.4367,
+      "step": 3952
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.668590272052762e-05,
+      "loss": 3.001,
+      "step": 3953
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.662407254740312e-05,
+      "loss": 3.0244,
+      "step": 3954
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.656224237427865e-05,
+      "loss": 2.8474,
+      "step": 3955
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.650041220115415e-05,
+      "loss": 2.9585,
+      "step": 3956
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.643858202802967e-05,
+      "loss": 2.8457,
+      "step": 3957
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.637675185490519e-05,
+      "loss": 2.8188,
+      "step": 3958
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.63149216817807e-05,
+      "loss": 2.8913,
+      "step": 3959
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.625309150865622e-05,
+      "loss": 3.0303,
+      "step": 3960
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.619126133553174e-05,
+      "loss": 2.7739,
+      "step": 3961
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.612943116240725e-05,
+      "loss": 2.9254,
+      "step": 3962
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.606760098928275e-05,
+      "loss": 3.184,
+      "step": 3963
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.600577081615828e-05,
+      "loss": 3.1448,
+      "step": 3964
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.594394064303379e-05,
+      "loss": 2.3476,
+      "step": 3965
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.58821104699093e-05,
+      "loss": 3.9869,
+      "step": 3966
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.582028029678482e-05,
+      "loss": 2.8861,
+      "step": 3967
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.575845012366034e-05,
+      "loss": 2.296,
+      "step": 3968
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.569661995053585e-05,
+      "loss": 3.6868,
+      "step": 3969
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.563478977741137e-05,
+      "loss": 3.5606,
+      "step": 3970
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.557295960428689e-05,
+      "loss": 3.6411,
+      "step": 3971
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.551112943116239e-05,
+      "loss": 3.2937,
+      "step": 3972
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.544929925803792e-05,
+      "loss": 3.7333,
+      "step": 3973
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.538746908491342e-05,
+      "loss": 3.2116,
+      "step": 3974
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.532563891178894e-05,
+      "loss": 3.5046,
+      "step": 3975
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.526380873866447e-05,
+      "loss": 3.5656,
+      "step": 3976
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.520197856553997e-05,
+      "loss": 3.3548,
+      "step": 3977
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.51401483924155e-05,
+      "loss": 3.2425,
+      "step": 3978
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.5078318219291e-05,
+      "loss": 3.4317,
+      "step": 3979
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.501648804616652e-05,
+      "loss": 3.4316,
+      "step": 3980
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.495465787304203e-05,
+      "loss": 3.1479,
+      "step": 3981
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.489282769991756e-05,
+      "loss": 3.1023,
+      "step": 3982
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.483099752679306e-05,
+      "loss": 3.3337,
+      "step": 3983
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.476916735366858e-05,
+      "loss": 3.3066,
+      "step": 3984
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.47073371805441e-05,
+      "loss": 3.3939,
+      "step": 3985
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.464550700741961e-05,
+      "loss": 3.1962,
+      "step": 3986
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.458367683429514e-05,
+      "loss": 3.2274,
+      "step": 3987
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.452184666117064e-05,
+      "loss": 3.2484,
+      "step": 3988
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.446001648804616e-05,
+      "loss": 3.0839,
+      "step": 3989
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.439818631492166e-05,
+      "loss": 3.4136,
+      "step": 3990
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.433635614179719e-05,
+      "loss": 3.1985,
+      "step": 3991
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.427452596867271e-05,
+      "loss": 3.3841,
+      "step": 3992
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.421269579554821e-05,
+      "loss": 3.006,
+      "step": 3993
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.415086562242374e-05,
+      "loss": 3.5481,
+      "step": 3994
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.408903544929924e-05,
+      "loss": 3.2807,
+      "step": 3995
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.402720527617478e-05,
+      "loss": 3.0266,
+      "step": 3996
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.396537510305028e-05,
+      "loss": 3.8601,
+      "step": 3997
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.39035449299258e-05,
+      "loss": 3.5225,
+      "step": 3998
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.38417147568013e-05,
+      "loss": 3.2227,
+      "step": 3999
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.377988458367683e-05,
+      "loss": 3.127,
+      "step": 4000
+    },
+    {
+      "epoch": 2.24,
+      "eval_loss": 4.365481853485107,
+      "eval_runtime": 953.9698,
+      "eval_samples_per_second": 2.769,
+      "eval_steps_per_second": 0.347,
+      "eval_wer": 1.7747520825069416,
+      "step": 4000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.371805441055234e-05,
+      "loss": 3.0929,
+      "step": 4001
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.365622423742786e-05,
+      "loss": 3.2353,
+      "step": 4002
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.359439406430338e-05,
+      "loss": 3.1433,
+      "step": 4003
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.353256389117888e-05,
+      "loss": 3.1448,
+      "step": 4004
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.347073371805441e-05,
+      "loss": 3.3881,
+      "step": 4005
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.340890354492991e-05,
+      "loss": 3.1303,
+      "step": 4006
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.334707337180543e-05,
+      "loss": 3.156,
+      "step": 4007
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.328524319868096e-05,
+      "loss": 3.2417,
+      "step": 4008
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.322341302555646e-05,
+      "loss": 2.8995,
+      "step": 4009
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.316158285243198e-05,
+      "loss": 2.8984,
+      "step": 4010
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.30997526793075e-05,
+      "loss": 3.1557,
+      "step": 4011
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.303792250618301e-05,
+      "loss": 2.9232,
+      "step": 4012
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.297609233305852e-05,
+      "loss": 2.7583,
+      "step": 4013
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.291426215993405e-05,
+      "loss": 2.6998,
+      "step": 4014
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.285243198680955e-05,
+      "loss": 2.4058,
+      "step": 4015
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.279060181368507e-05,
+      "loss": 2.9813,
+      "step": 4016
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.27287716405606e-05,
+      "loss": 2.5554,
+      "step": 4017
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.26669414674361e-05,
+      "loss": 2.5419,
+      "step": 4018
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.260511129431162e-05,
+      "loss": 3.6087,
+      "step": 4019
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.254328112118713e-05,
+      "loss": 3.2973,
+      "step": 4020
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.248145094806265e-05,
+      "loss": 3.1956,
+      "step": 4021
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.241962077493815e-05,
+      "loss": 3.2953,
+      "step": 4022
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.235779060181368e-05,
+      "loss": 3.2274,
+      "step": 4023
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.22959604286892e-05,
+      "loss": 3.4215,
+      "step": 4024
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.22341302555647e-05,
+      "loss": 3.4185,
+      "step": 4025
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.217230008244023e-05,
+      "loss": 3.454,
+      "step": 4026
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.211046990931574e-05,
+      "loss": 3.3528,
+      "step": 4027
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.204863973619125e-05,
+      "loss": 3.357,
+      "step": 4028
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.198680956306677e-05,
+      "loss": 3.2106,
+      "step": 4029
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.192497938994228e-05,
+      "loss": 3.2308,
+      "step": 4030
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.186314921681779e-05,
+      "loss": 3.0057,
+      "step": 4031
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.180131904369332e-05,
+      "loss": 3.1853,
+      "step": 4032
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.173948887056883e-05,
+      "loss": 3.546,
+      "step": 4033
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.167765869744434e-05,
+      "loss": 3.1603,
+      "step": 4034
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.161582852431987e-05,
+      "loss": 3.2683,
+      "step": 4035
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.155399835119537e-05,
+      "loss": 2.8928,
+      "step": 4036
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.149216817807089e-05,
+      "loss": 3.1096,
+      "step": 4037
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.14303380049464e-05,
+      "loss": 3.3289,
+      "step": 4038
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.136850783182192e-05,
+      "loss": 2.9157,
+      "step": 4039
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.130667765869745e-05,
+      "loss": 3.2545,
+      "step": 4040
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.124484748557295e-05,
+      "loss": 2.9959,
+      "step": 4041
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.118301731244847e-05,
+      "loss": 3.0071,
+      "step": 4042
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.112118713932397e-05,
+      "loss": 3.5522,
+      "step": 4043
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.10593569661995e-05,
+      "loss": 3.1922,
+      "step": 4044
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.0997526793075e-05,
+      "loss": 3.3715,
+      "step": 4045
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.093569661995052e-05,
+      "loss": 3.3027,
+      "step": 4046
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.087386644682605e-05,
+      "loss": 3.0325,
+      "step": 4047
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.081203627370156e-05,
+      "loss": 3.3967,
+      "step": 4048
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.075020610057709e-05,
+      "loss": 2.9456,
+      "step": 4049
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.068837592745259e-05,
+      "loss": 3.4321,
+      "step": 4050
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.06265457543281e-05,
+      "loss": 3.1231,
+      "step": 4051
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.056471558120361e-05,
+      "loss": 3.0589,
+      "step": 4052
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.050288540807914e-05,
+      "loss": 3.5179,
+      "step": 4053
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.044105523495464e-05,
+      "loss": 3.1119,
+      "step": 4054
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.037922506183017e-05,
+      "loss": 3.245,
+      "step": 4055
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.031739488870569e-05,
+      "loss": 3.1236,
+      "step": 4056
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.025556471558119e-05,
+      "loss": 3.494,
+      "step": 4057
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.019373454245672e-05,
+      "loss": 3.117,
+      "step": 4058
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.013190436933223e-05,
+      "loss": 3.0161,
+      "step": 4059
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.007007419620774e-05,
+      "loss": 2.8693,
+      "step": 4060
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.000824402308324e-05,
+      "loss": 3.0279,
+      "step": 4061
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.994641384995877e-05,
+      "loss": 2.8154,
+      "step": 4062
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.988458367683429e-05,
+      "loss": 3.2745,
+      "step": 4063
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.982275350370981e-05,
+      "loss": 3.3073,
+      "step": 4064
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.976092333058532e-05,
+      "loss": 2.6443,
+      "step": 4065
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.969909315746083e-05,
+      "loss": 2.8525,
+      "step": 4066
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.963726298433636e-05,
+      "loss": 2.5841,
+      "step": 4067
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.957543281121186e-05,
+      "loss": 2.7073,
+      "step": 4068
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.951360263808738e-05,
+      "loss": 3.4761,
+      "step": 4069
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.945177246496288e-05,
+      "loss": 3.2784,
+      "step": 4070
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.938994229183841e-05,
+      "loss": 3.5623,
+      "step": 4071
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.932811211871393e-05,
+      "loss": 3.2853,
+      "step": 4072
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.926628194558944e-05,
+      "loss": 3.1443,
+      "step": 4073
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.920445177246496e-05,
+      "loss": 3.421,
+      "step": 4074
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.914262159934046e-05,
+      "loss": 3.323,
+      "step": 4075
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.9080791426216e-05,
+      "loss": 3.6676,
+      "step": 4076
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.90189612530915e-05,
+      "loss": 3.4574,
+      "step": 4077
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.895713107996701e-05,
+      "loss": 3.2152,
+      "step": 4078
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.889530090684254e-05,
+      "loss": 3.2306,
+      "step": 4079
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.883347073371805e-05,
+      "loss": 3.2829,
+      "step": 4080
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.877164056059356e-05,
+      "loss": 3.1137,
+      "step": 4081
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.870981038746908e-05,
+      "loss": 3.0418,
+      "step": 4082
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.86479802143446e-05,
+      "loss": 3.4164,
+      "step": 4083
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.85861500412201e-05,
+      "loss": 3.2104,
+      "step": 4084
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.852431986809563e-05,
+      "loss": 3.27,
+      "step": 4085
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.846248969497113e-05,
+      "loss": 3.3998,
+      "step": 4086
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.840065952184665e-05,
+      "loss": 3.1777,
+      "step": 4087
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.833882934872218e-05,
+      "loss": 3.3282,
+      "step": 4088
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.827699917559768e-05,
+      "loss": 3.1694,
+      "step": 4089
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.82151690024732e-05,
+      "loss": 3.2421,
+      "step": 4090
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.815333882934872e-05,
+      "loss": 3.1488,
+      "step": 4091
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.809150865622423e-05,
+      "loss": 3.1707,
+      "step": 4092
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.802967848309973e-05,
+      "loss": 3.3217,
+      "step": 4093
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.796784830997527e-05,
+      "loss": 3.1917,
+      "step": 4094
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.790601813685078e-05,
+      "loss": 3.24,
+      "step": 4095
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.784418796372628e-05,
+      "loss": 3.2747,
+      "step": 4096
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.778235779060181e-05,
+      "loss": 3.4902,
+      "step": 4097
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.772052761747732e-05,
+      "loss": 3.3167,
+      "step": 4098
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.765869744435283e-05,
+      "loss": 3.2043,
+      "step": 4099
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.759686727122835e-05,
+      "loss": 3.4043,
+      "step": 4100
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.753503709810387e-05,
+      "loss": 3.2309,
+      "step": 4101
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.747320692497937e-05,
+      "loss": 3.0189,
+      "step": 4102
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.74113767518549e-05,
+      "loss": 2.9703,
+      "step": 4103
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.734954657873042e-05,
+      "loss": 2.8792,
+      "step": 4104
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.728771640560592e-05,
+      "loss": 2.8996,
+      "step": 4105
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.722588623248145e-05,
+      "loss": 2.876,
+      "step": 4106
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.716405605935695e-05,
+      "loss": 3.0201,
+      "step": 4107
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.710222588623248e-05,
+      "loss": 2.9135,
+      "step": 4108
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.704039571310799e-05,
+      "loss": 3.0777,
+      "step": 4109
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.69785655399835e-05,
+      "loss": 2.8282,
+      "step": 4110
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.691673536685903e-05,
+      "loss": 3.0229,
+      "step": 4111
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.685490519373454e-05,
+      "loss": 2.6066,
+      "step": 4112
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.679307502061005e-05,
+      "loss": 2.2096,
+      "step": 4113
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.673124484748556e-05,
+      "loss": 3.2611,
+      "step": 4114
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.666941467436109e-05,
+      "loss": 2.7507,
+      "step": 4115
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.660758450123659e-05,
+      "loss": 2.284,
+      "step": 4116
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.654575432811212e-05,
+      "loss": 2.2504,
+      "step": 4117
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.648392415498762e-05,
+      "loss": 2.0742,
+      "step": 4118
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.642209398186314e-05,
+      "loss": 3.3596,
+      "step": 4119
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.636026380873867e-05,
+      "loss": 3.5353,
+      "step": 4120
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.629843363561417e-05,
+      "loss": 3.5245,
+      "step": 4121
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.623660346248969e-05,
+      "loss": 3.2213,
+      "step": 4122
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.617477328936519e-05,
+      "loss": 3.2582,
+      "step": 4123
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.611294311624072e-05,
+      "loss": 3.2871,
+      "step": 4124
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.605111294311623e-05,
+      "loss": 3.1604,
+      "step": 4125
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.598928276999176e-05,
+      "loss": 3.2256,
+      "step": 4126
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.592745259686727e-05,
+      "loss": 3.2356,
+      "step": 4127
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.586562242374277e-05,
+      "loss": 3.59,
+      "step": 4128
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.58037922506183e-05,
+      "loss": 3.2621,
+      "step": 4129
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.574196207749381e-05,
+      "loss": 3.3906,
+      "step": 4130
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.568013190436932e-05,
+      "loss": 3.5278,
+      "step": 4131
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.561830173124484e-05,
+      "loss": 3.3629,
+      "step": 4132
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.555647155812036e-05,
+      "loss": 3.7326,
+      "step": 4133
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.549464138499586e-05,
+      "loss": 3.041,
+      "step": 4134
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.543281121187139e-05,
+      "loss": 3.064,
+      "step": 4135
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.537098103874691e-05,
+      "loss": 3.179,
+      "step": 4136
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.530915086562241e-05,
+      "loss": 3.0753,
+      "step": 4137
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.524732069249794e-05,
+      "loss": 3.2402,
+      "step": 4138
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.518549051937344e-05,
+      "loss": 3.0641,
+      "step": 4139
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.512366034624896e-05,
+      "loss": 3.1425,
+      "step": 4140
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.506183017312448e-05,
+      "loss": 2.7272,
+      "step": 4141
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.5e-05,
+      "loss": 3.1175,
+      "step": 4142
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.493816982687551e-05,
+      "loss": 2.9514,
+      "step": 4143
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.487633965375103e-05,
+      "loss": 3.5639,
+      "step": 4144
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.481450948062653e-05,
+      "loss": 3.1846,
+      "step": 4145
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.475267930750206e-05,
+      "loss": 3.3599,
+      "step": 4146
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.469084913437758e-05,
+      "loss": 3.5331,
+      "step": 4147
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.462901896125309e-05,
+      "loss": 3.3052,
+      "step": 4148
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.45671887881286e-05,
+      "loss": 3.3156,
+      "step": 4149
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.450535861500411e-05,
+      "loss": 2.8739,
+      "step": 4150
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.444352844187963e-05,
+      "loss": 2.8909,
+      "step": 4151
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.438169826875515e-05,
+      "loss": 3.0827,
+      "step": 4152
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.431986809563066e-05,
+      "loss": 3.1531,
+      "step": 4153
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.425803792250618e-05,
+      "loss": 2.9697,
+      "step": 4154
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.41962077493817e-05,
+      "loss": 2.7243,
+      "step": 4155
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.413437757625721e-05,
+      "loss": 3.0678,
+      "step": 4156
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.407254740313273e-05,
+      "loss": 2.7595,
+      "step": 4157
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.401071723000823e-05,
+      "loss": 2.8036,
+      "step": 4158
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.394888705688375e-05,
+      "loss": 2.9385,
+      "step": 4159
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.388705688375926e-05,
+      "loss": 3.1748,
+      "step": 4160
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.382522671063478e-05,
+      "loss": 2.869,
+      "step": 4161
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.37633965375103e-05,
+      "loss": 2.7645,
+      "step": 4162
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.370156636438581e-05,
+      "loss": 2.7321,
+      "step": 4163
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.363973619126133e-05,
+      "loss": 2.6396,
+      "step": 4164
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.357790601813685e-05,
+      "loss": 2.5265,
+      "step": 4165
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.351607584501236e-05,
+      "loss": 2.6081,
+      "step": 4166
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.345424567188787e-05,
+      "loss": 2.421,
+      "step": 4167
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.339241549876338e-05,
+      "loss": 2.0822,
+      "step": 4168
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.33305853256389e-05,
+      "loss": 3.3767,
+      "step": 4169
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.326875515251443e-05,
+      "loss": 3.2604,
+      "step": 4170
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.320692497938993e-05,
+      "loss": 3.5281,
+      "step": 4171
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.314509480626545e-05,
+      "loss": 3.2066,
+      "step": 4172
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.308326463314097e-05,
+      "loss": 3.2605,
+      "step": 4173
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.302143446001648e-05,
+      "loss": 3.2554,
+      "step": 4174
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.2959604286892e-05,
+      "loss": 3.3472,
+      "step": 4175
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.28977741137675e-05,
+      "loss": 3.1971,
+      "step": 4176
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.283594394064302e-05,
+      "loss": 3.4962,
+      "step": 4177
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.277411376751855e-05,
+      "loss": 3.47,
+      "step": 4178
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.271228359439407e-05,
+      "loss": 2.9754,
+      "step": 4179
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.265045342126957e-05,
+      "loss": 3.505,
+      "step": 4180
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.258862324814509e-05,
+      "loss": 3.7074,
+      "step": 4181
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.25267930750206e-05,
+      "loss": 3.1846,
+      "step": 4182
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.246496290189612e-05,
+      "loss": 3.2564,
+      "step": 4183
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.240313272877164e-05,
+      "loss": 3.6639,
+      "step": 4184
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.234130255564715e-05,
+      "loss": 2.9189,
+      "step": 4185
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.227947238252267e-05,
+      "loss": 3.2179,
+      "step": 4186
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.221764220939819e-05,
+      "loss": 3.1127,
+      "step": 4187
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.21558120362737e-05,
+      "loss": 3.0759,
+      "step": 4188
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.20939818631492e-05,
+      "loss": 3.3053,
+      "step": 4189
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.203215169002472e-05,
+      "loss": 3.046,
+      "step": 4190
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.197032151690024e-05,
+      "loss": 2.9471,
+      "step": 4191
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.190849134377575e-05,
+      "loss": 2.942,
+      "step": 4192
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.184666117065127e-05,
+      "loss": 3.1881,
+      "step": 4193
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.178483099752679e-05,
+      "loss": 3.1647,
+      "step": 4194
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.17230008244023e-05,
+      "loss": 3.1877,
+      "step": 4195
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.166117065127782e-05,
+      "loss": 2.9435,
+      "step": 4196
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.159934047815334e-05,
+      "loss": 3.0469,
+      "step": 4197
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.153751030502884e-05,
+      "loss": 3.1415,
+      "step": 4198
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.147568013190436e-05,
+      "loss": 2.9366,
+      "step": 4199
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.141384995877987e-05,
+      "loss": 2.9751,
+      "step": 4200
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.135201978565539e-05,
+      "loss": 2.844,
+      "step": 4201
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.129018961253091e-05,
+      "loss": 3.1055,
+      "step": 4202
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.122835943940642e-05,
+      "loss": 3.0242,
+      "step": 4203
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.116652926628194e-05,
+      "loss": 3.1161,
+      "step": 4204
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.110469909315746e-05,
+      "loss": 3.2008,
+      "step": 4205
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.104286892003297e-05,
+      "loss": 2.7854,
+      "step": 4206
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.098103874690849e-05,
+      "loss": 2.9278,
+      "step": 4207
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.0919208573784e-05,
+      "loss": 2.7048,
+      "step": 4208
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.085737840065951e-05,
+      "loss": 3.0044,
+      "step": 4209
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.079554822753504e-05,
+      "loss": 3.1149,
+      "step": 4210
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.073371805441054e-05,
+      "loss": 2.8857,
+      "step": 4211
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.067188788128606e-05,
+      "loss": 2.4852,
+      "step": 4212
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.061005770816158e-05,
+      "loss": 2.5397,
+      "step": 4213
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.054822753503709e-05,
+      "loss": 2.1577,
+      "step": 4214
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.048639736191261e-05,
+      "loss": 2.6559,
+      "step": 4215
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.042456718878813e-05,
+      "loss": 2.3933,
+      "step": 4216
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.036273701566363e-05,
+      "loss": 2.4391,
+      "step": 4217
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.030090684253916e-05,
+      "loss": 2.4007,
+      "step": 4218
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.023907666941468e-05,
+      "loss": 3.5164,
+      "step": 4219
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.017724649629018e-05,
+      "loss": 3.5551,
+      "step": 4220
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.01154163231657e-05,
+      "loss": 3.4778,
+      "step": 4221
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.005358615004121e-05,
+      "loss": 3.4131,
+      "step": 4222
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.999175597691673e-05,
+      "loss": 3.0805,
+      "step": 4223
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.992992580379225e-05,
+      "loss": 3.0583,
+      "step": 4224
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.986809563066776e-05,
+      "loss": 2.9939,
+      "step": 4225
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.980626545754328e-05,
+      "loss": 3.2132,
+      "step": 4226
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.97444352844188e-05,
+      "loss": 3.0794,
+      "step": 4227
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.968260511129431e-05,
+      "loss": 3.4748,
+      "step": 4228
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.962077493816981e-05,
+      "loss": 3.3425,
+      "step": 4229
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.955894476504533e-05,
+      "loss": 3.0167,
+      "step": 4230
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.949711459192085e-05,
+      "loss": 3.1168,
+      "step": 4231
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.943528441879636e-05,
+      "loss": 2.9952,
+      "step": 4232
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.937345424567188e-05,
+      "loss": 2.8095,
+      "step": 4233
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.93116240725474e-05,
+      "loss": 3.3434,
+      "step": 4234
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.924979389942291e-05,
+      "loss": 3.1505,
+      "step": 4235
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.918796372629843e-05,
+      "loss": 3.1836,
+      "step": 4236
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.912613355317395e-05,
+      "loss": 3.0789,
+      "step": 4237
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.906430338004946e-05,
+      "loss": 3.2141,
+      "step": 4238
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.900247320692497e-05,
+      "loss": 3.0417,
+      "step": 4239
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.894064303380048e-05,
+      "loss": 3.0002,
+      "step": 4240
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.8878812860676e-05,
+      "loss": 2.9734,
+      "step": 4241
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.881698268755152e-05,
+      "loss": 3.2705,
+      "step": 4242
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.875515251442703e-05,
+      "loss": 3.0544,
+      "step": 4243
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.869332234130255e-05,
+      "loss": 3.3997,
+      "step": 4244
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.863149216817807e-05,
+      "loss": 2.9657,
+      "step": 4245
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.856966199505358e-05,
+      "loss": 2.9739,
+      "step": 4246
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.85078318219291e-05,
+      "loss": 2.9608,
+      "step": 4247
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.84460016488046e-05,
+      "loss": 3.0208,
+      "step": 4248
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.838417147568012e-05,
+      "loss": 3.1654,
+      "step": 4249
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.832234130255565e-05,
+      "loss": 2.8145,
+      "step": 4250
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.826051112943115e-05,
+      "loss": 2.971,
+      "step": 4251
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.819868095630667e-05,
+      "loss": 2.7202,
+      "step": 4252
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.813685078318219e-05,
+      "loss": 3.028,
+      "step": 4253
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.80750206100577e-05,
+      "loss": 2.7389,
+      "step": 4254
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.801319043693322e-05,
+      "loss": 3.1671,
+      "step": 4255
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.795136026380874e-05,
+      "loss": 2.6911,
+      "step": 4256
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.788953009068425e-05,
+      "loss": 2.6474,
+      "step": 4257
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.782769991755977e-05,
+      "loss": 2.8674,
+      "step": 4258
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.776586974443528e-05,
+      "loss": 2.9297,
+      "step": 4259
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.770403957131079e-05,
+      "loss": 2.5962,
+      "step": 4260
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.76422093981863e-05,
+      "loss": 2.4087,
+      "step": 4261
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.758037922506182e-05,
+      "loss": 2.9253,
+      "step": 4262
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.751854905193734e-05,
+      "loss": 2.5915,
+      "step": 4263
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.745671887881285e-05,
+      "loss": 2.2664,
+      "step": 4264
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.739488870568837e-05,
+      "loss": 2.2531,
+      "step": 4265
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.733305853256389e-05,
+      "loss": 2.1501,
+      "step": 4266
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.72712283594394e-05,
+      "loss": 2.0019,
+      "step": 4267
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.720939818631492e-05,
+      "loss": 2.4163,
+      "step": 4268
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.714756801319044e-05,
+      "loss": 3.5249,
+      "step": 4269
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.708573784006594e-05,
+      "loss": 3.3567,
+      "step": 4270
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.702390766694146e-05,
+      "loss": 3.2061,
+      "step": 4271
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.696207749381697e-05,
+      "loss": 3.0563,
+      "step": 4272
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.690024732069249e-05,
+      "loss": 3.3419,
+      "step": 4273
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.6838417147568e-05,
+      "loss": 3.3259,
+      "step": 4274
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.677658697444352e-05,
+      "loss": 3.016,
+      "step": 4275
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.671475680131904e-05,
+      "loss": 3.5349,
+      "step": 4276
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.665292662819456e-05,
+      "loss": 3.0497,
+      "step": 4277
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.659109645507007e-05,
+      "loss": 3.0836,
+      "step": 4278
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.652926628194558e-05,
+      "loss": 3.1686,
+      "step": 4279
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.646743610882109e-05,
+      "loss": 3.1911,
+      "step": 4280
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.640560593569662e-05,
+      "loss": 3.2283,
+      "step": 4281
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.634377576257213e-05,
+      "loss": 2.9775,
+      "step": 4282
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.628194558944764e-05,
+      "loss": 3.2172,
+      "step": 4283
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.622011541632316e-05,
+      "loss": 3.1539,
+      "step": 4284
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.615828524319868e-05,
+      "loss": 3.2606,
+      "step": 4285
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.609645507007419e-05,
+      "loss": 2.9443,
+      "step": 4286
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.603462489694971e-05,
+      "loss": 3.0077,
+      "step": 4287
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.597279472382521e-05,
+      "loss": 3.0443,
+      "step": 4288
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.591096455070074e-05,
+      "loss": 3.1648,
+      "step": 4289
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.584913437757626e-05,
+      "loss": 3.0035,
+      "step": 4290
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.578730420445178e-05,
+      "loss": 3.1067,
+      "step": 4291
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.572547403132728e-05,
+      "loss": 2.832,
+      "step": 4292
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.56636438582028e-05,
+      "loss": 3.0184,
+      "step": 4293
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.560181368507831e-05,
+      "loss": 3.2485,
+      "step": 4294
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.553998351195383e-05,
+      "loss": 2.8581,
+      "step": 4295
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.547815333882934e-05,
+      "loss": 2.9704,
+      "step": 4296
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.541632316570486e-05,
+      "loss": 2.8147,
+      "step": 4297
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.535449299258038e-05,
+      "loss": 3.0873,
+      "step": 4298
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.52926628194559e-05,
+      "loss": 2.9297,
+      "step": 4299
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.523083264633141e-05,
+      "loss": 2.9503,
+      "step": 4300
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.516900247320691e-05,
+      "loss": 2.9404,
+      "step": 4301
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.510717230008243e-05,
+      "loss": 2.8408,
+      "step": 4302
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.504534212695795e-05,
+      "loss": 3.2252,
+      "step": 4303
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.498351195383346e-05,
+      "loss": 2.7425,
+      "step": 4304
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.492168178070898e-05,
+      "loss": 2.8488,
+      "step": 4305
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.48598516075845e-05,
+      "loss": 2.8738,
+      "step": 4306
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.479802143446001e-05,
+      "loss": 2.74,
+      "step": 4307
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.473619126133553e-05,
+      "loss": 2.6953,
+      "step": 4308
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.467436108821105e-05,
+      "loss": 2.8753,
+      "step": 4309
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.461253091508655e-05,
+      "loss": 2.8678,
+      "step": 4310
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.455070074196207e-05,
+      "loss": 2.9642,
+      "step": 4311
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.448887056883758e-05,
+      "loss": 2.8474,
+      "step": 4312
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.44270403957131e-05,
+      "loss": 2.8369,
+      "step": 4313
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.436521022258862e-05,
+      "loss": 3.0333,
+      "step": 4314
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.430338004946413e-05,
+      "loss": 2.8876,
+      "step": 4315
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.424154987633965e-05,
+      "loss": 2.4712,
+      "step": 4316
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.417971970321517e-05,
+      "loss": 2.0241,
+      "step": 4317
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.411788953009068e-05,
+      "loss": 2.7272,
+      "step": 4318
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.405605935696619e-05,
+      "loss": 3.2331,
+      "step": 4319
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.39942291838417e-05,
+      "loss": 3.6629,
+      "step": 4320
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.393239901071723e-05,
+      "loss": 3.0776,
+      "step": 4321
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.387056883759275e-05,
+      "loss": 3.1348,
+      "step": 4322
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.380873866446825e-05,
+      "loss": 3.2838,
+      "step": 4323
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.374690849134377e-05,
+      "loss": 3.1067,
+      "step": 4324
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.368507831821928e-05,
+      "loss": 2.9964,
+      "step": 4325
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.36232481450948e-05,
+      "loss": 3.3205,
+      "step": 4326
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.356141797197032e-05,
+      "loss": 3.2565,
+      "step": 4327
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.349958779884582e-05,
+      "loss": 2.8825,
+      "step": 4328
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.343775762572135e-05,
+      "loss": 2.9554,
+      "step": 4329
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.337592745259687e-05,
+      "loss": 3.0533,
+      "step": 4330
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.331409727947238e-05,
+      "loss": 3.3964,
+      "step": 4331
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.325226710634789e-05,
+      "loss": 3.1936,
+      "step": 4332
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.31904369332234e-05,
+      "loss": 2.9076,
+      "step": 4333
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.312860676009892e-05,
+      "loss": 3.1237,
+      "step": 4334
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.306677658697444e-05,
+      "loss": 3.0512,
+      "step": 4335
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.300494641384995e-05,
+      "loss": 3.26,
+      "step": 4336
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.294311624072547e-05,
+      "loss": 2.9103,
+      "step": 4337
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.288128606760099e-05,
+      "loss": 3.2102,
+      "step": 4338
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.28194558944765e-05,
+      "loss": 3.3061,
+      "step": 4339
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.275762572135202e-05,
+      "loss": 3.0018,
+      "step": 4340
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.269579554822752e-05,
+      "loss": 2.9103,
+      "step": 4341
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.263396537510304e-05,
+      "loss": 2.7719,
+      "step": 4342
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.257213520197856e-05,
+      "loss": 2.8795,
+      "step": 4343
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.251030502885407e-05,
+      "loss": 3.1277,
+      "step": 4344
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.244847485572959e-05,
+      "loss": 2.958,
+      "step": 4345
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.23866446826051e-05,
+      "loss": 2.6366,
+      "step": 4346
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.232481450948062e-05,
+      "loss": 2.919,
+      "step": 4347
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.226298433635614e-05,
+      "loss": 2.7514,
+      "step": 4348
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.220115416323166e-05,
+      "loss": 3.1008,
+      "step": 4349
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.213932399010716e-05,
+      "loss": 2.8435,
+      "step": 4350
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.207749381698268e-05,
+      "loss": 2.6619,
+      "step": 4351
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.201566364385819e-05,
+      "loss": 2.6519,
+      "step": 4352
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.195383347073372e-05,
+      "loss": 3.0077,
+      "step": 4353
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.189200329760923e-05,
+      "loss": 2.9372,
+      "step": 4354
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.183017312448474e-05,
+      "loss": 2.7806,
+      "step": 4355
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.176834295136026e-05,
+      "loss": 2.8649,
+      "step": 4356
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.170651277823577e-05,
+      "loss": 2.7698,
+      "step": 4357
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.164468260511129e-05,
+      "loss": 2.6476,
+      "step": 4358
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.15828524319868e-05,
+      "loss": 2.8705,
+      "step": 4359
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.152102225886231e-05,
+      "loss": 2.7952,
+      "step": 4360
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.145919208573784e-05,
+      "loss": 2.6176,
+      "step": 4361
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.139736191261336e-05,
+      "loss": 3.0368,
+      "step": 4362
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.133553173948886e-05,
+      "loss": 2.5747,
+      "step": 4363
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.127370156636438e-05,
+      "loss": 2.7214,
+      "step": 4364
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.12118713932399e-05,
+      "loss": 2.5695,
+      "step": 4365
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.115004122011541e-05,
+      "loss": 2.264,
+      "step": 4366
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.108821104699093e-05,
+      "loss": 2.5578,
+      "step": 4367
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.1026380873866444e-05,
+      "loss": 1.8504,
+      "step": 4368
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.096455070074196e-05,
+      "loss": 3.5033,
+      "step": 4369
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.090272052761748e-05,
+      "loss": 3.362,
+      "step": 4370
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.084089035449299e-05,
+      "loss": 3.2318,
+      "step": 4371
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.07790601813685e-05,
+      "loss": 3.2593,
+      "step": 4372
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.071723000824402e-05,
+      "loss": 2.9283,
+      "step": 4373
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.065539983511953e-05,
+      "loss": 3.2243,
+      "step": 4374
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.0593569661995046e-05,
+      "loss": 3.0926,
+      "step": 4375
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.053173948887057e-05,
+      "loss": 3.0931,
+      "step": 4376
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.046990931574608e-05,
+      "loss": 2.9576,
+      "step": 4377
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.0408079142621596e-05,
+      "loss": 3.2375,
+      "step": 4378
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.034624896949711e-05,
+      "loss": 3.349,
+      "step": 4379
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.028441879637262e-05,
+      "loss": 3.0157,
+      "step": 4380
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.022258862324814e-05,
+      "loss": 2.8173,
+      "step": 4381
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.0160758450123656e-05,
+      "loss": 2.9657,
+      "step": 4382
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.0098928276999165e-05,
+      "loss": 2.8938,
+      "step": 4383
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.003709810387469e-05,
+      "loss": 2.9943,
+      "step": 4384
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.9975267930750205e-05,
+      "loss": 3.2376,
+      "step": 4385
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.9913437757625715e-05,
+      "loss": 2.6532,
+      "step": 4386
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.985160758450123e-05,
+      "loss": 2.949,
+      "step": 4387
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.978977741137675e-05,
+      "loss": 2.8895,
+      "step": 4388
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.972794723825226e-05,
+      "loss": 3.168,
+      "step": 4389
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.9666117065127775e-05,
+      "loss": 2.9617,
+      "step": 4390
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.960428689200329e-05,
+      "loss": 2.6789,
+      "step": 4391
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.9542456718878815e-05,
+      "loss": 2.8416,
+      "step": 4392
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.9480626545754325e-05,
+      "loss": 2.861,
+      "step": 4393
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.941879637262984e-05,
+      "loss": 2.7134,
+      "step": 4394
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.935696619950535e-05,
+      "loss": 2.7129,
+      "step": 4395
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.929513602638087e-05,
+      "loss": 2.6623,
+      "step": 4396
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.9233305853256384e-05,
+      "loss": 2.9888,
+      "step": 4397
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.9171475680131894e-05,
+      "loss": 3.0149,
+      "step": 4398
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.910964550700741e-05,
+      "loss": 2.994,
+      "step": 4399
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.9047815333882934e-05,
+      "loss": 2.801,
+      "step": 4400
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.898598516075845e-05,
+      "loss": 2.7023,
+      "step": 4401
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.892415498763396e-05,
+      "loss": 2.5649,
+      "step": 4402
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.886232481450948e-05,
+      "loss": 2.826,
+      "step": 4403
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.8800494641384993e-05,
+      "loss": 3.2777,
+      "step": 4404
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.87386644682605e-05,
+      "loss": 2.6568,
+      "step": 4405
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.867683429513602e-05,
+      "loss": 2.5558,
+      "step": 4406
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.861500412201153e-05,
+      "loss": 2.7846,
+      "step": 4407
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.855317394888705e-05,
+      "loss": 2.491,
+      "step": 4408
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.849134377576257e-05,
+      "loss": 2.5184,
+      "step": 4409
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.8429513602638086e-05,
+      "loss": 2.8387,
+      "step": 4410
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.8367683429513596e-05,
+      "loss": 2.687,
+      "step": 4411
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.830585325638911e-05,
+      "loss": 2.2254,
+      "step": 4412
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.824402308326463e-05,
+      "loss": 2.4563,
+      "step": 4413
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.818219291014014e-05,
+      "loss": 2.7193,
+      "step": 4414
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.8120362737015656e-05,
+      "loss": 2.4387,
+      "step": 4415
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.805853256389118e-05,
+      "loss": 2.3309,
+      "step": 4416
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.799670239076669e-05,
+      "loss": 2.2579,
+      "step": 4417
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7934872217642205e-05,
+      "loss": 2.4245,
+      "step": 4418
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.787304204451772e-05,
+      "loss": 3.3669,
+      "step": 4419
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.781121187139323e-05,
+      "loss": 3.1223,
+      "step": 4420
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.774938169826875e-05,
+      "loss": 3.3126,
+      "step": 4421
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7687551525144265e-05,
+      "loss": 3.029,
+      "step": 4422
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7625721352019775e-05,
+      "loss": 3.0238,
+      "step": 4423
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.75638911788953e-05,
+      "loss": 2.8049,
+      "step": 4424
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7502061005770815e-05,
+      "loss": 3.017,
+      "step": 4425
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7440230832646325e-05,
+      "loss": 3.0431,
+      "step": 4426
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.737840065952184e-05,
+      "loss": 3.0598,
+      "step": 4427
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.731657048639736e-05,
+      "loss": 3.126,
+      "step": 4428
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.725474031327287e-05,
+      "loss": 3.1638,
+      "step": 4429
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7192910140148384e-05,
+      "loss": 2.9854,
+      "step": 4430
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.71310799670239e-05,
+      "loss": 2.8264,
+      "step": 4431
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7069249793899424e-05,
+      "loss": 2.8829,
+      "step": 4432
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7007419620774934e-05,
+      "loss": 2.779,
+      "step": 4433
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.694558944765045e-05,
+      "loss": 2.8545,
+      "step": 4434
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.688375927452597e-05,
+      "loss": 2.8469,
+      "step": 4435
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.682192910140148e-05,
+      "loss": 2.8068,
+      "step": 4436
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.6760098928276993e-05,
+      "loss": 2.6541,
+      "step": 4437
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.669826875515251e-05,
+      "loss": 3.0378,
+      "step": 4438
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.663643858202802e-05,
+      "loss": 2.9508,
+      "step": 4439
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.657460840890354e-05,
+      "loss": 2.9777,
+      "step": 4440
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.651277823577906e-05,
+      "loss": 2.946,
+      "step": 4441
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.645094806265457e-05,
+      "loss": 2.6227,
+      "step": 4442
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.6389117889530086e-05,
+      "loss": 3.0479,
+      "step": 4443
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.63272877164056e-05,
+      "loss": 2.9185,
+      "step": 4444
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.626545754328111e-05,
+      "loss": 2.7994,
+      "step": 4445
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.620362737015663e-05,
+      "loss": 2.7842,
+      "step": 4446
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.6141797197032146e-05,
+      "loss": 2.6849,
+      "step": 4447
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.607996702390766e-05,
+      "loss": 2.9479,
+      "step": 4448
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.601813685078318e-05,
+      "loss": 2.8821,
+      "step": 4449
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.5956306677658695e-05,
+      "loss": 2.7478,
+      "step": 4450
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.5894476504534205e-05,
+      "loss": 2.8078,
+      "step": 4451
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.583264633140972e-05,
+      "loss": 2.5962,
+      "step": 4452
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.577081615828524e-05,
+      "loss": 2.6587,
+      "step": 4453
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.570898598516075e-05,
+      "loss": 2.8726,
+      "step": 4454
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.5647155812036265e-05,
+      "loss": 3.1467,
+      "step": 4455
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.558532563891179e-05,
+      "loss": 2.4351,
+      "step": 4456
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.5523495465787305e-05,
+      "loss": 3.024,
+      "step": 4457
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.5461665292662815e-05,
+      "loss": 2.4876,
+      "step": 4458
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.539983511953833e-05,
+      "loss": 2.9127,
+      "step": 4459
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.533800494641384e-05,
+      "loss": 2.8874,
+      "step": 4460
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.527617477328936e-05,
+      "loss": 2.5589,
+      "step": 4461
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.5214344600164874e-05,
+      "loss": 2.2105,
+      "step": 4462
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.5152514427040384e-05,
+      "loss": 2.3833,
+      "step": 4463
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.509068425391591e-05,
+      "loss": 2.3841,
+      "step": 4464
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.5028854080791424e-05,
+      "loss": 2.8943,
+      "step": 4465
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.496702390766694e-05,
+      "loss": 2.446,
+      "step": 4466
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.490519373454245e-05,
+      "loss": 2.2773,
+      "step": 4467
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.484336356141797e-05,
+      "loss": 2.2707,
+      "step": 4468
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.4781533388293484e-05,
+      "loss": 3.2915,
+      "step": 4469
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.471970321516899e-05,
+      "loss": 3.1647,
+      "step": 4470
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.465787304204451e-05,
+      "loss": 3.1907,
+      "step": 4471
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.459604286892003e-05,
+      "loss": 3.3811,
+      "step": 4472
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.453421269579554e-05,
+      "loss": 2.9428,
+      "step": 4473
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.447238252267106e-05,
+      "loss": 2.8376,
+      "step": 4474
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.4410552349546576e-05,
+      "loss": 2.8014,
+      "step": 4475
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.4348722176422086e-05,
+      "loss": 2.7154,
+      "step": 4476
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.42868920032976e-05,
+      "loss": 2.7652,
+      "step": 4477
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.422506183017312e-05,
+      "loss": 3.0546,
+      "step": 4478
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.4163231657048636e-05,
+      "loss": 3.1427,
+      "step": 4479
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.410140148392415e-05,
+      "loss": 2.915,
+      "step": 4480
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.403957131079967e-05,
+      "loss": 2.973,
+      "step": 4481
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.397774113767518e-05,
+      "loss": 2.8372,
+      "step": 4482
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.3915910964550695e-05,
+      "loss": 2.654,
+      "step": 4483
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.385408079142621e-05,
+      "loss": 2.629,
+      "step": 4484
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.379225061830172e-05,
+      "loss": 2.7615,
+      "step": 4485
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.373042044517724e-05,
+      "loss": 2.8757,
+      "step": 4486
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.366859027205276e-05,
+      "loss": 2.744,
+      "step": 4487
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.360676009892828e-05,
+      "loss": 2.8477,
+      "step": 4488
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.354492992580379e-05,
+      "loss": 2.7085,
+      "step": 4489
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.3483099752679305e-05,
+      "loss": 2.8363,
+      "step": 4490
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.3421269579554815e-05,
+      "loss": 2.9144,
+      "step": 4491
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.335943940643033e-05,
+      "loss": 3.1668,
+      "step": 4492
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.329760923330585e-05,
+      "loss": 2.7029,
+      "step": 4493
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.323577906018136e-05,
+      "loss": 2.9121,
+      "step": 4494
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.317394888705688e-05,
+      "loss": 2.7248,
+      "step": 4495
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.31121187139324e-05,
+      "loss": 2.9665,
+      "step": 4496
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.3050288540807914e-05,
+      "loss": 2.755,
+      "step": 4497
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.2988458367683424e-05,
+      "loss": 2.5388,
+      "step": 4498
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.292662819455894e-05,
+      "loss": 2.6493,
+      "step": 4499
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.286479802143446e-05,
+      "loss": 2.7945,
+      "step": 4500
+    },
+    {
+      "epoch": 2.52,
+      "eval_loss": 3.852566719055176,
+      "eval_runtime": 957.5847,
+      "eval_samples_per_second": 2.759,
+      "eval_steps_per_second": 0.346,
+      "eval_wer": 1.6525188417294725,
+      "step": 4500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.280296784830997e-05,
+      "loss": 2.7852,
+      "step": 4501
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.2741137675185483e-05,
+      "loss": 2.8306,
+      "step": 4502
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.267930750206101e-05,
+      "loss": 2.9382,
+      "step": 4503
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.2617477328936517e-05,
+      "loss": 2.7517,
+      "step": 4504
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.255564715581203e-05,
+      "loss": 2.8425,
+      "step": 4505
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.249381698268755e-05,
+      "loss": 2.4246,
+      "step": 4506
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.243198680956306e-05,
+      "loss": 2.431,
+      "step": 4507
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.2370156636438576e-05,
+      "loss": 2.7013,
+      "step": 4508
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.230832646331409e-05,
+      "loss": 2.7106,
+      "step": 4509
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.22464962901896e-05,
+      "loss": 2.6687,
+      "step": 4510
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.2184666117065126e-05,
+      "loss": 2.4774,
+      "step": 4511
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.212283594394064e-05,
+      "loss": 2.703,
+      "step": 4512
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.206100577081615e-05,
+      "loss": 2.7679,
+      "step": 4513
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.199917559769167e-05,
+      "loss": 2.5795,
+      "step": 4514
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.1937345424567185e-05,
+      "loss": 2.2884,
+      "step": 4515
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.1875515251442695e-05,
+      "loss": 2.3184,
+      "step": 4516
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.181368507831821e-05,
+      "loss": 1.7158,
+      "step": 4517
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.175185490519373e-05,
+      "loss": 1.6335,
+      "step": 4518
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.169002473206925e-05,
+      "loss": 3.6333,
+      "step": 4519
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.162819455894476e-05,
+      "loss": 3.4132,
+      "step": 4520
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.156636438582028e-05,
+      "loss": 3.5185,
+      "step": 4521
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.1504534212695795e-05,
+      "loss": 3.1491,
+      "step": 4522
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.1442704039571305e-05,
+      "loss": 2.9642,
+      "step": 4523
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.138087386644682e-05,
+      "loss": 3.1275,
+      "step": 4524
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.131904369332233e-05,
+      "loss": 3.6762,
+      "step": 4525
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.125721352019785e-05,
+      "loss": 3.0063,
+      "step": 4526
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.119538334707337e-05,
+      "loss": 3.0853,
+      "step": 4527
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.113355317394889e-05,
+      "loss": 2.9904,
+      "step": 4528
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.10717230008244e-05,
+      "loss": 2.8939,
+      "step": 4529
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.1009892827699914e-05,
+      "loss": 2.9946,
+      "step": 4530
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.094806265457543e-05,
+      "loss": 3.0017,
+      "step": 4531
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.088623248145094e-05,
+      "loss": 2.9877,
+      "step": 4532
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.082440230832646e-05,
+      "loss": 2.6371,
+      "step": 4533
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.0762572135201974e-05,
+      "loss": 2.8214,
+      "step": 4534
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.070074196207749e-05,
+      "loss": 2.663,
+      "step": 4535
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.063891178895301e-05,
+      "loss": 2.8598,
+      "step": 4536
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.057708161582852e-05,
+      "loss": 3.0817,
+      "step": 4537
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.051525144270403e-05,
+      "loss": 2.7946,
+      "step": 4538
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.045342126957955e-05,
+      "loss": 2.7669,
+      "step": 4539
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.0391591096455066e-05,
+      "loss": 2.6988,
+      "step": 4540
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.0329760923330576e-05,
+      "loss": 2.8264,
+      "step": 4541
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.026793075020609e-05,
+      "loss": 2.3879,
+      "step": 4542
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.0206100577081616e-05,
+      "loss": 2.7147,
+      "step": 4543
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.0144270403957126e-05,
+      "loss": 2.6809,
+      "step": 4544
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.008244023083264e-05,
+      "loss": 2.6203,
+      "step": 4545
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.002061005770816e-05,
+      "loss": 2.3467,
+      "step": 4546
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.995877988458367e-05,
+      "loss": 2.8625,
+      "step": 4547
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.9896949711459185e-05,
+      "loss": 3.2189,
+      "step": 4548
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.98351195383347e-05,
+      "loss": 2.5224,
+      "step": 4549
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.977328936521021e-05,
+      "loss": 2.7725,
+      "step": 4550
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.9711459192085735e-05,
+      "loss": 2.6561,
+      "step": 4551
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.964962901896125e-05,
+      "loss": 2.8226,
+      "step": 4552
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.958779884583677e-05,
+      "loss": 2.6813,
+      "step": 4553
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.952596867271228e-05,
+      "loss": 2.793,
+      "step": 4554
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.9464138499587795e-05,
+      "loss": 2.4102,
+      "step": 4555
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.940230832646331e-05,
+      "loss": 2.429,
+      "step": 4556
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.934047815333882e-05,
+      "loss": 2.7131,
+      "step": 4557
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.927864798021434e-05,
+      "loss": 2.5584,
+      "step": 4558
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.921681780708986e-05,
+      "loss": 2.2782,
+      "step": 4559
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.915498763396537e-05,
+      "loss": 2.4103,
+      "step": 4560
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.909315746084089e-05,
+      "loss": 2.4571,
+      "step": 4561
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.9031327287716404e-05,
+      "loss": 2.5779,
+      "step": 4562
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.8969497114591914e-05,
+      "loss": 2.2856,
+      "step": 4563
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.890766694146743e-05,
+      "loss": 2.1105,
+      "step": 4564
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.884583676834295e-05,
+      "loss": 2.2943,
+      "step": 4565
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.878400659521846e-05,
+      "loss": 2.3816,
+      "step": 4566
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.872217642209398e-05,
+      "loss": 1.9024,
+      "step": 4567
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.86603462489695e-05,
+      "loss": 2.4274,
+      "step": 4568
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.859851607584501e-05,
+      "loss": 3.1449,
+      "step": 4569
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.853668590272052e-05,
+      "loss": 3.024,
+      "step": 4570
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.847485572959604e-05,
+      "loss": 3.1558,
+      "step": 4571
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.841302555647155e-05,
+      "loss": 2.9254,
+      "step": 4572
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.8351195383347066e-05,
+      "loss": 2.9484,
+      "step": 4573
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.828936521022258e-05,
+      "loss": 2.7554,
+      "step": 4574
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.8227535037098106e-05,
+      "loss": 2.896,
+      "step": 4575
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.8165704863973616e-05,
+      "loss": 2.8857,
+      "step": 4576
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.810387469084913e-05,
+      "loss": 2.7939,
+      "step": 4577
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.804204451772464e-05,
+      "loss": 2.8969,
+      "step": 4578
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.798021434460016e-05,
+      "loss": 2.9502,
+      "step": 4579
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.7918384171475676e-05,
+      "loss": 2.5842,
+      "step": 4580
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.7856553998351185e-05,
+      "loss": 3.1005,
+      "step": 4581
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.77947238252267e-05,
+      "loss": 3.0111,
+      "step": 4582
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.7732893652102225e-05,
+      "loss": 2.7306,
+      "step": 4583
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.767106347897774e-05,
+      "loss": 2.6806,
+      "step": 4584
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.760923330585325e-05,
+      "loss": 2.9298,
+      "step": 4585
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.754740313272877e-05,
+      "loss": 2.6023,
+      "step": 4586
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.7485572959604285e-05,
+      "loss": 3.1236,
+      "step": 4587
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.7423742786479795e-05,
+      "loss": 2.8151,
+      "step": 4588
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.736191261335531e-05,
+      "loss": 2.8053,
+      "step": 4589
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.7300082440230835e-05,
+      "loss": 2.6956,
+      "step": 4590
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.7238252267106344e-05,
+      "loss": 2.7383,
+      "step": 4591
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.717642209398186e-05,
+      "loss": 2.8282,
+      "step": 4592
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.711459192085738e-05,
+      "loss": 2.6568,
+      "step": 4593
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.705276174773289e-05,
+      "loss": 2.657,
+      "step": 4594
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.6990931574608404e-05,
+      "loss": 2.5007,
+      "step": 4595
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.692910140148392e-05,
+      "loss": 2.6343,
+      "step": 4596
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.686727122835943e-05,
+      "loss": 2.4248,
+      "step": 4597
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.6805441055234954e-05,
+      "loss": 2.7383,
+      "step": 4598
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.674361088211047e-05,
+      "loss": 2.6001,
+      "step": 4599
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.668178070898598e-05,
+      "loss": 2.4447,
+      "step": 4600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.66199505358615e-05,
+      "loss": 2.7069,
+      "step": 4601
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.655812036273701e-05,
+      "loss": 2.49,
+      "step": 4602
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.649629018961252e-05,
+      "loss": 2.6669,
+      "step": 4603
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.643446001648804e-05,
+      "loss": 2.3444,
+      "step": 4604
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.6372629843363556e-05,
+      "loss": 2.6531,
+      "step": 4605
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.631079967023908e-05,
+      "loss": 2.9661,
+      "step": 4606
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.624896949711459e-05,
+      "loss": 2.384,
+      "step": 4607
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.6187139323990106e-05,
+      "loss": 2.2609,
+      "step": 4608
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.6125309150865616e-05,
+      "loss": 2.5252,
+      "step": 4609
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.606347897774113e-05,
+      "loss": 2.4915,
+      "step": 4610
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.600164880461665e-05,
+      "loss": 2.3605,
+      "step": 4611
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.593981863149216e-05,
+      "loss": 2.2118,
+      "step": 4612
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5877988458367675e-05,
+      "loss": 2.6777,
+      "step": 4613
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.58161582852432e-05,
+      "loss": 1.8004,
+      "step": 4614
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5754328112118715e-05,
+      "loss": 1.9484,
+      "step": 4615
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5692497938994225e-05,
+      "loss": 1.9537,
+      "step": 4616
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.563066776586974e-05,
+      "loss": 2.16,
+      "step": 4617
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.556883759274526e-05,
+      "loss": 1.4459,
+      "step": 4618
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.550700741962077e-05,
+      "loss": 3.7187,
+      "step": 4619
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5445177246496285e-05,
+      "loss": 3.2841,
+      "step": 4620
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.53833470733718e-05,
+      "loss": 2.9766,
+      "step": 4621
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.532151690024732e-05,
+      "loss": 3.0657,
+      "step": 4622
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5259686727122835e-05,
+      "loss": 3.0327,
+      "step": 4623
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.519785655399835e-05,
+      "loss": 3.1043,
+      "step": 4624
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.513602638087386e-05,
+      "loss": 3.1761,
+      "step": 4625
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.507419620774938e-05,
+      "loss": 2.9237,
+      "step": 4626
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5012366034624894e-05,
+      "loss": 2.9775,
+      "step": 4627
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.4950535861500404e-05,
+      "loss": 2.7047,
+      "step": 4628
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.488870568837592e-05,
+      "loss": 3.076,
+      "step": 4629
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.4826875515251444e-05,
+      "loss": 2.9181,
+      "step": 4630
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.4765045342126954e-05,
+      "loss": 3.5651,
+      "step": 4631
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.470321516900247e-05,
+      "loss": 2.8954,
+      "step": 4632
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.464138499587799e-05,
+      "loss": 2.7204,
+      "step": 4633
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.45795548227535e-05,
+      "loss": 2.6864,
+      "step": 4634
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.451772464962901e-05,
+      "loss": 2.5516,
+      "step": 4635
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.445589447650453e-05,
+      "loss": 2.6053,
+      "step": 4636
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.439406430338004e-05,
+      "loss": 2.5995,
+      "step": 4637
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.433223413025556e-05,
+      "loss": 2.4533,
+      "step": 4638
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.427040395713108e-05,
+      "loss": 2.8681,
+      "step": 4639
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.4208573784006596e-05,
+      "loss": 2.5179,
+      "step": 4640
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.4146743610882106e-05,
+      "loss": 2.4083,
+      "step": 4641
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.408491343775762e-05,
+      "loss": 2.6893,
+      "step": 4642
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.402308326463313e-05,
+      "loss": 2.8324,
+      "step": 4643
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.396125309150865e-05,
+      "loss": 2.5786,
+      "step": 4644
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.3899422918384166e-05,
+      "loss": 2.4145,
+      "step": 4645
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.383759274525969e-05,
+      "loss": 2.7197,
+      "step": 4646
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.37757625721352e-05,
+      "loss": 2.922,
+      "step": 4647
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.3713932399010715e-05,
+      "loss": 2.4017,
+      "step": 4648
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.365210222588623e-05,
+      "loss": 2.7147,
+      "step": 4649
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.359027205276174e-05,
+      "loss": 2.1929,
+      "step": 4650
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.352844187963726e-05,
+      "loss": 2.5413,
+      "step": 4651
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.3466611706512775e-05,
+      "loss": 2.2158,
+      "step": 4652
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.3404781533388285e-05,
+      "loss": 2.7897,
+      "step": 4653
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.334295136026381e-05,
+      "loss": 2.7851,
+      "step": 4654
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.3281121187139325e-05,
+      "loss": 2.4268,
+      "step": 4655
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.3219291014014834e-05,
+      "loss": 2.6637,
+      "step": 4656
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.315746084089035e-05,
+      "loss": 2.5212,
+      "step": 4657
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.309563066776587e-05,
+      "loss": 2.2737,
+      "step": 4658
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.303380049464138e-05,
+      "loss": 2.3447,
+      "step": 4659
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.2971970321516894e-05,
+      "loss": 2.2942,
+      "step": 4660
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.291014014839241e-05,
+      "loss": 2.4288,
+      "step": 4661
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.284830997526793e-05,
+      "loss": 2.1177,
+      "step": 4662
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.2786479802143444e-05,
+      "loss": 2.0666,
+      "step": 4663
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.272464962901896e-05,
+      "loss": 2.8058,
+      "step": 4664
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.266281945589447e-05,
+      "loss": 2.0432,
+      "step": 4665
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.260098928276999e-05,
+      "loss": 1.9506,
+      "step": 4666
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.25391591096455e-05,
+      "loss": 2.3168,
+      "step": 4667
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.247732893652101e-05,
+      "loss": 2.1326,
+      "step": 4668
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.241549876339653e-05,
+      "loss": 3.1807,
+      "step": 4669
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.235366859027205e-05,
+      "loss": 3.1906,
+      "step": 4670
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.229183841714757e-05,
+      "loss": 3.2083,
+      "step": 4671
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.223000824402308e-05,
+      "loss": 3.2056,
+      "step": 4672
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.2168178070898596e-05,
+      "loss": 2.5949,
+      "step": 4673
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.2106347897774106e-05,
+      "loss": 2.8736,
+      "step": 4674
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.204451772464962e-05,
+      "loss": 2.8456,
+      "step": 4675
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.198268755152514e-05,
+      "loss": 2.772,
+      "step": 4676
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.192085737840065e-05,
+      "loss": 2.8082,
+      "step": 4677
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.185902720527617e-05,
+      "loss": 2.5831,
+      "step": 4678
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.179719703215169e-05,
+      "loss": 2.6529,
+      "step": 4679
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.1735366859027205e-05,
+      "loss": 2.8102,
+      "step": 4680
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.1673536685902715e-05,
+      "loss": 2.7144,
+      "step": 4681
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.161170651277823e-05,
+      "loss": 2.5331,
+      "step": 4682
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.154987633965375e-05,
+      "loss": 2.8789,
+      "step": 4683
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.148804616652926e-05,
+      "loss": 2.641,
+      "step": 4684
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.1426215993404775e-05,
+      "loss": 2.8375,
+      "step": 4685
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.13643858202803e-05,
+      "loss": 2.5851,
+      "step": 4686
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.130255564715581e-05,
+      "loss": 2.5214,
+      "step": 4687
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.1240725474031325e-05,
+      "loss": 2.3558,
+      "step": 4688
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.117889530090684e-05,
+      "loss": 2.5347,
+      "step": 4689
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.111706512778235e-05,
+      "loss": 2.37,
+      "step": 4690
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.105523495465787e-05,
+      "loss": 2.6989,
+      "step": 4691
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.0993404781533384e-05,
+      "loss": 2.6576,
+      "step": 4692
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.0931574608408894e-05,
+      "loss": 2.5548,
+      "step": 4693
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.086974443528442e-05,
+      "loss": 2.7144,
+      "step": 4694
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.0807914262159934e-05,
+      "loss": 2.7306,
+      "step": 4695
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.0746084089035444e-05,
+      "loss": 2.5748,
+      "step": 4696
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.068425391591096e-05,
+      "loss": 2.1664,
+      "step": 4697
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.062242374278648e-05,
+      "loss": 2.6646,
+      "step": 4698
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.056059356966199e-05,
+      "loss": 2.7077,
+      "step": 4699
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.04987633965375e-05,
+      "loss": 2.4308,
+      "step": 4700
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.043693322341303e-05,
+      "loss": 2.5496,
+      "step": 4701
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.037510305028854e-05,
+      "loss": 2.5744,
+      "step": 4702
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.031327287716405e-05,
+      "loss": 2.5613,
+      "step": 4703
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.025144270403957e-05,
+      "loss": 2.3339,
+      "step": 4704
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.0189612530915086e-05,
+      "loss": 2.6334,
+      "step": 4705
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.0127782357790596e-05,
+      "loss": 2.4655,
+      "step": 4706
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.006595218466611e-05,
+      "loss": 2.3778,
+      "step": 4707
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.000412201154162e-05,
+      "loss": 2.5315,
+      "step": 4708
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.9942291838417146e-05,
+      "loss": 2.1054,
+      "step": 4709
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.988046166529266e-05,
+      "loss": 2.3057,
+      "step": 4710
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.981863149216818e-05,
+      "loss": 2.1374,
+      "step": 4711
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.975680131904369e-05,
+      "loss": 2.4521,
+      "step": 4712
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.9694971145919205e-05,
+      "loss": 2.0084,
+      "step": 4713
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.963314097279472e-05,
+      "loss": 1.8124,
+      "step": 4714
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.957131079967023e-05,
+      "loss": 1.6021,
+      "step": 4715
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.950948062654575e-05,
+      "loss": 2.1155,
+      "step": 4716
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.944765045342127e-05,
+      "loss": 2.4512,
+      "step": 4717
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.938582028029678e-05,
+      "loss": 2.211,
+      "step": 4718
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.93239901071723e-05,
+      "loss": 3.383,
+      "step": 4719
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.9262159934047815e-05,
+      "loss": 3.2197,
+      "step": 4720
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.9200329760923324e-05,
+      "loss": 3.0011,
+      "step": 4721
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.913849958779884e-05,
+      "loss": 2.9956,
+      "step": 4722
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.907666941467436e-05,
+      "loss": 2.7763,
+      "step": 4723
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.901483924154987e-05,
+      "loss": 3.0211,
+      "step": 4724
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.895300906842539e-05,
+      "loss": 2.9111,
+      "step": 4725
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.889117889530091e-05,
+      "loss": 2.6724,
+      "step": 4726
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.882934872217642e-05,
+      "loss": 2.6607,
+      "step": 4727
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.8767518549051934e-05,
+      "loss": 2.4162,
+      "step": 4728
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.870568837592745e-05,
+      "loss": 2.6734,
+      "step": 4729
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.864385820280296e-05,
+      "loss": 3.0497,
+      "step": 4730
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.858202802967848e-05,
+      "loss": 2.6221,
+      "step": 4731
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.852019785655399e-05,
+      "loss": 2.9954,
+      "step": 4732
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.845836768342952e-05,
+      "loss": 2.525,
+      "step": 4733
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.8396537510305027e-05,
+      "loss": 2.7271,
+      "step": 4734
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.833470733718054e-05,
+      "loss": 2.6061,
+      "step": 4735
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.827287716405606e-05,
+      "loss": 2.7418,
+      "step": 4736
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.821104699093157e-05,
+      "loss": 2.3035,
+      "step": 4737
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.8149216817807086e-05,
+      "loss": 2.5574,
+      "step": 4738
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.8087386644682596e-05,
+      "loss": 2.5307,
+      "step": 4739
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.802555647155811e-05,
+      "loss": 2.3911,
+      "step": 4740
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7963726298433636e-05,
+      "loss": 2.5415,
+      "step": 4741
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.790189612530915e-05,
+      "loss": 2.2778,
+      "step": 4742
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.784006595218466e-05,
+      "loss": 2.6256,
+      "step": 4743
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.777823577906018e-05,
+      "loss": 2.4907,
+      "step": 4744
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7716405605935695e-05,
+      "loss": 2.405,
+      "step": 4745
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7654575432811205e-05,
+      "loss": 2.5248,
+      "step": 4746
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.759274525968672e-05,
+      "loss": 2.8151,
+      "step": 4747
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.753091508656224e-05,
+      "loss": 2.3464,
+      "step": 4748
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7469084913437755e-05,
+      "loss": 2.421,
+      "step": 4749
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7407254740313265e-05,
+      "loss": 2.8065,
+      "step": 4750
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.734542456718879e-05,
+      "loss": 2.5848,
+      "step": 4751
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.72835943940643e-05,
+      "loss": 2.2319,
+      "step": 4752
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7221764220939815e-05,
+      "loss": 2.4661,
+      "step": 4753
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.715993404781533e-05,
+      "loss": 2.3682,
+      "step": 4754
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.709810387469085e-05,
+      "loss": 2.3123,
+      "step": 4755
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7036273701566364e-05,
+      "loss": 2.2596,
+      "step": 4756
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6974443528441874e-05,
+      "loss": 2.2163,
+      "step": 4757
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.691261335531739e-05,
+      "loss": 2.0975,
+      "step": 4758
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.685078318219291e-05,
+      "loss": 2.2092,
+      "step": 4759
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6788953009068424e-05,
+      "loss": 2.5865,
+      "step": 4760
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6727122835943934e-05,
+      "loss": 2.11,
+      "step": 4761
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.666529266281945e-05,
+      "loss": 2.1952,
+      "step": 4762
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.660346248969497e-05,
+      "loss": 1.9664,
+      "step": 4763
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6541632316570483e-05,
+      "loss": 2.1208,
+      "step": 4764
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6479802143446e-05,
+      "loss": 2.2405,
+      "step": 4765
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.641797197032151e-05,
+      "loss": 1.9772,
+      "step": 4766
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.635614179719703e-05,
+      "loss": 2.3371,
+      "step": 4767
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.629431162407254e-05,
+      "loss": 1.4525,
+      "step": 4768
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.623248145094806e-05,
+      "loss": 3.574,
+      "step": 4769
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6170651277823576e-05,
+      "loss": 3.2175,
+      "step": 4770
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.610882110469909e-05,
+      "loss": 3.0184,
+      "step": 4771
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.60469909315746e-05,
+      "loss": 3.0491,
+      "step": 4772
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.598516075845012e-05,
+      "loss": 2.7517,
+      "step": 4773
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.5923330585325636e-05,
+      "loss": 2.7001,
+      "step": 4774
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.586150041220115e-05,
+      "loss": 2.5992,
+      "step": 4775
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.579967023907667e-05,
+      "loss": 2.6358,
+      "step": 4776
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.573784006595218e-05,
+      "loss": 2.5642,
+      "step": 4777
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.5676009892827695e-05,
+      "loss": 2.2134,
+      "step": 4778
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.561417971970321e-05,
+      "loss": 2.5497,
+      "step": 4779
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.555234954657873e-05,
+      "loss": 2.673,
+      "step": 4780
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.5490519373454245e-05,
+      "loss": 2.5508,
+      "step": 4781
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.5428689200329755e-05,
+      "loss": 2.712,
+      "step": 4782
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.536685902720527e-05,
+      "loss": 2.2926,
+      "step": 4783
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.530502885408079e-05,
+      "loss": 2.2524,
+      "step": 4784
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.5243198680956305e-05,
+      "loss": 2.5812,
+      "step": 4785
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.5181368507831814e-05,
+      "loss": 2.4048,
+      "step": 4786
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.511953833470734e-05,
+      "loss": 2.6271,
+      "step": 4787
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.505770816158285e-05,
+      "loss": 2.4278,
+      "step": 4788
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.4995877988458364e-05,
+      "loss": 2.5206,
+      "step": 4789
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.493404781533388e-05,
+      "loss": 2.5512,
+      "step": 4790
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.48722176422094e-05,
+      "loss": 2.4426,
+      "step": 4791
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.481038746908491e-05,
+      "loss": 2.5293,
+      "step": 4792
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4748557295960424e-05,
+      "loss": 2.6085,
+      "step": 4793
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.468672712283594e-05,
+      "loss": 2.5805,
+      "step": 4794
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.462489694971146e-05,
+      "loss": 2.6573,
+      "step": 4795
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4563066776586974e-05,
+      "loss": 2.4171,
+      "step": 4796
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4501236603462483e-05,
+      "loss": 2.5044,
+      "step": 4797
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4439406430338e-05,
+      "loss": 2.5976,
+      "step": 4798
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4377576257213517e-05,
+      "loss": 2.6034,
+      "step": 4799
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.431574608408903e-05,
+      "loss": 2.5019,
+      "step": 4800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.425391591096455e-05,
+      "loss": 2.3946,
+      "step": 4801
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.419208573784006e-05,
+      "loss": 2.646,
+      "step": 4802
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4130255564715576e-05,
+      "loss": 2.0102,
+      "step": 4803
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.406842539159109e-05,
+      "loss": 2.1768,
+      "step": 4804
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.400659521846661e-05,
+      "loss": 2.6267,
+      "step": 4805
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.3944765045342126e-05,
+      "loss": 2.1553,
+      "step": 4806
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.388293487221764e-05,
+      "loss": 2.2305,
+      "step": 4807
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.382110469909315e-05,
+      "loss": 2.093,
+      "step": 4808
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.375927452596867e-05,
+      "loss": 2.151,
+      "step": 4809
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.3697444352844185e-05,
+      "loss": 2.4038,
+      "step": 4810
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.36356141797197e-05,
+      "loss": 2.3513,
+      "step": 4811
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.357378400659522e-05,
+      "loss": 2.4647,
+      "step": 4812
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.351195383347073e-05,
+      "loss": 2.4894,
+      "step": 4813
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.3450123660346245e-05,
+      "loss": 2.1475,
+      "step": 4814
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.338829348722176e-05,
+      "loss": 1.7932,
+      "step": 4815
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.332646331409728e-05,
+      "loss": 1.9429,
+      "step": 4816
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.326463314097279e-05,
+      "loss": 2.0317,
+      "step": 4817
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.320280296784831e-05,
+      "loss": 2.2461,
+      "step": 4818
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.314097279472382e-05,
+      "loss": 3.0817,
+      "step": 4819
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.307914262159934e-05,
+      "loss": 3.1512,
+      "step": 4820
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.3017312448474854e-05,
+      "loss": 2.9575,
+      "step": 4821
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.295548227535037e-05,
+      "loss": 2.7542,
+      "step": 4822
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.289365210222589e-05,
+      "loss": 2.6783,
+      "step": 4823
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.28318219291014e-05,
+      "loss": 2.6967,
+      "step": 4824
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.2769991755976914e-05,
+      "loss": 2.6866,
+      "step": 4825
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.270816158285243e-05,
+      "loss": 2.4747,
+      "step": 4826
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.264633140972795e-05,
+      "loss": 2.8821,
+      "step": 4827
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.258450123660346e-05,
+      "loss": 2.4276,
+      "step": 4828
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2522671063478974e-05,
+      "loss": 2.5887,
+      "step": 4829
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.246084089035449e-05,
+      "loss": 2.6465,
+      "step": 4830
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.239901071723001e-05,
+      "loss": 2.2808,
+      "step": 4831
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.233718054410552e-05,
+      "loss": 2.4797,
+      "step": 4832
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.227535037098103e-05,
+      "loss": 2.7049,
+      "step": 4833
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.221352019785655e-05,
+      "loss": 2.529,
+      "step": 4834
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2151690024732066e-05,
+      "loss": 2.4463,
+      "step": 4835
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.208985985160758e-05,
+      "loss": 2.5628,
+      "step": 4836
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.202802967848309e-05,
+      "loss": 2.4689,
+      "step": 4837
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.1966199505358616e-05,
+      "loss": 2.5123,
+      "step": 4838
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.1904369332234126e-05,
+      "loss": 2.3253,
+      "step": 4839
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.184253915910964e-05,
+      "loss": 2.4048,
+      "step": 4840
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.178070898598516e-05,
+      "loss": 2.6418,
+      "step": 4841
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.1718878812860676e-05,
+      "loss": 2.506,
+      "step": 4842
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.165704863973619e-05,
+      "loss": 2.2683,
+      "step": 4843
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.15952184666117e-05,
+      "loss": 2.244,
+      "step": 4844
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.153338829348722e-05,
+      "loss": 2.2749,
+      "step": 4845
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1471558120362735e-05,
+      "loss": 2.577,
+      "step": 4846
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.140972794723825e-05,
+      "loss": 1.9871,
+      "step": 4847
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.134789777411376e-05,
+      "loss": 2.2812,
+      "step": 4848
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.128606760098928e-05,
+      "loss": 2.8674,
+      "step": 4849
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1224237427864795e-05,
+      "loss": 2.6919,
+      "step": 4850
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.116240725474031e-05,
+      "loss": 2.2075,
+      "step": 4851
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.110057708161583e-05,
+      "loss": 2.547,
+      "step": 4852
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.103874690849134e-05,
+      "loss": 2.1415,
+      "step": 4853
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.097691673536686e-05,
+      "loss": 2.9388,
+      "step": 4854
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.091508656224237e-05,
+      "loss": 2.4326,
+      "step": 4855
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.085325638911789e-05,
+      "loss": 2.2764,
+      "step": 4856
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.07914262159934e-05,
+      "loss": 2.4224,
+      "step": 4857
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.072959604286892e-05,
+      "loss": 2.1159,
+      "step": 4858
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.066776586974443e-05,
+      "loss": 2.44,
+      "step": 4859
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.060593569661995e-05,
+      "loss": 2.2266,
+      "step": 4860
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.0544105523495464e-05,
+      "loss": 2.0131,
+      "step": 4861
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.048227535037098e-05,
+      "loss": 2.5229,
+      "step": 4862
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0420445177246493e-05,
+      "loss": 2.0271,
+      "step": 4863
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.035861500412201e-05,
+      "loss": 2.1364,
+      "step": 4864
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0296784830997523e-05,
+      "loss": 2.2715,
+      "step": 4865
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.023495465787304e-05,
+      "loss": 1.8669,
+      "step": 4866
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0173124484748556e-05,
+      "loss": 2.2305,
+      "step": 4867
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.011129431162407e-05,
+      "loss": 1.7802,
+      "step": 4868
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0049464138499583e-05,
+      "loss": 3.237,
+      "step": 4869
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9987633965375103e-05,
+      "loss": 3.0692,
+      "step": 4870
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9925803792250616e-05,
+      "loss": 2.9976,
+      "step": 4871
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.986397361912613e-05,
+      "loss": 2.9877,
+      "step": 4872
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9802143446001646e-05,
+      "loss": 2.9223,
+      "step": 4873
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9740313272877162e-05,
+      "loss": 2.5597,
+      "step": 4874
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9678483099752675e-05,
+      "loss": 2.7015,
+      "step": 4875
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9616652926628192e-05,
+      "loss": 2.3995,
+      "step": 4876
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9554822753503705e-05,
+      "loss": 2.5696,
+      "step": 4877
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9492992580379225e-05,
+      "loss": 2.3406,
+      "step": 4878
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.943116240725474e-05,
+      "loss": 2.4358,
+      "step": 4879
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.936933223413025e-05,
+      "loss": 2.7956,
+      "step": 4880
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9307502061005765e-05,
+      "loss": 2.3773,
+      "step": 4881
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9245671887881285e-05,
+      "loss": 2.6716,
+      "step": 4882
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9183841714756798e-05,
+      "loss": 2.5529,
+      "step": 4883
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9122011541632315e-05,
+      "loss": 2.3156,
+      "step": 4884
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9060181368507828e-05,
+      "loss": 2.8119,
+      "step": 4885
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.8998351195383344e-05,
+      "loss": 2.2968,
+      "step": 4886
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.893652102225886e-05,
+      "loss": 2.5526,
+      "step": 4887
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.8874690849134374e-05,
+      "loss": 2.333,
+      "step": 4888
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.8812860676009887e-05,
+      "loss": 2.7631,
+      "step": 4889
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.8751030502885407e-05,
+      "loss": 2.256,
+      "step": 4890
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.868920032976092e-05,
+      "loss": 2.5115,
+      "step": 4891
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.8627370156636434e-05,
+      "loss": 2.387,
+      "step": 4892
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.856553998351195e-05,
+      "loss": 2.3612,
+      "step": 4893
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.8503709810387467e-05,
+      "loss": 2.3087,
+      "step": 4894
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.8441879637262984e-05,
+      "loss": 2.4176,
+      "step": 4895
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.8380049464138497e-05,
+      "loss": 2.289,
+      "step": 4896
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.831821929101401e-05,
+      "loss": 2.4161,
+      "step": 4897
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.825638911788953e-05,
+      "loss": 2.4878,
+      "step": 4898
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8194558944765043e-05,
+      "loss": 2.2953,
+      "step": 4899
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8132728771640556e-05,
+      "loss": 2.3446,
+      "step": 4900
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8070898598516073e-05,
+      "loss": 2.2198,
+      "step": 4901
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.800906842539159e-05,
+      "loss": 2.6391,
+      "step": 4902
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7947238252267103e-05,
+      "loss": 2.2114,
+      "step": 4903
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.788540807914262e-05,
+      "loss": 2.0506,
+      "step": 4904
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7823577906018132e-05,
+      "loss": 2.1686,
+      "step": 4905
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7761747732893652e-05,
+      "loss": 2.1363,
+      "step": 4906
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7699917559769166e-05,
+      "loss": 2.0503,
+      "step": 4907
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.763808738664468e-05,
+      "loss": 2.1743,
+      "step": 4908
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7576257213520192e-05,
+      "loss": 2.037,
+      "step": 4909
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7514427040395712e-05,
+      "loss": 2.4086,
+      "step": 4910
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7452596867271225e-05,
+      "loss": 2.0288,
+      "step": 4911
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7390766694146742e-05,
+      "loss": 2.0298,
+      "step": 4912
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7328936521022255e-05,
+      "loss": 2.0398,
+      "step": 4913
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.726710634789777e-05,
+      "loss": 1.9425,
+      "step": 4914
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7205276174773288e-05,
+      "loss": 1.7449,
+      "step": 4915
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.71434460016488e-05,
+      "loss": 1.9494,
+      "step": 4916
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7081615828524318e-05,
+      "loss": 2.004,
+      "step": 4917
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7019785655399834e-05,
+      "loss": 1.7176,
+      "step": 4918
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.6957955482275348e-05,
+      "loss": 2.8199,
+      "step": 4919
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.689612530915086e-05,
+      "loss": 3.0494,
+      "step": 4920
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.683429513602638e-05,
+      "loss": 2.9699,
+      "step": 4921
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.6772464962901894e-05,
+      "loss": 3.1728,
+      "step": 4922
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.6710634789777407e-05,
+      "loss": 2.7063,
+      "step": 4923
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.6648804616652924e-05,
+      "loss": 2.985,
+      "step": 4924
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.658697444352844e-05,
+      "loss": 2.5656,
+      "step": 4925
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.6525144270403957e-05,
+      "loss": 2.51,
+      "step": 4926
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.646331409727947e-05,
+      "loss": 2.5251,
+      "step": 4927
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.6401483924154983e-05,
+      "loss": 2.7384,
+      "step": 4928
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.6339653751030503e-05,
+      "loss": 2.4867,
+      "step": 4929
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.6277823577906017e-05,
+      "loss": 2.3838,
+      "step": 4930
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.621599340478153e-05,
+      "loss": 2.5365,
+      "step": 4931
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.6154163231657046e-05,
+      "loss": 2.6853,
+      "step": 4932
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.6092333058532563e-05,
+      "loss": 2.4821,
+      "step": 4933
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.6030502885408076e-05,
+      "loss": 2.4777,
+      "step": 4934
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.5968672712283593e-05,
+      "loss": 2.5796,
+      "step": 4935
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.5906842539159106e-05,
+      "loss": 2.6101,
+      "step": 4936
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.5845012366034626e-05,
+      "loss": 2.5678,
+      "step": 4937
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.578318219291014e-05,
+      "loss": 2.3493,
+      "step": 4938
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.5721352019785652e-05,
+      "loss": 2.3418,
+      "step": 4939
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.5659521846661166e-05,
+      "loss": 2.3677,
+      "step": 4940
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.5597691673536685e-05,
+      "loss": 2.1589,
+      "step": 4941
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.55358615004122e-05,
+      "loss": 2.3393,
+      "step": 4942
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.5474031327287715e-05,
+      "loss": 2.2681,
+      "step": 4943
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.541220115416323e-05,
+      "loss": 2.8163,
+      "step": 4944
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.5350370981038745e-05,
+      "loss": 2.2083,
+      "step": 4945
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.528854080791426e-05,
+      "loss": 2.3843,
+      "step": 4946
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.5226710634789775e-05,
+      "loss": 2.1472,
+      "step": 4947
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.5164880461665288e-05,
+      "loss": 2.6504,
+      "step": 4948
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.5103050288540808e-05,
+      "loss": 2.5889,
+      "step": 4949
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.504122011541632e-05,
+      "loss": 2.1493,
+      "step": 4950
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4979389942291834e-05,
+      "loss": 2.2822,
+      "step": 4951
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.491755976916735e-05,
+      "loss": 2.2039,
+      "step": 4952
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4855729596042868e-05,
+      "loss": 2.2423,
+      "step": 4953
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4793899422918384e-05,
+      "loss": 2.1596,
+      "step": 4954
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4732069249793897e-05,
+      "loss": 2.4269,
+      "step": 4955
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.467023907666941e-05,
+      "loss": 2.0631,
+      "step": 4956
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.460840890354493e-05,
+      "loss": 2.4088,
+      "step": 4957
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4546578730420444e-05,
+      "loss": 2.1334,
+      "step": 4958
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4484748557295957e-05,
+      "loss": 2.0167,
+      "step": 4959
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4422918384171474e-05,
+      "loss": 2.3687,
+      "step": 4960
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.436108821104699e-05,
+      "loss": 1.8568,
+      "step": 4961
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4299258037922503e-05,
+      "loss": 1.9449,
+      "step": 4962
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.423742786479802e-05,
+      "loss": 2.0782,
+      "step": 4963
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4175597691673533e-05,
+      "loss": 2.4926,
+      "step": 4964
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4113767518549053e-05,
+      "loss": 1.7861,
+      "step": 4965
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4051937345424566e-05,
+      "loss": 2.1621,
+      "step": 4966
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.399010717230008e-05,
+      "loss": 1.9034,
+      "step": 4967
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.3928276999175593e-05,
+      "loss": 1.5154,
+      "step": 4968
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3866446826051113e-05,
+      "loss": 2.8764,
+      "step": 4969
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3804616652926626e-05,
+      "loss": 3.1387,
+      "step": 4970
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3742786479802142e-05,
+      "loss": 2.9437,
+      "step": 4971
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3680956306677656e-05,
+      "loss": 2.6799,
+      "step": 4972
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3619126133553172e-05,
+      "loss": 2.7616,
+      "step": 4973
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.355729596042869e-05,
+      "loss": 2.5292,
+      "step": 4974
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3495465787304202e-05,
+      "loss": 2.5198,
+      "step": 4975
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3433635614179715e-05,
+      "loss": 2.7156,
+      "step": 4976
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3371805441055235e-05,
+      "loss": 2.4036,
+      "step": 4977
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.330997526793075e-05,
+      "loss": 2.6325,
+      "step": 4978
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.324814509480626e-05,
+      "loss": 2.5165,
+      "step": 4979
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3186314921681778e-05,
+      "loss": 2.4127,
+      "step": 4980
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3124484748557295e-05,
+      "loss": 2.4572,
+      "step": 4981
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3062654575432808e-05,
+      "loss": 2.305,
+      "step": 4982
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3000824402308325e-05,
+      "loss": 2.4946,
+      "step": 4983
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2938994229183838e-05,
+      "loss": 2.2038,
+      "step": 4984
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2877164056059358e-05,
+      "loss": 2.5263,
+      "step": 4985
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.281533388293487e-05,
+      "loss": 2.5821,
+      "step": 4986
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2753503709810384e-05,
+      "loss": 1.955,
+      "step": 4987
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.26916735366859e-05,
+      "loss": 2.4464,
+      "step": 4988
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2629843363561417e-05,
+      "loss": 2.4424,
+      "step": 4989
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.256801319043693e-05,
+      "loss": 2.5486,
+      "step": 4990
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2506183017312447e-05,
+      "loss": 2.3116,
+      "step": 4991
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.244435284418796e-05,
+      "loss": 2.2767,
+      "step": 4992
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2382522671063477e-05,
+      "loss": 2.2643,
+      "step": 4993
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2320692497938993e-05,
+      "loss": 2.3188,
+      "step": 4994
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2258862324814507e-05,
+      "loss": 2.4178,
+      "step": 4995
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.219703215169002e-05,
+      "loss": 2.0053,
+      "step": 4996
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.213520197856554e-05,
+      "loss": 2.2529,
+      "step": 4997
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2073371805441053e-05,
+      "loss": 2.3586,
+      "step": 4998
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2011541632316566e-05,
+      "loss": 2.7795,
+      "step": 4999
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1949711459192083e-05,
+      "loss": 2.1143,
+      "step": 5000
+    },
+    {
+      "epoch": 2.8,
+      "eval_loss": 3.2317845821380615,
+      "eval_runtime": 948.4333,
+      "eval_samples_per_second": 2.786,
+      "eval_steps_per_second": 0.349,
+      "eval_wer": 1.5486513288377628,
+      "step": 5000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.18878812860676e-05,
+      "loss": 2.2169,
+      "step": 5001
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1826051112943116e-05,
+      "loss": 2.3032,
+      "step": 5002
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.176422093981863e-05,
+      "loss": 2.0241,
+      "step": 5003
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1702390766694142e-05,
+      "loss": 2.257,
+      "step": 5004
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1640560593569662e-05,
+      "loss": 2.4208,
+      "step": 5005
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1578730420445176e-05,
+      "loss": 2.2187,
+      "step": 5006
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.151690024732069e-05,
+      "loss": 2.0534,
+      "step": 5007
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1455070074196205e-05,
+      "loss": 2.3813,
+      "step": 5008
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1393239901071722e-05,
+      "loss": 1.8973,
+      "step": 5009
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1331409727947235e-05,
+      "loss": 1.8752,
+      "step": 5010
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.126957955482275e-05,
+      "loss": 2.237,
+      "step": 5011
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1207749381698265e-05,
+      "loss": 2.1558,
+      "step": 5012
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1145919208573785e-05,
+      "loss": 1.9943,
+      "step": 5013
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1084089035449298e-05,
+      "loss": 2.0337,
+      "step": 5014
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.102225886232481e-05,
+      "loss": 1.9167,
+      "step": 5015
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0960428689200324e-05,
+      "loss": 2.1969,
+      "step": 5016
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0898598516075844e-05,
+      "loss": 1.8445,
+      "step": 5017
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0836768342951358e-05,
+      "loss": 1.7434,
+      "step": 5018
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0774938169826874e-05,
+      "loss": 3.0665,
+      "step": 5019
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0713107996702387e-05,
+      "loss": 2.8906,
+      "step": 5020
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0651277823577904e-05,
+      "loss": 3.0285,
+      "step": 5021
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.058944765045342e-05,
+      "loss": 3.1055,
+      "step": 5022
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0527617477328934e-05,
+      "loss": 3.0028,
+      "step": 5023
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0465787304204447e-05,
+      "loss": 2.7006,
+      "step": 5024
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0403957131079967e-05,
+      "loss": 2.529,
+      "step": 5025
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.034212695795548e-05,
+      "loss": 2.827,
+      "step": 5026
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0280296784830993e-05,
+      "loss": 2.2632,
+      "step": 5027
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0218466611706513e-05,
+      "loss": 2.4587,
+      "step": 5028
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0156636438582027e-05,
+      "loss": 2.6087,
+      "step": 5029
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0094806265457543e-05,
+      "loss": 2.3163,
+      "step": 5030
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0032976092333056e-05,
+      "loss": 2.5762,
+      "step": 5031
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9971145919208573e-05,
+      "loss": 2.2134,
+      "step": 5032
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.990931574608409e-05,
+      "loss": 2.2199,
+      "step": 5033
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9847485572959603e-05,
+      "loss": 2.1187,
+      "step": 5034
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9785655399835116e-05,
+      "loss": 2.3099,
+      "step": 5035
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9723825226710636e-05,
+      "loss": 2.2752,
+      "step": 5036
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.966199505358615e-05,
+      "loss": 2.3528,
+      "step": 5037
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9600164880461662e-05,
+      "loss": 2.2677,
+      "step": 5038
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.953833470733718e-05,
+      "loss": 2.3986,
+      "step": 5039
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9476504534212695e-05,
+      "loss": 2.3142,
+      "step": 5040
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.941467436108821e-05,
+      "loss": 2.1428,
+      "step": 5041
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9352844187963725e-05,
+      "loss": 2.436,
+      "step": 5042
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.929101401483924e-05,
+      "loss": 2.4329,
+      "step": 5043
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.922918384171476e-05,
+      "loss": 2.5694,
+      "step": 5044
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.916735366859027e-05,
+      "loss": 2.033,
+      "step": 5045
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9105523495465785e-05,
+      "loss": 2.275,
+      "step": 5046
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9043693322341298e-05,
+      "loss": 2.2277,
+      "step": 5047
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8981863149216818e-05,
+      "loss": 2.2289,
+      "step": 5048
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.892003297609233e-05,
+      "loss": 2.004,
+      "step": 5049
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8858202802967848e-05,
+      "loss": 2.5225,
+      "step": 5050
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.879637262984336e-05,
+      "loss": 2.1989,
+      "step": 5051
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8734542456718878e-05,
+      "loss": 2.2927,
+      "step": 5052
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8672712283594394e-05,
+      "loss": 2.2553,
+      "step": 5053
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8610882110469907e-05,
+      "loss": 2.1151,
+      "step": 5054
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8549051937345424e-05,
+      "loss": 2.4077,
+      "step": 5055
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8487221764220937e-05,
+      "loss": 1.9916,
+      "step": 5056
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8425391591096454e-05,
+      "loss": 2.6951,
+      "step": 5057
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8363561417971967e-05,
+      "loss": 2.1636,
+      "step": 5058
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8301731244847483e-05,
+      "loss": 2.3092,
+      "step": 5059
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8239901071723e-05,
+      "loss": 2.1181,
+      "step": 5060
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8178070898598517e-05,
+      "loss": 2.3036,
+      "step": 5061
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.811624072547403e-05,
+      "loss": 2.1974,
+      "step": 5062
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8054410552349546e-05,
+      "loss": 2.4448,
+      "step": 5063
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.799258037922506e-05,
+      "loss": 2.0804,
+      "step": 5064
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7930750206100576e-05,
+      "loss": 1.8604,
+      "step": 5065
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.786892003297609e-05,
+      "loss": 1.8091,
+      "step": 5066
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7807089859851606e-05,
+      "loss": 1.5319,
+      "step": 5067
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7745259686727123e-05,
+      "loss": 2.1672,
+      "step": 5068
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7683429513602636e-05,
+      "loss": 2.6822,
+      "step": 5069
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7621599340478152e-05,
+      "loss": 3.1044,
+      "step": 5070
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.755976916735367e-05,
+      "loss": 2.9838,
+      "step": 5071
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7497938994229182e-05,
+      "loss": 2.688,
+      "step": 5072
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.74361088211047e-05,
+      "loss": 2.7374,
+      "step": 5073
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7374278647980212e-05,
+      "loss": 2.7019,
+      "step": 5074
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.731244847485573e-05,
+      "loss": 2.4291,
+      "step": 5075
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7250618301731242e-05,
+      "loss": 2.1496,
+      "step": 5076
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7188788128606758e-05,
+      "loss": 2.6527,
+      "step": 5077
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7126957955482275e-05,
+      "loss": 2.496,
+      "step": 5078
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7065127782357788e-05,
+      "loss": 2.3012,
+      "step": 5079
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7003297609233305e-05,
+      "loss": 2.5714,
+      "step": 5080
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.694146743610882e-05,
+      "loss": 2.2526,
+      "step": 5081
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6879637262984334e-05,
+      "loss": 2.104,
+      "step": 5082
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.681780708985985e-05,
+      "loss": 2.4734,
+      "step": 5083
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6755976916735364e-05,
+      "loss": 2.5679,
+      "step": 5084
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.669414674361088e-05,
+      "loss": 2.2989,
+      "step": 5085
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6632316570486394e-05,
+      "loss": 2.2684,
+      "step": 5086
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.657048639736191e-05,
+      "loss": 2.098,
+      "step": 5087
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6508656224237427e-05,
+      "loss": 2.3435,
+      "step": 5088
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6446826051112944e-05,
+      "loss": 2.0616,
+      "step": 5089
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6384995877988457e-05,
+      "loss": 2.2738,
+      "step": 5090
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6323165704863974e-05,
+      "loss": 2.0683,
+      "step": 5091
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6261335531739487e-05,
+      "loss": 2.388,
+      "step": 5092
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6199505358615003e-05,
+      "loss": 2.3148,
+      "step": 5093
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6137675185490517e-05,
+      "loss": 2.3002,
+      "step": 5094
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6075845012366033e-05,
+      "loss": 2.0722,
+      "step": 5095
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6014014839241546e-05,
+      "loss": 2.2355,
+      "step": 5096
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5952184666117063e-05,
+      "loss": 2.0141,
+      "step": 5097
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.589035449299258e-05,
+      "loss": 2.2933,
+      "step": 5098
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5828524319868096e-05,
+      "loss": 2.3266,
+      "step": 5099
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.576669414674361e-05,
+      "loss": 2.1402,
+      "step": 5100
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5704863973619126e-05,
+      "loss": 2.1304,
+      "step": 5101
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.564303380049464e-05,
+      "loss": 2.238,
+      "step": 5102
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5581203627370156e-05,
+      "loss": 2.0104,
+      "step": 5103
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.551937345424567e-05,
+      "loss": 2.0661,
+      "step": 5104
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5457543281121185e-05,
+      "loss": 1.9962,
+      "step": 5105
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.53957131079967e-05,
+      "loss": 1.8585,
+      "step": 5106
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5333882934872215e-05,
+      "loss": 2.0366,
+      "step": 5107
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5272052761747732e-05,
+      "loss": 1.8373,
+      "step": 5108
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5210222588623247e-05,
+      "loss": 1.809,
+      "step": 5109
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5148392415498762e-05,
+      "loss": 2.1587,
+      "step": 5110
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5086562242374278e-05,
+      "loss": 1.9667,
+      "step": 5111
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5024732069249791e-05,
+      "loss": 1.8559,
+      "step": 5112
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4962901896125308e-05,
+      "loss": 1.683,
+      "step": 5113
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4901071723000823e-05,
+      "loss": 1.9702,
+      "step": 5114
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4839241549876338e-05,
+      "loss": 2.0826,
+      "step": 5115
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4777411376751853e-05,
+      "loss": 1.7309,
+      "step": 5116
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.471558120362737e-05,
+      "loss": 1.8815,
+      "step": 5117
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4653751030502882e-05,
+      "loss": 1.4094,
+      "step": 5118
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4591920857378399e-05,
+      "loss": 2.9086,
+      "step": 5119
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4530090684253914e-05,
+      "loss": 2.9554,
+      "step": 5120
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.446826051112943e-05,
+      "loss": 2.4842,
+      "step": 5121
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4406430338004944e-05,
+      "loss": 2.6614,
+      "step": 5122
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.434460016488046e-05,
+      "loss": 2.5313,
+      "step": 5123
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4282769991755975e-05,
+      "loss": 2.4357,
+      "step": 5124
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4220939818631492e-05,
+      "loss": 2.7676,
+      "step": 5125
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4159109645507005e-05,
+      "loss": 2.6628,
+      "step": 5126
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4097279472382522e-05,
+      "loss": 2.2963,
+      "step": 5127
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4035449299258036e-05,
+      "loss": 2.3948,
+      "step": 5128
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3973619126133551e-05,
+      "loss": 2.4447,
+      "step": 5129
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3911788953009066e-05,
+      "loss": 2.3338,
+      "step": 5130
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3849958779884583e-05,
+      "loss": 2.5885,
+      "step": 5131
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3788128606760096e-05,
+      "loss": 2.2513,
+      "step": 5132
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3726298433635613e-05,
+      "loss": 2.4206,
+      "step": 5133
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3664468260511127e-05,
+      "loss": 2.0851,
+      "step": 5134
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3602638087386644e-05,
+      "loss": 2.3233,
+      "step": 5135
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3540807914262159e-05,
+      "loss": 2.2493,
+      "step": 5136
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3478977741137674e-05,
+      "loss": 2.3537,
+      "step": 5137
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.341714756801319e-05,
+      "loss": 2.0938,
+      "step": 5138
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3355317394888704e-05,
+      "loss": 2.1243,
+      "step": 5139
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.329348722176422e-05,
+      "loss": 2.2873,
+      "step": 5140
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3231657048639735e-05,
+      "loss": 2.3587,
+      "step": 5141
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3169826875515252e-05,
+      "loss": 2.4031,
+      "step": 5142
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3107996702390765e-05,
+      "loss": 2.5001,
+      "step": 5143
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3046166529266281e-05,
+      "loss": 2.1788,
+      "step": 5144
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2984336356141796e-05,
+      "loss": 2.1788,
+      "step": 5145
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2922506183017313e-05,
+      "loss": 2.1469,
+      "step": 5146
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2860676009892826e-05,
+      "loss": 2.2705,
+      "step": 5147
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2798845836768343e-05,
+      "loss": 1.933,
+      "step": 5148
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2737015663643858e-05,
+      "loss": 2.1987,
+      "step": 5149
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2675185490519373e-05,
+      "loss": 2.3922,
+      "step": 5150
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2613355317394887e-05,
+      "loss": 1.936,
+      "step": 5151
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2551525144270404e-05,
+      "loss": 2.2429,
+      "step": 5152
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2489694971145917e-05,
+      "loss": 2.3664,
+      "step": 5153
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2427864798021434e-05,
+      "loss": 2.2968,
+      "step": 5154
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2366034624896949e-05,
+      "loss": 2.0839,
+      "step": 5155
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2304204451772465e-05,
+      "loss": 1.7959,
+      "step": 5156
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2242374278647978e-05,
+      "loss": 2.0235,
+      "step": 5157
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2180544105523495e-05,
+      "loss": 2.9132,
+      "step": 5158
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.211871393239901e-05,
+      "loss": 1.8626,
+      "step": 5159
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2056883759274527e-05,
+      "loss": 2.1285,
+      "step": 5160
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.199505358615004e-05,
+      "loss": 2.264,
+      "step": 5161
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1933223413025556e-05,
+      "loss": 2.1299,
+      "step": 5162
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1871393239901071e-05,
+      "loss": 1.751,
+      "step": 5163
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1809563066776586e-05,
+      "loss": 1.8856,
+      "step": 5164
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1747732893652101e-05,
+      "loss": 2.1998,
+      "step": 5165
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1685902720527618e-05,
+      "loss": 1.9353,
+      "step": 5166
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.162407254740313e-05,
+      "loss": 1.6399,
+      "step": 5167
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1562242374278647e-05,
+      "loss": 1.9308,
+      "step": 5168
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1500412201154162e-05,
+      "loss": 2.6961,
+      "step": 5169
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1438582028029679e-05,
+      "loss": 2.4171,
+      "step": 5170
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1376751854905192e-05,
+      "loss": 2.4758,
+      "step": 5171
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1314921681780709e-05,
+      "loss": 2.5269,
+      "step": 5172
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1253091508656224e-05,
+      "loss": 2.1876,
+      "step": 5173
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1191261335531738e-05,
+      "loss": 2.6851,
+      "step": 5174
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1129431162407253e-05,
+      "loss": 2.3189,
+      "step": 5175
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.106760098928277e-05,
+      "loss": 2.5802,
+      "step": 5176
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1005770816158283e-05,
+      "loss": 2.2753,
+      "step": 5177
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.09439406430338e-05,
+      "loss": 2.3881,
+      "step": 5178
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0882110469909315e-05,
+      "loss": 2.1024,
+      "step": 5179
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0820280296784831e-05,
+      "loss": 2.4391,
+      "step": 5180
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0758450123660344e-05,
+      "loss": 2.3527,
+      "step": 5181
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0696619950535861e-05,
+      "loss": 2.3839,
+      "step": 5182
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0634789777411376e-05,
+      "loss": 2.2073,
+      "step": 5183
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0572959604286892e-05,
+      "loss": 2.1523,
+      "step": 5184
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0511129431162406e-05,
+      "loss": 2.2609,
+      "step": 5185
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0449299258037922e-05,
+      "loss": 2.2032,
+      "step": 5186
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0387469084913437e-05,
+      "loss": 2.3994,
+      "step": 5187
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0325638911788952e-05,
+      "loss": 1.9978,
+      "step": 5188
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0263808738664467e-05,
+      "loss": 2.0159,
+      "step": 5189
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0201978565539983e-05,
+      "loss": 2.0758,
+      "step": 5190
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0140148392415497e-05,
+      "loss": 2.201,
+      "step": 5191
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0078318219291013e-05,
+      "loss": 2.0158,
+      "step": 5192
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0016488046166528e-05,
+      "loss": 2.0916,
+      "step": 5193
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.954657873042045e-06,
+      "loss": 2.3911,
+      "step": 5194
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.892827699917558e-06,
+      "loss": 2.3981,
+      "step": 5195
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.830997526793075e-06,
+      "loss": 2.2624,
+      "step": 5196
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.76916735366859e-06,
+      "loss": 2.4231,
+      "step": 5197
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.707337180544104e-06,
+      "loss": 2.0824,
+      "step": 5198
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.64550700741962e-06,
+      "loss": 2.4005,
+      "step": 5199
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.583676834295136e-06,
+      "loss": 2.1623,
+      "step": 5200
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.521846661170649e-06,
+      "loss": 2.2918,
+      "step": 5201
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.460016488046166e-06,
+      "loss": 1.9937,
+      "step": 5202
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.39818631492168e-06,
+      "loss": 2.1187,
+      "step": 5203
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.336356141797197e-06,
+      "loss": 1.9529,
+      "step": 5204
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.274525968672712e-06,
+      "loss": 1.9932,
+      "step": 5205
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.212695795548227e-06,
+      "loss": 1.8874,
+      "step": 5206
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.150865622423742e-06,
+      "loss": 1.9051,
+      "step": 5207
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.089035449299258e-06,
+      "loss": 2.1852,
+      "step": 5208
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.027205276174773e-06,
+      "loss": 2.1594,
+      "step": 5209
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.965375103050288e-06,
+      "loss": 1.9314,
+      "step": 5210
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.903544929925803e-06,
+      "loss": 1.8818,
+      "step": 5211
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.841714756801318e-06,
+      "loss": 1.968,
+      "step": 5212
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.779884583676834e-06,
+      "loss": 1.6102,
+      "step": 5213
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.71805441055235e-06,
+      "loss": 1.7799,
+      "step": 5214
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.656224237427864e-06,
+      "loss": 2.1957,
+      "step": 5215
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.594394064303379e-06,
+      "loss": 1.5883,
+      "step": 5216
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.532563891178894e-06,
+      "loss": 1.9527,
+      "step": 5217
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.47073371805441e-06,
+      "loss": 1.5981,
+      "step": 5218
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.408903544929926e-06,
+      "loss": 2.4788,
+      "step": 5219
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.34707337180544e-06,
+      "loss": 3.3907,
+      "step": 5220
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.285243198680955e-06,
+      "loss": 2.5009,
+      "step": 5221
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.223413025556472e-06,
+      "loss": 2.2684,
+      "step": 5222
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.161582852431987e-06,
+      "loss": 2.7057,
+      "step": 5223
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.099752679307502e-06,
+      "loss": 2.518,
+      "step": 5224
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.037922506183017e-06,
+      "loss": 2.7743,
+      "step": 5225
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.976092333058531e-06,
+      "loss": 2.354,
+      "step": 5226
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.914262159934048e-06,
+      "loss": 2.4532,
+      "step": 5227
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.852431986809563e-06,
+      "loss": 2.241,
+      "step": 5228
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.790601813685078e-06,
+      "loss": 2.0373,
+      "step": 5229
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.728771640560593e-06,
+      "loss": 2.326,
+      "step": 5230
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.666941467436108e-06,
+      "loss": 1.9683,
+      "step": 5231
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.605111294311623e-06,
+      "loss": 2.4748,
+      "step": 5232
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.543281121187139e-06,
+      "loss": 2.095,
+      "step": 5233
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.481450948062654e-06,
+      "loss": 1.9974,
+      "step": 5234
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.419620774938169e-06,
+      "loss": 2.265,
+      "step": 5235
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.357790601813685e-06,
+      "loss": 2.4797,
+      "step": 5236
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.2959604286891995e-06,
+      "loss": 1.9012,
+      "step": 5237
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.234130255564715e-06,
+      "loss": 2.5605,
+      "step": 5238
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.17230008244023e-06,
+      "loss": 2.1313,
+      "step": 5239
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.110469909315746e-06,
+      "loss": 1.9898,
+      "step": 5240
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.048639736191261e-06,
+      "loss": 2.3625,
+      "step": 5241
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.986809563066776e-06,
+      "loss": 1.9239,
+      "step": 5242
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.924979389942291e-06,
+      "loss": 2.1728,
+      "step": 5243
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.863149216817806e-06,
+      "loss": 2.2624,
+      "step": 5244
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.801319043693322e-06,
+      "loss": 1.9828,
+      "step": 5245
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.739488870568837e-06,
+      "loss": 2.069,
+      "step": 5246
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.677658697444352e-06,
+      "loss": 2.4456,
+      "step": 5247
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.6158285243198676e-06,
+      "loss": 1.8945,
+      "step": 5248
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.5539983511953824e-06,
+      "loss": 2.1312,
+      "step": 5249
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.492168178070898e-06,
+      "loss": 2.1015,
+      "step": 5250
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.430338004946413e-06,
+      "loss": 2.0997,
+      "step": 5251
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.368507831821929e-06,
+      "loss": 2.0262,
+      "step": 5252
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.306677658697444e-06,
+      "loss": 2.2074,
+      "step": 5253
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.244847485572959e-06,
+      "loss": 2.084,
+      "step": 5254
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.183017312448474e-06,
+      "loss": 2.0037,
+      "step": 5255
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.121187139323989e-06,
+      "loss": 1.9047,
+      "step": 5256
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.059356966199505e-06,
+      "loss": 2.1899,
+      "step": 5257
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.99752679307502e-06,
+      "loss": 1.8427,
+      "step": 5258
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.935696619950536e-06,
+      "loss": 1.8846,
+      "step": 5259
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.8738664468260505e-06,
+      "loss": 2.204,
+      "step": 5260
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.812036273701565e-06,
+      "loss": 1.6547,
+      "step": 5261
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.750206100577081e-06,
+      "loss": 1.7926,
+      "step": 5262
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.688375927452596e-06,
+      "loss": 1.9742,
+      "step": 5263
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.626545754328112e-06,
+      "loss": 1.6238,
+      "step": 5264
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.564715581203627e-06,
+      "loss": 1.419,
+      "step": 5265
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.5028854080791415e-06,
+      "loss": 1.6485,
+      "step": 5266
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.441055234954657e-06,
+      "loss": 1.8574,
+      "step": 5267
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.379225061830172e-06,
+      "loss": 1.8871,
+      "step": 5268
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.317394888705688e-06,
+      "loss": 2.7201,
+      "step": 5269
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.255564715581203e-06,
+      "loss": 2.2427,
+      "step": 5270
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.1937345424567185e-06,
+      "loss": 2.5061,
+      "step": 5271
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.1319043693322334e-06,
+      "loss": 2.5382,
+      "step": 5272
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.070074196207748e-06,
+      "loss": 2.2722,
+      "step": 5273
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.008244023083264e-06,
+      "loss": 2.5137,
+      "step": 5274
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.946413849958779e-06,
+      "loss": 2.1276,
+      "step": 5275
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.884583676834295e-06,
+      "loss": 2.1236,
+      "step": 5276
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.82275350370981e-06,
+      "loss": 2.5583,
+      "step": 5277
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.7609233305853245e-06,
+      "loss": 2.5587,
+      "step": 5278
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.69909315746084e-06,
+      "loss": 2.0686,
+      "step": 5279
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.637262984336356e-06,
+      "loss": 1.9335,
+      "step": 5280
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.575432811211871e-06,
+      "loss": 2.2854,
+      "step": 5281
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.513602638087387e-06,
+      "loss": 2.2711,
+      "step": 5282
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4517724649629015e-06,
+      "loss": 2.2849,
+      "step": 5283
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.389942291838417e-06,
+      "loss": 2.248,
+      "step": 5284
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.328112118713932e-06,
+      "loss": 2.1879,
+      "step": 5285
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.266281945589447e-06,
+      "loss": 2.38,
+      "step": 5286
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.204451772464963e-06,
+      "loss": 2.0492,
+      "step": 5287
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.142621599340478e-06,
+      "loss": 1.9724,
+      "step": 5288
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.080791426215993e-06,
+      "loss": 2.2804,
+      "step": 5289
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.018961253091508e-06,
+      "loss": 2.1429,
+      "step": 5290
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.957131079967024e-06,
+      "loss": 2.0707,
+      "step": 5291
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.895300906842539e-06,
+      "loss": 2.1888,
+      "step": 5292
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.833470733718054e-06,
+      "loss": 2.1909,
+      "step": 5293
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.7716405605935695e-06,
+      "loss": 2.1978,
+      "step": 5294
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.7098103874690844e-06,
+      "loss": 2.315,
+      "step": 5295
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6479802143445998e-06,
+      "loss": 2.2572,
+      "step": 5296
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.586150041220115e-06,
+      "loss": 2.025,
+      "step": 5297
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5243198680956304e-06,
+      "loss": 2.3348,
+      "step": 5298
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.4624896949711457e-06,
+      "loss": 2.0629,
+      "step": 5299
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.400659521846661e-06,
+      "loss": 2.0995,
+      "step": 5300
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.338829348722176e-06,
+      "loss": 2.7019,
+      "step": 5301
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2769991755976912e-06,
+      "loss": 2.0792,
+      "step": 5302
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2151690024732065e-06,
+      "loss": 2.3832,
+      "step": 5303
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.153338829348722e-06,
+      "loss": 2.1449,
+      "step": 5304
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.091508656224237e-06,
+      "loss": 1.8144,
+      "step": 5305
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.0296784830997525e-06,
+      "loss": 1.8389,
+      "step": 5306
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.967848309975268e-06,
+      "loss": 2.0717,
+      "step": 5307
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9060181368507827e-06,
+      "loss": 2.0828,
+      "step": 5308
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.844187963726298e-06,
+      "loss": 1.9322,
+      "step": 5309
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7823577906018133e-06,
+      "loss": 2.013,
+      "step": 5310
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7205276174773286e-06,
+      "loss": 1.7999,
+      "step": 5311
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.658697444352844e-06,
+      "loss": 1.9776,
+      "step": 5312
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5968672712283593e-06,
+      "loss": 1.9969,
+      "step": 5313
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.535037098103874e-06,
+      "loss": 1.8942,
+      "step": 5314
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4732069249793895e-06,
+      "loss": 1.806,
+      "step": 5315
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.411376751854905e-06,
+      "loss": 1.2892,
+      "step": 5316
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.34954657873042e-06,
+      "loss": 1.7242,
+      "step": 5317
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2877164056059354e-06,
+      "loss": 1.1839,
+      "step": 5318
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2258862324814507e-06,
+      "loss": 2.499,
+      "step": 5319
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.164056059356966e-06,
+      "loss": 2.4693,
+      "step": 5320
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1022258862324814e-06,
+      "loss": 2.2657,
+      "step": 5321
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0403957131079967e-06,
+      "loss": 2.3127,
+      "step": 5322
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.978565539983512e-06,
+      "loss": 2.1621,
+      "step": 5323
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.916735366859027e-06,
+      "loss": 2.3706,
+      "step": 5324
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8549051937345422e-06,
+      "loss": 2.323,
+      "step": 5325
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7930750206100575e-06,
+      "loss": 2.2049,
+      "step": 5326
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7312448474855728e-06,
+      "loss": 2.4171,
+      "step": 5327
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.669414674361088e-06,
+      "loss": 2.2707,
+      "step": 5328
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6075845012366033e-06,
+      "loss": 2.3797,
+      "step": 5329
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5457543281121186e-06,
+      "loss": 2.2425,
+      "step": 5330
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.483924154987634e-06,
+      "loss": 2.4373,
+      "step": 5331
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.422093981863149e-06,
+      "loss": 2.0473,
+      "step": 5332
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3602638087386643e-06,
+      "loss": 2.2637,
+      "step": 5333
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2984336356141796e-06,
+      "loss": 2.0147,
+      "step": 5334
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2366034624896947e-06,
+      "loss": 2.1401,
+      "step": 5335
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.17477328936521e-06,
+      "loss": 1.9557,
+      "step": 5336
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1129431162407254e-06,
+      "loss": 2.3407,
+      "step": 5337
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0511129431162407e-06,
+      "loss": 2.1183,
+      "step": 5338
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.89282769991756e-07,
+      "loss": 2.1582,
+      "step": 5339
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.274525968672711e-07,
+      "loss": 2.0458,
+      "step": 5340
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.656224237427864e-07,
+      "loss": 2.0737,
+      "step": 5341
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.037922506183016e-07,
+      "loss": 2.0112,
+      "step": 5342
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.41962077493817e-07,
+      "loss": 1.8716,
+      "step": 5343
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.801319043693322e-07,
+      "loss": 2.0319,
+      "step": 5344
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.183017312448474e-07,
+      "loss": 1.9679,
+      "step": 5345
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.564715581203627e-07,
+      "loss": 2.1353,
+      "step": 5346
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.94641384995878e-07,
+      "loss": 2.0299,
+      "step": 5347
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.328112118713932e-07,
+      "loss": 2.045,
+      "step": 5348
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.709810387469085e-07,
+      "loss": 1.99,
+      "step": 5349
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.091508656224237e-07,
+      "loss": 1.7072,
+      "step": 5350
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.47320692497939e-07,
+      "loss": 1.5116,
+      "step": 5351
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8549051937345424e-07,
+      "loss": 1.4431,
+      "step": 5352
+    },
+    {
+      "epoch": 3.0,
+      "step": 5352,
+      "total_flos": 0.0,
+      "train_loss": 4.03557997189117,
+      "train_runtime": 27620.0868,
+      "train_samples_per_second": 3.1,
+      "train_steps_per_second": 0.194
+    }
+  ],
+  "max_steps": 5352,
+  "num_train_epochs": 3,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}