diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,60825 @@
+{
+  "best_metric": 0.8935389133627019,
+  "best_model_checkpoint": "xtreme_s_xlsr_t5lephone-small_minds14.en-all/checkpoint-9400",
+  "epoch": 149.99628252788105,
+  "global_step": 10050,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999999999999996e-07,
+      "loss": 2.6343,
+      "step": 1
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999999999999993e-07,
+      "loss": 2.6511,
+      "step": 2
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6e-07,
+      "loss": 2.637,
+      "step": 3
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.999999999999999e-07,
+      "loss": 2.6419,
+      "step": 4
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1e-06,
+      "loss": 2.6541,
+      "step": 5
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2e-06,
+      "loss": 2.6411,
+      "step": 6
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4e-06,
+      "loss": 2.6381,
+      "step": 7
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5999999999999997e-06,
+      "loss": 2.6385,
+      "step": 8
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8e-06,
+      "loss": 2.6384,
+      "step": 9
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-06,
+      "loss": 2.6368,
+      "step": 10
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1999999999999997e-06,
+      "loss": 2.6385,
+      "step": 11
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4e-06,
+      "loss": 2.6431,
+      "step": 12
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5999999999999997e-06,
+      "loss": 2.6428,
+      "step": 13
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8e-06,
+      "loss": 2.6378,
+      "step": 14
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9999999999999997e-06,
+      "loss": 2.6362,
+      "step": 15
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.1999999999999994e-06,
+      "loss": 2.6277,
+      "step": 16
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4e-06,
+      "loss": 2.6533,
+      "step": 17
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6e-06,
+      "loss": 2.645,
+      "step": 18
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7999999999999996e-06,
+      "loss": 2.644,
+      "step": 19
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4e-06,
+      "loss": 2.6494,
+      "step": 20
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.2e-06,
+      "loss": 2.6442,
+      "step": 21
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.399999999999999e-06,
+      "loss": 2.638,
+      "step": 22
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.599999999999999e-06,
+      "loss": 2.6362,
+      "step": 23
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8e-06,
+      "loss": 2.6443,
+      "step": 24
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9999999999999996e-06,
+      "loss": 2.6295,
+      "step": 25
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.199999999999999e-06,
+      "loss": 2.6204,
+      "step": 26
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.399999999999999e-06,
+      "loss": 2.6425,
+      "step": 27
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.6e-06,
+      "loss": 2.6329,
+      "step": 28
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7999999999999995e-06,
+      "loss": 2.6295,
+      "step": 29
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 2.6281,
+      "step": 30
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.199999999999999e-06,
+      "loss": 2.6481,
+      "step": 31
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.399999999999999e-06,
+      "loss": 2.6351,
+      "step": 32
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.599999999999999e-06,
+      "loss": 2.6362,
+      "step": 33
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.8e-06,
+      "loss": 2.6711,
+      "step": 34
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7e-06,
+      "loss": 2.6709,
+      "step": 35
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.2e-06,
+      "loss": 2.6567,
+      "step": 36
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.3999999999999995e-06,
+      "loss": 2.6572,
+      "step": 37
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.599999999999999e-06,
+      "loss": 2.6388,
+      "step": 38
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.799999999999998e-06,
+      "loss": 2.6439,
+      "step": 39
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8e-06,
+      "loss": 2.6409,
+      "step": 40
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.2e-06,
+      "loss": 2.6417,
+      "step": 41
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.4e-06,
+      "loss": 2.6226,
+      "step": 42
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 2.6615,
+      "step": 43
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.799999999999999e-06,
+      "loss": 2.6323,
+      "step": 44
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.999999999999999e-06,
+      "loss": 2.6342,
+      "step": 45
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.199999999999998e-06,
+      "loss": 2.6494,
+      "step": 46
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.399999999999998e-06,
+      "loss": 2.6318,
+      "step": 47
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.6e-06,
+      "loss": 2.6485,
+      "step": 48
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.799999999999998e-06,
+      "loss": 2.6277,
+      "step": 49
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.999999999999999e-06,
+      "loss": 2.644,
+      "step": 50
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.02e-05,
+      "loss": 2.6556,
+      "step": 51
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.0399999999999999e-05,
+      "loss": 2.6486,
+      "step": 52
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.06e-05,
+      "loss": 2.632,
+      "step": 53
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0799999999999998e-05,
+      "loss": 2.6526,
+      "step": 54
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1e-05,
+      "loss": 2.6667,
+      "step": 55
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.12e-05,
+      "loss": 2.6432,
+      "step": 56
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.14e-05,
+      "loss": 2.6647,
+      "step": 57
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1599999999999999e-05,
+      "loss": 2.6317,
+      "step": 58
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1799999999999999e-05,
+      "loss": 2.6349,
+      "step": 59
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 2.6423,
+      "step": 60
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2199999999999998e-05,
+      "loss": 2.6268,
+      "step": 61
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.2399999999999998e-05,
+      "loss": 2.6271,
+      "step": 62
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.26e-05,
+      "loss": 2.603,
+      "step": 63
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2799999999999998e-05,
+      "loss": 2.5958,
+      "step": 64
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3e-05,
+      "loss": 2.6169,
+      "step": 65
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3199999999999997e-05,
+      "loss": 2.5735,
+      "step": 66
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3399999999999999e-05,
+      "loss": 2.6467,
+      "step": 67
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.36e-05,
+      "loss": 3.3148,
+      "step": 68
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3799999999999998e-05,
+      "loss": 2.6297,
+      "step": 69
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4e-05,
+      "loss": 2.6577,
+      "step": 70
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4199999999999998e-05,
+      "loss": 2.6311,
+      "step": 71
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.44e-05,
+      "loss": 2.616,
+      "step": 72
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4599999999999997e-05,
+      "loss": 2.6987,
+      "step": 73
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4799999999999999e-05,
+      "loss": 2.6268,
+      "step": 74
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 2.565,
+      "step": 75
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5199999999999998e-05,
+      "loss": 2.6689,
+      "step": 76
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5399999999999998e-05,
+      "loss": 2.6792,
+      "step": 77
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5599999999999996e-05,
+      "loss": 2.5829,
+      "step": 78
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5799999999999998e-05,
+      "loss": 2.6033,
+      "step": 79
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6e-05,
+      "loss": 2.5901,
+      "step": 80
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6199999999999997e-05,
+      "loss": 2.5608,
+      "step": 81
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.64e-05,
+      "loss": 2.5485,
+      "step": 82
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6599999999999997e-05,
+      "loss": 2.5484,
+      "step": 83
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.68e-05,
+      "loss": 2.602,
+      "step": 84
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.6999999999999996e-05,
+      "loss": 2.6658,
+      "step": 85
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 2.6671,
+      "step": 86
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.74e-05,
+      "loss": 2.676,
+      "step": 87
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7599999999999998e-05,
+      "loss": 2.6539,
+      "step": 88
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.78e-05,
+      "loss": 2.6765,
+      "step": 89
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 2.6133,
+      "step": 90
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.82e-05,
+      "loss": 2.639,
+      "step": 91
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8399999999999997e-05,
+      "loss": 2.6448,
+      "step": 92
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8599999999999998e-05,
+      "loss": 2.6431,
+      "step": 93
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8799999999999996e-05,
+      "loss": 2.6401,
+      "step": 94
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9e-05,
+      "loss": 2.6407,
+      "step": 95
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.92e-05,
+      "loss": 2.617,
+      "step": 96
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9399999999999997e-05,
+      "loss": 2.6529,
+      "step": 97
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9599999999999995e-05,
+      "loss": 2.6185,
+      "step": 98
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.98e-05,
+      "loss": 2.6339,
+      "step": 99
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 2.6058,
+      "step": 100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.0199999999999996e-05,
+      "loss": 2.6031,
+      "step": 101
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.04e-05,
+      "loss": 2.6588,
+      "step": 102
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.06e-05,
+      "loss": 2.6377,
+      "step": 103
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.0799999999999997e-05,
+      "loss": 2.6863,
+      "step": 104
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1e-05,
+      "loss": 2.6617,
+      "step": 105
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.12e-05,
+      "loss": 2.6473,
+      "step": 106
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.14e-05,
+      "loss": 2.6711,
+      "step": 107
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1599999999999996e-05,
+      "loss": 2.6132,
+      "step": 108
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1799999999999998e-05,
+      "loss": 2.6237,
+      "step": 109
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2e-05,
+      "loss": 2.645,
+      "step": 110
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2199999999999998e-05,
+      "loss": 2.6069,
+      "step": 111
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.24e-05,
+      "loss": 2.642,
+      "step": 112
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2599999999999997e-05,
+      "loss": 2.6126,
+      "step": 113
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.28e-05,
+      "loss": 2.6048,
+      "step": 114
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.2999999999999997e-05,
+      "loss": 2.6229,
+      "step": 115
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3199999999999998e-05,
+      "loss": 2.5935,
+      "step": 116
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.34e-05,
+      "loss": 2.5616,
+      "step": 117
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.3599999999999998e-05,
+      "loss": 2.6095,
+      "step": 118
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.38e-05,
+      "loss": 2.5821,
+      "step": 119
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 2.6188,
+      "step": 120
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.42e-05,
+      "loss": 2.6443,
+      "step": 121
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4399999999999997e-05,
+      "loss": 2.6284,
+      "step": 122
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 2.6273,
+      "step": 123
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.4799999999999996e-05,
+      "loss": 2.6608,
+      "step": 124
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4999999999999998e-05,
+      "loss": 2.6192,
+      "step": 125
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.52e-05,
+      "loss": 2.6312,
+      "step": 126
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5399999999999997e-05,
+      "loss": 2.6786,
+      "step": 127
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.5599999999999995e-05,
+      "loss": 2.6266,
+      "step": 128
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.5799999999999997e-05,
+      "loss": 2.6833,
+      "step": 129
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6e-05,
+      "loss": 2.6481,
+      "step": 130
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.6199999999999996e-05,
+      "loss": 2.5106,
+      "step": 131
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.6399999999999995e-05,
+      "loss": 2.5399,
+      "step": 132
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.66e-05,
+      "loss": 2.5729,
+      "step": 133
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.6799999999999998e-05,
+      "loss": 2.6409,
+      "step": 134
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 3.1924,
+      "step": 135
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.72e-05,
+      "loss": 2.5826,
+      "step": 136
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.74e-05,
+      "loss": 2.6108,
+      "step": 137
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.7599999999999997e-05,
+      "loss": 2.5618,
+      "step": 138
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.7799999999999995e-05,
+      "loss": 2.6187,
+      "step": 139
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.8e-05,
+      "loss": 2.5749,
+      "step": 140
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 2.5772,
+      "step": 141
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8399999999999996e-05,
+      "loss": 2.6725,
+      "step": 142
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.86e-05,
+      "loss": 2.5364,
+      "step": 143
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.88e-05,
+      "loss": 2.6227,
+      "step": 144
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8999999999999997e-05,
+      "loss": 2.6355,
+      "step": 145
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.9199999999999995e-05,
+      "loss": 2.4874,
+      "step": 146
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.94e-05,
+      "loss": 2.5832,
+      "step": 147
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9599999999999998e-05,
+      "loss": 2.5628,
+      "step": 148
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.9799999999999996e-05,
+      "loss": 2.5375,
+      "step": 149
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 2.5401,
+      "step": 150
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.02e-05,
+      "loss": 2.6058,
+      "step": 151
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0399999999999997e-05,
+      "loss": 2.6267,
+      "step": 152
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.06e-05,
+      "loss": 2.5839,
+      "step": 153
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.0799999999999996e-05,
+      "loss": 2.6529,
+      "step": 154
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0999999999999995e-05,
+      "loss": 2.6135,
+      "step": 155
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.119999999999999e-05,
+      "loss": 2.6257,
+      "step": 156
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.14e-05,
+      "loss": 2.7242,
+      "step": 157
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.1599999999999996e-05,
+      "loss": 2.6371,
+      "step": 158
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.1799999999999994e-05,
+      "loss": 2.5878,
+      "step": 159
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.2e-05,
+      "loss": 2.6008,
+      "step": 160
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.22e-05,
+      "loss": 2.5713,
+      "step": 161
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.2399999999999995e-05,
+      "loss": 2.6273,
+      "step": 162
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.259999999999999e-05,
+      "loss": 2.5838,
+      "step": 163
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.28e-05,
+      "loss": 2.5846,
+      "step": 164
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 2.562,
+      "step": 165
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.3199999999999994e-05,
+      "loss": 2.5598,
+      "step": 166
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.34e-05,
+      "loss": 2.4896,
+      "step": 167
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.36e-05,
+      "loss": 2.5917,
+      "step": 168
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3799999999999995e-05,
+      "loss": 2.6609,
+      "step": 169
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.399999999999999e-05,
+      "loss": 2.653,
+      "step": 170
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.42e-05,
+      "loss": 2.6373,
+      "step": 171
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 2.6079,
+      "step": 172
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.4599999999999994e-05,
+      "loss": 2.616,
+      "step": 173
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.48e-05,
+      "loss": 2.6555,
+      "step": 174
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.5e-05,
+      "loss": 2.6044,
+      "step": 175
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.5199999999999995e-05,
+      "loss": 2.6301,
+      "step": 176
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.539999999999999e-05,
+      "loss": 2.6324,
+      "step": 177
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.56e-05,
+      "loss": 2.5571,
+      "step": 178
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.5799999999999996e-05,
+      "loss": 2.5637,
+      "step": 179
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 2.4853,
+      "step": 180
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.62e-05,
+      "loss": 2.5275,
+      "step": 181
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.64e-05,
+      "loss": 2.4241,
+      "step": 182
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6599999999999995e-05,
+      "loss": 2.4874,
+      "step": 183
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.679999999999999e-05,
+      "loss": 2.6265,
+      "step": 184
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.7e-05,
+      "loss": 2.5432,
+      "step": 185
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7199999999999996e-05,
+      "loss": 2.6158,
+      "step": 186
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.7399999999999994e-05,
+      "loss": 2.5841,
+      "step": 187
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.759999999999999e-05,
+      "loss": 2.5802,
+      "step": 188
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.78e-05,
+      "loss": 2.5986,
+      "step": 189
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.8e-05,
+      "loss": 2.499,
+      "step": 190
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.8199999999999993e-05,
+      "loss": 2.6219,
+      "step": 191
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.84e-05,
+      "loss": 2.6017,
+      "step": 192
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.86e-05,
+      "loss": 2.5413,
+      "step": 193
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.8799999999999994e-05,
+      "loss": 2.5604,
+      "step": 194
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.9e-05,
+      "loss": 2.5409,
+      "step": 195
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.919999999999999e-05,
+      "loss": 2.5019,
+      "step": 196
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.9399999999999995e-05,
+      "loss": 2.5966,
+      "step": 197
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.96e-05,
+      "loss": 2.5088,
+      "step": 198
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.979999999999999e-05,
+      "loss": 2.5385,
+      "step": 199
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.9999999999999996e-05,
+      "loss": 2.3561,
+      "step": 200
+    },
+    {
+      "epoch": 2.98,
+      "eval_accuracy": 0.13338228095937346,
+      "eval_f1": 0.06813158475964522,
+      "eval_loss": 2.5464296340942383,
+      "eval_runtime": 349.0,
+      "eval_samples_per_second": 11.708,
+      "eval_steps_per_second": 0.734,
+      "step": 200
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.02e-05,
+      "loss": 2.4665,
+      "step": 201
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.039999999999999e-05,
+      "loss": 3.2467,
+      "step": 202
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.06e-05,
+      "loss": 2.5004,
+      "step": 203
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.08e-05,
+      "loss": 2.4464,
+      "step": 204
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.0999999999999994e-05,
+      "loss": 2.5131,
+      "step": 205
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.12e-05,
+      "loss": 2.6149,
+      "step": 206
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.14e-05,
+      "loss": 2.4677,
+      "step": 207
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.1599999999999995e-05,
+      "loss": 2.5008,
+      "step": 208
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.18e-05,
+      "loss": 2.5037,
+      "step": 209
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.2e-05,
+      "loss": 2.4629,
+      "step": 210
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.2199999999999996e-05,
+      "loss": 2.4151,
+      "step": 211
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.24e-05,
+      "loss": 2.3735,
+      "step": 212
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.259999999999999e-05,
+      "loss": 2.4441,
+      "step": 213
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.28e-05,
+      "loss": 2.432,
+      "step": 214
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.3e-05,
+      "loss": 2.4883,
+      "step": 215
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.319999999999999e-05,
+      "loss": 2.4319,
+      "step": 216
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.34e-05,
+      "loss": 2.4026,
+      "step": 217
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.3599999999999996e-05,
+      "loss": 2.357,
+      "step": 218
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.3799999999999994e-05,
+      "loss": 2.5772,
+      "step": 219
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.4e-05,
+      "loss": 2.3947,
+      "step": 220
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.42e-05,
+      "loss": 2.4569,
+      "step": 221
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.4399999999999995e-05,
+      "loss": 2.3699,
+      "step": 222
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.46e-05,
+      "loss": 2.7355,
+      "step": 223
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.48e-05,
+      "loss": 2.7759,
+      "step": 224
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 2.6568,
+      "step": 225
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.5199999999999994e-05,
+      "loss": 2.5626,
+      "step": 226
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.539999999999999e-05,
+      "loss": 2.282,
+      "step": 227
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.56e-05,
+      "loss": 2.3982,
+      "step": 228
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.5799999999999995e-05,
+      "loss": 2.4393,
+      "step": 229
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.599999999999999e-05,
+      "loss": 2.5938,
+      "step": 230
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.62e-05,
+      "loss": 2.6461,
+      "step": 231
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.6399999999999996e-05,
+      "loss": 2.327,
+      "step": 232
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.6599999999999994e-05,
+      "loss": 2.3971,
+      "step": 233
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.68e-05,
+      "loss": 2.515,
+      "step": 234
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.7e-05,
+      "loss": 2.505,
+      "step": 235
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.7199999999999995e-05,
+      "loss": 2.5023,
+      "step": 236
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.7399999999999993e-05,
+      "loss": 2.4241,
+      "step": 237
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.76e-05,
+      "loss": 2.4735,
+      "step": 238
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.7799999999999996e-05,
+      "loss": 2.503,
+      "step": 239
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 2.468,
+      "step": 240
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.82e-05,
+      "loss": 2.4701,
+      "step": 241
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.84e-05,
+      "loss": 2.4992,
+      "step": 242
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.8599999999999995e-05,
+      "loss": 2.557,
+      "step": 243
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.8799999999999994e-05,
+      "loss": 2.4359,
+      "step": 244
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.899999999999999e-05,
+      "loss": 2.508,
+      "step": 245
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.9199999999999997e-05,
+      "loss": 2.4404,
+      "step": 246
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9399999999999995e-05,
+      "loss": 2.2592,
+      "step": 247
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.959999999999999e-05,
+      "loss": 2.4594,
+      "step": 248
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.98e-05,
+      "loss": 2.3305,
+      "step": 249
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.9999999999999996e-05,
+      "loss": 2.3883,
+      "step": 250
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.0199999999999994e-05,
+      "loss": 2.2801,
+      "step": 251
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 5.04e-05,
+      "loss": 2.4757,
+      "step": 252
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 5.06e-05,
+      "loss": 2.4579,
+      "step": 253
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 5.0799999999999995e-05,
+      "loss": 2.3734,
+      "step": 254
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5.1e-05,
+      "loss": 2.3411,
+      "step": 255
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5.119999999999999e-05,
+      "loss": 2.3687,
+      "step": 256
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.1399999999999996e-05,
+      "loss": 2.4481,
+      "step": 257
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.1599999999999994e-05,
+      "loss": 2.471,
+      "step": 258
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.179999999999999e-05,
+      "loss": 2.4718,
+      "step": 259
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.2e-05,
+      "loss": 2.4932,
+      "step": 260
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.2199999999999995e-05,
+      "loss": 2.3567,
+      "step": 261
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.239999999999999e-05,
+      "loss": 2.3381,
+      "step": 262
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.26e-05,
+      "loss": 2.3708,
+      "step": 263
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.279999999999999e-05,
+      "loss": 2.4944,
+      "step": 264
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.2999999999999994e-05,
+      "loss": 2.3488,
+      "step": 265
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.32e-05,
+      "loss": 2.4171,
+      "step": 266
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.339999999999999e-05,
+      "loss": 2.3411,
+      "step": 267
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.3599999999999995e-05,
+      "loss": 2.3492,
+      "step": 268
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.38e-05,
+      "loss": 3.0346,
+      "step": 269
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 2.2454,
+      "step": 270
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.4199999999999996e-05,
+      "loss": 2.2054,
+      "step": 271
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.44e-05,
+      "loss": 2.2441,
+      "step": 272
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5.459999999999999e-05,
+      "loss": 2.3293,
+      "step": 273
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.48e-05,
+      "loss": 2.2064,
+      "step": 274
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.499999999999999e-05,
+      "loss": 2.3777,
+      "step": 275
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5.519999999999999e-05,
+      "loss": 2.1567,
+      "step": 276
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5.54e-05,
+      "loss": 2.1665,
+      "step": 277
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5.559999999999999e-05,
+      "loss": 2.249,
+      "step": 278
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5.5799999999999994e-05,
+      "loss": 2.1989,
+      "step": 279
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5.6e-05,
+      "loss": 2.2401,
+      "step": 280
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.619999999999999e-05,
+      "loss": 2.1298,
+      "step": 281
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 2.2022,
+      "step": 282
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.66e-05,
+      "loss": 2.0053,
+      "step": 283
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.679999999999999e-05,
+      "loss": 2.1011,
+      "step": 284
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 2.0889,
+      "step": 285
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.72e-05,
+      "loss": 2.2418,
+      "step": 286
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.739999999999999e-05,
+      "loss": 2.1132,
+      "step": 287
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.76e-05,
+      "loss": 2.2312,
+      "step": 288
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.78e-05,
+      "loss": 1.9977,
+      "step": 289
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.7999999999999994e-05,
+      "loss": 2.1724,
+      "step": 290
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.82e-05,
+      "loss": 2.1606,
+      "step": 291
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.839999999999999e-05,
+      "loss": 2.1853,
+      "step": 292
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.8599999999999995e-05,
+      "loss": 2.1057,
+      "step": 293
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.88e-05,
+      "loss": 2.0778,
+      "step": 294
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.899999999999999e-05,
+      "loss": 1.967,
+      "step": 295
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.9199999999999996e-05,
+      "loss": 2.3167,
+      "step": 296
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.94e-05,
+      "loss": 2.0744,
+      "step": 297
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.959999999999999e-05,
+      "loss": 2.1545,
+      "step": 298
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.98e-05,
+      "loss": 2.2421,
+      "step": 299
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 1.9392,
+      "step": 300
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.019999999999999e-05,
+      "loss": 1.8914,
+      "step": 301
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.04e-05,
+      "loss": 2.2063,
+      "step": 302
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.0599999999999996e-05,
+      "loss": 2.0361,
+      "step": 303
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.0799999999999994e-05,
+      "loss": 2.1857,
+      "step": 304
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.1e-05,
+      "loss": 1.8373,
+      "step": 305
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.12e-05,
+      "loss": 1.9555,
+      "step": 306
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.139999999999999e-05,
+      "loss": 2.0524,
+      "step": 307
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.159999999999999e-05,
+      "loss": 1.9854,
+      "step": 308
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.18e-05,
+      "loss": 1.9668,
+      "step": 309
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.199999999999999e-05,
+      "loss": 1.8757,
+      "step": 310
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.22e-05,
+      "loss": 1.7739,
+      "step": 311
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.239999999999999e-05,
+      "loss": 1.9759,
+      "step": 312
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.259999999999999e-05,
+      "loss": 1.8669,
+      "step": 313
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.28e-05,
+      "loss": 1.9185,
+      "step": 314
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.299999999999999e-05,
+      "loss": 1.8514,
+      "step": 315
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.319999999999999e-05,
+      "loss": 1.5974,
+      "step": 316
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.34e-05,
+      "loss": 1.6711,
+      "step": 317
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.359999999999999e-05,
+      "loss": 1.8531,
+      "step": 318
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.379999999999999e-05,
+      "loss": 2.0521,
+      "step": 319
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.4e-05,
+      "loss": 1.7659,
+      "step": 320
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.419999999999999e-05,
+      "loss": 1.9653,
+      "step": 321
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.44e-05,
+      "loss": 1.9988,
+      "step": 322
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.459999999999998e-05,
+      "loss": 2.0249,
+      "step": 323
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.479999999999999e-05,
+      "loss": 1.783,
+      "step": 324
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.5e-05,
+      "loss": 2.1492,
+      "step": 325
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.519999999999999e-05,
+      "loss": 1.7947,
+      "step": 326
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.539999999999999e-05,
+      "loss": 1.9124,
+      "step": 327
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.56e-05,
+      "loss": 1.7927,
+      "step": 328
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.579999999999999e-05,
+      "loss": 1.7357,
+      "step": 329
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 1.733,
+      "step": 330
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.62e-05,
+      "loss": 2.0315,
+      "step": 331
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.639999999999999e-05,
+      "loss": 1.9838,
+      "step": 332
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.659999999999999e-05,
+      "loss": 1.9308,
+      "step": 333
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.68e-05,
+      "loss": 1.8901,
+      "step": 334
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.699999999999999e-05,
+      "loss": 1.9154,
+      "step": 335
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.72e-05,
+      "loss": 2.1272,
+      "step": 336
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.739999999999998e-05,
+      "loss": 1.7548,
+      "step": 337
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.759999999999999e-05,
+      "loss": 1.6528,
+      "step": 338
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.78e-05,
+      "loss": 1.7716,
+      "step": 339
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.799999999999999e-05,
+      "loss": 1.4649,
+      "step": 340
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.819999999999999e-05,
+      "loss": 1.6351,
+      "step": 341
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.84e-05,
+      "loss": 1.534,
+      "step": 342
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.859999999999999e-05,
+      "loss": 1.6599,
+      "step": 343
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.879999999999999e-05,
+      "loss": 1.6582,
+      "step": 344
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.9e-05,
+      "loss": 1.8148,
+      "step": 345
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.919999999999999e-05,
+      "loss": 1.3586,
+      "step": 346
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.939999999999999e-05,
+      "loss": 1.8631,
+      "step": 347
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.96e-05,
+      "loss": 1.785,
+      "step": 348
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.979999999999999e-05,
+      "loss": 1.913,
+      "step": 349
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.979999999999999e-05,
+      "loss": 1.6482,
+      "step": 350
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 7e-05,
+      "loss": 1.5916,
+      "step": 351
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.02e-05,
+      "loss": 1.6887,
+      "step": 352
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.039999999999999e-05,
+      "loss": 1.6632,
+      "step": 353
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.06e-05,
+      "loss": 1.4964,
+      "step": 354
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.079999999999999e-05,
+      "loss": 1.8404,
+      "step": 355
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.099999999999999e-05,
+      "loss": 1.6802,
+      "step": 356
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.12e-05,
+      "loss": 1.7256,
+      "step": 357
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.139999999999999e-05,
+      "loss": 1.7233,
+      "step": 358
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.159999999999999e-05,
+      "loss": 1.7826,
+      "step": 359
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 7.18e-05,
+      "loss": 1.7502,
+      "step": 360
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 1.5327,
+      "step": 361
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 7.219999999999999e-05,
+      "loss": 1.4375,
+      "step": 362
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.24e-05,
+      "loss": 1.6509,
+      "step": 363
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 7.259999999999999e-05,
+      "loss": 1.628,
+      "step": 364
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.28e-05,
+      "loss": 1.5856,
+      "step": 365
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 7.3e-05,
+      "loss": 1.4559,
+      "step": 366
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.319999999999999e-05,
+      "loss": 1.7826,
+      "step": 367
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 7.34e-05,
+      "loss": 1.5844,
+      "step": 368
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.359999999999999e-05,
+      "loss": 1.7874,
+      "step": 369
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.379999999999999e-05,
+      "loss": 1.4583,
+      "step": 370
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 7.4e-05,
+      "loss": 1.7617,
+      "step": 371
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.419999999999999e-05,
+      "loss": 1.445,
+      "step": 372
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 7.439999999999999e-05,
+      "loss": 1.5213,
+      "step": 373
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.46e-05,
+      "loss": 1.6538,
+      "step": 374
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.479999999999999e-05,
+      "loss": 1.642,
+      "step": 375
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.5e-05,
+      "loss": 1.5272,
+      "step": 376
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.519999999999998e-05,
+      "loss": 1.615,
+      "step": 377
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.54e-05,
+      "loss": 1.4302,
+      "step": 378
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 7.56e-05,
+      "loss": 1.8705,
+      "step": 379
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.579999999999999e-05,
+      "loss": 1.7639,
+      "step": 380
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.6e-05,
+      "loss": 1.3727,
+      "step": 381
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.62e-05,
+      "loss": 1.7963,
+      "step": 382
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.639999999999999e-05,
+      "loss": 1.3527,
+      "step": 383
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.66e-05,
+      "loss": 1.3017,
+      "step": 384
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.68e-05,
+      "loss": 1.5047,
+      "step": 385
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.699999999999999e-05,
+      "loss": 1.6173,
+      "step": 386
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.72e-05,
+      "loss": 1.3575,
+      "step": 387
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.74e-05,
+      "loss": 1.6024,
+      "step": 388
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.759999999999999e-05,
+      "loss": 1.3431,
+      "step": 389
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 1.6089,
+      "step": 390
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.8e-05,
+      "loss": 1.6595,
+      "step": 391
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.819999999999999e-05,
+      "loss": 1.4502,
+      "step": 392
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.839999999999998e-05,
+      "loss": 1.7518,
+      "step": 393
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.86e-05,
+      "loss": 1.4151,
+      "step": 394
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.879999999999999e-05,
+      "loss": 1.4656,
+      "step": 395
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.899999999999998e-05,
+      "loss": 1.6396,
+      "step": 396
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.92e-05,
+      "loss": 1.3383,
+      "step": 397
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.939999999999999e-05,
+      "loss": 1.5093,
+      "step": 398
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.959999999999998e-05,
+      "loss": 1.3226,
+      "step": 399
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.98e-05,
+      "loss": 1.1851,
+      "step": 400
+    },
+    {
+      "epoch": 5.97,
+      "eval_accuracy": 0.5861478218306412,
+      "eval_f1": 0.5583492858181879,
+      "eval_loss": 1.5055691003799438,
+      "eval_runtime": 343.8006,
+      "eval_samples_per_second": 11.885,
+      "eval_steps_per_second": 0.745,
+      "step": 400
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.999999999999999e-05,
+      "loss": 1.3192,
+      "step": 401
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 8.019999999999998e-05,
+      "loss": 1.6715,
+      "step": 402
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 8.04e-05,
+      "loss": 1.7022,
+      "step": 403
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 8.06e-05,
+      "loss": 1.2076,
+      "step": 404
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 8.079999999999999e-05,
+      "loss": 1.451,
+      "step": 405
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 8.1e-05,
+      "loss": 1.811,
+      "step": 406
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 8.12e-05,
+      "loss": 1.6629,
+      "step": 407
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 8.139999999999999e-05,
+      "loss": 1.3388,
+      "step": 408
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 8.16e-05,
+      "loss": 1.5293,
+      "step": 409
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 8.18e-05,
+      "loss": 1.4277,
+      "step": 410
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 8.199999999999999e-05,
+      "loss": 1.195,
+      "step": 411
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 8.22e-05,
+      "loss": 1.461,
+      "step": 412
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 8.24e-05,
+      "loss": 1.5956,
+      "step": 413
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 8.259999999999999e-05,
+      "loss": 1.2268,
+      "step": 414
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 8.28e-05,
+      "loss": 1.6036,
+      "step": 415
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 8.3e-05,
+      "loss": 1.4096,
+      "step": 416
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 8.319999999999999e-05,
+      "loss": 1.3096,
+      "step": 417
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 8.34e-05,
+      "loss": 1.1763,
+      "step": 418
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 8.36e-05,
+      "loss": 1.5893,
+      "step": 419
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 8.379999999999999e-05,
+      "loss": 1.2182,
+      "step": 420
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 8.4e-05,
+      "loss": 1.243,
+      "step": 421
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 8.42e-05,
+      "loss": 1.5265,
+      "step": 422
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 8.439999999999999e-05,
+      "loss": 1.4368,
+      "step": 423
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 8.459999999999998e-05,
+      "loss": 1.4227,
+      "step": 424
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 8.48e-05,
+      "loss": 1.4331,
+      "step": 425
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 8.499999999999999e-05,
+      "loss": 1.3622,
+      "step": 426
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 8.519999999999998e-05,
+      "loss": 1.5802,
+      "step": 427
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 8.54e-05,
+      "loss": 1.4871,
+      "step": 428
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 8.56e-05,
+      "loss": 1.207,
+      "step": 429
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 8.579999999999998e-05,
+      "loss": 1.2297,
+      "step": 430
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 8.6e-05,
+      "loss": 1.5138,
+      "step": 431
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 8.62e-05,
+      "loss": 1.6429,
+      "step": 432
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 8.639999999999999e-05,
+      "loss": 1.3145,
+      "step": 433
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 8.659999999999999e-05,
+      "loss": 1.2543,
+      "step": 434
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 8.68e-05,
+      "loss": 1.0758,
+      "step": 435
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 1.6667,
+      "step": 436
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 8.719999999999999e-05,
+      "loss": 1.3135,
+      "step": 437
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 8.74e-05,
+      "loss": 1.3914,
+      "step": 438
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 8.759999999999999e-05,
+      "loss": 1.3286,
+      "step": 439
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 8.779999999999999e-05,
+      "loss": 1.3077,
+      "step": 440
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 8.8e-05,
+      "loss": 1.3647,
+      "step": 441
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 8.819999999999999e-05,
+      "loss": 1.3532,
+      "step": 442
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 8.84e-05,
+      "loss": 1.5361,
+      "step": 443
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 8.86e-05,
+      "loss": 1.2559,
+      "step": 444
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 8.879999999999999e-05,
+      "loss": 1.4815,
+      "step": 445
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 8.9e-05,
+      "loss": 0.9722,
+      "step": 446
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 8.92e-05,
+      "loss": 1.6683,
+      "step": 447
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 8.939999999999999e-05,
+      "loss": 1.5446,
+      "step": 448
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 8.96e-05,
+      "loss": 1.5083,
+      "step": 449
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 8.98e-05,
+      "loss": 1.4544,
+      "step": 450
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 1.5404,
+      "step": 451
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 9.02e-05,
+      "loss": 1.3788,
+      "step": 452
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 9.039999999999999e-05,
+      "loss": 1.7929,
+      "step": 453
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 9.059999999999999e-05,
+      "loss": 1.6221,
+      "step": 454
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 9.079999999999998e-05,
+      "loss": 2.0289,
+      "step": 455
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 9.099999999999999e-05,
+      "loss": 1.3513,
+      "step": 456
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.12e-05,
+      "loss": 1.1809,
+      "step": 457
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.139999999999999e-05,
+      "loss": 1.659,
+      "step": 458
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.159999999999999e-05,
+      "loss": 1.3772,
+      "step": 459
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 9.18e-05,
+      "loss": 1.0775,
+      "step": 460
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.199999999999999e-05,
+      "loss": 1.6983,
+      "step": 461
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 9.219999999999999e-05,
+      "loss": 1.1782,
+      "step": 462
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.24e-05,
+      "loss": 1.3642,
+      "step": 463
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.259999999999999e-05,
+      "loss": 1.1298,
+      "step": 464
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.279999999999999e-05,
+      "loss": 1.3433,
+      "step": 465
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.3e-05,
+      "loss": 1.2165,
+      "step": 466
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.319999999999999e-05,
+      "loss": 1.5146,
+      "step": 467
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.34e-05,
+      "loss": 1.4164,
+      "step": 468
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.36e-05,
+      "loss": 1.2011,
+      "step": 469
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.379999999999999e-05,
+      "loss": 1.6677,
+      "step": 470
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.4e-05,
+      "loss": 1.0573,
+      "step": 471
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.419999999999999e-05,
+      "loss": 1.3889,
+      "step": 472
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.439999999999999e-05,
+      "loss": 1.4316,
+      "step": 473
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.46e-05,
+      "loss": 1.0497,
+      "step": 474
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.479999999999999e-05,
+      "loss": 1.203,
+      "step": 475
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.499999999999999e-05,
+      "loss": 1.543,
+      "step": 476
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.52e-05,
+      "loss": 1.3419,
+      "step": 477
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.539999999999999e-05,
+      "loss": 1.3901,
+      "step": 478
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.559999999999999e-05,
+      "loss": 1.152,
+      "step": 479
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.58e-05,
+      "loss": 1.1519,
+      "step": 480
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 1.2613,
+      "step": 481
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.62e-05,
+      "loss": 1.5289,
+      "step": 482
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.64e-05,
+      "loss": 1.3045,
+      "step": 483
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.659999999999999e-05,
+      "loss": 1.1174,
+      "step": 484
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 9.68e-05,
+      "loss": 0.9389,
+      "step": 485
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 9.699999999999999e-05,
+      "loss": 1.277,
+      "step": 486
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 9.719999999999999e-05,
+      "loss": 1.4324,
+      "step": 487
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.74e-05,
+      "loss": 1.3783,
+      "step": 488
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 9.759999999999999e-05,
+      "loss": 0.9759,
+      "step": 489
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 9.779999999999999e-05,
+      "loss": 0.9928,
+      "step": 490
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 9.799999999999998e-05,
+      "loss": 1.0354,
+      "step": 491
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 9.819999999999999e-05,
+      "loss": 1.4557,
+      "step": 492
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 9.839999999999999e-05,
+      "loss": 1.3679,
+      "step": 493
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 9.859999999999998e-05,
+      "loss": 1.3464,
+      "step": 494
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 9.879999999999999e-05,
+      "loss": 1.1559,
+      "step": 495
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 9.9e-05,
+      "loss": 1.1564,
+      "step": 496
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 9.919999999999999e-05,
+      "loss": 0.9569,
+      "step": 497
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 9.939999999999999e-05,
+      "loss": 1.3608,
+      "step": 498
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 9.96e-05,
+      "loss": 0.9687,
+      "step": 499
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 9.979999999999999e-05,
+      "loss": 1.1822,
+      "step": 500
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 9.999999999999999e-05,
+      "loss": 0.9472,
+      "step": 501
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.0001002,
+      "loss": 1.1478,
+      "step": 502
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.00010039999999999999,
+      "loss": 1.7424,
+      "step": 503
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.00010059999999999999,
+      "loss": 1.2375,
+      "step": 504
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.0001008,
+      "loss": 1.6025,
+      "step": 505
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.00010099999999999999,
+      "loss": 1.0611,
+      "step": 506
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.0001012,
+      "loss": 1.1944,
+      "step": 507
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.0001014,
+      "loss": 1.2099,
+      "step": 508
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.00010159999999999999,
+      "loss": 1.0249,
+      "step": 509
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.00010179999999999998,
+      "loss": 1.1262,
+      "step": 510
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.000102,
+      "loss": 1.1328,
+      "step": 511
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 0.00010219999999999999,
+      "loss": 0.8573,
+      "step": 512
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.00010239999999999998,
+      "loss": 1.3475,
+      "step": 513
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 0.0001026,
+      "loss": 1.0557,
+      "step": 514
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 0.00010279999999999999,
+      "loss": 1.043,
+      "step": 515
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.00010299999999999998,
+      "loss": 1.2895,
+      "step": 516
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 0.00010319999999999999,
+      "loss": 1.2999,
+      "step": 517
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.00010339999999999999,
+      "loss": 0.8422,
+      "step": 518
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 0.00010359999999999998,
+      "loss": 1.3048,
+      "step": 519
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.00010379999999999999,
+      "loss": 1.054,
+      "step": 520
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 0.000104,
+      "loss": 1.0085,
+      "step": 521
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.00010419999999999998,
+      "loss": 1.372,
+      "step": 522
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.00010439999999999999,
+      "loss": 1.2399,
+      "step": 523
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.0001046,
+      "loss": 1.1393,
+      "step": 524
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.00010479999999999999,
+      "loss": 1.1062,
+      "step": 525
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.00010499999999999999,
+      "loss": 1.1683,
+      "step": 526
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 0.0001052,
+      "loss": 1.4802,
+      "step": 527
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.00010539999999999999,
+      "loss": 0.9131,
+      "step": 528
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 0.00010559999999999998,
+      "loss": 0.8833,
+      "step": 529
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.0001058,
+      "loss": 0.8876,
+      "step": 530
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 0.00010599999999999999,
+      "loss": 1.2304,
+      "step": 531
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.00010619999999999998,
+      "loss": 1.0087,
+      "step": 532
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 0.0001064,
+      "loss": 1.0249,
+      "step": 533
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.00010659999999999999,
+      "loss": 1.4578,
+      "step": 534
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.00010679999999999998,
+      "loss": 0.8898,
+      "step": 535
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.000107,
+      "loss": 1.1781,
+      "step": 536
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 0.00010719999999999999,
+      "loss": 1.749,
+      "step": 537
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 0.00010739999999999998,
+      "loss": 0.9843,
+      "step": 538
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 0.0001076,
+      "loss": 1.1008,
+      "step": 539
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 0.00010779999999999999,
+      "loss": 1.1893,
+      "step": 540
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 0.9748,
+      "step": 541
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 0.0001082,
+      "loss": 1.2305,
+      "step": 542
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 0.00010839999999999999,
+      "loss": 1.154,
+      "step": 543
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 0.00010859999999999998,
+      "loss": 1.2386,
+      "step": 544
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 0.0001088,
+      "loss": 1.2909,
+      "step": 545
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 0.00010899999999999999,
+      "loss": 1.1747,
+      "step": 546
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 0.00010919999999999998,
+      "loss": 1.2138,
+      "step": 547
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 0.00010939999999999998,
+      "loss": 1.421,
+      "step": 548
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 0.0001096,
+      "loss": 1.2327,
+      "step": 549
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 0.00010979999999999999,
+      "loss": 1.1603,
+      "step": 550
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 0.00010999999999999998,
+      "loss": 1.0756,
+      "step": 551
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 0.0001102,
+      "loss": 0.8484,
+      "step": 552
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 0.00011039999999999999,
+      "loss": 0.8909,
+      "step": 553
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 0.00011059999999999998,
+      "loss": 1.1358,
+      "step": 554
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 0.0001108,
+      "loss": 1.3485,
+      "step": 555
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 0.00011099999999999999,
+      "loss": 0.9957,
+      "step": 556
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 0.00011119999999999998,
+      "loss": 1.0312,
+      "step": 557
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 0.0001114,
+      "loss": 1.3304,
+      "step": 558
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 0.0001114,
+      "loss": 1.3952,
+      "step": 559
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 0.00011159999999999999,
+      "loss": 1.0264,
+      "step": 560
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 0.00011179999999999998,
+      "loss": 1.1699,
+      "step": 561
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 0.000112,
+      "loss": 1.2597,
+      "step": 562
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 0.00011219999999999999,
+      "loss": 0.9144,
+      "step": 563
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 0.00011239999999999998,
+      "loss": 1.4595,
+      "step": 564
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 0.0001126,
+      "loss": 0.9706,
+      "step": 565
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 0.00011279999999999999,
+      "loss": 1.5886,
+      "step": 566
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 0.00011299999999999998,
+      "loss": 0.974,
+      "step": 567
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 0.0001132,
+      "loss": 1.135,
+      "step": 568
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 0.00011339999999999999,
+      "loss": 1.0261,
+      "step": 569
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 0.00011359999999999998,
+      "loss": 1.3876,
+      "step": 570
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 0.0001138,
+      "loss": 1.1007,
+      "step": 571
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 1.0671,
+      "step": 572
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 0.00011419999999999998,
+      "loss": 1.2932,
+      "step": 573
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 0.0001144,
+      "loss": 1.4211,
+      "step": 574
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 0.0001146,
+      "loss": 0.9986,
+      "step": 575
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 0.00011479999999999999,
+      "loss": 1.0899,
+      "step": 576
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 0.000115,
+      "loss": 0.8065,
+      "step": 577
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 0.0001152,
+      "loss": 0.6506,
+      "step": 578
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 0.00011539999999999999,
+      "loss": 0.8245,
+      "step": 579
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 0.0001156,
+      "loss": 0.948,
+      "step": 580
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 0.0001158,
+      "loss": 0.8349,
+      "step": 581
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 0.00011599999999999999,
+      "loss": 0.8581,
+      "step": 582
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 0.00011619999999999998,
+      "loss": 0.9276,
+      "step": 583
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 0.0001164,
+      "loss": 1.0779,
+      "step": 584
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 0.00011659999999999999,
+      "loss": 1.0055,
+      "step": 585
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 0.00011679999999999998,
+      "loss": 1.0188,
+      "step": 586
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 0.000117,
+      "loss": 0.9448,
+      "step": 587
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 0.00011719999999999999,
+      "loss": 0.9065,
+      "step": 588
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 0.00011739999999999998,
+      "loss": 1.3076,
+      "step": 589
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 0.0001176,
+      "loss": 1.124,
+      "step": 590
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 0.00011779999999999999,
+      "loss": 1.0797,
+      "step": 591
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 0.00011799999999999998,
+      "loss": 0.9384,
+      "step": 592
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 0.0001182,
+      "loss": 0.8262,
+      "step": 593
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 0.00011839999999999999,
+      "loss": 1.4642,
+      "step": 594
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 0.00011859999999999998,
+      "loss": 1.0777,
+      "step": 595
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 0.0001188,
+      "loss": 1.0487,
+      "step": 596
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 0.00011899999999999999,
+      "loss": 1.0663,
+      "step": 597
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 0.00011919999999999998,
+      "loss": 1.3841,
+      "step": 598
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 0.0001194,
+      "loss": 1.095,
+      "step": 599
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 0.0001196,
+      "loss": 1.2805,
+      "step": 600
+    },
+    {
+      "epoch": 8.95,
+      "eval_accuracy": 0.7043563387175722,
+      "eval_f1": 0.7105830295720407,
+      "eval_loss": 1.139738917350769,
+      "eval_runtime": 343.0859,
+      "eval_samples_per_second": 11.91,
+      "eval_steps_per_second": 0.746,
+      "step": 600
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 0.00011979999999999998,
+      "loss": 0.944,
+      "step": 601
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 0.8265,
+      "step": 602
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 0.0001202,
+      "loss": 0.9694,
+      "step": 603
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 0.00012039999999999999,
+      "loss": 1.4551,
+      "step": 604
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 0.00012059999999999999,
+      "loss": 1.2523,
+      "step": 605
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 0.0001208,
+      "loss": 0.7349,
+      "step": 606
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 0.00012099999999999999,
+      "loss": 1.3356,
+      "step": 607
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 0.00012119999999999999,
+      "loss": 1.3264,
+      "step": 608
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 0.0001214,
+      "loss": 0.9967,
+      "step": 609
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 0.00012159999999999999,
+      "loss": 1.0618,
+      "step": 610
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 0.00012179999999999999,
+      "loss": 1.1899,
+      "step": 611
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 0.000122,
+      "loss": 0.8158,
+      "step": 612
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 0.0001222,
+      "loss": 1.4054,
+      "step": 613
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 0.0001224,
+      "loss": 0.9819,
+      "step": 614
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 0.0001226,
+      "loss": 0.8747,
+      "step": 615
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 0.00012279999999999998,
+      "loss": 0.989,
+      "step": 616
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 0.00012299999999999998,
+      "loss": 1.0601,
+      "step": 617
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 0.00012319999999999999,
+      "loss": 1.0786,
+      "step": 618
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 0.0001234,
+      "loss": 0.8078,
+      "step": 619
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 0.0001236,
+      "loss": 1.2362,
+      "step": 620
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 0.0001238,
+      "loss": 0.7693,
+      "step": 621
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 0.00012399999999999998,
+      "loss": 0.8017,
+      "step": 622
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 0.00012419999999999998,
+      "loss": 1.3265,
+      "step": 623
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 0.0001244,
+      "loss": 0.7762,
+      "step": 624
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 0.0001246,
+      "loss": 1.02,
+      "step": 625
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 0.00012479999999999997,
+      "loss": 1.1217,
+      "step": 626
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 0.000125,
+      "loss": 1.0048,
+      "step": 627
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 0.00012519999999999998,
+      "loss": 0.9923,
+      "step": 628
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 0.00012539999999999999,
+      "loss": 0.7986,
+      "step": 629
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 0.0001256,
+      "loss": 0.8572,
+      "step": 630
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 0.0001258,
+      "loss": 0.7296,
+      "step": 631
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 1.1171,
+      "step": 632
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 0.0001262,
+      "loss": 0.9852,
+      "step": 633
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 0.00012639999999999998,
+      "loss": 0.9377,
+      "step": 634
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 0.0001266,
+      "loss": 0.8699,
+      "step": 635
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 0.0001268,
+      "loss": 0.9284,
+      "step": 636
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 0.000127,
+      "loss": 0.9991,
+      "step": 637
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 0.00012719999999999997,
+      "loss": 1.0121,
+      "step": 638
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 0.0001274,
+      "loss": 1.0221,
+      "step": 639
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 0.00012759999999999998,
+      "loss": 1.0897,
+      "step": 640
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 0.0001278,
+      "loss": 1.2973,
+      "step": 641
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 0.000128,
+      "loss": 1.0074,
+      "step": 642
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 0.0001282,
+      "loss": 1.2608,
+      "step": 643
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 0.00012839999999999998,
+      "loss": 1.2781,
+      "step": 644
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 0.00012859999999999998,
+      "loss": 1.029,
+      "step": 645
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 0.0001288,
+      "loss": 0.8011,
+      "step": 646
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 0.000129,
+      "loss": 0.7844,
+      "step": 647
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 0.00012919999999999997,
+      "loss": 1.3425,
+      "step": 648
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 0.0001294,
+      "loss": 1.4064,
+      "step": 649
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 0.00012959999999999998,
+      "loss": 1.0535,
+      "step": 650
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 0.00012979999999999998,
+      "loss": 1.2251,
+      "step": 651
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 0.00013,
+      "loss": 1.0629,
+      "step": 652
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 0.0001302,
+      "loss": 0.9377,
+      "step": 653
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 0.00013039999999999997,
+      "loss": 1.1107,
+      "step": 654
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 0.0001306,
+      "loss": 1.4752,
+      "step": 655
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 0.00013079999999999998,
+      "loss": 1.0769,
+      "step": 656
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 0.00013099999999999999,
+      "loss": 0.988,
+      "step": 657
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 0.0001312,
+      "loss": 1.4048,
+      "step": 658
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 0.0001314,
+      "loss": 1.0237,
+      "step": 659
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 0.00013159999999999997,
+      "loss": 0.6261,
+      "step": 660
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 0.0001318,
+      "loss": 0.7488,
+      "step": 661
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 1.1096,
+      "step": 662
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 0.0001322,
+      "loss": 0.8872,
+      "step": 663
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 0.0001324,
+      "loss": 0.9701,
+      "step": 664
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 0.0001326,
+      "loss": 0.7629,
+      "step": 665
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 0.00013279999999999998,
+      "loss": 0.8341,
+      "step": 666
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 0.000133,
+      "loss": 1.316,
+      "step": 667
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 0.00013319999999999999,
+      "loss": 0.8714,
+      "step": 668
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 0.0001334,
+      "loss": 0.8048,
+      "step": 669
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 0.0001336,
+      "loss": 1.276,
+      "step": 670
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 0.0001338,
+      "loss": 1.2213,
+      "step": 671
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 0.00013399999999999998,
+      "loss": 0.954,
+      "step": 672
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 0.0001342,
+      "loss": 0.7978,
+      "step": 673
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 0.0001344,
+      "loss": 0.9366,
+      "step": 674
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 0.0001346,
+      "loss": 0.9517,
+      "step": 675
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 0.00013479999999999997,
+      "loss": 0.8653,
+      "step": 676
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 0.000135,
+      "loss": 0.8692,
+      "step": 677
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 0.00013519999999999998,
+      "loss": 0.7885,
+      "step": 678
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 0.00013539999999999998,
+      "loss": 1.0205,
+      "step": 679
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 0.0001356,
+      "loss": 0.9433,
+      "step": 680
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 0.0001358,
+      "loss": 0.769,
+      "step": 681
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 0.00013599999999999997,
+      "loss": 0.7986,
+      "step": 682
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 0.0001362,
+      "loss": 0.8684,
+      "step": 683
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 0.00013639999999999998,
+      "loss": 1.1515,
+      "step": 684
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 0.00013659999999999999,
+      "loss": 0.6979,
+      "step": 685
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 0.0001368,
+      "loss": 0.6019,
+      "step": 686
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 0.000137,
+      "loss": 0.9636,
+      "step": 687
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 0.00013719999999999997,
+      "loss": 0.9187,
+      "step": 688
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 0.0001374,
+      "loss": 1.0534,
+      "step": 689
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 0.00013759999999999998,
+      "loss": 0.5844,
+      "step": 690
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 0.0001378,
+      "loss": 1.2123,
+      "step": 691
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 0.000138,
+      "loss": 1.0089,
+      "step": 692
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 0.0001382,
+      "loss": 1.1607,
+      "step": 693
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 0.00013839999999999998,
+      "loss": 1.3608,
+      "step": 694
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 0.0001386,
+      "loss": 0.8186,
+      "step": 695
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 0.00013879999999999999,
+      "loss": 1.2984,
+      "step": 696
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 0.000139,
+      "loss": 0.8887,
+      "step": 697
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 0.0001392,
+      "loss": 1.0305,
+      "step": 698
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 0.0001394,
+      "loss": 0.8821,
+      "step": 699
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 0.00013959999999999998,
+      "loss": 0.9324,
+      "step": 700
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 0.00013979999999999998,
+      "loss": 0.7349,
+      "step": 701
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 0.00014,
+      "loss": 0.9737,
+      "step": 702
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 0.0001402,
+      "loss": 0.7634,
+      "step": 703
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 0.0001404,
+      "loss": 1.2076,
+      "step": 704
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 0.0001406,
+      "loss": 0.6951,
+      "step": 705
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 0.00014079999999999998,
+      "loss": 0.8113,
+      "step": 706
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 0.00014099999999999998,
+      "loss": 0.6317,
+      "step": 707
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 0.0001412,
+      "loss": 1.0482,
+      "step": 708
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 0.0001414,
+      "loss": 0.8645,
+      "step": 709
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 0.00014159999999999997,
+      "loss": 1.0108,
+      "step": 710
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 0.0001418,
+      "loss": 0.9291,
+      "step": 711
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 0.00014199999999999998,
+      "loss": 0.7584,
+      "step": 712
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 0.0001422,
+      "loss": 0.9291,
+      "step": 713
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 0.0001424,
+      "loss": 0.9126,
+      "step": 714
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 0.0001426,
+      "loss": 0.7771,
+      "step": 715
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 0.00014279999999999997,
+      "loss": 1.5661,
+      "step": 716
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 0.00014299999999999998,
+      "loss": 1.2408,
+      "step": 717
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 0.00014319999999999998,
+      "loss": 0.7724,
+      "step": 718
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 0.0001434,
+      "loss": 0.8097,
+      "step": 719
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 0.0001436,
+      "loss": 0.4914,
+      "step": 720
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 0.0001438,
+      "loss": 0.8946,
+      "step": 721
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 0.876,
+      "step": 722
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 0.00014419999999999998,
+      "loss": 0.8784,
+      "step": 723
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 0.00014439999999999999,
+      "loss": 1.0133,
+      "step": 724
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 0.0001446,
+      "loss": 1.1433,
+      "step": 725
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 0.0001448,
+      "loss": 1.361,
+      "step": 726
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 0.000145,
+      "loss": 1.006,
+      "step": 727
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 0.00014519999999999998,
+      "loss": 1.2211,
+      "step": 728
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 0.00014539999999999998,
+      "loss": 0.9271,
+      "step": 729
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 0.0001456,
+      "loss": 1.0685,
+      "step": 730
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 0.0001458,
+      "loss": 1.1139,
+      "step": 731
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 0.000146,
+      "loss": 0.9508,
+      "step": 732
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 0.0001462,
+      "loss": 0.6874,
+      "step": 733
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 0.00014639999999999998,
+      "loss": 1.0912,
+      "step": 734
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 0.00014659999999999999,
+      "loss": 0.5254,
+      "step": 735
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 0.0001468,
+      "loss": 1.1663,
+      "step": 736
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 0.000147,
+      "loss": 0.7879,
+      "step": 737
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 0.00014719999999999997,
+      "loss": 1.0129,
+      "step": 738
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 0.00014739999999999998,
+      "loss": 0.8334,
+      "step": 739
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 0.00014759999999999998,
+      "loss": 0.649,
+      "step": 740
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 0.0001478,
+      "loss": 1.2923,
+      "step": 741
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 0.000148,
+      "loss": 0.7388,
+      "step": 742
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 0.0001482,
+      "loss": 1.1052,
+      "step": 743
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 0.00014839999999999998,
+      "loss": 0.8788,
+      "step": 744
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 0.00014859999999999998,
+      "loss": 0.8711,
+      "step": 745
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 0.00014879999999999998,
+      "loss": 1.0916,
+      "step": 746
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 0.000149,
+      "loss": 0.9535,
+      "step": 747
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 0.0001492,
+      "loss": 0.8849,
+      "step": 748
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 0.0001494,
+      "loss": 0.9825,
+      "step": 749
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 0.00014959999999999998,
+      "loss": 0.9475,
+      "step": 750
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 0.00014979999999999998,
+      "loss": 1.3699,
+      "step": 751
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 0.00015,
+      "loss": 0.7989,
+      "step": 752
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 0.0001502,
+      "loss": 1.1212,
+      "step": 753
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 0.00015039999999999997,
+      "loss": 0.7149,
+      "step": 754
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 0.00015059999999999997,
+      "loss": 0.9158,
+      "step": 755
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 0.0001508,
+      "loss": 0.7415,
+      "step": 756
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 0.00015099999999999998,
+      "loss": 0.8255,
+      "step": 757
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 0.0001512,
+      "loss": 1.0209,
+      "step": 758
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 0.0001514,
+      "loss": 0.8702,
+      "step": 759
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 0.00015159999999999997,
+      "loss": 1.0192,
+      "step": 760
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 0.00015179999999999998,
+      "loss": 0.6197,
+      "step": 761
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 0.000152,
+      "loss": 0.8326,
+      "step": 762
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 0.00015219999999999999,
+      "loss": 0.8132,
+      "step": 763
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 0.0001524,
+      "loss": 0.7399,
+      "step": 764
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 0.0001526,
+      "loss": 1.057,
+      "step": 765
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 0.00015279999999999997,
+      "loss": 0.7861,
+      "step": 766
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 0.00015299999999999998,
+      "loss": 0.9496,
+      "step": 767
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 0.0001532,
+      "loss": 1.1022,
+      "step": 768
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 0.0001534,
+      "loss": 0.7708,
+      "step": 769
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 0.0001536,
+      "loss": 0.5718,
+      "step": 770
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 0.0001538,
+      "loss": 1.4677,
+      "step": 771
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 0.00015399999999999998,
+      "loss": 0.9794,
+      "step": 772
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 0.00015419999999999998,
+      "loss": 1.1667,
+      "step": 773
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 0.0001544,
+      "loss": 1.3804,
+      "step": 774
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 0.0001546,
+      "loss": 1.3366,
+      "step": 775
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 0.0001548,
+      "loss": 0.749,
+      "step": 776
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 0.000155,
+      "loss": 0.6442,
+      "step": 777
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 0.00015519999999999998,
+      "loss": 1.1415,
+      "step": 778
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 0.00015539999999999998,
+      "loss": 1.4601,
+      "step": 779
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 0.00015560000000000001,
+      "loss": 1.3154,
+      "step": 780
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 0.0001558,
+      "loss": 0.5728,
+      "step": 781
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 0.000156,
+      "loss": 0.6963,
+      "step": 782
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 0.0001562,
+      "loss": 0.8572,
+      "step": 783
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 0.00015639999999999998,
+      "loss": 0.8292,
+      "step": 784
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 0.00015659999999999998,
+      "loss": 1.1899,
+      "step": 785
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 0.00015679999999999996,
+      "loss": 0.6802,
+      "step": 786
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 0.000157,
+      "loss": 1.5924,
+      "step": 787
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 0.0001572,
+      "loss": 1.3299,
+      "step": 788
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 0.00015739999999999998,
+      "loss": 1.2905,
+      "step": 789
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 0.00015759999999999998,
+      "loss": 0.7761,
+      "step": 790
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 0.0001578,
+      "loss": 0.7079,
+      "step": 791
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 0.00015799999999999996,
+      "loss": 1.2024,
+      "step": 792
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 0.00015819999999999997,
+      "loss": 1.0789,
+      "step": 793
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 0.0001584,
+      "loss": 0.8125,
+      "step": 794
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 0.00015859999999999998,
+      "loss": 0.7292,
+      "step": 795
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 0.00015879999999999998,
+      "loss": 0.7865,
+      "step": 796
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 0.000159,
+      "loss": 1.002,
+      "step": 797
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 0.00015919999999999997,
+      "loss": 0.8947,
+      "step": 798
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 0.00015939999999999997,
+      "loss": 1.1309,
+      "step": 799
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 0.0001596,
+      "loss": 1.0801,
+      "step": 800
+    },
+    {
+      "epoch": 11.94,
+      "eval_accuracy": 0.7197748409202154,
+      "eval_f1": 0.7132262117406672,
+      "eval_loss": 0.9862720370292664,
+      "eval_runtime": 343.503,
+      "eval_samples_per_second": 11.895,
+      "eval_steps_per_second": 0.745,
+      "step": 800
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 0.00015979999999999998,
+      "loss": 0.7474,
+      "step": 801
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 0.00015999999999999999,
+      "loss": 0.843,
+      "step": 802
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 0.0001602,
+      "loss": 0.7314,
+      "step": 803
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 0.00016039999999999997,
+      "loss": 0.9725,
+      "step": 804
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 0.00016059999999999997,
+      "loss": 1.4201,
+      "step": 805
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 0.0001608,
+      "loss": 0.5819,
+      "step": 806
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 0.00016099999999999998,
+      "loss": 1.0351,
+      "step": 807
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 0.0001612,
+      "loss": 0.8731,
+      "step": 808
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 0.0001614,
+      "loss": 0.8561,
+      "step": 809
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 0.00016159999999999997,
+      "loss": 0.9219,
+      "step": 810
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 0.00016179999999999998,
+      "loss": 0.7039,
+      "step": 811
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 0.000162,
+      "loss": 1.0735,
+      "step": 812
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 0.00016219999999999999,
+      "loss": 1.1316,
+      "step": 813
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 0.0001624,
+      "loss": 0.83,
+      "step": 814
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 0.0001626,
+      "loss": 0.6262,
+      "step": 815
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 0.00016279999999999997,
+      "loss": 0.6751,
+      "step": 816
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 0.00016299999999999998,
+      "loss": 0.7606,
+      "step": 817
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 0.0001632,
+      "loss": 0.5516,
+      "step": 818
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 0.0001634,
+      "loss": 0.8138,
+      "step": 819
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 0.0001636,
+      "loss": 0.3187,
+      "step": 820
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 0.0001638,
+      "loss": 0.5061,
+      "step": 821
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 0.00016399999999999997,
+      "loss": 0.9732,
+      "step": 822
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 0.00016419999999999998,
+      "loss": 0.8939,
+      "step": 823
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 0.0001644,
+      "loss": 0.5243,
+      "step": 824
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 0.0001646,
+      "loss": 0.7636,
+      "step": 825
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 0.0001648,
+      "loss": 0.8959,
+      "step": 826
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 0.000165,
+      "loss": 1.1794,
+      "step": 827
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 0.00016519999999999998,
+      "loss": 0.9718,
+      "step": 828
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 0.00016539999999999998,
+      "loss": 0.8994,
+      "step": 829
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 0.0001656,
+      "loss": 0.6492,
+      "step": 830
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 0.00016579999999999996,
+      "loss": 0.5952,
+      "step": 831
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 0.000166,
+      "loss": 0.7995,
+      "step": 832
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 0.0001662,
+      "loss": 1.0975,
+      "step": 833
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 0.00016639999999999998,
+      "loss": 1.332,
+      "step": 834
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 0.00016659999999999998,
+      "loss": 1.1362,
+      "step": 835
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 0.0001668,
+      "loss": 0.7806,
+      "step": 836
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 0.00016699999999999997,
+      "loss": 0.8691,
+      "step": 837
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 0.0001672,
+      "loss": 0.8649,
+      "step": 838
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 0.0001674,
+      "loss": 0.7305,
+      "step": 839
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 0.00016759999999999998,
+      "loss": 1.0438,
+      "step": 840
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 0.00016779999999999999,
+      "loss": 0.977,
+      "step": 841
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 0.000168,
+      "loss": 0.5909,
+      "step": 842
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 0.00016819999999999997,
+      "loss": 1.1691,
+      "step": 843
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 0.0001684,
+      "loss": 1.4127,
+      "step": 844
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 0.0001686,
+      "loss": 1.1452,
+      "step": 845
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 0.00016879999999999998,
+      "loss": 0.858,
+      "step": 846
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 0.000169,
+      "loss": 0.9272,
+      "step": 847
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 0.00016919999999999997,
+      "loss": 0.8182,
+      "step": 848
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 0.00016939999999999997,
+      "loss": 0.93,
+      "step": 849
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 0.0001696,
+      "loss": 1.305,
+      "step": 850
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 0.00016979999999999998,
+      "loss": 0.7852,
+      "step": 851
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 0.00016999999999999999,
+      "loss": 0.6363,
+      "step": 852
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 0.0001702,
+      "loss": 0.9209,
+      "step": 853
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 0.00017039999999999997,
+      "loss": 0.5253,
+      "step": 854
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 0.00017059999999999997,
+      "loss": 1.0001,
+      "step": 855
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 0.0001708,
+      "loss": 0.9678,
+      "step": 856
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 0.00017099999999999998,
+      "loss": 0.639,
+      "step": 857
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 0.0001712,
+      "loss": 0.8503,
+      "step": 858
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 0.0001714,
+      "loss": 0.6808,
+      "step": 859
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 0.00017159999999999997,
+      "loss": 0.7688,
+      "step": 860
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 0.00017179999999999997,
+      "loss": 1.008,
+      "step": 861
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 0.000172,
+      "loss": 0.9021,
+      "step": 862
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 0.00017219999999999998,
+      "loss": 0.9418,
+      "step": 863
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 0.0001724,
+      "loss": 0.9163,
+      "step": 864
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 0.0001726,
+      "loss": 1.0017,
+      "step": 865
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 0.00017279999999999997,
+      "loss": 0.8404,
+      "step": 866
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 0.00017299999999999998,
+      "loss": 0.6335,
+      "step": 867
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 0.00017319999999999998,
+      "loss": 0.7219,
+      "step": 868
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 0.00017339999999999996,
+      "loss": 1.103,
+      "step": 869
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 0.0001736,
+      "loss": 0.7049,
+      "step": 870
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 0.0001738,
+      "loss": 0.9147,
+      "step": 871
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 0.91,
+      "step": 872
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 0.00017419999999999998,
+      "loss": 0.8884,
+      "step": 873
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 0.00017439999999999998,
+      "loss": 0.6004,
+      "step": 874
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 0.00017459999999999996,
+      "loss": 0.5358,
+      "step": 875
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 0.0001748,
+      "loss": 0.8575,
+      "step": 876
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 0.000175,
+      "loss": 0.8457,
+      "step": 877
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 0.00017519999999999998,
+      "loss": 0.7373,
+      "step": 878
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 0.00017539999999999998,
+      "loss": 0.6553,
+      "step": 879
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 0.00017559999999999999,
+      "loss": 0.7267,
+      "step": 880
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 0.00017579999999999996,
+      "loss": 0.6778,
+      "step": 881
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 0.000176,
+      "loss": 0.4163,
+      "step": 882
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 0.0001762,
+      "loss": 1.0095,
+      "step": 883
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 0.00017639999999999998,
+      "loss": 1.2363,
+      "step": 884
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 0.00017659999999999998,
+      "loss": 0.9183,
+      "step": 885
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 0.0001768,
+      "loss": 0.8321,
+      "step": 886
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 0.00017699999999999997,
+      "loss": 0.9908,
+      "step": 887
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 0.0001772,
+      "loss": 0.7905,
+      "step": 888
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 0.0001774,
+      "loss": 1.2317,
+      "step": 889
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 0.00017759999999999998,
+      "loss": 0.8439,
+      "step": 890
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 0.00017779999999999998,
+      "loss": 1.0982,
+      "step": 891
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 0.000178,
+      "loss": 0.7464,
+      "step": 892
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 0.00017819999999999997,
+      "loss": 0.602,
+      "step": 893
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 0.0001784,
+      "loss": 0.9053,
+      "step": 894
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 0.0001786,
+      "loss": 0.6464,
+      "step": 895
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 0.00017879999999999998,
+      "loss": 1.0542,
+      "step": 896
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 0.000179,
+      "loss": 0.5522,
+      "step": 897
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 0.0001792,
+      "loss": 0.7257,
+      "step": 898
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 0.00017939999999999997,
+      "loss": 0.6098,
+      "step": 899
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 0.0001796,
+      "loss": 0.5454,
+      "step": 900
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 0.0001798,
+      "loss": 0.7657,
+      "step": 901
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 0.78,
+      "step": 902
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 0.0001802,
+      "loss": 0.81,
+      "step": 903
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 0.0001804,
+      "loss": 0.9637,
+      "step": 904
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 0.00018059999999999997,
+      "loss": 0.7132,
+      "step": 905
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 0.00018079999999999998,
+      "loss": 0.9006,
+      "step": 906
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 0.000181,
+      "loss": 0.7289,
+      "step": 907
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 0.00018119999999999999,
+      "loss": 0.8941,
+      "step": 908
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 0.00018119999999999999,
+      "loss": 0.8517,
+      "step": 909
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 0.0001814,
+      "loss": 0.6031,
+      "step": 910
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 0.00018159999999999997,
+      "loss": 0.7666,
+      "step": 911
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 0.00018179999999999997,
+      "loss": 0.8234,
+      "step": 912
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 0.00018199999999999998,
+      "loss": 1.0899,
+      "step": 913
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 0.00018219999999999996,
+      "loss": 0.2315,
+      "step": 914
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 0.0001824,
+      "loss": 0.5115,
+      "step": 915
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 0.0001826,
+      "loss": 0.9171,
+      "step": 916
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 0.00018279999999999997,
+      "loss": 1.2401,
+      "step": 917
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 0.00018299999999999998,
+      "loss": 0.5019,
+      "step": 918
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 0.00018319999999999998,
+      "loss": 0.6102,
+      "step": 919
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 0.00018339999999999996,
+      "loss": 0.5548,
+      "step": 920
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 0.0001836,
+      "loss": 0.8196,
+      "step": 921
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 0.0001838,
+      "loss": 0.4653,
+      "step": 922
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 0.00018399999999999997,
+      "loss": 0.6946,
+      "step": 923
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 0.00018419999999999998,
+      "loss": 0.8383,
+      "step": 924
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 0.00018439999999999998,
+      "loss": 0.757,
+      "step": 925
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 0.00018459999999999996,
+      "loss": 1.1668,
+      "step": 926
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 0.0001848,
+      "loss": 1.2177,
+      "step": 927
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 0.000185,
+      "loss": 1.1548,
+      "step": 928
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 0.00018519999999999998,
+      "loss": 0.7392,
+      "step": 929
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 0.00018539999999999998,
+      "loss": 0.659,
+      "step": 930
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 0.00018559999999999998,
+      "loss": 0.4055,
+      "step": 931
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 0.00018579999999999996,
+      "loss": 0.7573,
+      "step": 932
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 0.000186,
+      "loss": 1.0789,
+      "step": 933
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 0.0001862,
+      "loss": 1.0103,
+      "step": 934
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 0.00018639999999999998,
+      "loss": 0.626,
+      "step": 935
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 0.00018659999999999998,
+      "loss": 0.5344,
+      "step": 936
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 0.0001868,
+      "loss": 1.0478,
+      "step": 937
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 0.00018699999999999996,
+      "loss": 1.033,
+      "step": 938
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 0.0001872,
+      "loss": 1.1783,
+      "step": 939
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 0.0001874,
+      "loss": 1.0731,
+      "step": 940
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 0.00018759999999999998,
+      "loss": 0.9886,
+      "step": 941
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 0.00018779999999999998,
+      "loss": 0.5971,
+      "step": 942
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 0.000188,
+      "loss": 0.9326,
+      "step": 943
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 0.00018819999999999997,
+      "loss": 0.9478,
+      "step": 944
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 0.00018839999999999997,
+      "loss": 0.5437,
+      "step": 945
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 0.0001886,
+      "loss": 0.9006,
+      "step": 946
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 0.00018879999999999998,
+      "loss": 1.1322,
+      "step": 947
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 0.00018899999999999999,
+      "loss": 0.7737,
+      "step": 948
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 0.0001892,
+      "loss": 0.658,
+      "step": 949
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 0.00018939999999999997,
+      "loss": 0.8001,
+      "step": 950
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 0.00018959999999999997,
+      "loss": 1.0102,
+      "step": 951
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 0.0001898,
+      "loss": 0.6759,
+      "step": 952
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 0.00018999999999999998,
+      "loss": 0.6328,
+      "step": 953
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 0.0001902,
+      "loss": 0.4403,
+      "step": 954
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 0.0001904,
+      "loss": 0.6944,
+      "step": 955
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 0.00019059999999999997,
+      "loss": 0.5963,
+      "step": 956
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 0.00019079999999999998,
+      "loss": 0.5515,
+      "step": 957
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 0.000191,
+      "loss": 0.9895,
+      "step": 958
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 0.00019119999999999999,
+      "loss": 0.5896,
+      "step": 959
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 0.0001914,
+      "loss": 0.6857,
+      "step": 960
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 0.0001916,
+      "loss": 0.9114,
+      "step": 961
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 0.00019179999999999997,
+      "loss": 0.7469,
+      "step": 962
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 0.6013,
+      "step": 963
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 0.0001922,
+      "loss": 1.166,
+      "step": 964
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 0.0001924,
+      "loss": 0.5838,
+      "step": 965
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 0.0001926,
+      "loss": 1.0111,
+      "step": 966
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 0.0001928,
+      "loss": 0.9963,
+      "step": 967
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 0.00019299999999999997,
+      "loss": 0.9305,
+      "step": 968
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 0.00019319999999999998,
+      "loss": 0.6325,
+      "step": 969
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 0.0001934,
+      "loss": 0.6293,
+      "step": 970
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 0.0001936,
+      "loss": 0.6669,
+      "step": 971
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 0.0001938,
+      "loss": 0.4591,
+      "step": 972
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 0.00019399999999999997,
+      "loss": 0.6482,
+      "step": 973
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 0.00019419999999999998,
+      "loss": 0.5052,
+      "step": 974
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 0.00019439999999999998,
+      "loss": 0.5585,
+      "step": 975
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 0.00019459999999999996,
+      "loss": 1.0939,
+      "step": 976
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 0.0001948,
+      "loss": 0.5625,
+      "step": 977
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 0.000195,
+      "loss": 0.7092,
+      "step": 978
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 0.00019519999999999997,
+      "loss": 0.6617,
+      "step": 979
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 0.00019539999999999998,
+      "loss": 0.4754,
+      "step": 980
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 0.00019559999999999998,
+      "loss": 0.5685,
+      "step": 981
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 0.00019579999999999996,
+      "loss": 0.5839,
+      "step": 982
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 0.00019599999999999997,
+      "loss": 0.6545,
+      "step": 983
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 0.0001962,
+      "loss": 0.6199,
+      "step": 984
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 0.00019639999999999998,
+      "loss": 1.2447,
+      "step": 985
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 0.00019659999999999998,
+      "loss": 0.6477,
+      "step": 986
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 0.00019679999999999999,
+      "loss": 0.8531,
+      "step": 987
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 0.00019699999999999996,
+      "loss": 1.2488,
+      "step": 988
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 0.00019719999999999997,
+      "loss": 1.353,
+      "step": 989
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 0.0001974,
+      "loss": 0.9964,
+      "step": 990
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 0.00019759999999999998,
+      "loss": 1.2242,
+      "step": 991
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 0.00019779999999999998,
+      "loss": 1.2178,
+      "step": 992
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 0.000198,
+      "loss": 0.9707,
+      "step": 993
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 0.00019819999999999997,
+      "loss": 1.4878,
+      "step": 994
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 0.00019839999999999997,
+      "loss": 1.3396,
+      "step": 995
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 0.0001986,
+      "loss": 1.1865,
+      "step": 996
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 0.00019879999999999998,
+      "loss": 1.0701,
+      "step": 997
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 0.00019899999999999999,
+      "loss": 1.0805,
+      "step": 998
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 0.0001992,
+      "loss": 0.8146,
+      "step": 999
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 0.00019939999999999997,
+      "loss": 0.9285,
+      "step": 1000
+    },
+    {
+      "epoch": 14.92,
+      "eval_accuracy": 0.7139011257953989,
+      "eval_f1": 0.7036851398351651,
+      "eval_loss": 0.991243302822113,
+      "eval_runtime": 344.8445,
+      "eval_samples_per_second": 11.849,
+      "eval_steps_per_second": 0.742,
+      "step": 1000
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 0.00019959999999999997,
+      "loss": 0.7898,
+      "step": 1001
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 0.0001998,
+      "loss": 1.1387,
+      "step": 1002
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 0.00019999999999999998,
+      "loss": 0.2668,
+      "step": 1003
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 0.0002002,
+      "loss": 0.3939,
+      "step": 1004
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 0.0002004,
+      "loss": 0.7606,
+      "step": 1005
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 0.00020059999999999997,
+      "loss": 1.0102,
+      "step": 1006
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 0.00020079999999999997,
+      "loss": 0.6444,
+      "step": 1007
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 0.000201,
+      "loss": 1.0351,
+      "step": 1008
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 0.00020119999999999998,
+      "loss": 0.7079,
+      "step": 1009
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 0.0002014,
+      "loss": 0.281,
+      "step": 1010
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 0.0002016,
+      "loss": 0.6687,
+      "step": 1011
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 0.00020179999999999997,
+      "loss": 0.8258,
+      "step": 1012
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 0.00020199999999999998,
+      "loss": 0.7249,
+      "step": 1013
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 0.0002022,
+      "loss": 0.776,
+      "step": 1014
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 0.0002024,
+      "loss": 0.9554,
+      "step": 1015
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 0.0002026,
+      "loss": 0.364,
+      "step": 1016
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 0.0002028,
+      "loss": 0.7906,
+      "step": 1017
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 0.00020299999999999997,
+      "loss": 0.6387,
+      "step": 1018
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 0.00020319999999999998,
+      "loss": 0.4794,
+      "step": 1019
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 0.00020339999999999998,
+      "loss": 1.0287,
+      "step": 1020
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 0.00020359999999999996,
+      "loss": 0.7316,
+      "step": 1021
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 0.0002038,
+      "loss": 0.8346,
+      "step": 1022
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 0.000204,
+      "loss": 1.083,
+      "step": 1023
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 0.00020419999999999998,
+      "loss": 0.9167,
+      "step": 1024
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 0.00020439999999999998,
+      "loss": 0.9092,
+      "step": 1025
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 0.00020459999999999999,
+      "loss": 0.6361,
+      "step": 1026
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 0.00020479999999999996,
+      "loss": 0.8581,
+      "step": 1027
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 0.000205,
+      "loss": 1.1408,
+      "step": 1028
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 0.0002052,
+      "loss": 0.6805,
+      "step": 1029
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 0.00020539999999999998,
+      "loss": 1.0752,
+      "step": 1030
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 0.00020559999999999998,
+      "loss": 0.8579,
+      "step": 1031
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 0.0002058,
+      "loss": 0.7609,
+      "step": 1032
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 0.00020599999999999997,
+      "loss": 0.6567,
+      "step": 1033
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 0.0002062,
+      "loss": 0.8744,
+      "step": 1034
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 0.00020639999999999998,
+      "loss": 1.0182,
+      "step": 1035
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 0.00020659999999999998,
+      "loss": 0.5767,
+      "step": 1036
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 0.00020679999999999999,
+      "loss": 0.7308,
+      "step": 1037
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 0.00020699999999999996,
+      "loss": 0.6098,
+      "step": 1038
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 0.00020719999999999997,
+      "loss": 1.1058,
+      "step": 1039
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 0.0002074,
+      "loss": 0.471,
+      "step": 1040
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 0.00020759999999999998,
+      "loss": 0.9943,
+      "step": 1041
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 0.00020779999999999998,
+      "loss": 0.598,
+      "step": 1042
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 0.000208,
+      "loss": 0.8455,
+      "step": 1043
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 0.00020819999999999996,
+      "loss": 0.7184,
+      "step": 1044
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 0.00020839999999999997,
+      "loss": 1.0079,
+      "step": 1045
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 0.0002086,
+      "loss": 0.5931,
+      "step": 1046
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 0.00020879999999999998,
+      "loss": 0.6176,
+      "step": 1047
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 0.00020899999999999998,
+      "loss": 0.6809,
+      "step": 1048
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 0.0002092,
+      "loss": 0.3144,
+      "step": 1049
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 0.00020939999999999997,
+      "loss": 0.7481,
+      "step": 1050
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 0.00020959999999999997,
+      "loss": 0.5394,
+      "step": 1051
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 0.0002098,
+      "loss": 0.6844,
+      "step": 1052
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 0.6366,
+      "step": 1053
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 0.0002102,
+      "loss": 0.585,
+      "step": 1054
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 0.0002104,
+      "loss": 0.6458,
+      "step": 1055
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 0.00021059999999999997,
+      "loss": 1.1576,
+      "step": 1056
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 0.00021079999999999997,
+      "loss": 0.7424,
+      "step": 1057
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 0.00021099999999999998,
+      "loss": 0.8765,
+      "step": 1058
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 0.00021119999999999996,
+      "loss": 0.6492,
+      "step": 1059
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 0.0002114,
+      "loss": 0.9297,
+      "step": 1060
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 0.0002116,
+      "loss": 0.7282,
+      "step": 1061
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 0.00021179999999999997,
+      "loss": 1.1035,
+      "step": 1062
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 0.00021199999999999998,
+      "loss": 0.8628,
+      "step": 1063
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 0.00021219999999999998,
+      "loss": 0.7203,
+      "step": 1064
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 0.00021239999999999996,
+      "loss": 0.5733,
+      "step": 1065
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 0.0002126,
+      "loss": 0.6324,
+      "step": 1066
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 0.0002128,
+      "loss": 0.899,
+      "step": 1067
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 0.7308,
+      "step": 1068
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 0.00021319999999999998,
+      "loss": 0.8845,
+      "step": 1069
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 0.00021339999999999998,
+      "loss": 0.2766,
+      "step": 1070
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 0.00021359999999999996,
+      "loss": 0.7306,
+      "step": 1071
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 0.0002138,
+      "loss": 0.9311,
+      "step": 1072
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 0.000214,
+      "loss": 0.7444,
+      "step": 1073
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 0.00021419999999999998,
+      "loss": 0.9739,
+      "step": 1074
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 0.00021439999999999998,
+      "loss": 0.9668,
+      "step": 1075
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 0.00021459999999999998,
+      "loss": 0.7567,
+      "step": 1076
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 0.00021479999999999996,
+      "loss": 0.4541,
+      "step": 1077
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 0.000215,
+      "loss": 0.443,
+      "step": 1078
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 0.0002152,
+      "loss": 0.4377,
+      "step": 1079
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 0.00021539999999999998,
+      "loss": 0.4075,
+      "step": 1080
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 0.00021559999999999998,
+      "loss": 1.0803,
+      "step": 1081
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 0.0002158,
+      "loss": 0.7278,
+      "step": 1082
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 0.4957,
+      "step": 1083
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 0.0002162,
+      "loss": 0.4261,
+      "step": 1084
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 0.0002164,
+      "loss": 0.3906,
+      "step": 1085
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 0.00021659999999999998,
+      "loss": 0.7426,
+      "step": 1086
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 0.00021679999999999998,
+      "loss": 0.6193,
+      "step": 1087
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 0.000217,
+      "loss": 0.5344,
+      "step": 1088
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 0.00021719999999999997,
+      "loss": 0.9232,
+      "step": 1089
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 0.0002174,
+      "loss": 0.5786,
+      "step": 1090
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 0.0002176,
+      "loss": 0.8235,
+      "step": 1091
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 0.00021779999999999998,
+      "loss": 0.9143,
+      "step": 1092
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 0.00021799999999999999,
+      "loss": 0.3326,
+      "step": 1093
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 0.0002182,
+      "loss": 0.2456,
+      "step": 1094
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 0.00021839999999999997,
+      "loss": 0.858,
+      "step": 1095
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 0.00021859999999999997,
+      "loss": 1.022,
+      "step": 1096
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 0.00021879999999999995,
+      "loss": 0.5284,
+      "step": 1097
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 0.3627,
+      "step": 1098
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 0.0002192,
+      "loss": 0.9552,
+      "step": 1099
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 0.00021939999999999997,
+      "loss": 0.619,
+      "step": 1100
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 0.00021959999999999997,
+      "loss": 0.2164,
+      "step": 1101
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 0.00021979999999999998,
+      "loss": 0.7193,
+      "step": 1102
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 0.00021999999999999995,
+      "loss": 0.7335,
+      "step": 1103
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 0.00022019999999999999,
+      "loss": 0.367,
+      "step": 1104
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 0.0002204,
+      "loss": 0.4086,
+      "step": 1105
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 0.00022059999999999997,
+      "loss": 1.1209,
+      "step": 1106
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 0.00022079999999999997,
+      "loss": 0.4251,
+      "step": 1107
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 0.00022099999999999998,
+      "loss": 0.2345,
+      "step": 1108
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 0.00022119999999999996,
+      "loss": 0.9918,
+      "step": 1109
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 0.0002214,
+      "loss": 0.5629,
+      "step": 1110
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 0.0002216,
+      "loss": 0.9618,
+      "step": 1111
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 0.00022179999999999997,
+      "loss": 0.9031,
+      "step": 1112
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 0.7487,
+      "step": 1113
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 0.00022219999999999998,
+      "loss": 0.8206,
+      "step": 1114
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 0.00022239999999999996,
+      "loss": 0.3871,
+      "step": 1115
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 0.0002226,
+      "loss": 0.6073,
+      "step": 1116
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 0.0002228,
+      "loss": 0.3924,
+      "step": 1117
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 0.00022299999999999997,
+      "loss": 0.6858,
+      "step": 1118
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 0.00022319999999999998,
+      "loss": 0.51,
+      "step": 1119
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 0.00022339999999999998,
+      "loss": 0.8494,
+      "step": 1120
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 0.00022359999999999996,
+      "loss": 0.7321,
+      "step": 1121
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 0.0002238,
+      "loss": 0.9092,
+      "step": 1122
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 0.000224,
+      "loss": 0.8845,
+      "step": 1123
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 0.00022419999999999997,
+      "loss": 1.2335,
+      "step": 1124
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 0.00022439999999999998,
+      "loss": 1.1761,
+      "step": 1125
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 0.00022459999999999998,
+      "loss": 1.3628,
+      "step": 1126
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 0.00022479999999999996,
+      "loss": 0.6865,
+      "step": 1127
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 0.000225,
+      "loss": 1.1485,
+      "step": 1128
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 0.0002252,
+      "loss": 0.9444,
+      "step": 1129
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 0.00022539999999999998,
+      "loss": 0.3779,
+      "step": 1130
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 0.00022559999999999998,
+      "loss": 0.7367,
+      "step": 1131
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 0.00022579999999999999,
+      "loss": 0.7024,
+      "step": 1132
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 0.00022599999999999996,
+      "loss": 0.6302,
+      "step": 1133
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 0.00022619999999999997,
+      "loss": 0.7474,
+      "step": 1134
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 0.0002264,
+      "loss": 0.9909,
+      "step": 1135
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 0.00022659999999999998,
+      "loss": 0.8745,
+      "step": 1136
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 0.00022679999999999998,
+      "loss": 0.4185,
+      "step": 1137
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 0.000227,
+      "loss": 0.9698,
+      "step": 1138
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 0.00022719999999999997,
+      "loss": 1.022,
+      "step": 1139
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 0.00022739999999999997,
+      "loss": 1.537,
+      "step": 1140
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 0.0002276,
+      "loss": 0.7912,
+      "step": 1141
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 0.00022779999999999998,
+      "loss": 0.8698,
+      "step": 1142
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 0.7288,
+      "step": 1143
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 0.0002282,
+      "loss": 0.2158,
+      "step": 1144
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 0.00022839999999999997,
+      "loss": 0.631,
+      "step": 1145
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 0.00022859999999999997,
+      "loss": 0.8167,
+      "step": 1146
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 0.0002288,
+      "loss": 0.4729,
+      "step": 1147
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 0.00022899999999999998,
+      "loss": 0.8643,
+      "step": 1148
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 0.0002292,
+      "loss": 0.5403,
+      "step": 1149
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 0.0002294,
+      "loss": 0.7748,
+      "step": 1150
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 0.00022959999999999997,
+      "loss": 0.9571,
+      "step": 1151
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 0.00022979999999999997,
+      "loss": 0.5619,
+      "step": 1152
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 0.00023,
+      "loss": 0.6504,
+      "step": 1153
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 0.00023019999999999998,
+      "loss": 0.5057,
+      "step": 1154
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 0.0002304,
+      "loss": 0.481,
+      "step": 1155
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 0.0002306,
+      "loss": 1.1169,
+      "step": 1156
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 0.00023079999999999997,
+      "loss": 0.8684,
+      "step": 1157
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 0.00023099999999999998,
+      "loss": 0.5652,
+      "step": 1158
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 0.0002312,
+      "loss": 0.9157,
+      "step": 1159
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 0.0002314,
+      "loss": 0.912,
+      "step": 1160
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 0.0002316,
+      "loss": 1.064,
+      "step": 1161
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 0.00023179999999999997,
+      "loss": 0.481,
+      "step": 1162
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 0.00023199999999999997,
+      "loss": 1.0895,
+      "step": 1163
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 0.00023219999999999998,
+      "loss": 0.9731,
+      "step": 1164
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 0.00023239999999999996,
+      "loss": 0.7848,
+      "step": 1165
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 0.00023259999999999996,
+      "loss": 0.8208,
+      "step": 1166
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 0.0002328,
+      "loss": 0.3356,
+      "step": 1167
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 0.00023299999999999997,
+      "loss": 0.5094,
+      "step": 1168
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 0.00023319999999999998,
+      "loss": 0.5053,
+      "step": 1169
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 0.00023339999999999998,
+      "loss": 0.6877,
+      "step": 1170
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 0.00023359999999999996,
+      "loss": 0.4477,
+      "step": 1171
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 0.00023379999999999996,
+      "loss": 0.5053,
+      "step": 1172
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 0.000234,
+      "loss": 0.9771,
+      "step": 1173
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 0.00023419999999999997,
+      "loss": 1.0193,
+      "step": 1174
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 0.00023439999999999998,
+      "loss": 0.9896,
+      "step": 1175
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 0.00023459999999999998,
+      "loss": 0.7799,
+      "step": 1176
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 0.00023479999999999996,
+      "loss": 1.0455,
+      "step": 1177
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 0.00023499999999999997,
+      "loss": 0.5766,
+      "step": 1178
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 0.0002352,
+      "loss": 1.0561,
+      "step": 1179
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 0.00023539999999999998,
+      "loss": 0.6531,
+      "step": 1180
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 0.00023559999999999998,
+      "loss": 1.1587,
+      "step": 1181
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 0.00023579999999999999,
+      "loss": 0.3106,
+      "step": 1182
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 0.00023599999999999996,
+      "loss": 0.9023,
+      "step": 1183
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 0.00023619999999999997,
+      "loss": 0.5792,
+      "step": 1184
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 0.0002364,
+      "loss": 0.7693,
+      "step": 1185
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 0.00023659999999999998,
+      "loss": 0.4481,
+      "step": 1186
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 0.00023679999999999998,
+      "loss": 0.6329,
+      "step": 1187
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 0.000237,
+      "loss": 0.4823,
+      "step": 1188
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 0.00023719999999999997,
+      "loss": 0.3002,
+      "step": 1189
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 0.00023739999999999997,
+      "loss": 0.8866,
+      "step": 1190
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 0.0002376,
+      "loss": 0.6895,
+      "step": 1191
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 0.00023779999999999998,
+      "loss": 1.239,
+      "step": 1192
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 0.00023799999999999998,
+      "loss": 0.8187,
+      "step": 1193
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 0.0002382,
+      "loss": 0.6027,
+      "step": 1194
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 0.00023839999999999997,
+      "loss": 0.7884,
+      "step": 1195
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 0.00023859999999999997,
+      "loss": 0.9204,
+      "step": 1196
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 0.0002388,
+      "loss": 0.7918,
+      "step": 1197
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 0.00023899999999999998,
+      "loss": 0.7283,
+      "step": 1198
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 0.0002392,
+      "loss": 0.429,
+      "step": 1199
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 0.0002394,
+      "loss": 0.4164,
+      "step": 1200
+    },
+    {
+      "epoch": 17.91,
+      "eval_accuracy": 0.7741067058247675,
+      "eval_f1": 0.7742523404153536,
+      "eval_loss": 0.8225661516189575,
+      "eval_runtime": 343.498,
+      "eval_samples_per_second": 11.895,
+      "eval_steps_per_second": 0.745,
+      "step": 1200
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 0.00023959999999999997,
+      "loss": 0.2999,
+      "step": 1201
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 0.00023979999999999997,
+      "loss": 0.6004,
+      "step": 1202
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 0.4274,
+      "step": 1203
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 0.00024019999999999996,
+      "loss": 0.3801,
+      "step": 1204
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 0.0002404,
+      "loss": 0.6226,
+      "step": 1205
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 0.0002406,
+      "loss": 0.8777,
+      "step": 1206
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 0.00024079999999999997,
+      "loss": 1.1747,
+      "step": 1207
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 0.00024099999999999998,
+      "loss": 0.6821,
+      "step": 1208
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 0.00024119999999999998,
+      "loss": 0.64,
+      "step": 1209
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 0.00024139999999999996,
+      "loss": 0.4556,
+      "step": 1210
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 0.0002416,
+      "loss": 0.5366,
+      "step": 1211
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 0.0002418,
+      "loss": 0.3206,
+      "step": 1212
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 0.00024199999999999997,
+      "loss": 0.496,
+      "step": 1213
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 0.00024219999999999998,
+      "loss": 0.9585,
+      "step": 1214
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 0.00024239999999999998,
+      "loss": 0.331,
+      "step": 1215
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 0.00024259999999999996,
+      "loss": 0.4689,
+      "step": 1216
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 0.0002428,
+      "loss": 0.9751,
+      "step": 1217
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 0.000243,
+      "loss": 0.6266,
+      "step": 1218
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 0.00024319999999999998,
+      "loss": 0.7254,
+      "step": 1219
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 0.00024339999999999998,
+      "loss": 0.86,
+      "step": 1220
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 0.00024359999999999999,
+      "loss": 0.7694,
+      "step": 1221
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 0.00024379999999999996,
+      "loss": 0.5908,
+      "step": 1222
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 0.000244,
+      "loss": 0.7398,
+      "step": 1223
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 0.00024419999999999997,
+      "loss": 0.5606,
+      "step": 1224
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 0.0002444,
+      "loss": 0.9814,
+      "step": 1225
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 0.0002446,
+      "loss": 1.1783,
+      "step": 1226
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 0.0002448,
+      "loss": 0.5309,
+      "step": 1227
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 0.000245,
+      "loss": 0.5142,
+      "step": 1228
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 0.0002452,
+      "loss": 0.5806,
+      "step": 1229
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 0.00024539999999999995,
+      "loss": 0.6274,
+      "step": 1230
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 0.00024559999999999995,
+      "loss": 0.7508,
+      "step": 1231
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 0.0002458,
+      "loss": 0.2308,
+      "step": 1232
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 0.6213,
+      "step": 1233
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 0.00024619999999999997,
+      "loss": 0.8141,
+      "step": 1234
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 0.00024639999999999997,
+      "loss": 0.2922,
+      "step": 1235
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 0.0002466,
+      "loss": 0.5329,
+      "step": 1236
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 0.0002468,
+      "loss": 0.4933,
+      "step": 1237
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 0.000247,
+      "loss": 0.5091,
+      "step": 1238
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 0.0002472,
+      "loss": 0.6618,
+      "step": 1239
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 0.0002474,
+      "loss": 1.1074,
+      "step": 1240
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 0.0002476,
+      "loss": 0.3855,
+      "step": 1241
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 0.00024779999999999995,
+      "loss": 0.5161,
+      "step": 1242
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 0.00024799999999999996,
+      "loss": 0.7131,
+      "step": 1243
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 0.00024819999999999996,
+      "loss": 0.6932,
+      "step": 1244
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 0.00024839999999999997,
+      "loss": 0.8774,
+      "step": 1245
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 0.00024859999999999997,
+      "loss": 0.5768,
+      "step": 1246
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 0.0002488,
+      "loss": 0.981,
+      "step": 1247
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 0.000249,
+      "loss": 1.1701,
+      "step": 1248
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 0.0002492,
+      "loss": 0.735,
+      "step": 1249
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 0.0002494,
+      "loss": 0.7628,
+      "step": 1250
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 0.00024959999999999994,
+      "loss": 1.064,
+      "step": 1251
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 0.0002498,
+      "loss": 1.0694,
+      "step": 1252
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 0.00025,
+      "loss": 0.4311,
+      "step": 1253
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 0.00025019999999999996,
+      "loss": 0.7704,
+      "step": 1254
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 0.00025039999999999996,
+      "loss": 0.2663,
+      "step": 1255
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 0.00025059999999999997,
+      "loss": 0.3193,
+      "step": 1256
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 0.00025079999999999997,
+      "loss": 0.5628,
+      "step": 1257
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 0.000251,
+      "loss": 0.779,
+      "step": 1258
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 0.0002512,
+      "loss": 1.1226,
+      "step": 1259
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 0.0002514,
+      "loss": 0.8199,
+      "step": 1260
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 0.0002516,
+      "loss": 0.6151,
+      "step": 1261
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 0.0002518,
+      "loss": 0.5055,
+      "step": 1262
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 0.1621,
+      "step": 1263
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 0.0002522,
+      "loss": 0.5277,
+      "step": 1264
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 0.0002524,
+      "loss": 0.9161,
+      "step": 1265
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 0.00025259999999999996,
+      "loss": 0.6097,
+      "step": 1266
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 0.00025279999999999996,
+      "loss": 0.4727,
+      "step": 1267
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 0.00025299999999999997,
+      "loss": 0.3202,
+      "step": 1268
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 0.0002532,
+      "loss": 1.0695,
+      "step": 1269
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 0.0002534,
+      "loss": 0.6529,
+      "step": 1270
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 0.0002536,
+      "loss": 0.276,
+      "step": 1271
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 0.0002538,
+      "loss": 0.5075,
+      "step": 1272
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 0.000254,
+      "loss": 0.6272,
+      "step": 1273
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 0.0002542,
+      "loss": 1.1173,
+      "step": 1274
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 0.00025439999999999995,
+      "loss": 0.3821,
+      "step": 1275
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 0.0002546,
+      "loss": 0.544,
+      "step": 1276
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 0.0002548,
+      "loss": 0.3951,
+      "step": 1277
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 0.00025499999999999996,
+      "loss": 0.7183,
+      "step": 1278
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 0.00025519999999999997,
+      "loss": 0.5846,
+      "step": 1279
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 0.0002554,
+      "loss": 0.8813,
+      "step": 1280
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 0.0002556,
+      "loss": 0.2713,
+      "step": 1281
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 0.0002558,
+      "loss": 0.666,
+      "step": 1282
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 0.000256,
+      "loss": 0.7438,
+      "step": 1283
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 0.0002562,
+      "loss": 0.5591,
+      "step": 1284
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 0.0002564,
+      "loss": 0.4984,
+      "step": 1285
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 0.00025659999999999995,
+      "loss": 0.6012,
+      "step": 1286
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 0.00025679999999999995,
+      "loss": 0.6843,
+      "step": 1287
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 0.00025699999999999996,
+      "loss": 0.7454,
+      "step": 1288
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 0.00025719999999999996,
+      "loss": 0.5694,
+      "step": 1289
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 0.00025739999999999997,
+      "loss": 0.7526,
+      "step": 1290
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 0.0002576,
+      "loss": 0.7195,
+      "step": 1291
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 0.0002578,
+      "loss": 0.5612,
+      "step": 1292
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 0.000258,
+      "loss": 0.9403,
+      "step": 1293
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 0.0002582,
+      "loss": 0.5989,
+      "step": 1294
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 0.00025839999999999994,
+      "loss": 0.8607,
+      "step": 1295
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 0.0002586,
+      "loss": 0.5824,
+      "step": 1296
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 0.0002588,
+      "loss": 0.6549,
+      "step": 1297
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 0.00025899999999999995,
+      "loss": 0.6932,
+      "step": 1298
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 0.00025919999999999996,
+      "loss": 0.6864,
+      "step": 1299
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 0.00025939999999999996,
+      "loss": 0.5762,
+      "step": 1300
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 0.00025959999999999997,
+      "loss": 0.8634,
+      "step": 1301
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 0.00025979999999999997,
+      "loss": 0.6391,
+      "step": 1302
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 0.00026,
+      "loss": 0.7527,
+      "step": 1303
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 0.0002602,
+      "loss": 0.4871,
+      "step": 1304
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 0.0002604,
+      "loss": 0.2296,
+      "step": 1305
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 0.0002606,
+      "loss": 0.282,
+      "step": 1306
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 0.00026079999999999994,
+      "loss": 0.8365,
+      "step": 1307
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 0.000261,
+      "loss": 0.5651,
+      "step": 1308
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 0.0002612,
+      "loss": 0.8462,
+      "step": 1309
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 0.00026139999999999996,
+      "loss": 0.7099,
+      "step": 1310
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 0.00026159999999999996,
+      "loss": 0.6501,
+      "step": 1311
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 0.00026179999999999997,
+      "loss": 0.6466,
+      "step": 1312
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 0.00026199999999999997,
+      "loss": 0.5846,
+      "step": 1313
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 0.0002622,
+      "loss": 0.9175,
+      "step": 1314
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 0.0002624,
+      "loss": 0.4235,
+      "step": 1315
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 0.0002626,
+      "loss": 0.4004,
+      "step": 1316
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 0.0002628,
+      "loss": 0.5586,
+      "step": 1317
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 0.000263,
+      "loss": 0.5125,
+      "step": 1318
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 0.00026319999999999995,
+      "loss": 0.4819,
+      "step": 1319
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 0.00026339999999999995,
+      "loss": 0.3895,
+      "step": 1320
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 0.0002636,
+      "loss": 0.4449,
+      "step": 1321
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 0.00026379999999999996,
+      "loss": 0.6049,
+      "step": 1322
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 1.2677,
+      "step": 1323
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 0.00026419999999999997,
+      "loss": 1.0832,
+      "step": 1324
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 0.0002644,
+      "loss": 0.7723,
+      "step": 1325
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 0.0002646,
+      "loss": 1.1757,
+      "step": 1326
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 0.0002648,
+      "loss": 0.5295,
+      "step": 1327
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 0.000265,
+      "loss": 0.2148,
+      "step": 1328
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 0.0002652,
+      "loss": 0.3802,
+      "step": 1329
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 0.0002654,
+      "loss": 0.3808,
+      "step": 1330
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 0.00026559999999999995,
+      "loss": 0.6835,
+      "step": 1331
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 0.00026579999999999996,
+      "loss": 0.7268,
+      "step": 1332
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 0.000266,
+      "loss": 0.3596,
+      "step": 1333
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 0.00026619999999999997,
+      "loss": 0.6912,
+      "step": 1334
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 0.00026639999999999997,
+      "loss": 1.3149,
+      "step": 1335
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 0.0002666,
+      "loss": 0.7943,
+      "step": 1336
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 0.0002668,
+      "loss": 1.0217,
+      "step": 1337
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 0.000267,
+      "loss": 0.6082,
+      "step": 1338
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 0.0002672,
+      "loss": 0.436,
+      "step": 1339
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 0.0002674,
+      "loss": 0.7156,
+      "step": 1340
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 0.0002676,
+      "loss": 0.6298,
+      "step": 1341
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 0.0002678,
+      "loss": 0.5792,
+      "step": 1342
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 0.00026799999999999995,
+      "loss": 0.6893,
+      "step": 1343
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 0.00026819999999999996,
+      "loss": 0.6991,
+      "step": 1344
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 0.0002684,
+      "loss": 0.7287,
+      "step": 1345
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 0.00026859999999999997,
+      "loss": 0.4918,
+      "step": 1346
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 0.0002688,
+      "loss": 0.912,
+      "step": 1347
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 0.000269,
+      "loss": 0.5828,
+      "step": 1348
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 0.0002692,
+      "loss": 0.6996,
+      "step": 1349
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 0.0002694,
+      "loss": 0.46,
+      "step": 1350
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 0.00026959999999999994,
+      "loss": 0.3978,
+      "step": 1351
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 0.0002698,
+      "loss": 0.5182,
+      "step": 1352
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 0.00027,
+      "loss": 0.893,
+      "step": 1353
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 0.00027019999999999995,
+      "loss": 0.9816,
+      "step": 1354
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 0.00027039999999999996,
+      "loss": 0.4094,
+      "step": 1355
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 0.00027059999999999996,
+      "loss": 0.4442,
+      "step": 1356
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 0.00027079999999999997,
+      "loss": 0.4466,
+      "step": 1357
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 0.000271,
+      "loss": 0.415,
+      "step": 1358
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 0.0002712,
+      "loss": 0.9744,
+      "step": 1359
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 0.0002714,
+      "loss": 0.9508,
+      "step": 1360
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 0.0002716,
+      "loss": 0.635,
+      "step": 1361
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 0.0002718,
+      "loss": 0.7858,
+      "step": 1362
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 0.00027199999999999994,
+      "loss": 1.0854,
+      "step": 1363
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 0.00027219999999999995,
+      "loss": 0.9812,
+      "step": 1364
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 0.0002724,
+      "loss": 0.5966,
+      "step": 1365
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 0.00027259999999999996,
+      "loss": 0.4025,
+      "step": 1366
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 0.00027279999999999996,
+      "loss": 0.5009,
+      "step": 1367
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 0.00027299999999999997,
+      "loss": 0.1482,
+      "step": 1368
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 0.00027319999999999997,
+      "loss": 0.5557,
+      "step": 1369
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 0.0002734,
+      "loss": 0.6613,
+      "step": 1370
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 0.0002736,
+      "loss": 0.8169,
+      "step": 1371
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 0.0002738,
+      "loss": 0.5959,
+      "step": 1372
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 0.000274,
+      "loss": 0.4948,
+      "step": 1373
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 0.0002742,
+      "loss": 0.7418,
+      "step": 1374
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 0.00027439999999999995,
+      "loss": 0.5943,
+      "step": 1375
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 0.00027459999999999995,
+      "loss": 0.716,
+      "step": 1376
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 0.0002748,
+      "loss": 0.8324,
+      "step": 1377
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 0.00027499999999999996,
+      "loss": 0.2354,
+      "step": 1378
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 0.00027519999999999997,
+      "loss": 0.5413,
+      "step": 1379
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 0.00027539999999999997,
+      "loss": 0.6385,
+      "step": 1380
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 0.0002756,
+      "loss": 0.5612,
+      "step": 1381
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 0.0002758,
+      "loss": 0.9119,
+      "step": 1382
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 0.000276,
+      "loss": 0.5478,
+      "step": 1383
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 0.0002762,
+      "loss": 0.5139,
+      "step": 1384
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 0.0002764,
+      "loss": 0.7378,
+      "step": 1385
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 0.0002766,
+      "loss": 0.6912,
+      "step": 1386
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 0.00027679999999999995,
+      "loss": 0.391,
+      "step": 1387
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 0.00027699999999999996,
+      "loss": 0.3529,
+      "step": 1388
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 0.0002772,
+      "loss": 0.6294,
+      "step": 1389
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 0.00027739999999999997,
+      "loss": 0.5305,
+      "step": 1390
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 0.00027759999999999997,
+      "loss": 0.6153,
+      "step": 1391
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 0.0002778,
+      "loss": 0.7893,
+      "step": 1392
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 0.000278,
+      "loss": 0.7969,
+      "step": 1393
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 0.0002782,
+      "loss": 0.5978,
+      "step": 1394
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 0.0002784,
+      "loss": 0.7604,
+      "step": 1395
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 0.00027859999999999994,
+      "loss": 0.8762,
+      "step": 1396
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 0.0002788,
+      "loss": 0.727,
+      "step": 1397
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 0.000279,
+      "loss": 0.8943,
+      "step": 1398
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 0.00027919999999999996,
+      "loss": 0.6312,
+      "step": 1399
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 0.00027939999999999996,
+      "loss": 0.7669,
+      "step": 1400
+    },
+    {
+      "epoch": 20.89,
+      "eval_accuracy": 0.7787567302985805,
+      "eval_f1": 0.778299974393245,
+      "eval_loss": 0.813083827495575,
+      "eval_runtime": 344.4352,
+      "eval_samples_per_second": 11.863,
+      "eval_steps_per_second": 0.743,
+      "step": 1400
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 0.00027959999999999997,
+      "loss": 0.455,
+      "step": 1401
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 0.00027979999999999997,
+      "loss": 0.6066,
+      "step": 1402
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 0.00028,
+      "loss": 0.3945,
+      "step": 1403
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 0.0002802,
+      "loss": 0.7466,
+      "step": 1404
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 0.0002804,
+      "loss": 0.5893,
+      "step": 1405
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 0.0002806,
+      "loss": 0.6915,
+      "step": 1406
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 0.0002808,
+      "loss": 1.0937,
+      "step": 1407
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 0.00028099999999999995,
+      "loss": 0.8182,
+      "step": 1408
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 0.0002812,
+      "loss": 0.9176,
+      "step": 1409
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 0.00028139999999999996,
+      "loss": 0.8487,
+      "step": 1410
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 0.00028159999999999996,
+      "loss": 0.4835,
+      "step": 1411
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 0.00028179999999999997,
+      "loss": 0.9755,
+      "step": 1412
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 1.315,
+      "step": 1413
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 0.0002822,
+      "loss": 0.8134,
+      "step": 1414
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 0.0002824,
+      "loss": 0.7944,
+      "step": 1415
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 0.0002826,
+      "loss": 0.4791,
+      "step": 1416
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 0.0002828,
+      "loss": 0.6249,
+      "step": 1417
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 0.000283,
+      "loss": 0.8417,
+      "step": 1418
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 0.00028319999999999994,
+      "loss": 0.774,
+      "step": 1419
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 0.00028339999999999995,
+      "loss": 0.727,
+      "step": 1420
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 0.0002836,
+      "loss": 0.3706,
+      "step": 1421
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 0.00028379999999999996,
+      "loss": 0.3887,
+      "step": 1422
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 0.00028399999999999996,
+      "loss": 0.6382,
+      "step": 1423
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 0.00028419999999999997,
+      "loss": 0.718,
+      "step": 1424
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 0.0002844,
+      "loss": 0.8399,
+      "step": 1425
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 0.0002846,
+      "loss": 0.9529,
+      "step": 1426
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 0.0002848,
+      "loss": 0.9305,
+      "step": 1427
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 0.000285,
+      "loss": 0.7231,
+      "step": 1428
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 0.0002852,
+      "loss": 0.5729,
+      "step": 1429
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 0.0002854,
+      "loss": 0.5407,
+      "step": 1430
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 0.00028559999999999995,
+      "loss": 0.7901,
+      "step": 1431
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 0.00028579999999999995,
+      "loss": 0.575,
+      "step": 1432
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 0.00028599999999999996,
+      "loss": 0.219,
+      "step": 1433
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 0.00028619999999999996,
+      "loss": 0.7134,
+      "step": 1434
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 0.00028639999999999997,
+      "loss": 0.2605,
+      "step": 1435
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 0.0002866,
+      "loss": 0.5372,
+      "step": 1436
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 0.0002868,
+      "loss": 0.4434,
+      "step": 1437
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 0.000287,
+      "loss": 0.774,
+      "step": 1438
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 0.0002872,
+      "loss": 0.2877,
+      "step": 1439
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 0.00028739999999999994,
+      "loss": 0.2525,
+      "step": 1440
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 0.0002876,
+      "loss": 0.5474,
+      "step": 1441
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 0.0002878,
+      "loss": 0.4075,
+      "step": 1442
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 0.72,
+      "step": 1443
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 0.00028819999999999996,
+      "loss": 0.3995,
+      "step": 1444
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 0.00028839999999999996,
+      "loss": 0.5891,
+      "step": 1445
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 0.00028859999999999997,
+      "loss": 0.8102,
+      "step": 1446
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 0.00028879999999999997,
+      "loss": 0.6418,
+      "step": 1447
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 0.000289,
+      "loss": 0.3104,
+      "step": 1448
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 0.0002892,
+      "loss": 0.8299,
+      "step": 1449
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 0.0002894,
+      "loss": 0.2404,
+      "step": 1450
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 0.0002896,
+      "loss": 0.3771,
+      "step": 1451
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 0.00028979999999999994,
+      "loss": 0.7731,
+      "step": 1452
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 0.00029,
+      "loss": 0.2826,
+      "step": 1453
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 0.0002902,
+      "loss": 0.5406,
+      "step": 1454
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 0.00029039999999999996,
+      "loss": 0.5265,
+      "step": 1455
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 0.00029059999999999996,
+      "loss": 0.2196,
+      "step": 1456
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 0.00029079999999999997,
+      "loss": 0.5047,
+      "step": 1457
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 0.00029099999999999997,
+      "loss": 0.8224,
+      "step": 1458
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 0.0002912,
+      "loss": 0.7324,
+      "step": 1459
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 0.0002914,
+      "loss": 0.9723,
+      "step": 1460
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 0.0002916,
+      "loss": 0.7277,
+      "step": 1461
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 0.0002918,
+      "loss": 0.6583,
+      "step": 1462
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 0.000292,
+      "loss": 0.6693,
+      "step": 1463
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 0.00029219999999999995,
+      "loss": 0.5129,
+      "step": 1464
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 0.0002924,
+      "loss": 0.5233,
+      "step": 1465
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 0.0002926,
+      "loss": 0.2062,
+      "step": 1466
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 0.00029279999999999996,
+      "loss": 0.9001,
+      "step": 1467
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 0.00029299999999999997,
+      "loss": 0.3718,
+      "step": 1468
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 0.00029319999999999997,
+      "loss": 0.5696,
+      "step": 1469
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 0.0002934,
+      "loss": 0.7076,
+      "step": 1470
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 0.0002936,
+      "loss": 0.7528,
+      "step": 1471
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 0.00029379999999999993,
+      "loss": 0.4238,
+      "step": 1472
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 0.000294,
+      "loss": 0.5624,
+      "step": 1473
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 0.0002942,
+      "loss": 0.8636,
+      "step": 1474
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 0.00029439999999999995,
+      "loss": 0.3772,
+      "step": 1475
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 0.00029459999999999995,
+      "loss": 1.2659,
+      "step": 1476
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 0.00029479999999999996,
+      "loss": 0.4182,
+      "step": 1477
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 0.00029499999999999996,
+      "loss": 0.3934,
+      "step": 1478
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 0.00029519999999999997,
+      "loss": 0.4777,
+      "step": 1479
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 0.00029539999999999997,
+      "loss": 0.8398,
+      "step": 1480
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 0.0002956,
+      "loss": 0.7846,
+      "step": 1481
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 0.0002958,
+      "loss": 0.8802,
+      "step": 1482
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 0.000296,
+      "loss": 0.4906,
+      "step": 1483
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 0.00029619999999999994,
+      "loss": 0.5514,
+      "step": 1484
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 0.0002964,
+      "loss": 0.3834,
+      "step": 1485
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 0.0002966,
+      "loss": 0.3833,
+      "step": 1486
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 0.00029679999999999995,
+      "loss": 0.2823,
+      "step": 1487
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 0.00029699999999999996,
+      "loss": 0.7299,
+      "step": 1488
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 0.00029719999999999996,
+      "loss": 0.2089,
+      "step": 1489
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 0.00029739999999999996,
+      "loss": 0.103,
+      "step": 1490
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 0.00029759999999999997,
+      "loss": 0.4865,
+      "step": 1491
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 0.0002978,
+      "loss": 0.9131,
+      "step": 1492
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 0.000298,
+      "loss": 0.4465,
+      "step": 1493
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 0.0002982,
+      "loss": 0.3508,
+      "step": 1494
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 0.0002984,
+      "loss": 0.4053,
+      "step": 1495
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 0.00029859999999999994,
+      "loss": 1.0208,
+      "step": 1496
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 0.0002988,
+      "loss": 0.3976,
+      "step": 1497
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 0.000299,
+      "loss": 0.6894,
+      "step": 1498
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 0.00029919999999999995,
+      "loss": 0.1601,
+      "step": 1499
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 0.00029939999999999996,
+      "loss": 0.4907,
+      "step": 1500
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 0.00029959999999999996,
+      "loss": 0.3205,
+      "step": 1501
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 0.00029979999999999997,
+      "loss": 1.2744,
+      "step": 1502
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 0.0003,
+      "loss": 0.4245,
+      "step": 1503
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 0.0002999649122807017,
+      "loss": 0.7102,
+      "step": 1504
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 0.00029992982456140347,
+      "loss": 1.26,
+      "step": 1505
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 0.0002998947368421052,
+      "loss": 0.8482,
+      "step": 1506
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 0.00029985964912280697,
+      "loss": 0.6502,
+      "step": 1507
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 0.0002998245614035087,
+      "loss": 1.0905,
+      "step": 1508
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 0.0002997894736842105,
+      "loss": 0.9749,
+      "step": 1509
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 0.00029975438596491227,
+      "loss": 0.8889,
+      "step": 1510
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 0.000299719298245614,
+      "loss": 0.9885,
+      "step": 1511
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 0.00029968421052631577,
+      "loss": 0.9825,
+      "step": 1512
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 0.0002996491228070175,
+      "loss": 0.5989,
+      "step": 1513
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 0.00029961403508771926,
+      "loss": 0.6243,
+      "step": 1514
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 0.000299578947368421,
+      "loss": 0.9003,
+      "step": 1515
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 0.0002995438596491228,
+      "loss": 0.3962,
+      "step": 1516
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 0.00029950877192982457,
+      "loss": 0.6765,
+      "step": 1517
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 0.0002994736842105263,
+      "loss": 0.242,
+      "step": 1518
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 0.000299438596491228,
+      "loss": 0.456,
+      "step": 1519
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 0.0002994035087719298,
+      "loss": 0.4331,
+      "step": 1520
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 0.00029936842105263156,
+      "loss": 0.1624,
+      "step": 1521
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 0.0002993333333333333,
+      "loss": 0.7405,
+      "step": 1522
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 0.00029929824561403506,
+      "loss": 0.2843,
+      "step": 1523
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 0.0002992631578947368,
+      "loss": 0.7121,
+      "step": 1524
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 0.00029922807017543856,
+      "loss": 1.2846,
+      "step": 1525
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 0.0002991929824561403,
+      "loss": 0.6734,
+      "step": 1526
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 0.00029915789473684205,
+      "loss": 1.0198,
+      "step": 1527
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 0.00029912280701754386,
+      "loss": 0.5129,
+      "step": 1528
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 0.0002990877192982456,
+      "loss": 0.4385,
+      "step": 1529
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 0.00029905263157894735,
+      "loss": 0.8057,
+      "step": 1530
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 0.0002990175438596491,
+      "loss": 1.1243,
+      "step": 1531
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 0.00029898245614035085,
+      "loss": 1.09,
+      "step": 1532
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 0.0002989473684210526,
+      "loss": 0.4577,
+      "step": 1533
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 0.00029891228070175435,
+      "loss": 0.5147,
+      "step": 1534
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 0.00029887719298245615,
+      "loss": 0.5657,
+      "step": 1535
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 0.00029884210526315785,
+      "loss": 0.7621,
+      "step": 1536
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 0.0002988070175438596,
+      "loss": 0.4339,
+      "step": 1537
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 0.00029877192982456134,
+      "loss": 0.6648,
+      "step": 1538
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 0.00029873684210526315,
+      "loss": 0.4844,
+      "step": 1539
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 0.0002987017543859649,
+      "loss": 0.4958,
+      "step": 1540
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 0.00029866666666666664,
+      "loss": 0.6266,
+      "step": 1541
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 0.0002986315789473684,
+      "loss": 1.2849,
+      "step": 1542
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 0.00029859649122807014,
+      "loss": 0.8311,
+      "step": 1543
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 0.0002985614035087719,
+      "loss": 0.7814,
+      "step": 1544
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 0.00029852631578947364,
+      "loss": 0.7051,
+      "step": 1545
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 0.00029849122807017544,
+      "loss": 0.752,
+      "step": 1546
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 0.0002984561403508772,
+      "loss": 0.4662,
+      "step": 1547
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 0.00029842105263157894,
+      "loss": 0.4054,
+      "step": 1548
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 0.0002983859649122807,
+      "loss": 0.9495,
+      "step": 1549
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 0.00029835087719298244,
+      "loss": 0.4645,
+      "step": 1550
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 0.0002983157894736842,
+      "loss": 0.9708,
+      "step": 1551
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 0.00029828070175438593,
+      "loss": 0.5805,
+      "step": 1552
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 0.0002982456140350877,
+      "loss": 0.6846,
+      "step": 1553
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 0.00029821052631578943,
+      "loss": 0.7177,
+      "step": 1554
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 0.0002981754385964912,
+      "loss": 0.764,
+      "step": 1555
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 0.00029814035087719293,
+      "loss": 0.512,
+      "step": 1556
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 0.00029810526315789473,
+      "loss": 0.1168,
+      "step": 1557
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 0.0002980701754385965,
+      "loss": 0.3928,
+      "step": 1558
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 0.00029803508771929823,
+      "loss": 0.624,
+      "step": 1559
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 0.000298,
+      "loss": 0.7086,
+      "step": 1560
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 0.00029796491228070173,
+      "loss": 0.6273,
+      "step": 1561
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 0.0002979298245614035,
+      "loss": 0.6187,
+      "step": 1562
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 0.0002978947368421052,
+      "loss": 0.6805,
+      "step": 1563
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 0.000297859649122807,
+      "loss": 0.5361,
+      "step": 1564
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 0.0002978245614035088,
+      "loss": 0.6424,
+      "step": 1565
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 0.0002977894736842105,
+      "loss": 0.1724,
+      "step": 1566
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 0.0002977543859649122,
+      "loss": 0.936,
+      "step": 1567
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 0.000297719298245614,
+      "loss": 0.5024,
+      "step": 1568
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 0.00029768421052631577,
+      "loss": 0.4122,
+      "step": 1569
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 0.0002976491228070175,
+      "loss": 0.7544,
+      "step": 1570
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 0.00029761403508771927,
+      "loss": 0.3982,
+      "step": 1571
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 0.000297578947368421,
+      "loss": 0.2862,
+      "step": 1572
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 0.00029754385964912277,
+      "loss": 0.4515,
+      "step": 1573
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 0.0002975087719298245,
+      "loss": 0.1853,
+      "step": 1574
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 0.00029747368421052627,
+      "loss": 0.7664,
+      "step": 1575
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 0.00029743859649122807,
+      "loss": 0.6729,
+      "step": 1576
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 0.0002974035087719298,
+      "loss": 0.3235,
+      "step": 1577
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 0.00029736842105263157,
+      "loss": 0.5128,
+      "step": 1578
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 0.0002973333333333333,
+      "loss": 0.8515,
+      "step": 1579
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 0.00029729824561403506,
+      "loss": 0.8926,
+      "step": 1580
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 0.0002972631578947368,
+      "loss": 1.0383,
+      "step": 1581
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 0.00029722807017543856,
+      "loss": 1.0073,
+      "step": 1582
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 0.00029719298245614036,
+      "loss": 1.0494,
+      "step": 1583
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 0.00029715789473684206,
+      "loss": 0.7003,
+      "step": 1584
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 0.0002971228070175438,
+      "loss": 0.5168,
+      "step": 1585
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 0.00029708771929824556,
+      "loss": 0.6879,
+      "step": 1586
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 0.00029705263157894736,
+      "loss": 0.2879,
+      "step": 1587
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 0.0002970175438596491,
+      "loss": 0.8452,
+      "step": 1588
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 0.00029698245614035086,
+      "loss": 0.2214,
+      "step": 1589
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 0.0002969473684210526,
+      "loss": 0.5489,
+      "step": 1590
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 0.00029691228070175435,
+      "loss": 0.4714,
+      "step": 1591
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 0.0002968771929824561,
+      "loss": 1.0541,
+      "step": 1592
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 0.00029684210526315785,
+      "loss": 0.8526,
+      "step": 1593
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 0.00029680701754385965,
+      "loss": 0.5043,
+      "step": 1594
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 0.0002967719298245614,
+      "loss": 0.4541,
+      "step": 1595
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 0.00029673684210526315,
+      "loss": 0.4321,
+      "step": 1596
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 0.00029670175438596485,
+      "loss": 0.7054,
+      "step": 1597
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 0.00029666666666666665,
+      "loss": 0.8708,
+      "step": 1598
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 0.0002966315789473684,
+      "loss": 0.343,
+      "step": 1599
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 0.00029659649122807015,
+      "loss": 0.4606,
+      "step": 1600
+    },
+    {
+      "epoch": 23.88,
+      "eval_accuracy": 0.7792462065589819,
+      "eval_f1": 0.7878802780794968,
+      "eval_loss": 0.8314271569252014,
+      "eval_runtime": 344.0688,
+      "eval_samples_per_second": 11.876,
+      "eval_steps_per_second": 0.744,
+      "step": 1600
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 0.0002965614035087719,
+      "loss": 0.6198,
+      "step": 1601
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 0.00029652631578947364,
+      "loss": 0.6289,
+      "step": 1602
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 0.0002964912280701754,
+      "loss": 0.4563,
+      "step": 1603
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 0.00029645614035087714,
+      "loss": 0.9609,
+      "step": 1604
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 0.00029642105263157895,
+      "loss": 0.4432,
+      "step": 1605
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 0.0002963859649122807,
+      "loss": 0.3827,
+      "step": 1606
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 0.00029635087719298244,
+      "loss": 0.9992,
+      "step": 1607
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 0.0002963157894736842,
+      "loss": 0.4266,
+      "step": 1608
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 0.00029628070175438594,
+      "loss": 0.5823,
+      "step": 1609
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 0.0002962456140350877,
+      "loss": 0.5493,
+      "step": 1610
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 0.00029621052631578944,
+      "loss": 0.3707,
+      "step": 1611
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 0.0002961754385964912,
+      "loss": 0.4518,
+      "step": 1612
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 0.000296140350877193,
+      "loss": 0.7895,
+      "step": 1613
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 0.00029610526315789474,
+      "loss": 0.5102,
+      "step": 1614
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 0.00029607017543859643,
+      "loss": 0.6533,
+      "step": 1615
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 0.00029603508771929824,
+      "loss": 0.6582,
+      "step": 1616
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 0.000296,
+      "loss": 0.6249,
+      "step": 1617
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 0.00029596491228070173,
+      "loss": 0.4666,
+      "step": 1618
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 0.0002959298245614035,
+      "loss": 0.0862,
+      "step": 1619
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 0.00029589473684210523,
+      "loss": 0.2562,
+      "step": 1620
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 0.000295859649122807,
+      "loss": 0.5261,
+      "step": 1621
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 0.00029582456140350873,
+      "loss": 0.5443,
+      "step": 1622
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 0.0002957894736842105,
+      "loss": 0.2514,
+      "step": 1623
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 0.0002957543859649123,
+      "loss": 0.501,
+      "step": 1624
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 0.00029571929824561403,
+      "loss": 0.6602,
+      "step": 1625
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 0.0002956842105263158,
+      "loss": 1.0655,
+      "step": 1626
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 0.0002956491228070175,
+      "loss": 1.0767,
+      "step": 1627
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 0.0002956140350877193,
+      "loss": 1.6448,
+      "step": 1628
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 0.000295578947368421,
+      "loss": 0.8344,
+      "step": 1629
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 0.0002955438596491228,
+      "loss": 0.4289,
+      "step": 1630
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 0.0002955087719298246,
+      "loss": 0.4206,
+      "step": 1631
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 0.00029547368421052627,
+      "loss": 0.4214,
+      "step": 1632
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 0.000295438596491228,
+      "loss": 0.8347,
+      "step": 1633
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 0.00029540350877192977,
+      "loss": 0.4034,
+      "step": 1634
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 0.00029536842105263157,
+      "loss": 0.4719,
+      "step": 1635
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 0.0002953333333333333,
+      "loss": 0.3112,
+      "step": 1636
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 0.00029529824561403507,
+      "loss": 0.7053,
+      "step": 1637
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 0.0002952631578947368,
+      "loss": 0.2759,
+      "step": 1638
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 0.00029522807017543857,
+      "loss": 0.3891,
+      "step": 1639
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 0.0002951929824561403,
+      "loss": 0.2628,
+      "step": 1640
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 0.00029515789473684206,
+      "loss": 0.4246,
+      "step": 1641
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 0.00029512280701754387,
+      "loss": 0.8412,
+      "step": 1642
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 0.0002950877192982456,
+      "loss": 0.5919,
+      "step": 1643
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 0.00029505263157894736,
+      "loss": 0.5095,
+      "step": 1644
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 0.00029501754385964906,
+      "loss": 0.7387,
+      "step": 1645
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 0.00029498245614035086,
+      "loss": 0.3755,
+      "step": 1646
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 0.0002949473684210526,
+      "loss": 0.3604,
+      "step": 1647
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 0.00029491228070175436,
+      "loss": 0.3853,
+      "step": 1648
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 0.0002948771929824561,
+      "loss": 0.4088,
+      "step": 1649
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 0.00029484210526315786,
+      "loss": 0.652,
+      "step": 1650
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 0.0002948070175438596,
+      "loss": 0.8167,
+      "step": 1651
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 0.00029477192982456135,
+      "loss": 0.3586,
+      "step": 1652
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 0.0002947368421052631,
+      "loss": 0.2884,
+      "step": 1653
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 0.0002947017543859649,
+      "loss": 0.2747,
+      "step": 1654
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 0.00029466666666666666,
+      "loss": 0.3327,
+      "step": 1655
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 0.0002946315789473684,
+      "loss": 0.2504,
+      "step": 1656
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 0.00029459649122807015,
+      "loss": 0.3425,
+      "step": 1657
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 0.0002945614035087719,
+      "loss": 0.429,
+      "step": 1658
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 0.00029452631578947365,
+      "loss": 0.3557,
+      "step": 1659
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 0.0002944912280701754,
+      "loss": 0.6772,
+      "step": 1660
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 0.0002944561403508772,
+      "loss": 0.321,
+      "step": 1661
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 0.00029442105263157895,
+      "loss": 0.7565,
+      "step": 1662
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 0.00029438596491228065,
+      "loss": 0.2763,
+      "step": 1663
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 0.0002943508771929824,
+      "loss": 0.8887,
+      "step": 1664
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 0.0002943157894736842,
+      "loss": 0.5218,
+      "step": 1665
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 0.00029428070175438595,
+      "loss": 0.2061,
+      "step": 1666
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 0.0002942456140350877,
+      "loss": 0.2076,
+      "step": 1667
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 0.00029421052631578944,
+      "loss": 0.3881,
+      "step": 1668
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 0.0002941754385964912,
+      "loss": 0.5423,
+      "step": 1669
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 0.00029414035087719294,
+      "loss": 0.2679,
+      "step": 1670
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 0.0002941052631578947,
+      "loss": 0.3646,
+      "step": 1671
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 0.0002940701754385965,
+      "loss": 0.5029,
+      "step": 1672
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 0.00029403508771929824,
+      "loss": 0.3023,
+      "step": 1673
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 0.000294,
+      "loss": 0.5306,
+      "step": 1674
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 0.00029396491228070174,
+      "loss": 0.4512,
+      "step": 1675
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 0.0002939298245614035,
+      "loss": 0.7319,
+      "step": 1676
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 0.00029389473684210524,
+      "loss": 0.604,
+      "step": 1677
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 0.000293859649122807,
+      "loss": 0.3086,
+      "step": 1678
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 0.0002938245614035088,
+      "loss": 0.2008,
+      "step": 1679
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 0.0002937894736842105,
+      "loss": 0.3837,
+      "step": 1680
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 0.00029375438596491223,
+      "loss": 0.5445,
+      "step": 1681
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 0.000293719298245614,
+      "loss": 0.4024,
+      "step": 1682
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 0.0002936842105263158,
+      "loss": 0.8895,
+      "step": 1683
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 0.00029364912280701753,
+      "loss": 0.4681,
+      "step": 1684
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 0.0002936140350877193,
+      "loss": 0.2096,
+      "step": 1685
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 0.00029357894736842103,
+      "loss": 0.3158,
+      "step": 1686
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 0.0002935438596491228,
+      "loss": 0.3478,
+      "step": 1687
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 0.00029350877192982453,
+      "loss": 0.1883,
+      "step": 1688
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 0.0002934736842105263,
+      "loss": 0.3698,
+      "step": 1689
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 0.000293438596491228,
+      "loss": 0.4285,
+      "step": 1690
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 0.00029340350877192983,
+      "loss": 0.373,
+      "step": 1691
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 0.0002933684210526316,
+      "loss": 0.1485,
+      "step": 1692
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 0.00029333333333333327,
+      "loss": 0.447,
+      "step": 1693
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 0.0002932982456140351,
+      "loss": 0.6378,
+      "step": 1694
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 0.0002932631578947368,
+      "loss": 0.5661,
+      "step": 1695
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 0.00029322807017543857,
+      "loss": 0.5903,
+      "step": 1696
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 0.0002931929824561403,
+      "loss": 0.6971,
+      "step": 1697
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 0.00029315789473684207,
+      "loss": 0.3974,
+      "step": 1698
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 0.0002931228070175438,
+      "loss": 0.2634,
+      "step": 1699
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 0.00029308771929824557,
+      "loss": 0.7687,
+      "step": 1700
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 0.0002930526315789473,
+      "loss": 0.6014,
+      "step": 1701
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 0.0002930175438596491,
+      "loss": 0.5096,
+      "step": 1702
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 0.00029298245614035087,
+      "loss": 0.3946,
+      "step": 1703
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 0.0002929473684210526,
+      "loss": 0.3069,
+      "step": 1704
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 0.00029291228070175437,
+      "loss": 0.2154,
+      "step": 1705
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 0.0002928771929824561,
+      "loss": 0.2049,
+      "step": 1706
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 0.00029284210526315786,
+      "loss": 0.4638,
+      "step": 1707
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 0.0002928070175438596,
+      "loss": 0.4175,
+      "step": 1708
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 0.0002927719298245614,
+      "loss": 1.1591,
+      "step": 1709
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 0.00029273684210526316,
+      "loss": 0.2331,
+      "step": 1710
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 0.00029270175438596486,
+      "loss": 0.7711,
+      "step": 1711
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 0.0002926666666666666,
+      "loss": 0.3989,
+      "step": 1712
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 0.0002926315789473684,
+      "loss": 0.4343,
+      "step": 1713
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 0.00029259649122807016,
+      "loss": 0.5231,
+      "step": 1714
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 0.0002925614035087719,
+      "loss": 0.3829,
+      "step": 1715
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 0.00029252631578947366,
+      "loss": 0.5188,
+      "step": 1716
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 0.0002924912280701754,
+      "loss": 0.495,
+      "step": 1717
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 0.00029245614035087715,
+      "loss": 0.1815,
+      "step": 1718
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 0.0002924210526315789,
+      "loss": 0.2124,
+      "step": 1719
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 0.0002923859649122807,
+      "loss": 0.372,
+      "step": 1720
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 0.00029235087719298245,
+      "loss": 0.1183,
+      "step": 1721
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 0.0002923157894736842,
+      "loss": 0.7579,
+      "step": 1722
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 0.00029228070175438595,
+      "loss": 0.217,
+      "step": 1723
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 0.0002922456140350877,
+      "loss": 0.0824,
+      "step": 1724
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 0.00029221052631578945,
+      "loss": 0.1303,
+      "step": 1725
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 0.0002921754385964912,
+      "loss": 0.8698,
+      "step": 1726
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 0.00029214035087719295,
+      "loss": 0.7743,
+      "step": 1727
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 0.0002921052631578947,
+      "loss": 0.3436,
+      "step": 1728
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 0.00029207017543859644,
+      "loss": 0.9807,
+      "step": 1729
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 0.0002920350877192982,
+      "loss": 0.3922,
+      "step": 1730
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 0.000292,
+      "loss": 0.9146,
+      "step": 1731
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 0.00029196491228070174,
+      "loss": 0.2695,
+      "step": 1732
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 0.0002919298245614035,
+      "loss": 0.5178,
+      "step": 1733
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 0.00029189473684210524,
+      "loss": 0.4618,
+      "step": 1734
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 0.000291859649122807,
+      "loss": 0.3861,
+      "step": 1735
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 0.00029182456140350874,
+      "loss": 0.2787,
+      "step": 1736
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 0.0002917894736842105,
+      "loss": 0.4399,
+      "step": 1737
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 0.00029175438596491224,
+      "loss": 0.9383,
+      "step": 1738
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 0.00029171929824561404,
+      "loss": 0.3467,
+      "step": 1739
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 0.0002916842105263158,
+      "loss": 0.6328,
+      "step": 1740
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 0.0002916491228070175,
+      "loss": 0.5946,
+      "step": 1741
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 0.0002916140350877193,
+      "loss": 0.2688,
+      "step": 1742
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 0.00029157894736842104,
+      "loss": 0.3514,
+      "step": 1743
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 0.0002915438596491228,
+      "loss": 0.8932,
+      "step": 1744
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 0.00029150877192982453,
+      "loss": 0.4737,
+      "step": 1745
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 0.0002914736842105263,
+      "loss": 0.1499,
+      "step": 1746
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 0.00029143859649122803,
+      "loss": 0.5537,
+      "step": 1747
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 0.0002914035087719298,
+      "loss": 0.2274,
+      "step": 1748
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 0.00029136842105263153,
+      "loss": 0.2173,
+      "step": 1749
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 0.00029133333333333333,
+      "loss": 0.6862,
+      "step": 1750
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 0.0002912982456140351,
+      "loss": 0.4882,
+      "step": 1751
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 0.00029126315789473683,
+      "loss": 0.3082,
+      "step": 1752
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 0.0002912280701754386,
+      "loss": 0.4787,
+      "step": 1753
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 0.0002911929824561403,
+      "loss": 0.5084,
+      "step": 1754
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 0.0002911578947368421,
+      "loss": 0.5261,
+      "step": 1755
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 0.0002911228070175438,
+      "loss": 0.8972,
+      "step": 1756
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 0.0002910877192982456,
+      "loss": 0.0873,
+      "step": 1757
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 0.0002910526315789473,
+      "loss": 0.0688,
+      "step": 1758
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 0.00029101754385964907,
+      "loss": 0.5787,
+      "step": 1759
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 0.0002909824561403508,
+      "loss": 0.5109,
+      "step": 1760
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 0.0002909473684210526,
+      "loss": 0.9027,
+      "step": 1761
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 0.00029091228070175437,
+      "loss": 0.6658,
+      "step": 1762
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 0.0002908771929824561,
+      "loss": 0.3153,
+      "step": 1763
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 0.00029084210526315787,
+      "loss": 0.4754,
+      "step": 1764
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 0.0002908070175438596,
+      "loss": 0.8009,
+      "step": 1765
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 0.00029077192982456137,
+      "loss": 0.7278,
+      "step": 1766
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 0.0002907368421052631,
+      "loss": 0.9031,
+      "step": 1767
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 0.0002907017543859649,
+      "loss": 0.4986,
+      "step": 1768
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 0.00029066666666666667,
+      "loss": 0.6661,
+      "step": 1769
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 0.0002906315789473684,
+      "loss": 0.555,
+      "step": 1770
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 0.00029059649122807016,
+      "loss": 0.5531,
+      "step": 1771
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 0.0002905614035087719,
+      "loss": 0.3935,
+      "step": 1772
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 0.00029052631578947366,
+      "loss": 0.8509,
+      "step": 1773
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 0.0002904912280701754,
+      "loss": 0.3773,
+      "step": 1774
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 0.00029045614035087716,
+      "loss": 0.4842,
+      "step": 1775
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 0.0002904210526315789,
+      "loss": 0.5316,
+      "step": 1776
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 0.00029038596491228066,
+      "loss": 0.3468,
+      "step": 1777
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 0.0002903508771929824,
+      "loss": 0.6874,
+      "step": 1778
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 0.0002903157894736842,
+      "loss": 0.4743,
+      "step": 1779
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 0.00029028070175438596,
+      "loss": 0.551,
+      "step": 1780
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 0.0002902456140350877,
+      "loss": 0.1915,
+      "step": 1781
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 0.00029021052631578945,
+      "loss": 0.8392,
+      "step": 1782
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 0.0002901754385964912,
+      "loss": 0.4326,
+      "step": 1783
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 0.00029014035087719295,
+      "loss": 0.3187,
+      "step": 1784
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 0.0002901052631578947,
+      "loss": 0.3209,
+      "step": 1785
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 0.00029007017543859645,
+      "loss": 0.5229,
+      "step": 1786
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 0.00029003508771929825,
+      "loss": 0.5639,
+      "step": 1787
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 0.00029,
+      "loss": 0.2592,
+      "step": 1788
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 0.0002899649122807017,
+      "loss": 0.5842,
+      "step": 1789
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 0.00028992982456140345,
+      "loss": 0.3494,
+      "step": 1790
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 0.00028989473684210525,
+      "loss": 0.3164,
+      "step": 1791
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 0.000289859649122807,
+      "loss": 0.4838,
+      "step": 1792
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 0.00028982456140350875,
+      "loss": 0.2019,
+      "step": 1793
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 0.0002897894736842105,
+      "loss": 0.3821,
+      "step": 1794
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 0.00028975438596491224,
+      "loss": 0.5468,
+      "step": 1795
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 0.000289719298245614,
+      "loss": 0.1877,
+      "step": 1796
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 0.00028968421052631574,
+      "loss": 0.5564,
+      "step": 1797
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 0.00028964912280701754,
+      "loss": 0.9035,
+      "step": 1798
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 0.0002896140350877193,
+      "loss": 0.6982,
+      "step": 1799
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 0.00028957894736842104,
+      "loss": 0.6975,
+      "step": 1800
+    },
+    {
+      "epoch": 26.86,
+      "eval_accuracy": 0.793930494371023,
+      "eval_f1": 0.792651690068327,
+      "eval_loss": 0.7666684985160828,
+      "eval_runtime": 344.0857,
+      "eval_samples_per_second": 11.875,
+      "eval_steps_per_second": 0.744,
+      "step": 1800
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 0.0002895438596491228,
+      "loss": 0.2577,
+      "step": 1801
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 0.00028950877192982454,
+      "loss": 0.2534,
+      "step": 1802
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 0.0002894736842105263,
+      "loss": 0.2775,
+      "step": 1803
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 0.00028943859649122804,
+      "loss": 0.2636,
+      "step": 1804
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 0.00028940350877192984,
+      "loss": 0.6394,
+      "step": 1805
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 0.00028936842105263153,
+      "loss": 0.6804,
+      "step": 1806
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 0.0002893333333333333,
+      "loss": 0.4613,
+      "step": 1807
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 0.00028929824561403503,
+      "loss": 0.2651,
+      "step": 1808
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 0.00028926315789473683,
+      "loss": 0.3102,
+      "step": 1809
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 0.0002892280701754386,
+      "loss": 0.6897,
+      "step": 1810
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 0.00028919298245614033,
+      "loss": 0.3374,
+      "step": 1811
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 0.0002891578947368421,
+      "loss": 1.2131,
+      "step": 1812
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 0.00028912280701754383,
+      "loss": 0.4629,
+      "step": 1813
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 0.0002890877192982456,
+      "loss": 0.4115,
+      "step": 1814
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 0.00028905263157894733,
+      "loss": 0.4939,
+      "step": 1815
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 0.0002890175438596491,
+      "loss": 0.3057,
+      "step": 1816
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 0.0002889824561403509,
+      "loss": 0.7153,
+      "step": 1817
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 0.00028894736842105263,
+      "loss": 0.3378,
+      "step": 1818
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 0.0002889122807017544,
+      "loss": 0.3109,
+      "step": 1819
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 0.0002888771929824561,
+      "loss": 0.4494,
+      "step": 1820
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 0.0002888421052631579,
+      "loss": 0.2402,
+      "step": 1821
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 0.0002888070175438596,
+      "loss": 0.1107,
+      "step": 1822
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 0.00028877192982456137,
+      "loss": 0.7066,
+      "step": 1823
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 0.0002887368421052631,
+      "loss": 0.1625,
+      "step": 1824
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 0.00028870175438596487,
+      "loss": 0.2452,
+      "step": 1825
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 0.0002886666666666666,
+      "loss": 1.0513,
+      "step": 1826
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 0.00028863157894736837,
+      "loss": 0.5964,
+      "step": 1827
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 0.00028859649122807017,
+      "loss": 0.7036,
+      "step": 1828
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 0.0002885614035087719,
+      "loss": 0.4673,
+      "step": 1829
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 0.00028852631578947367,
+      "loss": 0.255,
+      "step": 1830
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 0.0002884912280701754,
+      "loss": 0.4344,
+      "step": 1831
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 0.00028845614035087716,
+      "loss": 0.2454,
+      "step": 1832
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 0.0002884210526315789,
+      "loss": 0.6276,
+      "step": 1833
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 0.00028838596491228066,
+      "loss": 0.3496,
+      "step": 1834
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 0.00028835087719298247,
+      "loss": 0.249,
+      "step": 1835
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 0.0002883157894736842,
+      "loss": 0.1035,
+      "step": 1836
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 0.0002882807017543859,
+      "loss": 0.2411,
+      "step": 1837
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 0.00028824561403508766,
+      "loss": 0.3164,
+      "step": 1838
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 0.00028821052631578946,
+      "loss": 0.0571,
+      "step": 1839
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 0.0002881754385964912,
+      "loss": 0.0659,
+      "step": 1840
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 0.00028814035087719296,
+      "loss": 0.363,
+      "step": 1841
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 0.0002881052631578947,
+      "loss": 0.6705,
+      "step": 1842
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 0.00028807017543859646,
+      "loss": 0.4213,
+      "step": 1843
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 0.0002880350877192982,
+      "loss": 0.4256,
+      "step": 1844
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 0.3606,
+      "step": 1845
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 0.00028796491228070176,
+      "loss": 0.7367,
+      "step": 1846
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 0.0002879298245614035,
+      "loss": 0.391,
+      "step": 1847
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 0.00028789473684210525,
+      "loss": 0.7125,
+      "step": 1848
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 0.000287859649122807,
+      "loss": 0.0586,
+      "step": 1849
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 0.00028782456140350875,
+      "loss": 0.3911,
+      "step": 1850
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 0.0002877894736842105,
+      "loss": 0.222,
+      "step": 1851
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 0.00028775438596491225,
+      "loss": 0.2349,
+      "step": 1852
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 0.000287719298245614,
+      "loss": 0.1077,
+      "step": 1853
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 0.00028768421052631575,
+      "loss": 0.1471,
+      "step": 1854
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 0.0002876491228070175,
+      "loss": 0.4215,
+      "step": 1855
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 0.00028761403508771924,
+      "loss": 0.3904,
+      "step": 1856
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 0.00028757894736842105,
+      "loss": 0.4847,
+      "step": 1857
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 0.0002875438596491228,
+      "loss": 0.4888,
+      "step": 1858
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 0.00028750877192982454,
+      "loss": 0.7409,
+      "step": 1859
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 0.0002874736842105263,
+      "loss": 0.3056,
+      "step": 1860
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 0.00028743859649122804,
+      "loss": 0.4178,
+      "step": 1861
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 0.0002874035087719298,
+      "loss": 0.2794,
+      "step": 1862
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 0.00028736842105263154,
+      "loss": 0.2874,
+      "step": 1863
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 0.0002873333333333333,
+      "loss": 0.6377,
+      "step": 1864
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 0.0002872982456140351,
+      "loss": 1.002,
+      "step": 1865
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 0.00028726315789473684,
+      "loss": 0.5173,
+      "step": 1866
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 0.0002872280701754386,
+      "loss": 0.4654,
+      "step": 1867
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 0.00028719298245614034,
+      "loss": 0.3534,
+      "step": 1868
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 0.0002871578947368421,
+      "loss": 0.6233,
+      "step": 1869
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 0.00028712280701754384,
+      "loss": 0.3434,
+      "step": 1870
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 0.0002870877192982456,
+      "loss": 0.2129,
+      "step": 1871
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 0.00028705263157894733,
+      "loss": 0.1183,
+      "step": 1872
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 0.0002870175438596491,
+      "loss": 0.233,
+      "step": 1873
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 0.00028698245614035083,
+      "loss": 0.5479,
+      "step": 1874
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 0.0002869473684210526,
+      "loss": 0.3382,
+      "step": 1875
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 0.0002869122807017544,
+      "loss": 0.9043,
+      "step": 1876
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 0.00028687719298245613,
+      "loss": 0.6242,
+      "step": 1877
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 0.0002868421052631579,
+      "loss": 0.6041,
+      "step": 1878
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 0.00028680701754385963,
+      "loss": 0.6087,
+      "step": 1879
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 0.0002867719298245614,
+      "loss": 0.2625,
+      "step": 1880
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 0.0002867368421052631,
+      "loss": 0.3646,
+      "step": 1881
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 0.0002867017543859649,
+      "loss": 0.4302,
+      "step": 1882
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 0.0002866666666666667,
+      "loss": 0.9889,
+      "step": 1883
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 0.0002866315789473684,
+      "loss": 0.4311,
+      "step": 1884
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 0.0002865964912280701,
+      "loss": 0.3208,
+      "step": 1885
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 0.00028656140350877187,
+      "loss": 0.7227,
+      "step": 1886
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 0.00028652631578947367,
+      "loss": 0.2999,
+      "step": 1887
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 0.0002864912280701754,
+      "loss": 0.1791,
+      "step": 1888
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 0.00028645614035087717,
+      "loss": 0.377,
+      "step": 1889
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 0.0002864210526315789,
+      "loss": 0.1628,
+      "step": 1890
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 0.00028638596491228067,
+      "loss": 0.3154,
+      "step": 1891
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 0.0002863508771929824,
+      "loss": 0.1039,
+      "step": 1892
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 0.00028631578947368417,
+      "loss": 0.2205,
+      "step": 1893
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 0.00028628070175438597,
+      "loss": 0.8305,
+      "step": 1894
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 0.0002862456140350877,
+      "loss": 0.2431,
+      "step": 1895
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 0.00028621052631578947,
+      "loss": 0.2169,
+      "step": 1896
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 0.0002861754385964912,
+      "loss": 0.2878,
+      "step": 1897
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 0.00028614035087719296,
+      "loss": 0.3567,
+      "step": 1898
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 0.0002861052631578947,
+      "loss": 0.7409,
+      "step": 1899
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 0.00028607017543859646,
+      "loss": 0.3751,
+      "step": 1900
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 0.0002860350877192982,
+      "loss": 0.154,
+      "step": 1901
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 0.00028599999999999996,
+      "loss": 0.1643,
+      "step": 1902
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 0.0002859649122807017,
+      "loss": 0.2874,
+      "step": 1903
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 0.00028592982456140346,
+      "loss": 0.6538,
+      "step": 1904
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 0.00028589473684210526,
+      "loss": 0.3018,
+      "step": 1905
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 0.000285859649122807,
+      "loss": 0.7342,
+      "step": 1906
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 0.00028582456140350876,
+      "loss": 0.3263,
+      "step": 1907
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 0.0002857894736842105,
+      "loss": 0.5486,
+      "step": 1908
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 0.00028575438596491225,
+      "loss": 0.6566,
+      "step": 1909
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 0.000285719298245614,
+      "loss": 0.645,
+      "step": 1910
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 0.00028568421052631575,
+      "loss": 0.657,
+      "step": 1911
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 0.0002856491228070175,
+      "loss": 0.5839,
+      "step": 1912
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 0.0002856140350877193,
+      "loss": 0.4959,
+      "step": 1913
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 0.00028557894736842105,
+      "loss": 0.6578,
+      "step": 1914
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 0.0002855438596491228,
+      "loss": 1.0061,
+      "step": 1915
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 0.0002855087719298245,
+      "loss": 0.6346,
+      "step": 1916
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 0.0002854736842105263,
+      "loss": 1.1097,
+      "step": 1917
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 0.00028543859649122805,
+      "loss": 0.45,
+      "step": 1918
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 0.0002854035087719298,
+      "loss": 0.2154,
+      "step": 1919
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 0.00028536842105263155,
+      "loss": 0.4049,
+      "step": 1920
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 0.0002853333333333333,
+      "loss": 0.742,
+      "step": 1921
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 0.00028529824561403504,
+      "loss": 0.336,
+      "step": 1922
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 0.0002852631578947368,
+      "loss": 0.4915,
+      "step": 1923
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 0.0002852280701754386,
+      "loss": 0.3299,
+      "step": 1924
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 0.00028519298245614034,
+      "loss": 0.1165,
+      "step": 1925
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 0.0002851578947368421,
+      "loss": 0.4511,
+      "step": 1926
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 0.00028512280701754384,
+      "loss": 0.36,
+      "step": 1927
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 0.0002850877192982456,
+      "loss": 0.9207,
+      "step": 1928
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 0.00028505263157894734,
+      "loss": 0.573,
+      "step": 1929
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 0.0002850175438596491,
+      "loss": 0.282,
+      "step": 1930
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 0.0002849824561403509,
+      "loss": 0.8202,
+      "step": 1931
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 0.00028494736842105264,
+      "loss": 0.6128,
+      "step": 1932
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 0.00028491228070175433,
+      "loss": 0.3645,
+      "step": 1933
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 0.0002848771929824561,
+      "loss": 0.1557,
+      "step": 1934
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 0.0002848421052631579,
+      "loss": 0.2146,
+      "step": 1935
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 0.00028480701754385963,
+      "loss": 0.7744,
+      "step": 1936
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 0.0002847719298245614,
+      "loss": 0.0497,
+      "step": 1937
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 0.00028473684210526313,
+      "loss": 0.2807,
+      "step": 1938
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 0.0002847017543859649,
+      "loss": 0.3621,
+      "step": 1939
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 0.00028466666666666663,
+      "loss": 0.4343,
+      "step": 1940
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 0.0002846315789473684,
+      "loss": 0.4488,
+      "step": 1941
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 0.0002845964912280702,
+      "loss": 0.6635,
+      "step": 1942
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 0.00028456140350877193,
+      "loss": 0.49,
+      "step": 1943
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 0.0002845263157894737,
+      "loss": 0.6394,
+      "step": 1944
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 0.00028449122807017543,
+      "loss": 0.289,
+      "step": 1945
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 0.0002844561403508772,
+      "loss": 0.4145,
+      "step": 1946
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 0.0002844210526315789,
+      "loss": 0.4305,
+      "step": 1947
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 0.0002843859649122807,
+      "loss": 0.6245,
+      "step": 1948
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 0.0002843508771929824,
+      "loss": 0.5812,
+      "step": 1949
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 0.00028431578947368417,
+      "loss": 0.3968,
+      "step": 1950
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 0.0002842807017543859,
+      "loss": 0.4733,
+      "step": 1951
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 0.00028424561403508767,
+      "loss": 0.32,
+      "step": 1952
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 0.0002842105263157894,
+      "loss": 0.1961,
+      "step": 1953
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 0.0002841754385964912,
+      "loss": 0.5253,
+      "step": 1954
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 0.00028414035087719297,
+      "loss": 0.1697,
+      "step": 1955
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 0.0002841052631578947,
+      "loss": 0.3636,
+      "step": 1956
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 0.00028407017543859647,
+      "loss": 0.4167,
+      "step": 1957
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 0.0002840350877192982,
+      "loss": 0.4889,
+      "step": 1958
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 0.00028399999999999996,
+      "loss": 0.4878,
+      "step": 1959
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 0.0002839649122807017,
+      "loss": 0.6979,
+      "step": 1960
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 0.0002839298245614035,
+      "loss": 0.9622,
+      "step": 1961
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 0.00028389473684210526,
+      "loss": 0.5342,
+      "step": 1962
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 0.000283859649122807,
+      "loss": 0.3308,
+      "step": 1963
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 0.0002838245614035087,
+      "loss": 0.3056,
+      "step": 1964
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 0.0002837894736842105,
+      "loss": 0.3,
+      "step": 1965
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 0.00028375438596491226,
+      "loss": 0.7208,
+      "step": 1966
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 0.000283719298245614,
+      "loss": 1.1475,
+      "step": 1967
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 0.00028368421052631576,
+      "loss": 0.2273,
+      "step": 1968
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 0.0002836491228070175,
+      "loss": 0.3318,
+      "step": 1969
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 0.00028361403508771926,
+      "loss": 0.1126,
+      "step": 1970
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 0.000283578947368421,
+      "loss": 0.2493,
+      "step": 1971
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 0.0002835438596491228,
+      "loss": 0.1019,
+      "step": 1972
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 0.00028350877192982456,
+      "loss": 0.3286,
+      "step": 1973
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 0.0002834736842105263,
+      "loss": 0.0936,
+      "step": 1974
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 0.00028343859649122805,
+      "loss": 0.7561,
+      "step": 1975
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 0.0002834035087719298,
+      "loss": 0.2806,
+      "step": 1976
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 0.00028336842105263155,
+      "loss": 0.4837,
+      "step": 1977
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 0.0002833333333333333,
+      "loss": 0.4605,
+      "step": 1978
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 0.00028329824561403505,
+      "loss": 0.7167,
+      "step": 1979
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 0.00028326315789473685,
+      "loss": 0.3975,
+      "step": 1980
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 0.00028322807017543855,
+      "loss": 0.3134,
+      "step": 1981
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 0.0002831929824561403,
+      "loss": 0.4018,
+      "step": 1982
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 0.0002831578947368421,
+      "loss": 0.1087,
+      "step": 1983
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 0.00028312280701754385,
+      "loss": 0.793,
+      "step": 1984
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 0.0002830877192982456,
+      "loss": 0.6217,
+      "step": 1985
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 0.00028305263157894734,
+      "loss": 0.3012,
+      "step": 1986
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 0.0002830175438596491,
+      "loss": 0.4654,
+      "step": 1987
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 0.00028298245614035084,
+      "loss": 0.0225,
+      "step": 1988
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 0.0002829473684210526,
+      "loss": 0.216,
+      "step": 1989
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 0.00028291228070175434,
+      "loss": 0.4494,
+      "step": 1990
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 0.00028287719298245614,
+      "loss": 0.5049,
+      "step": 1991
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 0.0002828421052631579,
+      "loss": 0.0648,
+      "step": 1992
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 0.00028280701754385964,
+      "loss": 0.4833,
+      "step": 1993
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 0.0002827719298245614,
+      "loss": 0.3664,
+      "step": 1994
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 0.00028273684210526314,
+      "loss": 0.5582,
+      "step": 1995
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 0.0002827017543859649,
+      "loss": 0.323,
+      "step": 1996
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 0.00028266666666666663,
+      "loss": 0.5679,
+      "step": 1997
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 0.0002826315789473684,
+      "loss": 0.2273,
+      "step": 1998
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 0.00028259649122807013,
+      "loss": 0.6848,
+      "step": 1999
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 0.0002825614035087719,
+      "loss": 0.9913,
+      "step": 2000
+    },
+    {
+      "epoch": 29.85,
+      "eval_accuracy": 0.7706803720019579,
+      "eval_f1": 0.7733909006780546,
+      "eval_loss": 0.9206514954566956,
+      "eval_runtime": 344.174,
+      "eval_samples_per_second": 11.872,
+      "eval_steps_per_second": 0.744,
+      "step": 2000
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 0.00028252631578947363,
+      "loss": 0.7397,
+      "step": 2001
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 0.00028249122807017543,
+      "loss": 0.4359,
+      "step": 2002
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 0.0002824561403508772,
+      "loss": 0.1009,
+      "step": 2003
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 0.00028242105263157893,
+      "loss": 0.2327,
+      "step": 2004
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 0.0002823859649122807,
+      "loss": 0.343,
+      "step": 2005
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 0.00028235087719298243,
+      "loss": 0.3399,
+      "step": 2006
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 0.0002823157894736842,
+      "loss": 0.5942,
+      "step": 2007
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 0.0002822807017543859,
+      "loss": 0.5096,
+      "step": 2008
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 0.00028224561403508773,
+      "loss": 0.2877,
+      "step": 2009
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 0.0002822105263157895,
+      "loss": 0.1841,
+      "step": 2010
+    },
+    {
+      "epoch": 30.01,
+      "learning_rate": 0.0002821754385964912,
+      "loss": 0.0952,
+      "step": 2011
+    },
+    {
+      "epoch": 30.03,
+      "learning_rate": 0.0002821403508771929,
+      "loss": 0.6921,
+      "step": 2012
+    },
+    {
+      "epoch": 30.04,
+      "learning_rate": 0.0002821052631578947,
+      "loss": 0.3671,
+      "step": 2013
+    },
+    {
+      "epoch": 30.06,
+      "learning_rate": 0.00028207017543859647,
+      "loss": 0.3699,
+      "step": 2014
+    },
+    {
+      "epoch": 30.07,
+      "learning_rate": 0.0002820350877192982,
+      "loss": 0.4393,
+      "step": 2015
+    },
+    {
+      "epoch": 30.09,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 0.0431,
+      "step": 2016
+    },
+    {
+      "epoch": 30.1,
+      "learning_rate": 0.0002819649122807017,
+      "loss": 0.4641,
+      "step": 2017
+    },
+    {
+      "epoch": 30.12,
+      "learning_rate": 0.00028192982456140347,
+      "loss": 0.4369,
+      "step": 2018
+    },
+    {
+      "epoch": 30.13,
+      "learning_rate": 0.0002818947368421052,
+      "loss": 0.1586,
+      "step": 2019
+    },
+    {
+      "epoch": 30.15,
+      "learning_rate": 0.000281859649122807,
+      "loss": 0.2343,
+      "step": 2020
+    },
+    {
+      "epoch": 30.16,
+      "learning_rate": 0.00028182456140350877,
+      "loss": 0.0312,
+      "step": 2021
+    },
+    {
+      "epoch": 30.18,
+      "learning_rate": 0.0002817894736842105,
+      "loss": 0.1787,
+      "step": 2022
+    },
+    {
+      "epoch": 30.19,
+      "learning_rate": 0.00028175438596491227,
+      "loss": 0.3184,
+      "step": 2023
+    },
+    {
+      "epoch": 30.21,
+      "learning_rate": 0.000281719298245614,
+      "loss": 0.3404,
+      "step": 2024
+    },
+    {
+      "epoch": 30.22,
+      "learning_rate": 0.00028168421052631576,
+      "loss": 0.2371,
+      "step": 2025
+    },
+    {
+      "epoch": 30.24,
+      "learning_rate": 0.0002816491228070175,
+      "loss": 0.3565,
+      "step": 2026
+    },
+    {
+      "epoch": 30.25,
+      "learning_rate": 0.00028161403508771926,
+      "loss": 0.6661,
+      "step": 2027
+    },
+    {
+      "epoch": 30.27,
+      "learning_rate": 0.00028157894736842106,
+      "loss": 0.9013,
+      "step": 2028
+    },
+    {
+      "epoch": 30.28,
+      "learning_rate": 0.00028154385964912276,
+      "loss": 0.3751,
+      "step": 2029
+    },
+    {
+      "epoch": 30.3,
+      "learning_rate": 0.0002815087719298245,
+      "loss": 0.1435,
+      "step": 2030
+    },
+    {
+      "epoch": 30.31,
+      "learning_rate": 0.0002814736842105263,
+      "loss": 0.2184,
+      "step": 2031
+    },
+    {
+      "epoch": 30.33,
+      "learning_rate": 0.00028143859649122806,
+      "loss": 0.0664,
+      "step": 2032
+    },
+    {
+      "epoch": 30.34,
+      "learning_rate": 0.0002814035087719298,
+      "loss": 0.227,
+      "step": 2033
+    },
+    {
+      "epoch": 30.36,
+      "learning_rate": 0.00028136842105263156,
+      "loss": 0.3569,
+      "step": 2034
+    },
+    {
+      "epoch": 30.37,
+      "learning_rate": 0.0002813333333333333,
+      "loss": 0.455,
+      "step": 2035
+    },
+    {
+      "epoch": 30.39,
+      "learning_rate": 0.00028129824561403505,
+      "loss": 0.3361,
+      "step": 2036
+    },
+    {
+      "epoch": 30.4,
+      "learning_rate": 0.0002812631578947368,
+      "loss": 0.2786,
+      "step": 2037
+    },
+    {
+      "epoch": 30.42,
+      "learning_rate": 0.00028122807017543855,
+      "loss": 0.0724,
+      "step": 2038
+    },
+    {
+      "epoch": 30.43,
+      "learning_rate": 0.00028119298245614035,
+      "loss": 0.7161,
+      "step": 2039
+    },
+    {
+      "epoch": 30.45,
+      "learning_rate": 0.0002811578947368421,
+      "loss": 0.0976,
+      "step": 2040
+    },
+    {
+      "epoch": 30.46,
+      "learning_rate": 0.00028112280701754385,
+      "loss": 0.4322,
+      "step": 2041
+    },
+    {
+      "epoch": 30.48,
+      "learning_rate": 0.00028108771929824555,
+      "loss": 0.4597,
+      "step": 2042
+    },
+    {
+      "epoch": 30.49,
+      "learning_rate": 0.00028105263157894735,
+      "loss": 0.6184,
+      "step": 2043
+    },
+    {
+      "epoch": 30.51,
+      "learning_rate": 0.0002810175438596491,
+      "loss": 0.7813,
+      "step": 2044
+    },
+    {
+      "epoch": 30.52,
+      "learning_rate": 0.00028098245614035085,
+      "loss": 0.3278,
+      "step": 2045
+    },
+    {
+      "epoch": 30.54,
+      "learning_rate": 0.0002809473684210526,
+      "loss": 0.1911,
+      "step": 2046
+    },
+    {
+      "epoch": 30.55,
+      "learning_rate": 0.00028091228070175434,
+      "loss": 0.3555,
+      "step": 2047
+    },
+    {
+      "epoch": 30.57,
+      "learning_rate": 0.0002808771929824561,
+      "loss": 0.6221,
+      "step": 2048
+    },
+    {
+      "epoch": 30.58,
+      "learning_rate": 0.00028084210526315784,
+      "loss": 0.376,
+      "step": 2049
+    },
+    {
+      "epoch": 30.59,
+      "learning_rate": 0.00028080701754385965,
+      "loss": 0.259,
+      "step": 2050
+    },
+    {
+      "epoch": 30.61,
+      "learning_rate": 0.0002807719298245614,
+      "loss": 0.3966,
+      "step": 2051
+    },
+    {
+      "epoch": 30.62,
+      "learning_rate": 0.00028073684210526314,
+      "loss": 0.5047,
+      "step": 2052
+    },
+    {
+      "epoch": 30.64,
+      "learning_rate": 0.0002807017543859649,
+      "loss": 0.4883,
+      "step": 2053
+    },
+    {
+      "epoch": 30.65,
+      "learning_rate": 0.00028066666666666664,
+      "loss": 0.0462,
+      "step": 2054
+    },
+    {
+      "epoch": 30.67,
+      "learning_rate": 0.0002806315789473684,
+      "loss": 0.152,
+      "step": 2055
+    },
+    {
+      "epoch": 30.68,
+      "learning_rate": 0.00028059649122807014,
+      "loss": 0.0942,
+      "step": 2056
+    },
+    {
+      "epoch": 30.7,
+      "learning_rate": 0.00028056140350877194,
+      "loss": 0.5002,
+      "step": 2057
+    },
+    {
+      "epoch": 30.71,
+      "learning_rate": 0.0002805263157894737,
+      "loss": 0.3678,
+      "step": 2058
+    },
+    {
+      "epoch": 30.73,
+      "learning_rate": 0.0002804912280701754,
+      "loss": 0.1678,
+      "step": 2059
+    },
+    {
+      "epoch": 30.74,
+      "learning_rate": 0.00028045614035087713,
+      "loss": 0.5979,
+      "step": 2060
+    },
+    {
+      "epoch": 30.76,
+      "learning_rate": 0.00028042105263157894,
+      "loss": 0.2498,
+      "step": 2061
+    },
+    {
+      "epoch": 30.77,
+      "learning_rate": 0.0002803859649122807,
+      "loss": 0.6051,
+      "step": 2062
+    },
+    {
+      "epoch": 30.79,
+      "learning_rate": 0.00028035087719298243,
+      "loss": 0.3782,
+      "step": 2063
+    },
+    {
+      "epoch": 30.8,
+      "learning_rate": 0.0002803157894736842,
+      "loss": 0.3381,
+      "step": 2064
+    },
+    {
+      "epoch": 30.82,
+      "learning_rate": 0.00028028070175438593,
+      "loss": 0.5742,
+      "step": 2065
+    },
+    {
+      "epoch": 30.83,
+      "learning_rate": 0.0002802456140350877,
+      "loss": 0.2356,
+      "step": 2066
+    },
+    {
+      "epoch": 30.85,
+      "learning_rate": 0.00028021052631578943,
+      "loss": 0.5509,
+      "step": 2067
+    },
+    {
+      "epoch": 30.86,
+      "learning_rate": 0.00028017543859649123,
+      "loss": 0.4965,
+      "step": 2068
+    },
+    {
+      "epoch": 30.88,
+      "learning_rate": 0.000280140350877193,
+      "loss": 0.1467,
+      "step": 2069
+    },
+    {
+      "epoch": 30.89,
+      "learning_rate": 0.00028010526315789473,
+      "loss": 0.4174,
+      "step": 2070
+    },
+    {
+      "epoch": 30.91,
+      "learning_rate": 0.0002800701754385965,
+      "loss": 0.2724,
+      "step": 2071
+    },
+    {
+      "epoch": 30.92,
+      "learning_rate": 0.0002800350877192982,
+      "loss": 0.4106,
+      "step": 2072
+    },
+    {
+      "epoch": 30.94,
+      "learning_rate": 0.00028,
+      "loss": 0.5703,
+      "step": 2073
+    },
+    {
+      "epoch": 30.95,
+      "learning_rate": 0.0002799649122807017,
+      "loss": 0.3324,
+      "step": 2074
+    },
+    {
+      "epoch": 30.97,
+      "learning_rate": 0.0002799298245614035,
+      "loss": 0.13,
+      "step": 2075
+    },
+    {
+      "epoch": 30.98,
+      "learning_rate": 0.0002798947368421053,
+      "loss": 0.2645,
+      "step": 2076
+    },
+    {
+      "epoch": 31.0,
+      "learning_rate": 0.00027985964912280697,
+      "loss": 0.1889,
+      "step": 2077
+    },
+    {
+      "epoch": 31.01,
+      "learning_rate": 0.0002798245614035087,
+      "loss": 0.3298,
+      "step": 2078
+    },
+    {
+      "epoch": 31.03,
+      "learning_rate": 0.00027978947368421047,
+      "loss": 0.1496,
+      "step": 2079
+    },
+    {
+      "epoch": 31.04,
+      "learning_rate": 0.00027975438596491227,
+      "loss": 0.3236,
+      "step": 2080
+    },
+    {
+      "epoch": 31.06,
+      "learning_rate": 0.000279719298245614,
+      "loss": 0.1155,
+      "step": 2081
+    },
+    {
+      "epoch": 31.07,
+      "learning_rate": 0.00027968421052631577,
+      "loss": 0.1675,
+      "step": 2082
+    },
+    {
+      "epoch": 31.09,
+      "learning_rate": 0.0002796491228070175,
+      "loss": 0.2023,
+      "step": 2083
+    },
+    {
+      "epoch": 31.1,
+      "learning_rate": 0.00027961403508771927,
+      "loss": 0.35,
+      "step": 2084
+    },
+    {
+      "epoch": 31.12,
+      "learning_rate": 0.000279578947368421,
+      "loss": 0.5572,
+      "step": 2085
+    },
+    {
+      "epoch": 31.13,
+      "learning_rate": 0.00027954385964912276,
+      "loss": 0.1961,
+      "step": 2086
+    },
+    {
+      "epoch": 31.15,
+      "learning_rate": 0.00027950877192982457,
+      "loss": 0.1702,
+      "step": 2087
+    },
+    {
+      "epoch": 31.16,
+      "learning_rate": 0.0002794736842105263,
+      "loss": 0.0372,
+      "step": 2088
+    },
+    {
+      "epoch": 31.18,
+      "learning_rate": 0.00027943859649122806,
+      "loss": 0.3766,
+      "step": 2089
+    },
+    {
+      "epoch": 31.19,
+      "learning_rate": 0.00027940350877192976,
+      "loss": 0.3156,
+      "step": 2090
+    },
+    {
+      "epoch": 31.21,
+      "learning_rate": 0.00027936842105263156,
+      "loss": 0.1704,
+      "step": 2091
+    },
+    {
+      "epoch": 31.22,
+      "learning_rate": 0.0002793333333333333,
+      "loss": 0.0498,
+      "step": 2092
+    },
+    {
+      "epoch": 31.24,
+      "learning_rate": 0.00027929824561403506,
+      "loss": 0.2717,
+      "step": 2093
+    },
+    {
+      "epoch": 31.25,
+      "learning_rate": 0.0002792631578947368,
+      "loss": 0.0501,
+      "step": 2094
+    },
+    {
+      "epoch": 31.27,
+      "learning_rate": 0.00027922807017543856,
+      "loss": 0.0992,
+      "step": 2095
+    },
+    {
+      "epoch": 31.28,
+      "learning_rate": 0.0002791929824561403,
+      "loss": 0.5988,
+      "step": 2096
+    },
+    {
+      "epoch": 31.3,
+      "learning_rate": 0.00027915789473684205,
+      "loss": 0.2703,
+      "step": 2097
+    },
+    {
+      "epoch": 31.31,
+      "learning_rate": 0.00027912280701754386,
+      "loss": 0.5195,
+      "step": 2098
+    },
+    {
+      "epoch": 31.33,
+      "learning_rate": 0.0002790877192982456,
+      "loss": 0.0573,
+      "step": 2099
+    },
+    {
+      "epoch": 31.34,
+      "learning_rate": 0.00027905263157894736,
+      "loss": 0.0256,
+      "step": 2100
+    },
+    {
+      "epoch": 31.36,
+      "learning_rate": 0.0002790175438596491,
+      "loss": 0.579,
+      "step": 2101
+    },
+    {
+      "epoch": 31.37,
+      "learning_rate": 0.00027898245614035085,
+      "loss": 0.0547,
+      "step": 2102
+    },
+    {
+      "epoch": 31.39,
+      "learning_rate": 0.0002789473684210526,
+      "loss": 0.0967,
+      "step": 2103
+    },
+    {
+      "epoch": 31.4,
+      "learning_rate": 0.00027891228070175435,
+      "loss": 0.258,
+      "step": 2104
+    },
+    {
+      "epoch": 31.42,
+      "learning_rate": 0.00027887719298245615,
+      "loss": 0.2852,
+      "step": 2105
+    },
+    {
+      "epoch": 31.43,
+      "learning_rate": 0.0002788421052631579,
+      "loss": 0.0816,
+      "step": 2106
+    },
+    {
+      "epoch": 31.45,
+      "learning_rate": 0.0002788070175438596,
+      "loss": 0.2592,
+      "step": 2107
+    },
+    {
+      "epoch": 31.46,
+      "learning_rate": 0.00027877192982456135,
+      "loss": 0.4082,
+      "step": 2108
+    },
+    {
+      "epoch": 31.48,
+      "learning_rate": 0.00027873684210526315,
+      "loss": 0.0363,
+      "step": 2109
+    },
+    {
+      "epoch": 31.49,
+      "learning_rate": 0.0002787017543859649,
+      "loss": 0.2759,
+      "step": 2110
+    },
+    {
+      "epoch": 31.51,
+      "learning_rate": 0.00027866666666666665,
+      "loss": 0.9859,
+      "step": 2111
+    },
+    {
+      "epoch": 31.52,
+      "learning_rate": 0.0002786315789473684,
+      "loss": 0.3908,
+      "step": 2112
+    },
+    {
+      "epoch": 31.54,
+      "learning_rate": 0.00027859649122807014,
+      "loss": 0.4191,
+      "step": 2113
+    },
+    {
+      "epoch": 31.55,
+      "learning_rate": 0.0002785614035087719,
+      "loss": 0.2121,
+      "step": 2114
+    },
+    {
+      "epoch": 31.57,
+      "learning_rate": 0.00027852631578947364,
+      "loss": 0.0456,
+      "step": 2115
+    },
+    {
+      "epoch": 31.58,
+      "learning_rate": 0.0002784912280701754,
+      "loss": 0.3328,
+      "step": 2116
+    },
+    {
+      "epoch": 31.59,
+      "learning_rate": 0.0002784561403508772,
+      "loss": 0.4704,
+      "step": 2117
+    },
+    {
+      "epoch": 31.61,
+      "learning_rate": 0.00027842105263157894,
+      "loss": 0.1809,
+      "step": 2118
+    },
+    {
+      "epoch": 31.62,
+      "learning_rate": 0.0002783859649122807,
+      "loss": 0.394,
+      "step": 2119
+    },
+    {
+      "epoch": 31.64,
+      "learning_rate": 0.00027835087719298244,
+      "loss": 0.0635,
+      "step": 2120
+    },
+    {
+      "epoch": 31.65,
+      "learning_rate": 0.0002783157894736842,
+      "loss": 0.0528,
+      "step": 2121
+    },
+    {
+      "epoch": 31.67,
+      "learning_rate": 0.00027828070175438594,
+      "loss": 0.0976,
+      "step": 2122
+    },
+    {
+      "epoch": 31.68,
+      "learning_rate": 0.0002782456140350877,
+      "loss": 0.5734,
+      "step": 2123
+    },
+    {
+      "epoch": 31.7,
+      "learning_rate": 0.0002782105263157895,
+      "loss": 0.2444,
+      "step": 2124
+    },
+    {
+      "epoch": 31.71,
+      "learning_rate": 0.0002781754385964912,
+      "loss": 0.1145,
+      "step": 2125
+    },
+    {
+      "epoch": 31.73,
+      "learning_rate": 0.00027814035087719293,
+      "loss": 0.2101,
+      "step": 2126
+    },
+    {
+      "epoch": 31.74,
+      "learning_rate": 0.0002781052631578947,
+      "loss": 0.3387,
+      "step": 2127
+    },
+    {
+      "epoch": 31.76,
+      "learning_rate": 0.0002780701754385965,
+      "loss": 0.1206,
+      "step": 2128
+    },
+    {
+      "epoch": 31.77,
+      "learning_rate": 0.00027803508771929823,
+      "loss": 0.3084,
+      "step": 2129
+    },
+    {
+      "epoch": 31.79,
+      "learning_rate": 0.000278,
+      "loss": 0.5995,
+      "step": 2130
+    },
+    {
+      "epoch": 31.8,
+      "learning_rate": 0.00027796491228070173,
+      "loss": 0.6344,
+      "step": 2131
+    },
+    {
+      "epoch": 31.82,
+      "learning_rate": 0.0002779298245614035,
+      "loss": 0.4073,
+      "step": 2132
+    },
+    {
+      "epoch": 31.83,
+      "learning_rate": 0.00027789473684210523,
+      "loss": 0.2271,
+      "step": 2133
+    },
+    {
+      "epoch": 31.85,
+      "learning_rate": 0.000277859649122807,
+      "loss": 0.1028,
+      "step": 2134
+    },
+    {
+      "epoch": 31.86,
+      "learning_rate": 0.0002778245614035088,
+      "loss": 0.1707,
+      "step": 2135
+    },
+    {
+      "epoch": 31.88,
+      "learning_rate": 0.00027778947368421053,
+      "loss": 0.1946,
+      "step": 2136
+    },
+    {
+      "epoch": 31.89,
+      "learning_rate": 0.0002777543859649123,
+      "loss": 0.5396,
+      "step": 2137
+    },
+    {
+      "epoch": 31.91,
+      "learning_rate": 0.00027771929824561397,
+      "loss": 0.0741,
+      "step": 2138
+    },
+    {
+      "epoch": 31.92,
+      "learning_rate": 0.0002776842105263158,
+      "loss": 0.3099,
+      "step": 2139
+    },
+    {
+      "epoch": 31.94,
+      "learning_rate": 0.0002776491228070175,
+      "loss": 0.2818,
+      "step": 2140
+    },
+    {
+      "epoch": 31.95,
+      "learning_rate": 0.00027761403508771927,
+      "loss": 0.3299,
+      "step": 2141
+    },
+    {
+      "epoch": 31.97,
+      "learning_rate": 0.000277578947368421,
+      "loss": 0.4072,
+      "step": 2142
+    },
+    {
+      "epoch": 31.98,
+      "learning_rate": 0.00027754385964912277,
+      "loss": 0.0182,
+      "step": 2143
+    },
+    {
+      "epoch": 32.0,
+      "learning_rate": 0.0002775087719298245,
+      "loss": 0.6486,
+      "step": 2144
+    },
+    {
+      "epoch": 32.01,
+      "learning_rate": 0.00027747368421052627,
+      "loss": 0.6269,
+      "step": 2145
+    },
+    {
+      "epoch": 32.03,
+      "learning_rate": 0.00027743859649122807,
+      "loss": 0.353,
+      "step": 2146
+    },
+    {
+      "epoch": 32.04,
+      "learning_rate": 0.0002774035087719298,
+      "loss": 0.4854,
+      "step": 2147
+    },
+    {
+      "epoch": 32.06,
+      "learning_rate": 0.00027736842105263157,
+      "loss": 0.2553,
+      "step": 2148
+    },
+    {
+      "epoch": 32.07,
+      "learning_rate": 0.0002773333333333333,
+      "loss": 0.158,
+      "step": 2149
+    },
+    {
+      "epoch": 32.09,
+      "learning_rate": 0.00027729824561403507,
+      "loss": 0.4838,
+      "step": 2150
+    },
+    {
+      "epoch": 32.1,
+      "learning_rate": 0.0002772631578947368,
+      "loss": 0.2814,
+      "step": 2151
+    },
+    {
+      "epoch": 32.12,
+      "learning_rate": 0.00027722807017543856,
+      "loss": 0.4563,
+      "step": 2152
+    },
+    {
+      "epoch": 32.13,
+      "learning_rate": 0.0002771929824561403,
+      "loss": 0.4573,
+      "step": 2153
+    },
+    {
+      "epoch": 32.15,
+      "learning_rate": 0.0002771578947368421,
+      "loss": 0.3097,
+      "step": 2154
+    },
+    {
+      "epoch": 32.16,
+      "learning_rate": 0.0002771228070175438,
+      "loss": 0.2421,
+      "step": 2155
+    },
+    {
+      "epoch": 32.18,
+      "learning_rate": 0.00027708771929824556,
+      "loss": 0.8276,
+      "step": 2156
+    },
+    {
+      "epoch": 32.19,
+      "learning_rate": 0.00027705263157894736,
+      "loss": 0.4142,
+      "step": 2157
+    },
+    {
+      "epoch": 32.21,
+      "learning_rate": 0.0002770175438596491,
+      "loss": 0.0869,
+      "step": 2158
+    },
+    {
+      "epoch": 32.22,
+      "learning_rate": 0.00027698245614035086,
+      "loss": 0.3774,
+      "step": 2159
+    },
+    {
+      "epoch": 32.24,
+      "learning_rate": 0.0002769473684210526,
+      "loss": 0.2919,
+      "step": 2160
+    },
+    {
+      "epoch": 32.25,
+      "learning_rate": 0.00027691228070175436,
+      "loss": 0.652,
+      "step": 2161
+    },
+    {
+      "epoch": 32.27,
+      "learning_rate": 0.0002768771929824561,
+      "loss": 0.2367,
+      "step": 2162
+    },
+    {
+      "epoch": 32.28,
+      "learning_rate": 0.00027684210526315785,
+      "loss": 0.7383,
+      "step": 2163
+    },
+    {
+      "epoch": 32.3,
+      "learning_rate": 0.0002768070175438596,
+      "loss": 0.4451,
+      "step": 2164
+    },
+    {
+      "epoch": 32.31,
+      "learning_rate": 0.0002767719298245614,
+      "loss": 0.2831,
+      "step": 2165
+    },
+    {
+      "epoch": 32.33,
+      "learning_rate": 0.00027673684210526315,
+      "loss": 0.1604,
+      "step": 2166
+    },
+    {
+      "epoch": 32.34,
+      "learning_rate": 0.0002767017543859649,
+      "loss": 0.0478,
+      "step": 2167
+    },
+    {
+      "epoch": 32.36,
+      "learning_rate": 0.00027666666666666665,
+      "loss": 0.2886,
+      "step": 2168
+    },
+    {
+      "epoch": 32.37,
+      "learning_rate": 0.0002766315789473684,
+      "loss": 0.625,
+      "step": 2169
+    },
+    {
+      "epoch": 32.39,
+      "learning_rate": 0.00027659649122807015,
+      "loss": 0.2271,
+      "step": 2170
+    },
+    {
+      "epoch": 32.4,
+      "learning_rate": 0.0002765614035087719,
+      "loss": 0.1946,
+      "step": 2171
+    },
+    {
+      "epoch": 32.42,
+      "learning_rate": 0.00027652631578947365,
+      "loss": 0.2989,
+      "step": 2172
+    },
+    {
+      "epoch": 32.43,
+      "learning_rate": 0.0002764912280701754,
+      "loss": 0.0619,
+      "step": 2173
+    },
+    {
+      "epoch": 32.45,
+      "learning_rate": 0.00027645614035087714,
+      "loss": 0.2763,
+      "step": 2174
+    },
+    {
+      "epoch": 32.46,
+      "learning_rate": 0.0002764210526315789,
+      "loss": 0.416,
+      "step": 2175
+    },
+    {
+      "epoch": 32.48,
+      "learning_rate": 0.0002763859649122807,
+      "loss": 0.2922,
+      "step": 2176
+    },
+    {
+      "epoch": 32.49,
+      "learning_rate": 0.00027635087719298244,
+      "loss": 0.2956,
+      "step": 2177
+    },
+    {
+      "epoch": 32.51,
+      "learning_rate": 0.0002763157894736842,
+      "loss": 0.369,
+      "step": 2178
+    },
+    {
+      "epoch": 32.52,
+      "learning_rate": 0.00027628070175438594,
+      "loss": 0.4236,
+      "step": 2179
+    },
+    {
+      "epoch": 32.54,
+      "learning_rate": 0.0002762456140350877,
+      "loss": 0.4884,
+      "step": 2180
+    },
+    {
+      "epoch": 32.55,
+      "learning_rate": 0.00027621052631578944,
+      "loss": 0.2356,
+      "step": 2181
+    },
+    {
+      "epoch": 32.57,
+      "learning_rate": 0.0002761754385964912,
+      "loss": 0.4257,
+      "step": 2182
+    },
+    {
+      "epoch": 32.58,
+      "learning_rate": 0.000276140350877193,
+      "loss": 0.3091,
+      "step": 2183
+    },
+    {
+      "epoch": 32.59,
+      "learning_rate": 0.00027610526315789474,
+      "loss": 0.2393,
+      "step": 2184
+    },
+    {
+      "epoch": 32.61,
+      "learning_rate": 0.0002760701754385965,
+      "loss": 0.2166,
+      "step": 2185
+    },
+    {
+      "epoch": 32.62,
+      "learning_rate": 0.0002760350877192982,
+      "loss": 0.0775,
+      "step": 2186
+    },
+    {
+      "epoch": 32.64,
+      "learning_rate": 0.000276,
+      "loss": 0.2225,
+      "step": 2187
+    },
+    {
+      "epoch": 32.65,
+      "learning_rate": 0.00027596491228070174,
+      "loss": 0.2278,
+      "step": 2188
+    },
+    {
+      "epoch": 32.67,
+      "learning_rate": 0.0002759298245614035,
+      "loss": 0.2411,
+      "step": 2189
+    },
+    {
+      "epoch": 32.68,
+      "learning_rate": 0.00027589473684210523,
+      "loss": 0.9149,
+      "step": 2190
+    },
+    {
+      "epoch": 32.7,
+      "learning_rate": 0.000275859649122807,
+      "loss": 0.186,
+      "step": 2191
+    },
+    {
+      "epoch": 32.71,
+      "learning_rate": 0.00027582456140350873,
+      "loss": 0.3165,
+      "step": 2192
+    },
+    {
+      "epoch": 32.73,
+      "learning_rate": 0.0002757894736842105,
+      "loss": 0.4464,
+      "step": 2193
+    },
+    {
+      "epoch": 32.74,
+      "learning_rate": 0.0002757543859649123,
+      "loss": 0.3044,
+      "step": 2194
+    },
+    {
+      "epoch": 32.76,
+      "learning_rate": 0.00027571929824561403,
+      "loss": 0.4148,
+      "step": 2195
+    },
+    {
+      "epoch": 32.77,
+      "learning_rate": 0.0002756842105263158,
+      "loss": 0.4251,
+      "step": 2196
+    },
+    {
+      "epoch": 32.79,
+      "learning_rate": 0.00027564912280701753,
+      "loss": 0.3422,
+      "step": 2197
+    },
+    {
+      "epoch": 32.8,
+      "learning_rate": 0.0002756140350877193,
+      "loss": 0.0342,
+      "step": 2198
+    },
+    {
+      "epoch": 32.82,
+      "learning_rate": 0.000275578947368421,
+      "loss": 0.2479,
+      "step": 2199
+    },
+    {
+      "epoch": 32.83,
+      "learning_rate": 0.0002755438596491228,
+      "loss": 0.2307,
+      "step": 2200
+    },
+    {
+      "epoch": 32.83,
+      "eval_accuracy": 0.8086147821830642,
+      "eval_f1": 0.8071662783486067,
+      "eval_loss": 0.7650861740112305,
+      "eval_runtime": 343.8365,
+      "eval_samples_per_second": 11.884,
+      "eval_steps_per_second": 0.745,
+      "step": 2200
+    },
+    {
+      "epoch": 32.85,
+      "learning_rate": 0.0002755087719298245,
+      "loss": 0.5653,
+      "step": 2201
+    },
+    {
+      "epoch": 32.86,
+      "learning_rate": 0.0002754736842105263,
+      "loss": 0.2916,
+      "step": 2202
+    },
+    {
+      "epoch": 32.88,
+      "learning_rate": 0.000275438596491228,
+      "loss": 0.0993,
+      "step": 2203
+    },
+    {
+      "epoch": 32.89,
+      "learning_rate": 0.00027540350877192977,
+      "loss": 0.3158,
+      "step": 2204
+    },
+    {
+      "epoch": 32.91,
+      "learning_rate": 0.0002753684210526315,
+      "loss": 0.0403,
+      "step": 2205
+    },
+    {
+      "epoch": 32.92,
+      "learning_rate": 0.0002753333333333333,
+      "loss": 0.1977,
+      "step": 2206
+    },
+    {
+      "epoch": 32.94,
+      "learning_rate": 0.00027529824561403507,
+      "loss": 0.2377,
+      "step": 2207
+    },
+    {
+      "epoch": 32.95,
+      "learning_rate": 0.0002752631578947368,
+      "loss": 0.1325,
+      "step": 2208
+    },
+    {
+      "epoch": 32.97,
+      "learning_rate": 0.00027522807017543857,
+      "loss": 0.4147,
+      "step": 2209
+    },
+    {
+      "epoch": 32.98,
+      "learning_rate": 0.0002751929824561403,
+      "loss": 0.1223,
+      "step": 2210
+    },
+    {
+      "epoch": 33.0,
+      "learning_rate": 0.00027515789473684207,
+      "loss": 0.1362,
+      "step": 2211
+    },
+    {
+      "epoch": 33.01,
+      "learning_rate": 0.0002751228070175438,
+      "loss": 0.3086,
+      "step": 2212
+    },
+    {
+      "epoch": 33.03,
+      "learning_rate": 0.0002750877192982456,
+      "loss": 0.3467,
+      "step": 2213
+    },
+    {
+      "epoch": 33.04,
+      "learning_rate": 0.00027505263157894737,
+      "loss": 0.1881,
+      "step": 2214
+    },
+    {
+      "epoch": 33.06,
+      "learning_rate": 0.0002750175438596491,
+      "loss": 0.2578,
+      "step": 2215
+    },
+    {
+      "epoch": 33.07,
+      "learning_rate": 0.00027498245614035086,
+      "loss": 0.2942,
+      "step": 2216
+    },
+    {
+      "epoch": 33.09,
+      "learning_rate": 0.0002749473684210526,
+      "loss": 0.3025,
+      "step": 2217
+    },
+    {
+      "epoch": 33.1,
+      "learning_rate": 0.00027491228070175436,
+      "loss": 0.1853,
+      "step": 2218
+    },
+    {
+      "epoch": 33.12,
+      "learning_rate": 0.0002748771929824561,
+      "loss": 0.052,
+      "step": 2219
+    },
+    {
+      "epoch": 33.13,
+      "learning_rate": 0.00027484210526315786,
+      "loss": 0.3404,
+      "step": 2220
+    },
+    {
+      "epoch": 33.15,
+      "learning_rate": 0.0002748070175438596,
+      "loss": 0.0831,
+      "step": 2221
+    },
+    {
+      "epoch": 33.16,
+      "learning_rate": 0.00027477192982456136,
+      "loss": 0.1659,
+      "step": 2222
+    },
+    {
+      "epoch": 33.18,
+      "learning_rate": 0.0002747368421052631,
+      "loss": 0.0441,
+      "step": 2223
+    },
+    {
+      "epoch": 33.19,
+      "learning_rate": 0.0002747017543859649,
+      "loss": 0.2571,
+      "step": 2224
+    },
+    {
+      "epoch": 33.21,
+      "learning_rate": 0.00027466666666666666,
+      "loss": 0.3079,
+      "step": 2225
+    },
+    {
+      "epoch": 33.22,
+      "learning_rate": 0.0002746315789473684,
+      "loss": 0.3602,
+      "step": 2226
+    },
+    {
+      "epoch": 33.24,
+      "learning_rate": 0.00027459649122807015,
+      "loss": 0.1547,
+      "step": 2227
+    },
+    {
+      "epoch": 33.25,
+      "learning_rate": 0.0002745614035087719,
+      "loss": 0.3108,
+      "step": 2228
+    },
+    {
+      "epoch": 33.27,
+      "learning_rate": 0.00027452631578947365,
+      "loss": 0.6134,
+      "step": 2229
+    },
+    {
+      "epoch": 33.28,
+      "learning_rate": 0.0002744912280701754,
+      "loss": 0.3319,
+      "step": 2230
+    },
+    {
+      "epoch": 33.3,
+      "learning_rate": 0.0002744561403508772,
+      "loss": 0.5913,
+      "step": 2231
+    },
+    {
+      "epoch": 33.31,
+      "learning_rate": 0.00027442105263157895,
+      "loss": 0.4172,
+      "step": 2232
+    },
+    {
+      "epoch": 33.33,
+      "learning_rate": 0.0002743859649122807,
+      "loss": 0.5581,
+      "step": 2233
+    },
+    {
+      "epoch": 33.34,
+      "learning_rate": 0.0002743508771929824,
+      "loss": 0.1744,
+      "step": 2234
+    },
+    {
+      "epoch": 33.36,
+      "learning_rate": 0.0002743157894736842,
+      "loss": 0.3023,
+      "step": 2235
+    },
+    {
+      "epoch": 33.37,
+      "learning_rate": 0.00027428070175438595,
+      "loss": 0.9059,
+      "step": 2236
+    },
+    {
+      "epoch": 33.39,
+      "learning_rate": 0.0002742456140350877,
+      "loss": 0.1562,
+      "step": 2237
+    },
+    {
+      "epoch": 33.4,
+      "learning_rate": 0.00027421052631578945,
+      "loss": 0.2438,
+      "step": 2238
+    },
+    {
+      "epoch": 33.42,
+      "learning_rate": 0.0002741754385964912,
+      "loss": 0.5985,
+      "step": 2239
+    },
+    {
+      "epoch": 33.43,
+      "learning_rate": 0.00027414035087719294,
+      "loss": 0.021,
+      "step": 2240
+    },
+    {
+      "epoch": 33.45,
+      "learning_rate": 0.0002741052631578947,
+      "loss": 0.4208,
+      "step": 2241
+    },
+    {
+      "epoch": 33.46,
+      "learning_rate": 0.00027407017543859644,
+      "loss": 0.2522,
+      "step": 2242
+    },
+    {
+      "epoch": 33.48,
+      "learning_rate": 0.00027403508771929824,
+      "loss": 0.2324,
+      "step": 2243
+    },
+    {
+      "epoch": 33.49,
+      "learning_rate": 0.000274,
+      "loss": 0.235,
+      "step": 2244
+    },
+    {
+      "epoch": 33.51,
+      "learning_rate": 0.00027396491228070174,
+      "loss": 0.3669,
+      "step": 2245
+    },
+    {
+      "epoch": 33.52,
+      "learning_rate": 0.0002739298245614035,
+      "loss": 0.3819,
+      "step": 2246
+    },
+    {
+      "epoch": 33.54,
+      "learning_rate": 0.00027389473684210524,
+      "loss": 0.1008,
+      "step": 2247
+    },
+    {
+      "epoch": 33.55,
+      "learning_rate": 0.000273859649122807,
+      "loss": 0.4268,
+      "step": 2248
+    },
+    {
+      "epoch": 33.57,
+      "learning_rate": 0.00027382456140350874,
+      "loss": 0.0733,
+      "step": 2249
+    },
+    {
+      "epoch": 33.58,
+      "learning_rate": 0.00027378947368421054,
+      "loss": 0.2913,
+      "step": 2250
+    },
+    {
+      "epoch": 33.59,
+      "learning_rate": 0.00027375438596491223,
+      "loss": 0.2841,
+      "step": 2251
+    },
+    {
+      "epoch": 33.61,
+      "learning_rate": 0.000273719298245614,
+      "loss": 0.0646,
+      "step": 2252
+    },
+    {
+      "epoch": 33.62,
+      "learning_rate": 0.00027368421052631573,
+      "loss": 0.4224,
+      "step": 2253
+    },
+    {
+      "epoch": 33.64,
+      "learning_rate": 0.00027364912280701753,
+      "loss": 0.1624,
+      "step": 2254
+    },
+    {
+      "epoch": 33.65,
+      "learning_rate": 0.0002736140350877193,
+      "loss": 0.1866,
+      "step": 2255
+    },
+    {
+      "epoch": 33.67,
+      "learning_rate": 0.00027357894736842103,
+      "loss": 0.0219,
+      "step": 2256
+    },
+    {
+      "epoch": 33.68,
+      "learning_rate": 0.0002735438596491228,
+      "loss": 0.123,
+      "step": 2257
+    },
+    {
+      "epoch": 33.7,
+      "learning_rate": 0.00027350877192982453,
+      "loss": 0.4814,
+      "step": 2258
+    },
+    {
+      "epoch": 33.71,
+      "learning_rate": 0.0002734736842105263,
+      "loss": 0.1772,
+      "step": 2259
+    },
+    {
+      "epoch": 33.73,
+      "learning_rate": 0.000273438596491228,
+      "loss": 0.2027,
+      "step": 2260
+    },
+    {
+      "epoch": 33.74,
+      "learning_rate": 0.00027340350877192983,
+      "loss": 0.2114,
+      "step": 2261
+    },
+    {
+      "epoch": 33.76,
+      "learning_rate": 0.0002733684210526316,
+      "loss": 0.4351,
+      "step": 2262
+    },
+    {
+      "epoch": 33.77,
+      "learning_rate": 0.00027333333333333333,
+      "loss": 0.4057,
+      "step": 2263
+    },
+    {
+      "epoch": 33.79,
+      "learning_rate": 0.0002732982456140351,
+      "loss": 0.2215,
+      "step": 2264
+    },
+    {
+      "epoch": 33.8,
+      "learning_rate": 0.0002732631578947368,
+      "loss": 0.1509,
+      "step": 2265
+    },
+    {
+      "epoch": 33.82,
+      "learning_rate": 0.0002732280701754386,
+      "loss": 0.4424,
+      "step": 2266
+    },
+    {
+      "epoch": 33.83,
+      "learning_rate": 0.0002731929824561403,
+      "loss": 0.2958,
+      "step": 2267
+    },
+    {
+      "epoch": 33.85,
+      "learning_rate": 0.00027315789473684207,
+      "loss": 0.0407,
+      "step": 2268
+    },
+    {
+      "epoch": 33.86,
+      "learning_rate": 0.0002731228070175438,
+      "loss": 0.2227,
+      "step": 2269
+    },
+    {
+      "epoch": 33.88,
+      "learning_rate": 0.00027308771929824557,
+      "loss": 0.6847,
+      "step": 2270
+    },
+    {
+      "epoch": 33.89,
+      "learning_rate": 0.0002730526315789473,
+      "loss": 0.1189,
+      "step": 2271
+    },
+    {
+      "epoch": 33.91,
+      "learning_rate": 0.0002730175438596491,
+      "loss": 0.4449,
+      "step": 2272
+    },
+    {
+      "epoch": 33.92,
+      "learning_rate": 0.00027298245614035087,
+      "loss": 0.1553,
+      "step": 2273
+    },
+    {
+      "epoch": 33.94,
+      "learning_rate": 0.0002729473684210526,
+      "loss": 0.0367,
+      "step": 2274
+    },
+    {
+      "epoch": 33.95,
+      "learning_rate": 0.00027291228070175437,
+      "loss": 0.0587,
+      "step": 2275
+    },
+    {
+      "epoch": 33.97,
+      "learning_rate": 0.0002728771929824561,
+      "loss": 0.4163,
+      "step": 2276
+    },
+    {
+      "epoch": 33.98,
+      "learning_rate": 0.00027284210526315786,
+      "loss": 0.3024,
+      "step": 2277
+    },
+    {
+      "epoch": 34.0,
+      "learning_rate": 0.0002728070175438596,
+      "loss": 0.6945,
+      "step": 2278
+    },
+    {
+      "epoch": 34.01,
+      "learning_rate": 0.00027277192982456136,
+      "loss": 0.4193,
+      "step": 2279
+    },
+    {
+      "epoch": 34.03,
+      "learning_rate": 0.00027273684210526317,
+      "loss": 0.2735,
+      "step": 2280
+    },
+    {
+      "epoch": 34.04,
+      "learning_rate": 0.0002727017543859649,
+      "loss": 0.1118,
+      "step": 2281
+    },
+    {
+      "epoch": 34.06,
+      "learning_rate": 0.0002726666666666666,
+      "loss": 0.6835,
+      "step": 2282
+    },
+    {
+      "epoch": 34.07,
+      "learning_rate": 0.0002726315789473684,
+      "loss": 0.3097,
+      "step": 2283
+    },
+    {
+      "epoch": 34.09,
+      "learning_rate": 0.00027259649122807016,
+      "loss": 0.2908,
+      "step": 2284
+    },
+    {
+      "epoch": 34.1,
+      "learning_rate": 0.0002725614035087719,
+      "loss": 0.168,
+      "step": 2285
+    },
+    {
+      "epoch": 34.12,
+      "learning_rate": 0.00027252631578947366,
+      "loss": 0.3362,
+      "step": 2286
+    },
+    {
+      "epoch": 34.13,
+      "learning_rate": 0.0002724912280701754,
+      "loss": 0.2603,
+      "step": 2287
+    },
+    {
+      "epoch": 34.15,
+      "learning_rate": 0.00027245614035087716,
+      "loss": 0.0215,
+      "step": 2288
+    },
+    {
+      "epoch": 34.16,
+      "learning_rate": 0.0002724210526315789,
+      "loss": 0.0313,
+      "step": 2289
+    },
+    {
+      "epoch": 34.18,
+      "learning_rate": 0.00027238596491228065,
+      "loss": 0.0771,
+      "step": 2290
+    },
+    {
+      "epoch": 34.19,
+      "learning_rate": 0.00027235087719298246,
+      "loss": 0.067,
+      "step": 2291
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 0.0002723157894736842,
+      "loss": 0.5942,
+      "step": 2292
+    },
+    {
+      "epoch": 34.22,
+      "learning_rate": 0.00027228070175438595,
+      "loss": 0.0307,
+      "step": 2293
+    },
+    {
+      "epoch": 34.24,
+      "learning_rate": 0.0002722456140350877,
+      "loss": 0.226,
+      "step": 2294
+    },
+    {
+      "epoch": 34.25,
+      "learning_rate": 0.00027221052631578945,
+      "loss": 0.0363,
+      "step": 2295
+    },
+    {
+      "epoch": 34.27,
+      "learning_rate": 0.0002721754385964912,
+      "loss": 0.1999,
+      "step": 2296
+    },
+    {
+      "epoch": 34.28,
+      "learning_rate": 0.00027214035087719295,
+      "loss": 0.1683,
+      "step": 2297
+    },
+    {
+      "epoch": 34.3,
+      "learning_rate": 0.00027210526315789475,
+      "loss": 0.0772,
+      "step": 2298
+    },
+    {
+      "epoch": 34.31,
+      "learning_rate": 0.00027207017543859645,
+      "loss": 0.1123,
+      "step": 2299
+    },
+    {
+      "epoch": 34.33,
+      "learning_rate": 0.0002720350877192982,
+      "loss": 0.4395,
+      "step": 2300
+    },
+    {
+      "epoch": 34.34,
+      "learning_rate": 0.00027199999999999994,
+      "loss": 0.4303,
+      "step": 2301
+    },
+    {
+      "epoch": 34.36,
+      "learning_rate": 0.00027196491228070175,
+      "loss": 0.3241,
+      "step": 2302
+    },
+    {
+      "epoch": 34.37,
+      "learning_rate": 0.0002719298245614035,
+      "loss": 0.1974,
+      "step": 2303
+    },
+    {
+      "epoch": 34.39,
+      "learning_rate": 0.00027189473684210524,
+      "loss": 0.3825,
+      "step": 2304
+    },
+    {
+      "epoch": 34.4,
+      "learning_rate": 0.000271859649122807,
+      "loss": 0.1469,
+      "step": 2305
+    },
+    {
+      "epoch": 34.42,
+      "learning_rate": 0.00027182456140350874,
+      "loss": 0.2059,
+      "step": 2306
+    },
+    {
+      "epoch": 34.43,
+      "learning_rate": 0.0002717894736842105,
+      "loss": 0.026,
+      "step": 2307
+    },
+    {
+      "epoch": 34.45,
+      "learning_rate": 0.00027175438596491224,
+      "loss": 0.2546,
+      "step": 2308
+    },
+    {
+      "epoch": 34.46,
+      "learning_rate": 0.00027171929824561404,
+      "loss": 0.1752,
+      "step": 2309
+    },
+    {
+      "epoch": 34.48,
+      "learning_rate": 0.0002716842105263158,
+      "loss": 0.1128,
+      "step": 2310
+    },
+    {
+      "epoch": 34.49,
+      "learning_rate": 0.00027164912280701754,
+      "loss": 0.2363,
+      "step": 2311
+    },
+    {
+      "epoch": 34.51,
+      "learning_rate": 0.0002716140350877193,
+      "loss": 0.2964,
+      "step": 2312
+    },
+    {
+      "epoch": 34.52,
+      "learning_rate": 0.00027157894736842104,
+      "loss": 0.6012,
+      "step": 2313
+    },
+    {
+      "epoch": 34.54,
+      "learning_rate": 0.0002715438596491228,
+      "loss": 0.4373,
+      "step": 2314
+    },
+    {
+      "epoch": 34.55,
+      "learning_rate": 0.00027150877192982453,
+      "loss": 0.3563,
+      "step": 2315
+    },
+    {
+      "epoch": 34.57,
+      "learning_rate": 0.0002714736842105263,
+      "loss": 0.264,
+      "step": 2316
+    },
+    {
+      "epoch": 34.58,
+      "learning_rate": 0.00027143859649122803,
+      "loss": 0.0819,
+      "step": 2317
+    },
+    {
+      "epoch": 34.59,
+      "learning_rate": 0.0002714035087719298,
+      "loss": 0.2622,
+      "step": 2318
+    },
+    {
+      "epoch": 34.61,
+      "learning_rate": 0.00027136842105263153,
+      "loss": 0.2915,
+      "step": 2319
+    },
+    {
+      "epoch": 34.62,
+      "learning_rate": 0.00027133333333333333,
+      "loss": 0.2508,
+      "step": 2320
+    },
+    {
+      "epoch": 34.64,
+      "learning_rate": 0.0002712982456140351,
+      "loss": 0.2723,
+      "step": 2321
+    },
+    {
+      "epoch": 34.65,
+      "learning_rate": 0.00027126315789473683,
+      "loss": 0.3011,
+      "step": 2322
+    },
+    {
+      "epoch": 34.67,
+      "learning_rate": 0.0002712280701754386,
+      "loss": 0.3078,
+      "step": 2323
+    },
+    {
+      "epoch": 34.68,
+      "learning_rate": 0.00027119298245614033,
+      "loss": 0.0938,
+      "step": 2324
+    },
+    {
+      "epoch": 34.7,
+      "learning_rate": 0.0002711578947368421,
+      "loss": 0.0643,
+      "step": 2325
+    },
+    {
+      "epoch": 34.71,
+      "learning_rate": 0.0002711228070175438,
+      "loss": 0.1048,
+      "step": 2326
+    },
+    {
+      "epoch": 34.73,
+      "learning_rate": 0.0002710877192982456,
+      "loss": 0.0173,
+      "step": 2327
+    },
+    {
+      "epoch": 34.74,
+      "learning_rate": 0.0002710526315789474,
+      "loss": 0.1349,
+      "step": 2328
+    },
+    {
+      "epoch": 34.76,
+      "learning_rate": 0.0002710175438596491,
+      "loss": 0.3766,
+      "step": 2329
+    },
+    {
+      "epoch": 34.77,
+      "learning_rate": 0.0002709824561403508,
+      "loss": 0.2583,
+      "step": 2330
+    },
+    {
+      "epoch": 34.79,
+      "learning_rate": 0.0002709473684210526,
+      "loss": 0.1862,
+      "step": 2331
+    },
+    {
+      "epoch": 34.8,
+      "learning_rate": 0.00027091228070175437,
+      "loss": 0.253,
+      "step": 2332
+    },
+    {
+      "epoch": 34.82,
+      "learning_rate": 0.0002708771929824561,
+      "loss": 0.7504,
+      "step": 2333
+    },
+    {
+      "epoch": 34.83,
+      "learning_rate": 0.00027084210526315787,
+      "loss": 0.6195,
+      "step": 2334
+    },
+    {
+      "epoch": 34.85,
+      "learning_rate": 0.0002708070175438596,
+      "loss": 0.163,
+      "step": 2335
+    },
+    {
+      "epoch": 34.86,
+      "learning_rate": 0.00027077192982456137,
+      "loss": 0.0618,
+      "step": 2336
+    },
+    {
+      "epoch": 34.88,
+      "learning_rate": 0.0002707368421052631,
+      "loss": 0.3675,
+      "step": 2337
+    },
+    {
+      "epoch": 34.89,
+      "learning_rate": 0.00027070175438596487,
+      "loss": 0.2927,
+      "step": 2338
+    },
+    {
+      "epoch": 34.91,
+      "learning_rate": 0.00027066666666666667,
+      "loss": 0.7439,
+      "step": 2339
+    },
+    {
+      "epoch": 34.92,
+      "learning_rate": 0.0002706315789473684,
+      "loss": 0.2834,
+      "step": 2340
+    },
+    {
+      "epoch": 34.94,
+      "learning_rate": 0.00027059649122807017,
+      "loss": 0.6386,
+      "step": 2341
+    },
+    {
+      "epoch": 34.95,
+      "learning_rate": 0.0002705614035087719,
+      "loss": 0.2333,
+      "step": 2342
+    },
+    {
+      "epoch": 34.97,
+      "learning_rate": 0.00027052631578947366,
+      "loss": 0.5232,
+      "step": 2343
+    },
+    {
+      "epoch": 34.98,
+      "learning_rate": 0.0002704912280701754,
+      "loss": 0.2471,
+      "step": 2344
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 0.00027045614035087716,
+      "loss": 1.1787,
+      "step": 2345
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 0.00027042105263157896,
+      "loss": 0.2669,
+      "step": 2346
+    },
+    {
+      "epoch": 35.03,
+      "learning_rate": 0.00027038596491228066,
+      "loss": 0.7957,
+      "step": 2347
+    },
+    {
+      "epoch": 35.04,
+      "learning_rate": 0.0002703508771929824,
+      "loss": 0.4517,
+      "step": 2348
+    },
+    {
+      "epoch": 35.06,
+      "learning_rate": 0.00027031578947368416,
+      "loss": 0.299,
+      "step": 2349
+    },
+    {
+      "epoch": 35.07,
+      "learning_rate": 0.00027028070175438596,
+      "loss": 0.2013,
+      "step": 2350
+    },
+    {
+      "epoch": 35.09,
+      "learning_rate": 0.0002702456140350877,
+      "loss": 0.2649,
+      "step": 2351
+    },
+    {
+      "epoch": 35.1,
+      "learning_rate": 0.00027021052631578946,
+      "loss": 0.1592,
+      "step": 2352
+    },
+    {
+      "epoch": 35.12,
+      "learning_rate": 0.0002701754385964912,
+      "loss": 0.2795,
+      "step": 2353
+    },
+    {
+      "epoch": 35.13,
+      "learning_rate": 0.00027014035087719295,
+      "loss": 0.3779,
+      "step": 2354
+    },
+    {
+      "epoch": 35.15,
+      "learning_rate": 0.0002701052631578947,
+      "loss": 0.2906,
+      "step": 2355
+    },
+    {
+      "epoch": 35.16,
+      "learning_rate": 0.00027007017543859645,
+      "loss": 0.2444,
+      "step": 2356
+    },
+    {
+      "epoch": 35.18,
+      "learning_rate": 0.00027003508771929825,
+      "loss": 0.3235,
+      "step": 2357
+    },
+    {
+      "epoch": 35.19,
+      "learning_rate": 0.00027,
+      "loss": 0.248,
+      "step": 2358
+    },
+    {
+      "epoch": 35.21,
+      "learning_rate": 0.00026996491228070175,
+      "loss": 0.2753,
+      "step": 2359
+    },
+    {
+      "epoch": 35.22,
+      "learning_rate": 0.00026992982456140345,
+      "loss": 0.2824,
+      "step": 2360
+    },
+    {
+      "epoch": 35.24,
+      "learning_rate": 0.00026989473684210525,
+      "loss": 0.0507,
+      "step": 2361
+    },
+    {
+      "epoch": 35.25,
+      "learning_rate": 0.000269859649122807,
+      "loss": 0.1854,
+      "step": 2362
+    },
+    {
+      "epoch": 35.27,
+      "learning_rate": 0.00026982456140350875,
+      "loss": 0.2716,
+      "step": 2363
+    },
+    {
+      "epoch": 35.28,
+      "learning_rate": 0.0002697894736842105,
+      "loss": 0.8457,
+      "step": 2364
+    },
+    {
+      "epoch": 35.3,
+      "learning_rate": 0.00026975438596491224,
+      "loss": 0.3006,
+      "step": 2365
+    },
+    {
+      "epoch": 35.31,
+      "learning_rate": 0.000269719298245614,
+      "loss": 0.1396,
+      "step": 2366
+    },
+    {
+      "epoch": 35.33,
+      "learning_rate": 0.00026968421052631574,
+      "loss": 0.3765,
+      "step": 2367
+    },
+    {
+      "epoch": 35.34,
+      "learning_rate": 0.0002696491228070175,
+      "loss": 0.1906,
+      "step": 2368
+    },
+    {
+      "epoch": 35.36,
+      "learning_rate": 0.0002696140350877193,
+      "loss": 0.5598,
+      "step": 2369
+    },
+    {
+      "epoch": 35.37,
+      "learning_rate": 0.00026957894736842104,
+      "loss": 0.3803,
+      "step": 2370
+    },
+    {
+      "epoch": 35.39,
+      "learning_rate": 0.0002695438596491228,
+      "loss": 0.2419,
+      "step": 2371
+    },
+    {
+      "epoch": 35.4,
+      "learning_rate": 0.00026950877192982454,
+      "loss": 0.0922,
+      "step": 2372
+    },
+    {
+      "epoch": 35.42,
+      "learning_rate": 0.0002694736842105263,
+      "loss": 0.261,
+      "step": 2373
+    },
+    {
+      "epoch": 35.43,
+      "learning_rate": 0.00026943859649122804,
+      "loss": 0.1247,
+      "step": 2374
+    },
+    {
+      "epoch": 35.45,
+      "learning_rate": 0.0002694035087719298,
+      "loss": 0.1853,
+      "step": 2375
+    },
+    {
+      "epoch": 35.46,
+      "learning_rate": 0.0002693684210526316,
+      "loss": 0.5109,
+      "step": 2376
+    },
+    {
+      "epoch": 35.48,
+      "learning_rate": 0.00026933333333333334,
+      "loss": 0.6572,
+      "step": 2377
+    },
+    {
+      "epoch": 35.49,
+      "learning_rate": 0.00026929824561403503,
+      "loss": 0.1821,
+      "step": 2378
+    },
+    {
+      "epoch": 35.51,
+      "learning_rate": 0.0002692631578947368,
+      "loss": 0.1235,
+      "step": 2379
+    },
+    {
+      "epoch": 35.52,
+      "learning_rate": 0.0002692280701754386,
+      "loss": 0.1746,
+      "step": 2380
+    },
+    {
+      "epoch": 35.54,
+      "learning_rate": 0.00026919298245614033,
+      "loss": 0.3027,
+      "step": 2381
+    },
+    {
+      "epoch": 35.55,
+      "learning_rate": 0.0002691578947368421,
+      "loss": 0.1361,
+      "step": 2382
+    },
+    {
+      "epoch": 35.57,
+      "learning_rate": 0.00026912280701754383,
+      "loss": 0.2752,
+      "step": 2383
+    },
+    {
+      "epoch": 35.58,
+      "learning_rate": 0.0002690877192982456,
+      "loss": 0.1813,
+      "step": 2384
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 0.00026905263157894733,
+      "loss": 0.2419,
+      "step": 2385
+    },
+    {
+      "epoch": 35.61,
+      "learning_rate": 0.0002690175438596491,
+      "loss": 0.542,
+      "step": 2386
+    },
+    {
+      "epoch": 35.62,
+      "learning_rate": 0.0002689824561403509,
+      "loss": 0.2718,
+      "step": 2387
+    },
+    {
+      "epoch": 35.64,
+      "learning_rate": 0.00026894736842105263,
+      "loss": 0.3817,
+      "step": 2388
+    },
+    {
+      "epoch": 35.65,
+      "learning_rate": 0.0002689122807017544,
+      "loss": 0.0568,
+      "step": 2389
+    },
+    {
+      "epoch": 35.67,
+      "learning_rate": 0.0002688771929824561,
+      "loss": 0.212,
+      "step": 2390
+    },
+    {
+      "epoch": 35.68,
+      "learning_rate": 0.0002688421052631579,
+      "loss": 0.0527,
+      "step": 2391
+    },
+    {
+      "epoch": 35.7,
+      "learning_rate": 0.0002688070175438596,
+      "loss": 0.2184,
+      "step": 2392
+    },
+    {
+      "epoch": 35.71,
+      "learning_rate": 0.0002687719298245614,
+      "loss": 0.0476,
+      "step": 2393
+    },
+    {
+      "epoch": 35.73,
+      "learning_rate": 0.0002687368421052632,
+      "loss": 0.138,
+      "step": 2394
+    },
+    {
+      "epoch": 35.74,
+      "learning_rate": 0.00026870175438596487,
+      "loss": 0.2552,
+      "step": 2395
+    },
+    {
+      "epoch": 35.76,
+      "learning_rate": 0.0002686666666666666,
+      "loss": 0.5772,
+      "step": 2396
+    },
+    {
+      "epoch": 35.77,
+      "learning_rate": 0.00026863157894736837,
+      "loss": 0.3532,
+      "step": 2397
+    },
+    {
+      "epoch": 35.79,
+      "learning_rate": 0.00026859649122807017,
+      "loss": 0.0265,
+      "step": 2398
+    },
+    {
+      "epoch": 35.8,
+      "learning_rate": 0.0002685614035087719,
+      "loss": 0.1063,
+      "step": 2399
+    },
+    {
+      "epoch": 35.82,
+      "learning_rate": 0.00026852631578947367,
+      "loss": 0.1412,
+      "step": 2400
+    },
+    {
+      "epoch": 35.82,
+      "eval_accuracy": 0.8311306901615272,
+      "eval_f1": 0.8352003849989705,
+      "eval_loss": 0.7131851315498352,
+      "eval_runtime": 344.3646,
+      "eval_samples_per_second": 11.865,
+      "eval_steps_per_second": 0.743,
+      "step": 2400
+    },
+    {
+      "epoch": 35.83,
+      "learning_rate": 0.0002684912280701754,
+      "loss": 0.3079,
+      "step": 2401
+    },
+    {
+      "epoch": 35.85,
+      "learning_rate": 0.00026845614035087717,
+      "loss": 0.4798,
+      "step": 2402
+    },
+    {
+      "epoch": 35.86,
+      "learning_rate": 0.0002684210526315789,
+      "loss": 0.058,
+      "step": 2403
+    },
+    {
+      "epoch": 35.88,
+      "learning_rate": 0.00026838596491228066,
+      "loss": 0.0388,
+      "step": 2404
+    },
+    {
+      "epoch": 35.89,
+      "learning_rate": 0.0002683508771929824,
+      "loss": 0.0393,
+      "step": 2405
+    },
+    {
+      "epoch": 35.91,
+      "learning_rate": 0.0002683157894736842,
+      "loss": 0.0151,
+      "step": 2406
+    },
+    {
+      "epoch": 35.92,
+      "learning_rate": 0.00026828070175438596,
+      "loss": 0.2027,
+      "step": 2407
+    },
+    {
+      "epoch": 35.94,
+      "learning_rate": 0.00026824561403508766,
+      "loss": 0.1488,
+      "step": 2408
+    },
+    {
+      "epoch": 35.95,
+      "learning_rate": 0.00026821052631578946,
+      "loss": 0.0274,
+      "step": 2409
+    },
+    {
+      "epoch": 35.97,
+      "learning_rate": 0.0002681754385964912,
+      "loss": 0.5102,
+      "step": 2410
+    },
+    {
+      "epoch": 35.98,
+      "learning_rate": 0.00026814035087719296,
+      "loss": 0.0618,
+      "step": 2411
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 0.0002681052631578947,
+      "loss": 0.0711,
+      "step": 2412
+    },
+    {
+      "epoch": 36.01,
+      "learning_rate": 0.00026807017543859646,
+      "loss": 0.333,
+      "step": 2413
+    },
+    {
+      "epoch": 36.03,
+      "learning_rate": 0.0002680350877192982,
+      "loss": 0.2717,
+      "step": 2414
+    },
+    {
+      "epoch": 36.04,
+      "learning_rate": 0.00026799999999999995,
+      "loss": 0.027,
+      "step": 2415
+    },
+    {
+      "epoch": 36.06,
+      "learning_rate": 0.0002679649122807017,
+      "loss": 0.2816,
+      "step": 2416
+    },
+    {
+      "epoch": 36.07,
+      "learning_rate": 0.0002679298245614035,
+      "loss": 0.2222,
+      "step": 2417
+    },
+    {
+      "epoch": 36.09,
+      "learning_rate": 0.00026789473684210526,
+      "loss": 0.2245,
+      "step": 2418
+    },
+    {
+      "epoch": 36.1,
+      "learning_rate": 0.000267859649122807,
+      "loss": 0.3236,
+      "step": 2419
+    },
+    {
+      "epoch": 36.12,
+      "learning_rate": 0.00026782456140350875,
+      "loss": 0.0227,
+      "step": 2420
+    },
+    {
+      "epoch": 36.13,
+      "learning_rate": 0.0002677894736842105,
+      "loss": 0.1237,
+      "step": 2421
+    },
+    {
+      "epoch": 36.15,
+      "learning_rate": 0.00026775438596491225,
+      "loss": 0.2968,
+      "step": 2422
+    },
+    {
+      "epoch": 36.16,
+      "learning_rate": 0.000267719298245614,
+      "loss": 0.4212,
+      "step": 2423
+    },
+    {
+      "epoch": 36.18,
+      "learning_rate": 0.0002676842105263158,
+      "loss": 0.5356,
+      "step": 2424
+    },
+    {
+      "epoch": 36.19,
+      "learning_rate": 0.00026764912280701755,
+      "loss": 0.1707,
+      "step": 2425
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 0.00026761403508771925,
+      "loss": 0.0538,
+      "step": 2426
+    },
+    {
+      "epoch": 36.22,
+      "learning_rate": 0.000267578947368421,
+      "loss": 0.5187,
+      "step": 2427
+    },
+    {
+      "epoch": 36.24,
+      "learning_rate": 0.0002675438596491228,
+      "loss": 0.1425,
+      "step": 2428
+    },
+    {
+      "epoch": 36.25,
+      "learning_rate": 0.00026750877192982455,
+      "loss": 0.2559,
+      "step": 2429
+    },
+    {
+      "epoch": 36.27,
+      "learning_rate": 0.0002674736842105263,
+      "loss": 0.2514,
+      "step": 2430
+    },
+    {
+      "epoch": 36.28,
+      "learning_rate": 0.00026743859649122804,
+      "loss": 0.3183,
+      "step": 2431
+    },
+    {
+      "epoch": 36.3,
+      "learning_rate": 0.0002674035087719298,
+      "loss": 0.4393,
+      "step": 2432
+    },
+    {
+      "epoch": 36.31,
+      "learning_rate": 0.00026736842105263154,
+      "loss": 0.2093,
+      "step": 2433
+    },
+    {
+      "epoch": 36.33,
+      "learning_rate": 0.0002673333333333333,
+      "loss": 0.3006,
+      "step": 2434
+    },
+    {
+      "epoch": 36.34,
+      "learning_rate": 0.0002672982456140351,
+      "loss": 0.1366,
+      "step": 2435
+    },
+    {
+      "epoch": 36.36,
+      "learning_rate": 0.00026726315789473684,
+      "loss": 0.1511,
+      "step": 2436
+    },
+    {
+      "epoch": 36.37,
+      "learning_rate": 0.0002672280701754386,
+      "loss": 0.3088,
+      "step": 2437
+    },
+    {
+      "epoch": 36.39,
+      "learning_rate": 0.00026719298245614034,
+      "loss": 0.2611,
+      "step": 2438
+    },
+    {
+      "epoch": 36.4,
+      "learning_rate": 0.0002671578947368421,
+      "loss": 0.0553,
+      "step": 2439
+    },
+    {
+      "epoch": 36.42,
+      "learning_rate": 0.00026712280701754384,
+      "loss": 0.0143,
+      "step": 2440
+    },
+    {
+      "epoch": 36.43,
+      "learning_rate": 0.0002670877192982456,
+      "loss": 0.2214,
+      "step": 2441
+    },
+    {
+      "epoch": 36.45,
+      "learning_rate": 0.00026705263157894733,
+      "loss": 0.3594,
+      "step": 2442
+    },
+    {
+      "epoch": 36.46,
+      "learning_rate": 0.0002670175438596491,
+      "loss": 0.0609,
+      "step": 2443
+    },
+    {
+      "epoch": 36.48,
+      "learning_rate": 0.00026698245614035083,
+      "loss": 0.1486,
+      "step": 2444
+    },
+    {
+      "epoch": 36.49,
+      "learning_rate": 0.0002669473684210526,
+      "loss": 0.033,
+      "step": 2445
+    },
+    {
+      "epoch": 36.51,
+      "learning_rate": 0.0002669122807017544,
+      "loss": 0.8025,
+      "step": 2446
+    },
+    {
+      "epoch": 36.52,
+      "learning_rate": 0.00026687719298245613,
+      "loss": 0.3945,
+      "step": 2447
+    },
+    {
+      "epoch": 36.54,
+      "learning_rate": 0.0002668421052631579,
+      "loss": 0.3908,
+      "step": 2448
+    },
+    {
+      "epoch": 36.55,
+      "learning_rate": 0.00026680701754385963,
+      "loss": 0.1322,
+      "step": 2449
+    },
+    {
+      "epoch": 36.57,
+      "learning_rate": 0.0002667719298245614,
+      "loss": 0.3136,
+      "step": 2450
+    },
+    {
+      "epoch": 36.58,
+      "learning_rate": 0.00026673684210526313,
+      "loss": 0.1262,
+      "step": 2451
+    },
+    {
+      "epoch": 36.59,
+      "learning_rate": 0.0002667017543859649,
+      "loss": 0.4639,
+      "step": 2452
+    },
+    {
+      "epoch": 36.61,
+      "learning_rate": 0.0002666666666666666,
+      "loss": 0.1867,
+      "step": 2453
+    },
+    {
+      "epoch": 36.62,
+      "learning_rate": 0.00026663157894736843,
+      "loss": 0.0463,
+      "step": 2454
+    },
+    {
+      "epoch": 36.64,
+      "learning_rate": 0.0002665964912280702,
+      "loss": 0.2519,
+      "step": 2455
+    },
+    {
+      "epoch": 36.65,
+      "learning_rate": 0.00026656140350877187,
+      "loss": 0.1951,
+      "step": 2456
+    },
+    {
+      "epoch": 36.67,
+      "learning_rate": 0.0002665263157894737,
+      "loss": 0.0287,
+      "step": 2457
+    },
+    {
+      "epoch": 36.68,
+      "learning_rate": 0.0002664912280701754,
+      "loss": 0.7845,
+      "step": 2458
+    },
+    {
+      "epoch": 36.7,
+      "learning_rate": 0.00026645614035087717,
+      "loss": 0.1309,
+      "step": 2459
+    },
+    {
+      "epoch": 36.71,
+      "learning_rate": 0.0002664210526315789,
+      "loss": 0.1365,
+      "step": 2460
+    },
+    {
+      "epoch": 36.73,
+      "learning_rate": 0.00026638596491228067,
+      "loss": 0.0312,
+      "step": 2461
+    },
+    {
+      "epoch": 36.74,
+      "learning_rate": 0.0002663508771929824,
+      "loss": 0.1098,
+      "step": 2462
+    },
+    {
+      "epoch": 36.76,
+      "learning_rate": 0.00026631578947368417,
+      "loss": 0.4963,
+      "step": 2463
+    },
+    {
+      "epoch": 36.77,
+      "learning_rate": 0.0002662807017543859,
+      "loss": 0.1178,
+      "step": 2464
+    },
+    {
+      "epoch": 36.79,
+      "learning_rate": 0.0002662456140350877,
+      "loss": 0.1513,
+      "step": 2465
+    },
+    {
+      "epoch": 36.8,
+      "learning_rate": 0.00026621052631578947,
+      "loss": 0.2865,
+      "step": 2466
+    },
+    {
+      "epoch": 36.82,
+      "learning_rate": 0.0002661754385964912,
+      "loss": 0.3721,
+      "step": 2467
+    },
+    {
+      "epoch": 36.83,
+      "learning_rate": 0.00026614035087719297,
+      "loss": 0.1291,
+      "step": 2468
+    },
+    {
+      "epoch": 36.85,
+      "learning_rate": 0.0002661052631578947,
+      "loss": 0.2043,
+      "step": 2469
+    },
+    {
+      "epoch": 36.86,
+      "learning_rate": 0.00026607017543859646,
+      "loss": 0.1493,
+      "step": 2470
+    },
+    {
+      "epoch": 36.88,
+      "learning_rate": 0.0002660350877192982,
+      "loss": 0.311,
+      "step": 2471
+    },
+    {
+      "epoch": 36.89,
+      "learning_rate": 0.000266,
+      "loss": 0.3776,
+      "step": 2472
+    },
+    {
+      "epoch": 36.91,
+      "learning_rate": 0.0002659649122807017,
+      "loss": 0.0136,
+      "step": 2473
+    },
+    {
+      "epoch": 36.92,
+      "learning_rate": 0.00026592982456140346,
+      "loss": 0.3089,
+      "step": 2474
+    },
+    {
+      "epoch": 36.94,
+      "learning_rate": 0.0002658947368421052,
+      "loss": 0.1018,
+      "step": 2475
+    },
+    {
+      "epoch": 36.95,
+      "learning_rate": 0.000265859649122807,
+      "loss": 0.2423,
+      "step": 2476
+    },
+    {
+      "epoch": 36.97,
+      "learning_rate": 0.00026582456140350876,
+      "loss": 0.3071,
+      "step": 2477
+    },
+    {
+      "epoch": 36.98,
+      "learning_rate": 0.0002657894736842105,
+      "loss": 0.2843,
+      "step": 2478
+    },
+    {
+      "epoch": 37.0,
+      "learning_rate": 0.00026575438596491226,
+      "loss": 0.4304,
+      "step": 2479
+    },
+    {
+      "epoch": 37.01,
+      "learning_rate": 0.000265719298245614,
+      "loss": 0.2623,
+      "step": 2480
+    },
+    {
+      "epoch": 37.03,
+      "learning_rate": 0.00026568421052631575,
+      "loss": 0.5204,
+      "step": 2481
+    },
+    {
+      "epoch": 37.04,
+      "learning_rate": 0.0002656491228070175,
+      "loss": 0.181,
+      "step": 2482
+    },
+    {
+      "epoch": 37.06,
+      "learning_rate": 0.0002656140350877193,
+      "loss": 0.2256,
+      "step": 2483
+    },
+    {
+      "epoch": 37.07,
+      "learning_rate": 0.00026557894736842105,
+      "loss": 0.6583,
+      "step": 2484
+    },
+    {
+      "epoch": 37.09,
+      "learning_rate": 0.0002655438596491228,
+      "loss": 0.3567,
+      "step": 2485
+    },
+    {
+      "epoch": 37.1,
+      "learning_rate": 0.00026550877192982455,
+      "loss": 0.2477,
+      "step": 2486
+    },
+    {
+      "epoch": 37.12,
+      "learning_rate": 0.0002654736842105263,
+      "loss": 0.4258,
+      "step": 2487
+    },
+    {
+      "epoch": 37.13,
+      "learning_rate": 0.00026543859649122805,
+      "loss": 0.0732,
+      "step": 2488
+    },
+    {
+      "epoch": 37.15,
+      "learning_rate": 0.0002654035087719298,
+      "loss": 0.0374,
+      "step": 2489
+    },
+    {
+      "epoch": 37.16,
+      "learning_rate": 0.00026536842105263155,
+      "loss": 0.1289,
+      "step": 2490
+    },
+    {
+      "epoch": 37.18,
+      "learning_rate": 0.0002653333333333333,
+      "loss": 0.0477,
+      "step": 2491
+    },
+    {
+      "epoch": 37.19,
+      "learning_rate": 0.00026529824561403504,
+      "loss": 0.1553,
+      "step": 2492
+    },
+    {
+      "epoch": 37.21,
+      "learning_rate": 0.0002652631578947368,
+      "loss": 0.1792,
+      "step": 2493
+    },
+    {
+      "epoch": 37.22,
+      "learning_rate": 0.0002652280701754386,
+      "loss": 0.0681,
+      "step": 2494
+    },
+    {
+      "epoch": 37.24,
+      "learning_rate": 0.00026519298245614034,
+      "loss": 0.3027,
+      "step": 2495
+    },
+    {
+      "epoch": 37.25,
+      "learning_rate": 0.0002651578947368421,
+      "loss": 0.2969,
+      "step": 2496
+    },
+    {
+      "epoch": 37.27,
+      "learning_rate": 0.00026512280701754384,
+      "loss": 0.4292,
+      "step": 2497
+    },
+    {
+      "epoch": 37.28,
+      "learning_rate": 0.0002650877192982456,
+      "loss": 0.285,
+      "step": 2498
+    },
+    {
+      "epoch": 37.3,
+      "learning_rate": 0.00026505263157894734,
+      "loss": 1.0064,
+      "step": 2499
+    },
+    {
+      "epoch": 37.31,
+      "learning_rate": 0.0002650175438596491,
+      "loss": 0.2859,
+      "step": 2500
+    },
+    {
+      "epoch": 37.33,
+      "learning_rate": 0.00026498245614035084,
+      "loss": 0.131,
+      "step": 2501
+    },
+    {
+      "epoch": 37.34,
+      "learning_rate": 0.00026494736842105264,
+      "loss": 0.3665,
+      "step": 2502
+    },
+    {
+      "epoch": 37.36,
+      "learning_rate": 0.0002649122807017544,
+      "loss": 0.4487,
+      "step": 2503
+    },
+    {
+      "epoch": 37.37,
+      "learning_rate": 0.0002648771929824561,
+      "loss": 0.722,
+      "step": 2504
+    },
+    {
+      "epoch": 37.39,
+      "learning_rate": 0.00026484210526315783,
+      "loss": 1.3793,
+      "step": 2505
+    },
+    {
+      "epoch": 37.4,
+      "learning_rate": 0.00026480701754385964,
+      "loss": 0.2841,
+      "step": 2506
+    },
+    {
+      "epoch": 37.42,
+      "learning_rate": 0.0002647719298245614,
+      "loss": 0.6317,
+      "step": 2507
+    },
+    {
+      "epoch": 37.43,
+      "learning_rate": 0.00026473684210526313,
+      "loss": 0.4639,
+      "step": 2508
+    },
+    {
+      "epoch": 37.45,
+      "learning_rate": 0.0002647017543859649,
+      "loss": 0.2349,
+      "step": 2509
+    },
+    {
+      "epoch": 37.46,
+      "learning_rate": 0.00026466666666666663,
+      "loss": 0.252,
+      "step": 2510
+    },
+    {
+      "epoch": 37.48,
+      "learning_rate": 0.0002646315789473684,
+      "loss": 0.2653,
+      "step": 2511
+    },
+    {
+      "epoch": 37.49,
+      "learning_rate": 0.00026459649122807013,
+      "loss": 0.1756,
+      "step": 2512
+    },
+    {
+      "epoch": 37.51,
+      "learning_rate": 0.00026456140350877193,
+      "loss": 0.6025,
+      "step": 2513
+    },
+    {
+      "epoch": 37.52,
+      "learning_rate": 0.0002645263157894737,
+      "loss": 0.1538,
+      "step": 2514
+    },
+    {
+      "epoch": 37.54,
+      "learning_rate": 0.00026449122807017543,
+      "loss": 0.7887,
+      "step": 2515
+    },
+    {
+      "epoch": 37.55,
+      "learning_rate": 0.0002644561403508772,
+      "loss": 0.1094,
+      "step": 2516
+    },
+    {
+      "epoch": 37.57,
+      "learning_rate": 0.0002644210526315789,
+      "loss": 0.1136,
+      "step": 2517
+    },
+    {
+      "epoch": 37.58,
+      "learning_rate": 0.0002643859649122807,
+      "loss": 0.054,
+      "step": 2518
+    },
+    {
+      "epoch": 37.59,
+      "learning_rate": 0.0002643508771929824,
+      "loss": 0.4868,
+      "step": 2519
+    },
+    {
+      "epoch": 37.61,
+      "learning_rate": 0.0002643157894736842,
+      "loss": 0.2704,
+      "step": 2520
+    },
+    {
+      "epoch": 37.62,
+      "learning_rate": 0.0002642807017543859,
+      "loss": 0.0936,
+      "step": 2521
+    },
+    {
+      "epoch": 37.64,
+      "learning_rate": 0.00026424561403508767,
+      "loss": 0.7214,
+      "step": 2522
+    },
+    {
+      "epoch": 37.65,
+      "learning_rate": 0.0002642105263157894,
+      "loss": 0.0592,
+      "step": 2523
+    },
+    {
+      "epoch": 37.67,
+      "learning_rate": 0.0002641754385964912,
+      "loss": 0.0877,
+      "step": 2524
+    },
+    {
+      "epoch": 37.68,
+      "learning_rate": 0.00026414035087719297,
+      "loss": 0.2035,
+      "step": 2525
+    },
+    {
+      "epoch": 37.7,
+      "learning_rate": 0.0002641052631578947,
+      "loss": 0.0444,
+      "step": 2526
+    },
+    {
+      "epoch": 37.71,
+      "learning_rate": 0.00026407017543859647,
+      "loss": 0.1678,
+      "step": 2527
+    },
+    {
+      "epoch": 37.73,
+      "learning_rate": 0.0002640350877192982,
+      "loss": 0.037,
+      "step": 2528
+    },
+    {
+      "epoch": 37.74,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 0.0603,
+      "step": 2529
+    },
+    {
+      "epoch": 37.76,
+      "learning_rate": 0.0002639649122807017,
+      "loss": 0.2909,
+      "step": 2530
+    },
+    {
+      "epoch": 37.77,
+      "learning_rate": 0.00026392982456140346,
+      "loss": 0.1873,
+      "step": 2531
+    },
+    {
+      "epoch": 37.79,
+      "learning_rate": 0.00026389473684210527,
+      "loss": 0.0805,
+      "step": 2532
+    },
+    {
+      "epoch": 37.8,
+      "learning_rate": 0.000263859649122807,
+      "loss": 0.1528,
+      "step": 2533
+    },
+    {
+      "epoch": 37.82,
+      "learning_rate": 0.00026382456140350876,
+      "loss": 0.1128,
+      "step": 2534
+    },
+    {
+      "epoch": 37.83,
+      "learning_rate": 0.0002637894736842105,
+      "loss": 0.0441,
+      "step": 2535
+    },
+    {
+      "epoch": 37.85,
+      "learning_rate": 0.00026375438596491226,
+      "loss": 0.3716,
+      "step": 2536
+    },
+    {
+      "epoch": 37.86,
+      "learning_rate": 0.000263719298245614,
+      "loss": 0.2316,
+      "step": 2537
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 0.00026368421052631576,
+      "loss": 0.397,
+      "step": 2538
+    },
+    {
+      "epoch": 37.89,
+      "learning_rate": 0.0002636491228070175,
+      "loss": 0.0878,
+      "step": 2539
+    },
+    {
+      "epoch": 37.91,
+      "learning_rate": 0.00026361403508771926,
+      "loss": 0.3564,
+      "step": 2540
+    },
+    {
+      "epoch": 37.92,
+      "learning_rate": 0.000263578947368421,
+      "loss": 0.2722,
+      "step": 2541
+    },
+    {
+      "epoch": 37.94,
+      "learning_rate": 0.00026354385964912275,
+      "loss": 0.2298,
+      "step": 2542
+    },
+    {
+      "epoch": 37.95,
+      "learning_rate": 0.00026350877192982456,
+      "loss": 0.2963,
+      "step": 2543
+    },
+    {
+      "epoch": 37.97,
+      "learning_rate": 0.0002634736842105263,
+      "loss": 0.2939,
+      "step": 2544
+    },
+    {
+      "epoch": 37.98,
+      "learning_rate": 0.00026343859649122805,
+      "loss": 0.1811,
+      "step": 2545
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 0.0002634035087719298,
+      "loss": 0.2587,
+      "step": 2546
+    },
+    {
+      "epoch": 38.01,
+      "learning_rate": 0.00026336842105263155,
+      "loss": 0.335,
+      "step": 2547
+    },
+    {
+      "epoch": 38.03,
+      "learning_rate": 0.0002633333333333333,
+      "loss": 0.3061,
+      "step": 2548
+    },
+    {
+      "epoch": 38.04,
+      "learning_rate": 0.00026329824561403505,
+      "loss": 0.2655,
+      "step": 2549
+    },
+    {
+      "epoch": 38.06,
+      "learning_rate": 0.00026326315789473685,
+      "loss": 0.2371,
+      "step": 2550
+    },
+    {
+      "epoch": 38.07,
+      "learning_rate": 0.0002632280701754386,
+      "loss": 0.5286,
+      "step": 2551
+    },
+    {
+      "epoch": 38.09,
+      "learning_rate": 0.0002631929824561403,
+      "loss": 0.3128,
+      "step": 2552
+    },
+    {
+      "epoch": 38.1,
+      "learning_rate": 0.00026315789473684205,
+      "loss": 0.0212,
+      "step": 2553
+    },
+    {
+      "epoch": 38.12,
+      "learning_rate": 0.00026312280701754385,
+      "loss": 0.5485,
+      "step": 2554
+    },
+    {
+      "epoch": 38.13,
+      "learning_rate": 0.0002630877192982456,
+      "loss": 0.2955,
+      "step": 2555
+    },
+    {
+      "epoch": 38.15,
+      "learning_rate": 0.00026305263157894735,
+      "loss": 0.0294,
+      "step": 2556
+    },
+    {
+      "epoch": 38.16,
+      "learning_rate": 0.0002630175438596491,
+      "loss": 0.2217,
+      "step": 2557
+    },
+    {
+      "epoch": 38.18,
+      "learning_rate": 0.00026298245614035084,
+      "loss": 0.2246,
+      "step": 2558
+    },
+    {
+      "epoch": 38.19,
+      "learning_rate": 0.0002629473684210526,
+      "loss": 0.4677,
+      "step": 2559
+    },
+    {
+      "epoch": 38.21,
+      "learning_rate": 0.00026291228070175434,
+      "loss": 0.1354,
+      "step": 2560
+    },
+    {
+      "epoch": 38.22,
+      "learning_rate": 0.00026287719298245614,
+      "loss": 0.0242,
+      "step": 2561
+    },
+    {
+      "epoch": 38.24,
+      "learning_rate": 0.0002628421052631579,
+      "loss": 0.1915,
+      "step": 2562
+    },
+    {
+      "epoch": 38.25,
+      "learning_rate": 0.00026280701754385964,
+      "loss": 0.494,
+      "step": 2563
+    },
+    {
+      "epoch": 38.27,
+      "learning_rate": 0.0002627719298245614,
+      "loss": 0.1895,
+      "step": 2564
+    },
+    {
+      "epoch": 38.28,
+      "learning_rate": 0.00026273684210526314,
+      "loss": 0.0574,
+      "step": 2565
+    },
+    {
+      "epoch": 38.3,
+      "learning_rate": 0.0002627017543859649,
+      "loss": 0.085,
+      "step": 2566
+    },
+    {
+      "epoch": 38.31,
+      "learning_rate": 0.00026266666666666664,
+      "loss": 0.317,
+      "step": 2567
+    },
+    {
+      "epoch": 38.33,
+      "learning_rate": 0.0002626315789473684,
+      "loss": 0.1981,
+      "step": 2568
+    },
+    {
+      "epoch": 38.34,
+      "learning_rate": 0.00026259649122807013,
+      "loss": 0.2723,
+      "step": 2569
+    },
+    {
+      "epoch": 38.36,
+      "learning_rate": 0.0002625614035087719,
+      "loss": 0.4135,
+      "step": 2570
+    },
+    {
+      "epoch": 38.37,
+      "learning_rate": 0.00026252631578947363,
+      "loss": 0.3537,
+      "step": 2571
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 0.00026249122807017543,
+      "loss": 0.223,
+      "step": 2572
+    },
+    {
+      "epoch": 38.4,
+      "learning_rate": 0.0002624561403508772,
+      "loss": 0.0888,
+      "step": 2573
+    },
+    {
+      "epoch": 38.42,
+      "learning_rate": 0.00026242105263157893,
+      "loss": 0.0888,
+      "step": 2574
+    },
+    {
+      "epoch": 38.43,
+      "learning_rate": 0.0002623859649122807,
+      "loss": 0.2686,
+      "step": 2575
+    },
+    {
+      "epoch": 38.45,
+      "learning_rate": 0.00026235087719298243,
+      "loss": 0.291,
+      "step": 2576
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 0.0002623157894736842,
+      "loss": 0.0836,
+      "step": 2577
+    },
+    {
+      "epoch": 38.48,
+      "learning_rate": 0.00026228070175438593,
+      "loss": 0.2366,
+      "step": 2578
+    },
+    {
+      "epoch": 38.49,
+      "learning_rate": 0.0002622456140350877,
+      "loss": 0.0858,
+      "step": 2579
+    },
+    {
+      "epoch": 38.51,
+      "learning_rate": 0.0002622105263157895,
+      "loss": 0.5286,
+      "step": 2580
+    },
+    {
+      "epoch": 38.52,
+      "learning_rate": 0.00026217543859649123,
+      "loss": 0.5125,
+      "step": 2581
+    },
+    {
+      "epoch": 38.54,
+      "learning_rate": 0.000262140350877193,
+      "loss": 0.0903,
+      "step": 2582
+    },
+    {
+      "epoch": 38.55,
+      "learning_rate": 0.0002621052631578947,
+      "loss": 0.457,
+      "step": 2583
+    },
+    {
+      "epoch": 38.57,
+      "learning_rate": 0.0002620701754385965,
+      "loss": 0.6476,
+      "step": 2584
+    },
+    {
+      "epoch": 38.58,
+      "learning_rate": 0.0002620350877192982,
+      "loss": 0.0249,
+      "step": 2585
+    },
+    {
+      "epoch": 38.59,
+      "learning_rate": 0.00026199999999999997,
+      "loss": 0.2783,
+      "step": 2586
+    },
+    {
+      "epoch": 38.61,
+      "learning_rate": 0.0002619649122807017,
+      "loss": 0.2174,
+      "step": 2587
+    },
+    {
+      "epoch": 38.62,
+      "learning_rate": 0.00026192982456140347,
+      "loss": 0.2787,
+      "step": 2588
+    },
+    {
+      "epoch": 38.64,
+      "learning_rate": 0.0002618947368421052,
+      "loss": 0.073,
+      "step": 2589
+    },
+    {
+      "epoch": 38.65,
+      "learning_rate": 0.00026185964912280697,
+      "loss": 0.0238,
+      "step": 2590
+    },
+    {
+      "epoch": 38.67,
+      "learning_rate": 0.00026182456140350877,
+      "loss": 0.0319,
+      "step": 2591
+    },
+    {
+      "epoch": 38.68,
+      "learning_rate": 0.0002617894736842105,
+      "loss": 0.1248,
+      "step": 2592
+    },
+    {
+      "epoch": 38.7,
+      "learning_rate": 0.00026175438596491227,
+      "loss": 0.049,
+      "step": 2593
+    },
+    {
+      "epoch": 38.71,
+      "learning_rate": 0.000261719298245614,
+      "loss": 0.3386,
+      "step": 2594
+    },
+    {
+      "epoch": 38.73,
+      "learning_rate": 0.00026168421052631576,
+      "loss": 0.3199,
+      "step": 2595
+    },
+    {
+      "epoch": 38.74,
+      "learning_rate": 0.0002616491228070175,
+      "loss": 0.1528,
+      "step": 2596
+    },
+    {
+      "epoch": 38.76,
+      "learning_rate": 0.00026161403508771926,
+      "loss": 0.1884,
+      "step": 2597
+    },
+    {
+      "epoch": 38.77,
+      "learning_rate": 0.00026157894736842107,
+      "loss": 0.3862,
+      "step": 2598
+    },
+    {
+      "epoch": 38.79,
+      "learning_rate": 0.0002615438596491228,
+      "loss": 0.1399,
+      "step": 2599
+    },
+    {
+      "epoch": 38.8,
+      "learning_rate": 0.0002615087719298245,
+      "loss": 0.2141,
+      "step": 2600
+    },
+    {
+      "epoch": 38.8,
+      "eval_accuracy": 0.8262359275575134,
+      "eval_f1": 0.8275994930387057,
+      "eval_loss": 0.7550894618034363,
+      "eval_runtime": 345.2152,
+      "eval_samples_per_second": 11.836,
+      "eval_steps_per_second": 0.742,
+      "step": 2600
+    },
+    {
+      "epoch": 38.82,
+      "learning_rate": 0.00026147368421052626,
+      "loss": 0.0499,
+      "step": 2601
+    },
+    {
+      "epoch": 38.83,
+      "learning_rate": 0.00026143859649122806,
+      "loss": 0.3111,
+      "step": 2602
+    },
+    {
+      "epoch": 38.85,
+      "learning_rate": 0.0002614035087719298,
+      "loss": 0.1958,
+      "step": 2603
+    },
+    {
+      "epoch": 38.86,
+      "learning_rate": 0.00026136842105263156,
+      "loss": 0.0505,
+      "step": 2604
+    },
+    {
+      "epoch": 38.88,
+      "learning_rate": 0.0002613333333333333,
+      "loss": 0.1559,
+      "step": 2605
+    },
+    {
+      "epoch": 38.89,
+      "learning_rate": 0.00026129824561403506,
+      "loss": 0.0148,
+      "step": 2606
+    },
+    {
+      "epoch": 38.91,
+      "learning_rate": 0.0002612631578947368,
+      "loss": 0.3198,
+      "step": 2607
+    },
+    {
+      "epoch": 38.92,
+      "learning_rate": 0.00026122807017543855,
+      "loss": 0.5153,
+      "step": 2608
+    },
+    {
+      "epoch": 38.94,
+      "learning_rate": 0.00026119298245614036,
+      "loss": 0.2361,
+      "step": 2609
+    },
+    {
+      "epoch": 38.95,
+      "learning_rate": 0.0002611578947368421,
+      "loss": 0.3028,
+      "step": 2610
+    },
+    {
+      "epoch": 38.97,
+      "learning_rate": 0.00026112280701754385,
+      "loss": 0.3361,
+      "step": 2611
+    },
+    {
+      "epoch": 38.98,
+      "learning_rate": 0.0002610877192982456,
+      "loss": 0.0152,
+      "step": 2612
+    },
+    {
+      "epoch": 39.0,
+      "learning_rate": 0.00026105263157894735,
+      "loss": 0.0569,
+      "step": 2613
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 0.0002610175438596491,
+      "loss": 0.2744,
+      "step": 2614
+    },
+    {
+      "epoch": 39.03,
+      "learning_rate": 0.00026098245614035085,
+      "loss": 0.2632,
+      "step": 2615
+    },
+    {
+      "epoch": 39.04,
+      "learning_rate": 0.0002609473684210526,
+      "loss": 0.0301,
+      "step": 2616
+    },
+    {
+      "epoch": 39.06,
+      "learning_rate": 0.00026091228070175435,
+      "loss": 0.2426,
+      "step": 2617
+    },
+    {
+      "epoch": 39.07,
+      "learning_rate": 0.0002608771929824561,
+      "loss": 0.4076,
+      "step": 2618
+    },
+    {
+      "epoch": 39.09,
+      "learning_rate": 0.00026084210526315784,
+      "loss": 0.2153,
+      "step": 2619
+    },
+    {
+      "epoch": 39.1,
+      "learning_rate": 0.00026080701754385965,
+      "loss": 0.0533,
+      "step": 2620
+    },
+    {
+      "epoch": 39.12,
+      "learning_rate": 0.0002607719298245614,
+      "loss": 0.4746,
+      "step": 2621
+    },
+    {
+      "epoch": 39.13,
+      "learning_rate": 0.00026073684210526314,
+      "loss": 0.71,
+      "step": 2622
+    },
+    {
+      "epoch": 39.15,
+      "learning_rate": 0.0002607017543859649,
+      "loss": 0.2725,
+      "step": 2623
+    },
+    {
+      "epoch": 39.16,
+      "learning_rate": 0.00026066666666666664,
+      "loss": 0.1129,
+      "step": 2624
+    },
+    {
+      "epoch": 39.18,
+      "learning_rate": 0.0002606315789473684,
+      "loss": 0.2319,
+      "step": 2625
+    },
+    {
+      "epoch": 39.19,
+      "learning_rate": 0.00026059649122807014,
+      "loss": 0.1732,
+      "step": 2626
+    },
+    {
+      "epoch": 39.21,
+      "learning_rate": 0.0002605614035087719,
+      "loss": 0.3672,
+      "step": 2627
+    },
+    {
+      "epoch": 39.22,
+      "learning_rate": 0.0002605263157894737,
+      "loss": 0.2615,
+      "step": 2628
+    },
+    {
+      "epoch": 39.24,
+      "learning_rate": 0.00026049122807017544,
+      "loss": 0.2643,
+      "step": 2629
+    },
+    {
+      "epoch": 39.25,
+      "learning_rate": 0.0002604561403508772,
+      "loss": 0.3267,
+      "step": 2630
+    },
+    {
+      "epoch": 39.27,
+      "learning_rate": 0.0002604210526315789,
+      "loss": 0.5538,
+      "step": 2631
+    },
+    {
+      "epoch": 39.28,
+      "learning_rate": 0.0002603859649122807,
+      "loss": 0.1607,
+      "step": 2632
+    },
+    {
+      "epoch": 39.3,
+      "learning_rate": 0.00026035087719298244,
+      "loss": 0.3511,
+      "step": 2633
+    },
+    {
+      "epoch": 39.31,
+      "learning_rate": 0.0002603157894736842,
+      "loss": 0.0744,
+      "step": 2634
+    },
+    {
+      "epoch": 39.33,
+      "learning_rate": 0.00026028070175438593,
+      "loss": 0.4115,
+      "step": 2635
+    },
+    {
+      "epoch": 39.34,
+      "learning_rate": 0.0002602456140350877,
+      "loss": 0.189,
+      "step": 2636
+    },
+    {
+      "epoch": 39.36,
+      "learning_rate": 0.00026021052631578943,
+      "loss": 0.437,
+      "step": 2637
+    },
+    {
+      "epoch": 39.37,
+      "learning_rate": 0.0002601754385964912,
+      "loss": 0.1349,
+      "step": 2638
+    },
+    {
+      "epoch": 39.39,
+      "learning_rate": 0.000260140350877193,
+      "loss": 0.1645,
+      "step": 2639
+    },
+    {
+      "epoch": 39.4,
+      "learning_rate": 0.00026010526315789473,
+      "loss": 0.7744,
+      "step": 2640
+    },
+    {
+      "epoch": 39.42,
+      "learning_rate": 0.0002600701754385965,
+      "loss": 0.0679,
+      "step": 2641
+    },
+    {
+      "epoch": 39.43,
+      "learning_rate": 0.00026003508771929823,
+      "loss": 0.1496,
+      "step": 2642
+    },
+    {
+      "epoch": 39.45,
+      "learning_rate": 0.00026,
+      "loss": 0.056,
+      "step": 2643
+    },
+    {
+      "epoch": 39.46,
+      "learning_rate": 0.0002599649122807017,
+      "loss": 0.3514,
+      "step": 2644
+    },
+    {
+      "epoch": 39.48,
+      "learning_rate": 0.0002599298245614035,
+      "loss": 0.112,
+      "step": 2645
+    },
+    {
+      "epoch": 39.49,
+      "learning_rate": 0.0002598947368421053,
+      "loss": 0.0599,
+      "step": 2646
+    },
+    {
+      "epoch": 39.51,
+      "learning_rate": 0.000259859649122807,
+      "loss": 0.4294,
+      "step": 2647
+    },
+    {
+      "epoch": 39.52,
+      "learning_rate": 0.0002598245614035087,
+      "loss": 0.1908,
+      "step": 2648
+    },
+    {
+      "epoch": 39.54,
+      "learning_rate": 0.00025978947368421047,
+      "loss": 0.4259,
+      "step": 2649
+    },
+    {
+      "epoch": 39.55,
+      "learning_rate": 0.00025975438596491227,
+      "loss": 0.2764,
+      "step": 2650
+    },
+    {
+      "epoch": 39.57,
+      "learning_rate": 0.000259719298245614,
+      "loss": 0.2793,
+      "step": 2651
+    },
+    {
+      "epoch": 39.58,
+      "learning_rate": 0.00025968421052631577,
+      "loss": 0.2748,
+      "step": 2652
+    },
+    {
+      "epoch": 39.59,
+      "learning_rate": 0.0002596491228070175,
+      "loss": 0.0622,
+      "step": 2653
+    },
+    {
+      "epoch": 39.61,
+      "learning_rate": 0.00025961403508771927,
+      "loss": 0.2978,
+      "step": 2654
+    },
+    {
+      "epoch": 39.62,
+      "learning_rate": 0.000259578947368421,
+      "loss": 0.0851,
+      "step": 2655
+    },
+    {
+      "epoch": 39.64,
+      "learning_rate": 0.00025954385964912277,
+      "loss": 0.2829,
+      "step": 2656
+    },
+    {
+      "epoch": 39.65,
+      "learning_rate": 0.00025950877192982457,
+      "loss": 0.1534,
+      "step": 2657
+    },
+    {
+      "epoch": 39.67,
+      "learning_rate": 0.0002594736842105263,
+      "loss": 0.4885,
+      "step": 2658
+    },
+    {
+      "epoch": 39.68,
+      "learning_rate": 0.00025943859649122807,
+      "loss": 0.3212,
+      "step": 2659
+    },
+    {
+      "epoch": 39.7,
+      "learning_rate": 0.0002594035087719298,
+      "loss": 0.054,
+      "step": 2660
+    },
+    {
+      "epoch": 39.71,
+      "learning_rate": 0.00025936842105263156,
+      "loss": 0.1702,
+      "step": 2661
+    },
+    {
+      "epoch": 39.73,
+      "learning_rate": 0.0002593333333333333,
+      "loss": 0.068,
+      "step": 2662
+    },
+    {
+      "epoch": 39.74,
+      "learning_rate": 0.00025929824561403506,
+      "loss": 0.4455,
+      "step": 2663
+    },
+    {
+      "epoch": 39.76,
+      "learning_rate": 0.0002592631578947368,
+      "loss": 0.5979,
+      "step": 2664
+    },
+    {
+      "epoch": 39.77,
+      "learning_rate": 0.00025922807017543856,
+      "loss": 0.1285,
+      "step": 2665
+    },
+    {
+      "epoch": 39.79,
+      "learning_rate": 0.0002591929824561403,
+      "loss": 0.259,
+      "step": 2666
+    },
+    {
+      "epoch": 39.8,
+      "learning_rate": 0.00025915789473684206,
+      "loss": 0.2704,
+      "step": 2667
+    },
+    {
+      "epoch": 39.82,
+      "learning_rate": 0.0002591228070175438,
+      "loss": 0.5382,
+      "step": 2668
+    },
+    {
+      "epoch": 39.83,
+      "learning_rate": 0.0002590877192982456,
+      "loss": 0.1596,
+      "step": 2669
+    },
+    {
+      "epoch": 39.85,
+      "learning_rate": 0.00025905263157894736,
+      "loss": 0.1808,
+      "step": 2670
+    },
+    {
+      "epoch": 39.86,
+      "learning_rate": 0.0002590175438596491,
+      "loss": 0.1089,
+      "step": 2671
+    },
+    {
+      "epoch": 39.88,
+      "learning_rate": 0.00025898245614035085,
+      "loss": 0.1954,
+      "step": 2672
+    },
+    {
+      "epoch": 39.89,
+      "learning_rate": 0.0002589473684210526,
+      "loss": 0.1062,
+      "step": 2673
+    },
+    {
+      "epoch": 39.91,
+      "learning_rate": 0.00025891228070175435,
+      "loss": 0.032,
+      "step": 2674
+    },
+    {
+      "epoch": 39.92,
+      "learning_rate": 0.0002588771929824561,
+      "loss": 0.2953,
+      "step": 2675
+    },
+    {
+      "epoch": 39.94,
+      "learning_rate": 0.0002588421052631579,
+      "loss": 0.2644,
+      "step": 2676
+    },
+    {
+      "epoch": 39.95,
+      "learning_rate": 0.00025880701754385965,
+      "loss": 0.0182,
+      "step": 2677
+    },
+    {
+      "epoch": 39.97,
+      "learning_rate": 0.0002587719298245614,
+      "loss": 0.0144,
+      "step": 2678
+    },
+    {
+      "epoch": 39.98,
+      "learning_rate": 0.0002587368421052631,
+      "loss": 0.0845,
+      "step": 2679
+    },
+    {
+      "epoch": 40.0,
+      "learning_rate": 0.0002587017543859649,
+      "loss": 0.3278,
+      "step": 2680
+    },
+    {
+      "epoch": 40.01,
+      "learning_rate": 0.00025866666666666665,
+      "loss": 0.1518,
+      "step": 2681
+    },
+    {
+      "epoch": 40.03,
+      "learning_rate": 0.0002586315789473684,
+      "loss": 0.3586,
+      "step": 2682
+    },
+    {
+      "epoch": 40.04,
+      "learning_rate": 0.00025859649122807015,
+      "loss": 0.0538,
+      "step": 2683
+    },
+    {
+      "epoch": 40.06,
+      "learning_rate": 0.0002585614035087719,
+      "loss": 0.211,
+      "step": 2684
+    },
+    {
+      "epoch": 40.07,
+      "learning_rate": 0.00025852631578947364,
+      "loss": 0.0149,
+      "step": 2685
+    },
+    {
+      "epoch": 40.09,
+      "learning_rate": 0.0002584912280701754,
+      "loss": 0.0996,
+      "step": 2686
+    },
+    {
+      "epoch": 40.1,
+      "learning_rate": 0.0002584561403508772,
+      "loss": 0.0451,
+      "step": 2687
+    },
+    {
+      "epoch": 40.12,
+      "learning_rate": 0.00025842105263157894,
+      "loss": 0.1934,
+      "step": 2688
+    },
+    {
+      "epoch": 40.13,
+      "learning_rate": 0.0002583859649122807,
+      "loss": 0.1372,
+      "step": 2689
+    },
+    {
+      "epoch": 40.15,
+      "learning_rate": 0.00025835087719298244,
+      "loss": 0.3718,
+      "step": 2690
+    },
+    {
+      "epoch": 40.16,
+      "learning_rate": 0.0002583157894736842,
+      "loss": 0.0081,
+      "step": 2691
+    },
+    {
+      "epoch": 40.18,
+      "learning_rate": 0.00025828070175438594,
+      "loss": 0.4675,
+      "step": 2692
+    },
+    {
+      "epoch": 40.19,
+      "learning_rate": 0.0002582456140350877,
+      "loss": 0.0657,
+      "step": 2693
+    },
+    {
+      "epoch": 40.21,
+      "learning_rate": 0.0002582105263157895,
+      "loss": 0.0938,
+      "step": 2694
+    },
+    {
+      "epoch": 40.22,
+      "learning_rate": 0.00025817543859649124,
+      "loss": 0.3227,
+      "step": 2695
+    },
+    {
+      "epoch": 40.24,
+      "learning_rate": 0.00025814035087719293,
+      "loss": 0.4954,
+      "step": 2696
+    },
+    {
+      "epoch": 40.25,
+      "learning_rate": 0.0002581052631578947,
+      "loss": 0.0188,
+      "step": 2697
+    },
+    {
+      "epoch": 40.27,
+      "learning_rate": 0.0002580701754385965,
+      "loss": 0.1594,
+      "step": 2698
+    },
+    {
+      "epoch": 40.28,
+      "learning_rate": 0.00025803508771929823,
+      "loss": 0.2826,
+      "step": 2699
+    },
+    {
+      "epoch": 40.3,
+      "learning_rate": 0.000258,
+      "loss": 0.135,
+      "step": 2700
+    },
+    {
+      "epoch": 40.31,
+      "learning_rate": 0.00025796491228070173,
+      "loss": 0.038,
+      "step": 2701
+    },
+    {
+      "epoch": 40.33,
+      "learning_rate": 0.0002579298245614035,
+      "loss": 0.2259,
+      "step": 2702
+    },
+    {
+      "epoch": 40.34,
+      "learning_rate": 0.00025789473684210523,
+      "loss": 0.0391,
+      "step": 2703
+    },
+    {
+      "epoch": 40.36,
+      "learning_rate": 0.000257859649122807,
+      "loss": 0.2152,
+      "step": 2704
+    },
+    {
+      "epoch": 40.37,
+      "learning_rate": 0.0002578245614035087,
+      "loss": 0.1713,
+      "step": 2705
+    },
+    {
+      "epoch": 40.39,
+      "learning_rate": 0.00025778947368421053,
+      "loss": 0.3175,
+      "step": 2706
+    },
+    {
+      "epoch": 40.4,
+      "learning_rate": 0.0002577543859649123,
+      "loss": 0.1026,
+      "step": 2707
+    },
+    {
+      "epoch": 40.42,
+      "learning_rate": 0.00025771929824561403,
+      "loss": 0.0078,
+      "step": 2708
+    },
+    {
+      "epoch": 40.43,
+      "learning_rate": 0.0002576842105263158,
+      "loss": 0.1131,
+      "step": 2709
+    },
+    {
+      "epoch": 40.45,
+      "learning_rate": 0.0002576491228070175,
+      "loss": 0.209,
+      "step": 2710
+    },
+    {
+      "epoch": 40.46,
+      "learning_rate": 0.0002576140350877193,
+      "loss": 0.161,
+      "step": 2711
+    },
+    {
+      "epoch": 40.48,
+      "learning_rate": 0.000257578947368421,
+      "loss": 0.096,
+      "step": 2712
+    },
+    {
+      "epoch": 40.49,
+      "learning_rate": 0.00025754385964912277,
+      "loss": 0.1569,
+      "step": 2713
+    },
+    {
+      "epoch": 40.51,
+      "learning_rate": 0.0002575087719298245,
+      "loss": 0.1204,
+      "step": 2714
+    },
+    {
+      "epoch": 40.52,
+      "learning_rate": 0.00025747368421052627,
+      "loss": 0.3505,
+      "step": 2715
+    },
+    {
+      "epoch": 40.54,
+      "learning_rate": 0.000257438596491228,
+      "loss": 0.172,
+      "step": 2716
+    },
+    {
+      "epoch": 40.55,
+      "learning_rate": 0.0002574035087719298,
+      "loss": 0.0452,
+      "step": 2717
+    },
+    {
+      "epoch": 40.57,
+      "learning_rate": 0.00025736842105263157,
+      "loss": 0.2946,
+      "step": 2718
+    },
+    {
+      "epoch": 40.58,
+      "learning_rate": 0.0002573333333333333,
+      "loss": 0.0231,
+      "step": 2719
+    },
+    {
+      "epoch": 40.59,
+      "learning_rate": 0.00025729824561403507,
+      "loss": 0.1766,
+      "step": 2720
+    },
+    {
+      "epoch": 40.61,
+      "learning_rate": 0.0002572631578947368,
+      "loss": 0.2658,
+      "step": 2721
+    },
+    {
+      "epoch": 40.62,
+      "learning_rate": 0.00025722807017543856,
+      "loss": 0.2161,
+      "step": 2722
+    },
+    {
+      "epoch": 40.64,
+      "learning_rate": 0.0002571929824561403,
+      "loss": 0.0936,
+      "step": 2723
+    },
+    {
+      "epoch": 40.65,
+      "learning_rate": 0.0002571578947368421,
+      "loss": 0.1312,
+      "step": 2724
+    },
+    {
+      "epoch": 40.67,
+      "learning_rate": 0.00025712280701754386,
+      "loss": 0.1671,
+      "step": 2725
+    },
+    {
+      "epoch": 40.68,
+      "learning_rate": 0.0002570877192982456,
+      "loss": 0.1141,
+      "step": 2726
+    },
+    {
+      "epoch": 40.7,
+      "learning_rate": 0.0002570526315789473,
+      "loss": 0.1463,
+      "step": 2727
+    },
+    {
+      "epoch": 40.71,
+      "learning_rate": 0.0002570175438596491,
+      "loss": 0.2518,
+      "step": 2728
+    },
+    {
+      "epoch": 40.73,
+      "learning_rate": 0.00025698245614035086,
+      "loss": 0.0229,
+      "step": 2729
+    },
+    {
+      "epoch": 40.74,
+      "learning_rate": 0.0002569473684210526,
+      "loss": 0.0613,
+      "step": 2730
+    },
+    {
+      "epoch": 40.76,
+      "learning_rate": 0.00025691228070175436,
+      "loss": 0.1352,
+      "step": 2731
+    },
+    {
+      "epoch": 40.77,
+      "learning_rate": 0.0002568771929824561,
+      "loss": 0.1684,
+      "step": 2732
+    },
+    {
+      "epoch": 40.79,
+      "learning_rate": 0.00025684210526315786,
+      "loss": 0.2984,
+      "step": 2733
+    },
+    {
+      "epoch": 40.8,
+      "learning_rate": 0.0002568070175438596,
+      "loss": 0.2261,
+      "step": 2734
+    },
+    {
+      "epoch": 40.82,
+      "learning_rate": 0.0002567719298245614,
+      "loss": 0.0484,
+      "step": 2735
+    },
+    {
+      "epoch": 40.83,
+      "learning_rate": 0.00025673684210526316,
+      "loss": 0.4353,
+      "step": 2736
+    },
+    {
+      "epoch": 40.85,
+      "learning_rate": 0.0002567017543859649,
+      "loss": 0.1866,
+      "step": 2737
+    },
+    {
+      "epoch": 40.86,
+      "learning_rate": 0.00025666666666666665,
+      "loss": 0.2467,
+      "step": 2738
+    },
+    {
+      "epoch": 40.88,
+      "learning_rate": 0.0002566315789473684,
+      "loss": 0.0329,
+      "step": 2739
+    },
+    {
+      "epoch": 40.89,
+      "learning_rate": 0.00025659649122807015,
+      "loss": 0.108,
+      "step": 2740
+    },
+    {
+      "epoch": 40.91,
+      "learning_rate": 0.0002565614035087719,
+      "loss": 0.025,
+      "step": 2741
+    },
+    {
+      "epoch": 40.92,
+      "learning_rate": 0.00025652631578947365,
+      "loss": 0.1324,
+      "step": 2742
+    },
+    {
+      "epoch": 40.94,
+      "learning_rate": 0.00025649122807017545,
+      "loss": 0.2477,
+      "step": 2743
+    },
+    {
+      "epoch": 40.95,
+      "learning_rate": 0.00025645614035087715,
+      "loss": 0.0161,
+      "step": 2744
+    },
+    {
+      "epoch": 40.97,
+      "learning_rate": 0.0002564210526315789,
+      "loss": 0.1783,
+      "step": 2745
+    },
+    {
+      "epoch": 40.98,
+      "learning_rate": 0.0002563859649122807,
+      "loss": 0.0191,
+      "step": 2746
+    },
+    {
+      "epoch": 41.0,
+      "learning_rate": 0.00025635087719298245,
+      "loss": 0.1988,
+      "step": 2747
+    },
+    {
+      "epoch": 41.01,
+      "learning_rate": 0.0002563157894736842,
+      "loss": 0.9461,
+      "step": 2748
+    },
+    {
+      "epoch": 41.03,
+      "learning_rate": 0.00025628070175438594,
+      "loss": 0.6253,
+      "step": 2749
+    },
+    {
+      "epoch": 41.04,
+      "learning_rate": 0.0002562456140350877,
+      "loss": 0.3213,
+      "step": 2750
+    },
+    {
+      "epoch": 41.06,
+      "learning_rate": 0.00025621052631578944,
+      "loss": 0.3665,
+      "step": 2751
+    },
+    {
+      "epoch": 41.07,
+      "learning_rate": 0.0002561754385964912,
+      "loss": 0.1163,
+      "step": 2752
+    },
+    {
+      "epoch": 41.09,
+      "learning_rate": 0.00025614035087719294,
+      "loss": 0.0786,
+      "step": 2753
+    },
+    {
+      "epoch": 41.1,
+      "learning_rate": 0.00025610526315789474,
+      "loss": 0.2727,
+      "step": 2754
+    },
+    {
+      "epoch": 41.12,
+      "learning_rate": 0.0002560701754385965,
+      "loss": 0.1635,
+      "step": 2755
+    },
+    {
+      "epoch": 41.13,
+      "learning_rate": 0.00025603508771929824,
+      "loss": 0.029,
+      "step": 2756
+    },
+    {
+      "epoch": 41.15,
+      "learning_rate": 0.000256,
+      "loss": 0.0872,
+      "step": 2757
+    },
+    {
+      "epoch": 41.16,
+      "learning_rate": 0.00025596491228070174,
+      "loss": 0.3625,
+      "step": 2758
+    },
+    {
+      "epoch": 41.18,
+      "learning_rate": 0.0002559298245614035,
+      "loss": 0.0278,
+      "step": 2759
+    },
+    {
+      "epoch": 41.19,
+      "learning_rate": 0.00025589473684210523,
+      "loss": 0.0368,
+      "step": 2760
+    },
+    {
+      "epoch": 41.21,
+      "learning_rate": 0.000255859649122807,
+      "loss": 0.6355,
+      "step": 2761
+    },
+    {
+      "epoch": 41.22,
+      "learning_rate": 0.00025582456140350873,
+      "loss": 0.0689,
+      "step": 2762
+    },
+    {
+      "epoch": 41.24,
+      "learning_rate": 0.0002557894736842105,
+      "loss": 0.0832,
+      "step": 2763
+    },
+    {
+      "epoch": 41.25,
+      "learning_rate": 0.00025575438596491223,
+      "loss": 0.3266,
+      "step": 2764
+    },
+    {
+      "epoch": 41.27,
+      "learning_rate": 0.00025571929824561403,
+      "loss": 0.0166,
+      "step": 2765
+    },
+    {
+      "epoch": 41.28,
+      "learning_rate": 0.0002556842105263158,
+      "loss": 0.3294,
+      "step": 2766
+    },
+    {
+      "epoch": 41.3,
+      "learning_rate": 0.00025564912280701753,
+      "loss": 0.3561,
+      "step": 2767
+    },
+    {
+      "epoch": 41.31,
+      "learning_rate": 0.0002556140350877193,
+      "loss": 0.4908,
+      "step": 2768
+    },
+    {
+      "epoch": 41.33,
+      "learning_rate": 0.00025557894736842103,
+      "loss": 0.3531,
+      "step": 2769
+    },
+    {
+      "epoch": 41.34,
+      "learning_rate": 0.0002555438596491228,
+      "loss": 0.3064,
+      "step": 2770
+    },
+    {
+      "epoch": 41.36,
+      "learning_rate": 0.0002555087719298245,
+      "loss": 0.2952,
+      "step": 2771
+    },
+    {
+      "epoch": 41.37,
+      "learning_rate": 0.00025547368421052633,
+      "loss": 0.0237,
+      "step": 2772
+    },
+    {
+      "epoch": 41.39,
+      "learning_rate": 0.0002554385964912281,
+      "loss": 0.0403,
+      "step": 2773
+    },
+    {
+      "epoch": 41.4,
+      "learning_rate": 0.0002554035087719298,
+      "loss": 0.0116,
+      "step": 2774
+    },
+    {
+      "epoch": 41.42,
+      "learning_rate": 0.0002553684210526315,
+      "loss": 0.0635,
+      "step": 2775
+    },
+    {
+      "epoch": 41.43,
+      "learning_rate": 0.0002553333333333333,
+      "loss": 0.0113,
+      "step": 2776
+    },
+    {
+      "epoch": 41.45,
+      "learning_rate": 0.00025529824561403507,
+      "loss": 0.0259,
+      "step": 2777
+    },
+    {
+      "epoch": 41.46,
+      "learning_rate": 0.0002552631578947368,
+      "loss": 0.025,
+      "step": 2778
+    },
+    {
+      "epoch": 41.48,
+      "learning_rate": 0.00025522807017543857,
+      "loss": 0.1297,
+      "step": 2779
+    },
+    {
+      "epoch": 41.49,
+      "learning_rate": 0.0002551929824561403,
+      "loss": 0.4284,
+      "step": 2780
+    },
+    {
+      "epoch": 41.51,
+      "learning_rate": 0.00025515789473684207,
+      "loss": 0.2254,
+      "step": 2781
+    },
+    {
+      "epoch": 41.52,
+      "learning_rate": 0.0002551228070175438,
+      "loss": 0.246,
+      "step": 2782
+    },
+    {
+      "epoch": 41.54,
+      "learning_rate": 0.0002550877192982456,
+      "loss": 0.1542,
+      "step": 2783
+    },
+    {
+      "epoch": 41.55,
+      "learning_rate": 0.00025505263157894737,
+      "loss": 0.0179,
+      "step": 2784
+    },
+    {
+      "epoch": 41.57,
+      "learning_rate": 0.0002550175438596491,
+      "loss": 0.1614,
+      "step": 2785
+    },
+    {
+      "epoch": 41.58,
+      "learning_rate": 0.00025498245614035087,
+      "loss": 0.4609,
+      "step": 2786
+    },
+    {
+      "epoch": 41.59,
+      "learning_rate": 0.0002549473684210526,
+      "loss": 0.24,
+      "step": 2787
+    },
+    {
+      "epoch": 41.61,
+      "learning_rate": 0.00025491228070175436,
+      "loss": 0.4677,
+      "step": 2788
+    },
+    {
+      "epoch": 41.62,
+      "learning_rate": 0.0002548771929824561,
+      "loss": 0.2245,
+      "step": 2789
+    },
+    {
+      "epoch": 41.64,
+      "learning_rate": 0.00025484210526315786,
+      "loss": 0.1657,
+      "step": 2790
+    },
+    {
+      "epoch": 41.65,
+      "learning_rate": 0.00025480701754385966,
+      "loss": 0.0554,
+      "step": 2791
+    },
+    {
+      "epoch": 41.67,
+      "learning_rate": 0.00025477192982456136,
+      "loss": 0.0357,
+      "step": 2792
+    },
+    {
+      "epoch": 41.68,
+      "learning_rate": 0.0002547368421052631,
+      "loss": 0.0624,
+      "step": 2793
+    },
+    {
+      "epoch": 41.7,
+      "learning_rate": 0.00025470175438596486,
+      "loss": 0.0994,
+      "step": 2794
+    },
+    {
+      "epoch": 41.71,
+      "learning_rate": 0.00025466666666666666,
+      "loss": 0.1196,
+      "step": 2795
+    },
+    {
+      "epoch": 41.73,
+      "learning_rate": 0.0002546315789473684,
+      "loss": 0.2126,
+      "step": 2796
+    },
+    {
+      "epoch": 41.74,
+      "learning_rate": 0.00025459649122807016,
+      "loss": 0.2458,
+      "step": 2797
+    },
+    {
+      "epoch": 41.76,
+      "learning_rate": 0.0002545614035087719,
+      "loss": 0.0517,
+      "step": 2798
+    },
+    {
+      "epoch": 41.77,
+      "learning_rate": 0.00025452631578947365,
+      "loss": 0.0656,
+      "step": 2799
+    },
+    {
+      "epoch": 41.79,
+      "learning_rate": 0.0002544912280701754,
+      "loss": 0.2169,
+      "step": 2800
+    },
+    {
+      "epoch": 41.79,
+      "eval_accuracy": 0.8159569260890847,
+      "eval_f1": 0.8147649383825286,
+      "eval_loss": 0.789986789226532,
+      "eval_runtime": 344.2728,
+      "eval_samples_per_second": 11.868,
+      "eval_steps_per_second": 0.744,
+      "step": 2800
+    },
+    {
+      "epoch": 41.8,
+      "learning_rate": 0.00025445614035087715,
+      "loss": 0.472,
+      "step": 2801
+    },
+    {
+      "epoch": 41.82,
+      "learning_rate": 0.00025442105263157895,
+      "loss": 0.3,
+      "step": 2802
+    },
+    {
+      "epoch": 41.83,
+      "learning_rate": 0.0002543859649122807,
+      "loss": 0.2579,
+      "step": 2803
+    },
+    {
+      "epoch": 41.85,
+      "learning_rate": 0.00025435087719298245,
+      "loss": 0.0847,
+      "step": 2804
+    },
+    {
+      "epoch": 41.86,
+      "learning_rate": 0.00025431578947368415,
+      "loss": 0.014,
+      "step": 2805
+    },
+    {
+      "epoch": 41.88,
+      "learning_rate": 0.00025428070175438595,
+      "loss": 0.0145,
+      "step": 2806
+    },
+    {
+      "epoch": 41.89,
+      "learning_rate": 0.0002542456140350877,
+      "loss": 0.1619,
+      "step": 2807
+    },
+    {
+      "epoch": 41.91,
+      "learning_rate": 0.00025421052631578945,
+      "loss": 0.1608,
+      "step": 2808
+    },
+    {
+      "epoch": 41.92,
+      "learning_rate": 0.0002541754385964912,
+      "loss": 0.0117,
+      "step": 2809
+    },
+    {
+      "epoch": 41.94,
+      "learning_rate": 0.00025414035087719294,
+      "loss": 0.3642,
+      "step": 2810
+    },
+    {
+      "epoch": 41.95,
+      "learning_rate": 0.0002541052631578947,
+      "loss": 0.0423,
+      "step": 2811
+    },
+    {
+      "epoch": 41.97,
+      "learning_rate": 0.00025407017543859644,
+      "loss": 0.0056,
+      "step": 2812
+    },
+    {
+      "epoch": 41.98,
+      "learning_rate": 0.00025403508771929825,
+      "loss": 0.0841,
+      "step": 2813
+    },
+    {
+      "epoch": 42.0,
+      "learning_rate": 0.000254,
+      "loss": 0.2236,
+      "step": 2814
+    },
+    {
+      "epoch": 42.01,
+      "learning_rate": 0.00025396491228070174,
+      "loss": 0.0366,
+      "step": 2815
+    },
+    {
+      "epoch": 42.03,
+      "learning_rate": 0.0002539298245614035,
+      "loss": 0.1367,
+      "step": 2816
+    },
+    {
+      "epoch": 42.04,
+      "learning_rate": 0.00025389473684210524,
+      "loss": 0.0367,
+      "step": 2817
+    },
+    {
+      "epoch": 42.06,
+      "learning_rate": 0.000253859649122807,
+      "loss": 0.0964,
+      "step": 2818
+    },
+    {
+      "epoch": 42.07,
+      "learning_rate": 0.00025382456140350874,
+      "loss": 0.3533,
+      "step": 2819
+    },
+    {
+      "epoch": 42.09,
+      "learning_rate": 0.00025378947368421054,
+      "loss": 0.1794,
+      "step": 2820
+    },
+    {
+      "epoch": 42.1,
+      "learning_rate": 0.0002537543859649123,
+      "loss": 0.3442,
+      "step": 2821
+    },
+    {
+      "epoch": 42.12,
+      "learning_rate": 0.000253719298245614,
+      "loss": 0.0575,
+      "step": 2822
+    },
+    {
+      "epoch": 42.13,
+      "learning_rate": 0.00025368421052631573,
+      "loss": 0.2326,
+      "step": 2823
+    },
+    {
+      "epoch": 42.15,
+      "learning_rate": 0.00025364912280701754,
+      "loss": 0.4745,
+      "step": 2824
+    },
+    {
+      "epoch": 42.16,
+      "learning_rate": 0.0002536140350877193,
+      "loss": 0.0171,
+      "step": 2825
+    },
+    {
+      "epoch": 42.18,
+      "learning_rate": 0.00025357894736842103,
+      "loss": 0.1074,
+      "step": 2826
+    },
+    {
+      "epoch": 42.19,
+      "learning_rate": 0.0002535438596491228,
+      "loss": 0.2265,
+      "step": 2827
+    },
+    {
+      "epoch": 42.21,
+      "learning_rate": 0.00025350877192982453,
+      "loss": 0.0297,
+      "step": 2828
+    },
+    {
+      "epoch": 42.22,
+      "learning_rate": 0.0002534736842105263,
+      "loss": 0.0187,
+      "step": 2829
+    },
+    {
+      "epoch": 42.24,
+      "learning_rate": 0.00025343859649122803,
+      "loss": 0.0116,
+      "step": 2830
+    },
+    {
+      "epoch": 42.25,
+      "learning_rate": 0.0002534035087719298,
+      "loss": 0.3781,
+      "step": 2831
+    },
+    {
+      "epoch": 42.27,
+      "learning_rate": 0.0002533684210526316,
+      "loss": 0.5599,
+      "step": 2832
+    },
+    {
+      "epoch": 42.28,
+      "learning_rate": 0.00025333333333333333,
+      "loss": 0.0833,
+      "step": 2833
+    },
+    {
+      "epoch": 42.3,
+      "learning_rate": 0.0002532982456140351,
+      "loss": 0.0891,
+      "step": 2834
+    },
+    {
+      "epoch": 42.31,
+      "learning_rate": 0.0002532631578947368,
+      "loss": 0.1705,
+      "step": 2835
+    },
+    {
+      "epoch": 42.33,
+      "learning_rate": 0.0002532280701754386,
+      "loss": 0.5718,
+      "step": 2836
+    },
+    {
+      "epoch": 42.34,
+      "learning_rate": 0.0002531929824561403,
+      "loss": 0.2741,
+      "step": 2837
+    },
+    {
+      "epoch": 42.36,
+      "learning_rate": 0.0002531578947368421,
+      "loss": 0.1348,
+      "step": 2838
+    },
+    {
+      "epoch": 42.37,
+      "learning_rate": 0.0002531228070175439,
+      "loss": 0.0928,
+      "step": 2839
+    },
+    {
+      "epoch": 42.39,
+      "learning_rate": 0.00025308771929824557,
+      "loss": 0.1806,
+      "step": 2840
+    },
+    {
+      "epoch": 42.4,
+      "learning_rate": 0.0002530526315789473,
+      "loss": 0.1709,
+      "step": 2841
+    },
+    {
+      "epoch": 42.42,
+      "learning_rate": 0.00025301754385964907,
+      "loss": 0.2734,
+      "step": 2842
+    },
+    {
+      "epoch": 42.43,
+      "learning_rate": 0.00025298245614035087,
+      "loss": 0.3118,
+      "step": 2843
+    },
+    {
+      "epoch": 42.45,
+      "learning_rate": 0.0002529473684210526,
+      "loss": 0.0629,
+      "step": 2844
+    },
+    {
+      "epoch": 42.46,
+      "learning_rate": 0.00025291228070175437,
+      "loss": 0.6791,
+      "step": 2845
+    },
+    {
+      "epoch": 42.48,
+      "learning_rate": 0.0002528771929824561,
+      "loss": 0.2058,
+      "step": 2846
+    },
+    {
+      "epoch": 42.49,
+      "learning_rate": 0.00025284210526315787,
+      "loss": 0.0604,
+      "step": 2847
+    },
+    {
+      "epoch": 42.51,
+      "learning_rate": 0.0002528070175438596,
+      "loss": 0.1178,
+      "step": 2848
+    },
+    {
+      "epoch": 42.52,
+      "learning_rate": 0.00025277192982456136,
+      "loss": 0.1147,
+      "step": 2849
+    },
+    {
+      "epoch": 42.54,
+      "learning_rate": 0.00025273684210526317,
+      "loss": 0.126,
+      "step": 2850
+    },
+    {
+      "epoch": 42.55,
+      "learning_rate": 0.0002527017543859649,
+      "loss": 0.0091,
+      "step": 2851
+    },
+    {
+      "epoch": 42.57,
+      "learning_rate": 0.00025266666666666666,
+      "loss": 0.1277,
+      "step": 2852
+    },
+    {
+      "epoch": 42.58,
+      "learning_rate": 0.00025263157894736836,
+      "loss": 0.5943,
+      "step": 2853
+    },
+    {
+      "epoch": 42.59,
+      "learning_rate": 0.00025259649122807016,
+      "loss": 0.1464,
+      "step": 2854
+    },
+    {
+      "epoch": 42.61,
+      "learning_rate": 0.0002525614035087719,
+      "loss": 0.6232,
+      "step": 2855
+    },
+    {
+      "epoch": 42.62,
+      "learning_rate": 0.00025252631578947366,
+      "loss": 0.2973,
+      "step": 2856
+    },
+    {
+      "epoch": 42.64,
+      "learning_rate": 0.0002524912280701754,
+      "loss": 0.1586,
+      "step": 2857
+    },
+    {
+      "epoch": 42.65,
+      "learning_rate": 0.00025245614035087716,
+      "loss": 0.0124,
+      "step": 2858
+    },
+    {
+      "epoch": 42.67,
+      "learning_rate": 0.0002524210526315789,
+      "loss": 0.0061,
+      "step": 2859
+    },
+    {
+      "epoch": 42.68,
+      "learning_rate": 0.00025238596491228065,
+      "loss": 0.015,
+      "step": 2860
+    },
+    {
+      "epoch": 42.7,
+      "learning_rate": 0.00025235087719298246,
+      "loss": 0.337,
+      "step": 2861
+    },
+    {
+      "epoch": 42.71,
+      "learning_rate": 0.0002523157894736842,
+      "loss": 0.1753,
+      "step": 2862
+    },
+    {
+      "epoch": 42.73,
+      "learning_rate": 0.00025228070175438596,
+      "loss": 0.0205,
+      "step": 2863
+    },
+    {
+      "epoch": 42.74,
+      "learning_rate": 0.0002522456140350877,
+      "loss": 0.1048,
+      "step": 2864
+    },
+    {
+      "epoch": 42.76,
+      "learning_rate": 0.00025221052631578945,
+      "loss": 0.6739,
+      "step": 2865
+    },
+    {
+      "epoch": 42.77,
+      "learning_rate": 0.0002521754385964912,
+      "loss": 0.0291,
+      "step": 2866
+    },
+    {
+      "epoch": 42.79,
+      "learning_rate": 0.00025214035087719295,
+      "loss": 0.0214,
+      "step": 2867
+    },
+    {
+      "epoch": 42.8,
+      "learning_rate": 0.0002521052631578947,
+      "loss": 0.2474,
+      "step": 2868
+    },
+    {
+      "epoch": 42.82,
+      "learning_rate": 0.0002520701754385965,
+      "loss": 0.3674,
+      "step": 2869
+    },
+    {
+      "epoch": 42.83,
+      "learning_rate": 0.0002520350877192982,
+      "loss": 0.2382,
+      "step": 2870
+    },
+    {
+      "epoch": 42.85,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 0.2272,
+      "step": 2871
+    },
+    {
+      "epoch": 42.86,
+      "learning_rate": 0.00025196491228070175,
+      "loss": 0.0438,
+      "step": 2872
+    },
+    {
+      "epoch": 42.88,
+      "learning_rate": 0.0002519298245614035,
+      "loss": 0.2005,
+      "step": 2873
+    },
+    {
+      "epoch": 42.89,
+      "learning_rate": 0.00025189473684210525,
+      "loss": 0.1324,
+      "step": 2874
+    },
+    {
+      "epoch": 42.91,
+      "learning_rate": 0.000251859649122807,
+      "loss": 0.1544,
+      "step": 2875
+    },
+    {
+      "epoch": 42.92,
+      "learning_rate": 0.00025182456140350874,
+      "loss": 0.1731,
+      "step": 2876
+    },
+    {
+      "epoch": 42.94,
+      "learning_rate": 0.0002517894736842105,
+      "loss": 0.0689,
+      "step": 2877
+    },
+    {
+      "epoch": 42.95,
+      "learning_rate": 0.00025175438596491224,
+      "loss": 0.114,
+      "step": 2878
+    },
+    {
+      "epoch": 42.97,
+      "learning_rate": 0.000251719298245614,
+      "loss": 0.0568,
+      "step": 2879
+    },
+    {
+      "epoch": 42.98,
+      "learning_rate": 0.0002516842105263158,
+      "loss": 0.3209,
+      "step": 2880
+    },
+    {
+      "epoch": 43.0,
+      "learning_rate": 0.00025164912280701754,
+      "loss": 0.2897,
+      "step": 2881
+    },
+    {
+      "epoch": 43.01,
+      "learning_rate": 0.0002516140350877193,
+      "loss": 0.2322,
+      "step": 2882
+    },
+    {
+      "epoch": 43.03,
+      "learning_rate": 0.00025157894736842104,
+      "loss": 0.1146,
+      "step": 2883
+    },
+    {
+      "epoch": 43.04,
+      "learning_rate": 0.0002515438596491228,
+      "loss": 0.0472,
+      "step": 2884
+    },
+    {
+      "epoch": 43.06,
+      "learning_rate": 0.00025150877192982454,
+      "loss": 0.0171,
+      "step": 2885
+    },
+    {
+      "epoch": 43.07,
+      "learning_rate": 0.0002514736842105263,
+      "loss": 0.2853,
+      "step": 2886
+    },
+    {
+      "epoch": 43.09,
+      "learning_rate": 0.0002514385964912281,
+      "loss": 0.1367,
+      "step": 2887
+    },
+    {
+      "epoch": 43.1,
+      "learning_rate": 0.0002514035087719298,
+      "loss": 0.5374,
+      "step": 2888
+    },
+    {
+      "epoch": 43.12,
+      "learning_rate": 0.00025136842105263153,
+      "loss": 0.0226,
+      "step": 2889
+    },
+    {
+      "epoch": 43.13,
+      "learning_rate": 0.0002513333333333333,
+      "loss": 0.3714,
+      "step": 2890
+    },
+    {
+      "epoch": 43.15,
+      "learning_rate": 0.0002512982456140351,
+      "loss": 0.0909,
+      "step": 2891
+    },
+    {
+      "epoch": 43.16,
+      "learning_rate": 0.00025126315789473683,
+      "loss": 0.5632,
+      "step": 2892
+    },
+    {
+      "epoch": 43.18,
+      "learning_rate": 0.0002512280701754386,
+      "loss": 0.1337,
+      "step": 2893
+    },
+    {
+      "epoch": 43.19,
+      "learning_rate": 0.00025119298245614033,
+      "loss": 0.0837,
+      "step": 2894
+    },
+    {
+      "epoch": 43.21,
+      "learning_rate": 0.0002511578947368421,
+      "loss": 0.0137,
+      "step": 2895
+    },
+    {
+      "epoch": 43.22,
+      "learning_rate": 0.00025112280701754383,
+      "loss": 0.0378,
+      "step": 2896
+    },
+    {
+      "epoch": 43.24,
+      "learning_rate": 0.0002510877192982456,
+      "loss": 0.0816,
+      "step": 2897
+    },
+    {
+      "epoch": 43.25,
+      "learning_rate": 0.0002510526315789474,
+      "loss": 0.039,
+      "step": 2898
+    },
+    {
+      "epoch": 43.27,
+      "learning_rate": 0.00025101754385964913,
+      "loss": 0.4815,
+      "step": 2899
+    },
+    {
+      "epoch": 43.28,
+      "learning_rate": 0.0002509824561403509,
+      "loss": 0.0348,
+      "step": 2900
+    },
+    {
+      "epoch": 43.3,
+      "learning_rate": 0.00025094736842105257,
+      "loss": 0.3811,
+      "step": 2901
+    },
+    {
+      "epoch": 43.31,
+      "learning_rate": 0.0002509122807017544,
+      "loss": 0.1087,
+      "step": 2902
+    },
+    {
+      "epoch": 43.33,
+      "learning_rate": 0.0002508771929824561,
+      "loss": 0.0497,
+      "step": 2903
+    },
+    {
+      "epoch": 43.34,
+      "learning_rate": 0.00025084210526315787,
+      "loss": 0.1322,
+      "step": 2904
+    },
+    {
+      "epoch": 43.36,
+      "learning_rate": 0.0002508070175438596,
+      "loss": 0.2186,
+      "step": 2905
+    },
+    {
+      "epoch": 43.37,
+      "learning_rate": 0.00025077192982456137,
+      "loss": 0.1789,
+      "step": 2906
+    },
+    {
+      "epoch": 43.39,
+      "learning_rate": 0.0002507368421052631,
+      "loss": 0.3261,
+      "step": 2907
+    },
+    {
+      "epoch": 43.4,
+      "learning_rate": 0.00025070175438596487,
+      "loss": 0.0251,
+      "step": 2908
+    },
+    {
+      "epoch": 43.42,
+      "learning_rate": 0.00025066666666666667,
+      "loss": 0.3344,
+      "step": 2909
+    },
+    {
+      "epoch": 43.43,
+      "learning_rate": 0.0002506315789473684,
+      "loss": 0.2145,
+      "step": 2910
+    },
+    {
+      "epoch": 43.45,
+      "learning_rate": 0.00025059649122807017,
+      "loss": 0.2443,
+      "step": 2911
+    },
+    {
+      "epoch": 43.46,
+      "learning_rate": 0.0002505614035087719,
+      "loss": 0.1077,
+      "step": 2912
+    },
+    {
+      "epoch": 43.48,
+      "learning_rate": 0.00025052631578947367,
+      "loss": 0.0922,
+      "step": 2913
+    },
+    {
+      "epoch": 43.49,
+      "learning_rate": 0.0002504912280701754,
+      "loss": 0.2037,
+      "step": 2914
+    },
+    {
+      "epoch": 43.51,
+      "learning_rate": 0.00025045614035087716,
+      "loss": 0.1559,
+      "step": 2915
+    },
+    {
+      "epoch": 43.52,
+      "learning_rate": 0.0002504210526315789,
+      "loss": 0.1043,
+      "step": 2916
+    },
+    {
+      "epoch": 43.54,
+      "learning_rate": 0.0002503859649122807,
+      "loss": 0.2706,
+      "step": 2917
+    },
+    {
+      "epoch": 43.55,
+      "learning_rate": 0.0002503508771929824,
+      "loss": 0.0522,
+      "step": 2918
+    },
+    {
+      "epoch": 43.57,
+      "learning_rate": 0.00025031578947368416,
+      "loss": 0.0158,
+      "step": 2919
+    },
+    {
+      "epoch": 43.58,
+      "learning_rate": 0.00025028070175438596,
+      "loss": 0.3399,
+      "step": 2920
+    },
+    {
+      "epoch": 43.59,
+      "learning_rate": 0.0002502456140350877,
+      "loss": 0.1641,
+      "step": 2921
+    },
+    {
+      "epoch": 43.61,
+      "learning_rate": 0.00025021052631578946,
+      "loss": 0.3141,
+      "step": 2922
+    },
+    {
+      "epoch": 43.62,
+      "learning_rate": 0.0002501754385964912,
+      "loss": 0.0162,
+      "step": 2923
+    },
+    {
+      "epoch": 43.64,
+      "learning_rate": 0.00025014035087719296,
+      "loss": 0.561,
+      "step": 2924
+    },
+    {
+      "epoch": 43.65,
+      "learning_rate": 0.0002501052631578947,
+      "loss": 0.1644,
+      "step": 2925
+    },
+    {
+      "epoch": 43.67,
+      "learning_rate": 0.00025007017543859645,
+      "loss": 0.2219,
+      "step": 2926
+    },
+    {
+      "epoch": 43.68,
+      "learning_rate": 0.0002500350877192982,
+      "loss": 0.1341,
+      "step": 2927
+    },
+    {
+      "epoch": 43.7,
+      "learning_rate": 0.00025,
+      "loss": 0.33,
+      "step": 2928
+    },
+    {
+      "epoch": 43.71,
+      "learning_rate": 0.00024996491228070175,
+      "loss": 0.1807,
+      "step": 2929
+    },
+    {
+      "epoch": 43.73,
+      "learning_rate": 0.0002499298245614035,
+      "loss": 0.1551,
+      "step": 2930
+    },
+    {
+      "epoch": 43.74,
+      "learning_rate": 0.00024989473684210525,
+      "loss": 0.1855,
+      "step": 2931
+    },
+    {
+      "epoch": 43.76,
+      "learning_rate": 0.000249859649122807,
+      "loss": 0.0575,
+      "step": 2932
+    },
+    {
+      "epoch": 43.77,
+      "learning_rate": 0.00024982456140350875,
+      "loss": 0.2949,
+      "step": 2933
+    },
+    {
+      "epoch": 43.79,
+      "learning_rate": 0.0002497894736842105,
+      "loss": 0.1351,
+      "step": 2934
+    },
+    {
+      "epoch": 43.8,
+      "learning_rate": 0.00024975438596491225,
+      "loss": 0.1623,
+      "step": 2935
+    },
+    {
+      "epoch": 43.82,
+      "learning_rate": 0.000249719298245614,
+      "loss": 0.3443,
+      "step": 2936
+    },
+    {
+      "epoch": 43.83,
+      "learning_rate": 0.00024968421052631574,
+      "loss": 0.1855,
+      "step": 2937
+    },
+    {
+      "epoch": 43.85,
+      "learning_rate": 0.0002496491228070175,
+      "loss": 0.498,
+      "step": 2938
+    },
+    {
+      "epoch": 43.86,
+      "learning_rate": 0.0002496140350877193,
+      "loss": 0.1217,
+      "step": 2939
+    },
+    {
+      "epoch": 43.88,
+      "learning_rate": 0.00024957894736842104,
+      "loss": 0.021,
+      "step": 2940
+    },
+    {
+      "epoch": 43.89,
+      "learning_rate": 0.0002495438596491228,
+      "loss": 0.0545,
+      "step": 2941
+    },
+    {
+      "epoch": 43.91,
+      "learning_rate": 0.00024950877192982454,
+      "loss": 0.0702,
+      "step": 2942
+    },
+    {
+      "epoch": 43.92,
+      "learning_rate": 0.0002494736842105263,
+      "loss": 0.2528,
+      "step": 2943
+    },
+    {
+      "epoch": 43.94,
+      "learning_rate": 0.00024943859649122804,
+      "loss": 0.0393,
+      "step": 2944
+    },
+    {
+      "epoch": 43.95,
+      "learning_rate": 0.0002494035087719298,
+      "loss": 0.2664,
+      "step": 2945
+    },
+    {
+      "epoch": 43.97,
+      "learning_rate": 0.0002493684210526316,
+      "loss": 0.3916,
+      "step": 2946
+    },
+    {
+      "epoch": 43.98,
+      "learning_rate": 0.00024933333333333334,
+      "loss": 0.08,
+      "step": 2947
+    },
+    {
+      "epoch": 44.0,
+      "learning_rate": 0.0002492982456140351,
+      "loss": 0.07,
+      "step": 2948
+    },
+    {
+      "epoch": 44.01,
+      "learning_rate": 0.0002492631578947368,
+      "loss": 0.4027,
+      "step": 2949
+    },
+    {
+      "epoch": 44.03,
+      "learning_rate": 0.0002492280701754386,
+      "loss": 0.1744,
+      "step": 2950
+    },
+    {
+      "epoch": 44.04,
+      "learning_rate": 0.00024919298245614034,
+      "loss": 0.1237,
+      "step": 2951
+    },
+    {
+      "epoch": 44.06,
+      "learning_rate": 0.0002491578947368421,
+      "loss": 0.1386,
+      "step": 2952
+    },
+    {
+      "epoch": 44.07,
+      "learning_rate": 0.00024912280701754383,
+      "loss": 0.1239,
+      "step": 2953
+    },
+    {
+      "epoch": 44.09,
+      "learning_rate": 0.0002490877192982456,
+      "loss": 0.3292,
+      "step": 2954
+    },
+    {
+      "epoch": 44.1,
+      "learning_rate": 0.00024905263157894733,
+      "loss": 0.1744,
+      "step": 2955
+    },
+    {
+      "epoch": 44.12,
+      "learning_rate": 0.0002490175438596491,
+      "loss": 0.2993,
+      "step": 2956
+    },
+    {
+      "epoch": 44.13,
+      "learning_rate": 0.00024898245614035083,
+      "loss": 0.0331,
+      "step": 2957
+    },
+    {
+      "epoch": 44.15,
+      "learning_rate": 0.00024894736842105263,
+      "loss": 0.1431,
+      "step": 2958
+    },
+    {
+      "epoch": 44.16,
+      "learning_rate": 0.0002489122807017544,
+      "loss": 0.0802,
+      "step": 2959
+    },
+    {
+      "epoch": 44.18,
+      "learning_rate": 0.00024887719298245613,
+      "loss": 0.131,
+      "step": 2960
+    },
+    {
+      "epoch": 44.19,
+      "learning_rate": 0.0002488421052631579,
+      "loss": 0.098,
+      "step": 2961
+    },
+    {
+      "epoch": 44.21,
+      "learning_rate": 0.0002488070175438596,
+      "loss": 0.0511,
+      "step": 2962
+    },
+    {
+      "epoch": 44.22,
+      "learning_rate": 0.0002487719298245614,
+      "loss": 0.2082,
+      "step": 2963
+    },
+    {
+      "epoch": 44.24,
+      "learning_rate": 0.0002487368421052631,
+      "loss": 0.0368,
+      "step": 2964
+    },
+    {
+      "epoch": 44.25,
+      "learning_rate": 0.0002487017543859649,
+      "loss": 0.0232,
+      "step": 2965
+    },
+    {
+      "epoch": 44.27,
+      "learning_rate": 0.0002486666666666666,
+      "loss": 0.1291,
+      "step": 2966
+    },
+    {
+      "epoch": 44.28,
+      "learning_rate": 0.00024863157894736837,
+      "loss": 0.015,
+      "step": 2967
+    },
+    {
+      "epoch": 44.3,
+      "learning_rate": 0.0002485964912280701,
+      "loss": 0.3891,
+      "step": 2968
+    },
+    {
+      "epoch": 44.31,
+      "learning_rate": 0.0002485614035087719,
+      "loss": 0.0984,
+      "step": 2969
+    },
+    {
+      "epoch": 44.33,
+      "learning_rate": 0.00024852631578947367,
+      "loss": 0.3002,
+      "step": 2970
+    },
+    {
+      "epoch": 44.34,
+      "learning_rate": 0.0002484912280701754,
+      "loss": 0.0087,
+      "step": 2971
+    },
+    {
+      "epoch": 44.36,
+      "learning_rate": 0.00024845614035087717,
+      "loss": 0.2016,
+      "step": 2972
+    },
+    {
+      "epoch": 44.37,
+      "learning_rate": 0.0002484210526315789,
+      "loss": 0.0873,
+      "step": 2973
+    },
+    {
+      "epoch": 44.39,
+      "learning_rate": 0.00024838596491228067,
+      "loss": 0.0506,
+      "step": 2974
+    },
+    {
+      "epoch": 44.4,
+      "learning_rate": 0.0002483508771929824,
+      "loss": 0.2401,
+      "step": 2975
+    },
+    {
+      "epoch": 44.42,
+      "learning_rate": 0.0002483157894736842,
+      "loss": 0.0794,
+      "step": 2976
+    },
+    {
+      "epoch": 44.43,
+      "learning_rate": 0.00024828070175438597,
+      "loss": 0.0242,
+      "step": 2977
+    },
+    {
+      "epoch": 44.45,
+      "learning_rate": 0.0002482456140350877,
+      "loss": 0.0321,
+      "step": 2978
+    },
+    {
+      "epoch": 44.46,
+      "learning_rate": 0.00024821052631578946,
+      "loss": 0.1454,
+      "step": 2979
+    },
+    {
+      "epoch": 44.48,
+      "learning_rate": 0.0002481754385964912,
+      "loss": 0.0272,
+      "step": 2980
+    },
+    {
+      "epoch": 44.49,
+      "learning_rate": 0.00024814035087719296,
+      "loss": 0.0047,
+      "step": 2981
+    },
+    {
+      "epoch": 44.51,
+      "learning_rate": 0.0002481052631578947,
+      "loss": 0.0876,
+      "step": 2982
+    },
+    {
+      "epoch": 44.52,
+      "learning_rate": 0.00024807017543859646,
+      "loss": 0.053,
+      "step": 2983
+    },
+    {
+      "epoch": 44.54,
+      "learning_rate": 0.0002480350877192982,
+      "loss": 0.0717,
+      "step": 2984
+    },
+    {
+      "epoch": 44.55,
+      "learning_rate": 0.00024799999999999996,
+      "loss": 0.5333,
+      "step": 2985
+    },
+    {
+      "epoch": 44.57,
+      "learning_rate": 0.0002479649122807017,
+      "loss": 0.0326,
+      "step": 2986
+    },
+    {
+      "epoch": 44.58,
+      "learning_rate": 0.0002479298245614035,
+      "loss": 0.1626,
+      "step": 2987
+    },
+    {
+      "epoch": 44.59,
+      "learning_rate": 0.00024789473684210526,
+      "loss": 0.1837,
+      "step": 2988
+    },
+    {
+      "epoch": 44.61,
+      "learning_rate": 0.000247859649122807,
+      "loss": 0.3757,
+      "step": 2989
+    },
+    {
+      "epoch": 44.62,
+      "learning_rate": 0.00024782456140350875,
+      "loss": 0.065,
+      "step": 2990
+    },
+    {
+      "epoch": 44.64,
+      "learning_rate": 0.0002477894736842105,
+      "loss": 0.21,
+      "step": 2991
+    },
+    {
+      "epoch": 44.65,
+      "learning_rate": 0.00024775438596491225,
+      "loss": 0.0765,
+      "step": 2992
+    },
+    {
+      "epoch": 44.67,
+      "learning_rate": 0.000247719298245614,
+      "loss": 0.1718,
+      "step": 2993
+    },
+    {
+      "epoch": 44.68,
+      "learning_rate": 0.00024768421052631575,
+      "loss": 0.0413,
+      "step": 2994
+    },
+    {
+      "epoch": 44.7,
+      "learning_rate": 0.00024764912280701755,
+      "loss": 0.0597,
+      "step": 2995
+    },
+    {
+      "epoch": 44.71,
+      "learning_rate": 0.0002476140350877193,
+      "loss": 0.2591,
+      "step": 2996
+    },
+    {
+      "epoch": 44.73,
+      "learning_rate": 0.000247578947368421,
+      "loss": 0.0086,
+      "step": 2997
+    },
+    {
+      "epoch": 44.74,
+      "learning_rate": 0.0002475438596491228,
+      "loss": 0.2637,
+      "step": 2998
+    },
+    {
+      "epoch": 44.76,
+      "learning_rate": 0.00024750877192982455,
+      "loss": 0.7918,
+      "step": 2999
+    },
+    {
+      "epoch": 44.77,
+      "learning_rate": 0.0002474736842105263,
+      "loss": 0.3942,
+      "step": 3000
+    },
+    {
+      "epoch": 44.77,
+      "eval_accuracy": 0.8042094958394518,
+      "eval_f1": 0.8129959248685728,
+      "eval_loss": 0.8620648384094238,
+      "eval_runtime": 344.2906,
+      "eval_samples_per_second": 11.868,
+      "eval_steps_per_second": 0.744,
+      "step": 3000
+    },
+    {
+      "epoch": 44.79,
+      "learning_rate": 0.00024743859649122805,
+      "loss": 0.4602,
+      "step": 3001
+    },
+    {
+      "epoch": 44.8,
+      "learning_rate": 0.0002474035087719298,
+      "loss": 0.1656,
+      "step": 3002
+    },
+    {
+      "epoch": 44.82,
+      "learning_rate": 0.00024736842105263154,
+      "loss": 0.3406,
+      "step": 3003
+    },
+    {
+      "epoch": 44.83,
+      "learning_rate": 0.0002473333333333333,
+      "loss": 0.2357,
+      "step": 3004
+    },
+    {
+      "epoch": 44.85,
+      "learning_rate": 0.00024729824561403504,
+      "loss": 0.5817,
+      "step": 3005
+    },
+    {
+      "epoch": 44.86,
+      "learning_rate": 0.00024726315789473684,
+      "loss": 0.0355,
+      "step": 3006
+    },
+    {
+      "epoch": 44.88,
+      "learning_rate": 0.0002472280701754386,
+      "loss": 0.332,
+      "step": 3007
+    },
+    {
+      "epoch": 44.89,
+      "learning_rate": 0.00024719298245614034,
+      "loss": 0.0443,
+      "step": 3008
+    },
+    {
+      "epoch": 44.91,
+      "learning_rate": 0.0002471578947368421,
+      "loss": 0.3544,
+      "step": 3009
+    },
+    {
+      "epoch": 44.92,
+      "learning_rate": 0.00024712280701754384,
+      "loss": 0.2409,
+      "step": 3010
+    },
+    {
+      "epoch": 44.94,
+      "learning_rate": 0.0002470877192982456,
+      "loss": 0.025,
+      "step": 3011
+    },
+    {
+      "epoch": 44.95,
+      "learning_rate": 0.00024705263157894734,
+      "loss": 0.1748,
+      "step": 3012
+    },
+    {
+      "epoch": 44.97,
+      "learning_rate": 0.00024701754385964914,
+      "loss": 0.0428,
+      "step": 3013
+    },
+    {
+      "epoch": 44.98,
+      "learning_rate": 0.00024698245614035083,
+      "loss": 0.1191,
+      "step": 3014
+    },
+    {
+      "epoch": 45.0,
+      "learning_rate": 0.0002469473684210526,
+      "loss": 0.2148,
+      "step": 3015
+    },
+    {
+      "epoch": 45.01,
+      "learning_rate": 0.00024691228070175433,
+      "loss": 0.3834,
+      "step": 3016
+    },
+    {
+      "epoch": 45.03,
+      "learning_rate": 0.00024687719298245613,
+      "loss": 0.2043,
+      "step": 3017
+    },
+    {
+      "epoch": 45.04,
+      "learning_rate": 0.0002468421052631579,
+      "loss": 0.0231,
+      "step": 3018
+    },
+    {
+      "epoch": 45.06,
+      "learning_rate": 0.00024680701754385963,
+      "loss": 0.17,
+      "step": 3019
+    },
+    {
+      "epoch": 45.07,
+      "learning_rate": 0.0002467719298245614,
+      "loss": 0.2151,
+      "step": 3020
+    },
+    {
+      "epoch": 45.09,
+      "learning_rate": 0.00024673684210526313,
+      "loss": 0.0963,
+      "step": 3021
+    },
+    {
+      "epoch": 45.1,
+      "learning_rate": 0.0002467017543859649,
+      "loss": 0.0191,
+      "step": 3022
+    },
+    {
+      "epoch": 45.12,
+      "learning_rate": 0.0002466666666666666,
+      "loss": 0.1489,
+      "step": 3023
+    },
+    {
+      "epoch": 45.13,
+      "learning_rate": 0.00024663157894736843,
+      "loss": 0.3331,
+      "step": 3024
+    },
+    {
+      "epoch": 45.15,
+      "learning_rate": 0.0002465964912280702,
+      "loss": 0.3295,
+      "step": 3025
+    },
+    {
+      "epoch": 45.16,
+      "learning_rate": 0.00024656140350877193,
+      "loss": 0.2693,
+      "step": 3026
+    },
+    {
+      "epoch": 45.18,
+      "learning_rate": 0.0002465263157894737,
+      "loss": 0.4132,
+      "step": 3027
+    },
+    {
+      "epoch": 45.19,
+      "learning_rate": 0.0002464912280701754,
+      "loss": 0.0197,
+      "step": 3028
+    },
+    {
+      "epoch": 45.21,
+      "learning_rate": 0.0002464561403508772,
+      "loss": 0.0577,
+      "step": 3029
+    },
+    {
+      "epoch": 45.22,
+      "learning_rate": 0.0002464210526315789,
+      "loss": 0.1374,
+      "step": 3030
+    },
+    {
+      "epoch": 45.24,
+      "learning_rate": 0.00024638596491228067,
+      "loss": 0.1153,
+      "step": 3031
+    },
+    {
+      "epoch": 45.25,
+      "learning_rate": 0.0002463508771929824,
+      "loss": 0.1013,
+      "step": 3032
+    },
+    {
+      "epoch": 45.27,
+      "learning_rate": 0.00024631578947368417,
+      "loss": 0.2066,
+      "step": 3033
+    },
+    {
+      "epoch": 45.28,
+      "learning_rate": 0.0002462807017543859,
+      "loss": 0.0493,
+      "step": 3034
+    },
+    {
+      "epoch": 45.3,
+      "learning_rate": 0.0002462456140350877,
+      "loss": 0.1365,
+      "step": 3035
+    },
+    {
+      "epoch": 45.31,
+      "learning_rate": 0.00024621052631578947,
+      "loss": 0.088,
+      "step": 3036
+    },
+    {
+      "epoch": 45.33,
+      "learning_rate": 0.0002461754385964912,
+      "loss": 0.0117,
+      "step": 3037
+    },
+    {
+      "epoch": 45.34,
+      "learning_rate": 0.00024614035087719297,
+      "loss": 0.0095,
+      "step": 3038
+    },
+    {
+      "epoch": 45.36,
+      "learning_rate": 0.0002461052631578947,
+      "loss": 0.1326,
+      "step": 3039
+    },
+    {
+      "epoch": 45.37,
+      "learning_rate": 0.00024607017543859646,
+      "loss": 0.1197,
+      "step": 3040
+    },
+    {
+      "epoch": 45.39,
+      "learning_rate": 0.0002460350877192982,
+      "loss": 0.029,
+      "step": 3041
+    },
+    {
+      "epoch": 45.4,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 0.1934,
+      "step": 3042
+    },
+    {
+      "epoch": 45.42,
+      "learning_rate": 0.00024596491228070177,
+      "loss": 0.1238,
+      "step": 3043
+    },
+    {
+      "epoch": 45.43,
+      "learning_rate": 0.0002459298245614035,
+      "loss": 0.0603,
+      "step": 3044
+    },
+    {
+      "epoch": 45.45,
+      "learning_rate": 0.0002458947368421052,
+      "loss": 0.0289,
+      "step": 3045
+    },
+    {
+      "epoch": 45.46,
+      "learning_rate": 0.000245859649122807,
+      "loss": 0.3161,
+      "step": 3046
+    },
+    {
+      "epoch": 45.48,
+      "learning_rate": 0.00024582456140350876,
+      "loss": 0.1879,
+      "step": 3047
+    },
+    {
+      "epoch": 45.49,
+      "learning_rate": 0.0002457894736842105,
+      "loss": 0.2013,
+      "step": 3048
+    },
+    {
+      "epoch": 45.51,
+      "learning_rate": 0.00024575438596491226,
+      "loss": 0.2212,
+      "step": 3049
+    },
+    {
+      "epoch": 45.52,
+      "learning_rate": 0.000245719298245614,
+      "loss": 0.4477,
+      "step": 3050
+    },
+    {
+      "epoch": 45.54,
+      "learning_rate": 0.00024568421052631576,
+      "loss": 0.2636,
+      "step": 3051
+    },
+    {
+      "epoch": 45.55,
+      "learning_rate": 0.0002456491228070175,
+      "loss": 0.3108,
+      "step": 3052
+    },
+    {
+      "epoch": 45.57,
+      "learning_rate": 0.00024561403508771925,
+      "loss": 0.2828,
+      "step": 3053
+    },
+    {
+      "epoch": 45.58,
+      "learning_rate": 0.00024557894736842106,
+      "loss": 0.0839,
+      "step": 3054
+    },
+    {
+      "epoch": 45.59,
+      "learning_rate": 0.0002455438596491228,
+      "loss": 0.1829,
+      "step": 3055
+    },
+    {
+      "epoch": 45.61,
+      "learning_rate": 0.00024550877192982455,
+      "loss": 0.0478,
+      "step": 3056
+    },
+    {
+      "epoch": 45.62,
+      "learning_rate": 0.0002454736842105263,
+      "loss": 0.2361,
+      "step": 3057
+    },
+    {
+      "epoch": 45.64,
+      "learning_rate": 0.00024543859649122805,
+      "loss": 0.1828,
+      "step": 3058
+    },
+    {
+      "epoch": 45.65,
+      "learning_rate": 0.0002454035087719298,
+      "loss": 0.245,
+      "step": 3059
+    },
+    {
+      "epoch": 45.67,
+      "learning_rate": 0.00024536842105263155,
+      "loss": 0.139,
+      "step": 3060
+    },
+    {
+      "epoch": 45.68,
+      "learning_rate": 0.00024533333333333335,
+      "loss": 0.1808,
+      "step": 3061
+    },
+    {
+      "epoch": 45.7,
+      "learning_rate": 0.00024529824561403505,
+      "loss": 0.0683,
+      "step": 3062
+    },
+    {
+      "epoch": 45.71,
+      "learning_rate": 0.0002452631578947368,
+      "loss": 0.0367,
+      "step": 3063
+    },
+    {
+      "epoch": 45.73,
+      "learning_rate": 0.00024522807017543854,
+      "loss": 0.1152,
+      "step": 3064
+    },
+    {
+      "epoch": 45.74,
+      "learning_rate": 0.00024519298245614035,
+      "loss": 0.1368,
+      "step": 3065
+    },
+    {
+      "epoch": 45.76,
+      "learning_rate": 0.0002451578947368421,
+      "loss": 0.0907,
+      "step": 3066
+    },
+    {
+      "epoch": 45.77,
+      "learning_rate": 0.00024512280701754384,
+      "loss": 0.0304,
+      "step": 3067
+    },
+    {
+      "epoch": 45.79,
+      "learning_rate": 0.0002450877192982456,
+      "loss": 0.156,
+      "step": 3068
+    },
+    {
+      "epoch": 45.8,
+      "learning_rate": 0.00024505263157894734,
+      "loss": 0.3528,
+      "step": 3069
+    },
+    {
+      "epoch": 45.82,
+      "learning_rate": 0.0002450175438596491,
+      "loss": 0.0618,
+      "step": 3070
+    },
+    {
+      "epoch": 45.83,
+      "learning_rate": 0.00024498245614035084,
+      "loss": 0.2586,
+      "step": 3071
+    },
+    {
+      "epoch": 45.85,
+      "learning_rate": 0.00024494736842105264,
+      "loss": 0.2432,
+      "step": 3072
+    },
+    {
+      "epoch": 45.86,
+      "learning_rate": 0.0002449122807017544,
+      "loss": 0.1289,
+      "step": 3073
+    },
+    {
+      "epoch": 45.88,
+      "learning_rate": 0.00024487719298245614,
+      "loss": 0.0413,
+      "step": 3074
+    },
+    {
+      "epoch": 45.89,
+      "learning_rate": 0.0002448421052631579,
+      "loss": 0.228,
+      "step": 3075
+    },
+    {
+      "epoch": 45.91,
+      "learning_rate": 0.00024480701754385964,
+      "loss": 0.2158,
+      "step": 3076
+    },
+    {
+      "epoch": 45.92,
+      "learning_rate": 0.0002447719298245614,
+      "loss": 0.0989,
+      "step": 3077
+    },
+    {
+      "epoch": 45.94,
+      "learning_rate": 0.00024473684210526314,
+      "loss": 0.3633,
+      "step": 3078
+    },
+    {
+      "epoch": 45.95,
+      "learning_rate": 0.0002447017543859649,
+      "loss": 0.2802,
+      "step": 3079
+    },
+    {
+      "epoch": 45.97,
+      "learning_rate": 0.00024466666666666663,
+      "loss": 0.1715,
+      "step": 3080
+    },
+    {
+      "epoch": 45.98,
+      "learning_rate": 0.0002446315789473684,
+      "loss": 0.0474,
+      "step": 3081
+    },
+    {
+      "epoch": 46.0,
+      "learning_rate": 0.00024459649122807013,
+      "loss": 0.3234,
+      "step": 3082
+    },
+    {
+      "epoch": 46.01,
+      "learning_rate": 0.00024456140350877193,
+      "loss": 0.3223,
+      "step": 3083
+    },
+    {
+      "epoch": 46.03,
+      "learning_rate": 0.0002445263157894737,
+      "loss": 0.0872,
+      "step": 3084
+    },
+    {
+      "epoch": 46.04,
+      "learning_rate": 0.00024449122807017543,
+      "loss": 0.0712,
+      "step": 3085
+    },
+    {
+      "epoch": 46.06,
+      "learning_rate": 0.0002444561403508772,
+      "loss": 0.1009,
+      "step": 3086
+    },
+    {
+      "epoch": 46.07,
+      "learning_rate": 0.00024442105263157893,
+      "loss": 0.3466,
+      "step": 3087
+    },
+    {
+      "epoch": 46.09,
+      "learning_rate": 0.0002443859649122807,
+      "loss": 0.3214,
+      "step": 3088
+    },
+    {
+      "epoch": 46.1,
+      "learning_rate": 0.0002443508771929824,
+      "loss": 0.507,
+      "step": 3089
+    },
+    {
+      "epoch": 46.12,
+      "learning_rate": 0.0002443157894736842,
+      "loss": 0.0643,
+      "step": 3090
+    },
+    {
+      "epoch": 46.13,
+      "learning_rate": 0.000244280701754386,
+      "loss": 0.0152,
+      "step": 3091
+    },
+    {
+      "epoch": 46.15,
+      "learning_rate": 0.0002442456140350877,
+      "loss": 0.0998,
+      "step": 3092
+    },
+    {
+      "epoch": 46.16,
+      "learning_rate": 0.0002442105263157894,
+      "loss": 0.0875,
+      "step": 3093
+    },
+    {
+      "epoch": 46.18,
+      "learning_rate": 0.00024417543859649117,
+      "loss": 0.1965,
+      "step": 3094
+    },
+    {
+      "epoch": 46.19,
+      "learning_rate": 0.00024414035087719297,
+      "loss": 0.124,
+      "step": 3095
+    },
+    {
+      "epoch": 46.21,
+      "learning_rate": 0.00024410526315789472,
+      "loss": 0.0267,
+      "step": 3096
+    },
+    {
+      "epoch": 46.22,
+      "learning_rate": 0.00024407017543859647,
+      "loss": 0.0135,
+      "step": 3097
+    },
+    {
+      "epoch": 46.24,
+      "learning_rate": 0.00024403508771929825,
+      "loss": 0.1685,
+      "step": 3098
+    },
+    {
+      "epoch": 46.25,
+      "learning_rate": 0.000244,
+      "loss": 0.2474,
+      "step": 3099
+    },
+    {
+      "epoch": 46.27,
+      "learning_rate": 0.00024396491228070172,
+      "loss": 0.1671,
+      "step": 3100
+    },
+    {
+      "epoch": 46.28,
+      "learning_rate": 0.00024392982456140347,
+      "loss": 0.1471,
+      "step": 3101
+    },
+    {
+      "epoch": 46.3,
+      "learning_rate": 0.00024389473684210524,
+      "loss": 0.2128,
+      "step": 3102
+    },
+    {
+      "epoch": 46.31,
+      "learning_rate": 0.000243859649122807,
+      "loss": 0.0124,
+      "step": 3103
+    },
+    {
+      "epoch": 46.33,
+      "learning_rate": 0.00024382456140350874,
+      "loss": 0.3404,
+      "step": 3104
+    },
+    {
+      "epoch": 46.34,
+      "learning_rate": 0.0002437894736842105,
+      "loss": 0.1251,
+      "step": 3105
+    },
+    {
+      "epoch": 46.36,
+      "learning_rate": 0.00024375438596491226,
+      "loss": 0.1597,
+      "step": 3106
+    },
+    {
+      "epoch": 46.37,
+      "learning_rate": 0.000243719298245614,
+      "loss": 0.0564,
+      "step": 3107
+    },
+    {
+      "epoch": 46.39,
+      "learning_rate": 0.00024368421052631576,
+      "loss": 0.0055,
+      "step": 3108
+    },
+    {
+      "epoch": 46.4,
+      "learning_rate": 0.00024364912280701754,
+      "loss": 0.206,
+      "step": 3109
+    },
+    {
+      "epoch": 46.42,
+      "learning_rate": 0.00024361403508771929,
+      "loss": 0.1958,
+      "step": 3110
+    },
+    {
+      "epoch": 46.43,
+      "learning_rate": 0.00024357894736842103,
+      "loss": 0.2692,
+      "step": 3111
+    },
+    {
+      "epoch": 46.45,
+      "learning_rate": 0.00024354385964912278,
+      "loss": 0.045,
+      "step": 3112
+    },
+    {
+      "epoch": 46.46,
+      "learning_rate": 0.00024350877192982456,
+      "loss": 0.1451,
+      "step": 3113
+    },
+    {
+      "epoch": 46.48,
+      "learning_rate": 0.0002434736842105263,
+      "loss": 0.066,
+      "step": 3114
+    },
+    {
+      "epoch": 46.49,
+      "learning_rate": 0.00024343859649122806,
+      "loss": 0.0914,
+      "step": 3115
+    },
+    {
+      "epoch": 46.51,
+      "learning_rate": 0.00024340350877192978,
+      "loss": 0.4548,
+      "step": 3116
+    },
+    {
+      "epoch": 46.52,
+      "learning_rate": 0.00024336842105263158,
+      "loss": 0.2507,
+      "step": 3117
+    },
+    {
+      "epoch": 46.54,
+      "learning_rate": 0.0002433333333333333,
+      "loss": 0.0096,
+      "step": 3118
+    },
+    {
+      "epoch": 46.55,
+      "learning_rate": 0.00024329824561403505,
+      "loss": 0.1576,
+      "step": 3119
+    },
+    {
+      "epoch": 46.57,
+      "learning_rate": 0.0002432631578947368,
+      "loss": 0.2727,
+      "step": 3120
+    },
+    {
+      "epoch": 46.58,
+      "learning_rate": 0.00024322807017543858,
+      "loss": 0.0633,
+      "step": 3121
+    },
+    {
+      "epoch": 46.59,
+      "learning_rate": 0.00024319298245614033,
+      "loss": 0.0677,
+      "step": 3122
+    },
+    {
+      "epoch": 46.61,
+      "learning_rate": 0.00024315789473684207,
+      "loss": 0.2125,
+      "step": 3123
+    },
+    {
+      "epoch": 46.62,
+      "learning_rate": 0.00024312280701754385,
+      "loss": 0.06,
+      "step": 3124
+    },
+    {
+      "epoch": 46.64,
+      "learning_rate": 0.0002430877192982456,
+      "loss": 0.0515,
+      "step": 3125
+    },
+    {
+      "epoch": 46.65,
+      "learning_rate": 0.00024305263157894735,
+      "loss": 0.0243,
+      "step": 3126
+    },
+    {
+      "epoch": 46.67,
+      "learning_rate": 0.0002430175438596491,
+      "loss": 0.1532,
+      "step": 3127
+    },
+    {
+      "epoch": 46.68,
+      "learning_rate": 0.00024298245614035087,
+      "loss": 0.2456,
+      "step": 3128
+    },
+    {
+      "epoch": 46.7,
+      "learning_rate": 0.00024294736842105262,
+      "loss": 0.034,
+      "step": 3129
+    },
+    {
+      "epoch": 46.71,
+      "learning_rate": 0.00024291228070175437,
+      "loss": 0.1093,
+      "step": 3130
+    },
+    {
+      "epoch": 46.73,
+      "learning_rate": 0.0002428771929824561,
+      "loss": 0.0393,
+      "step": 3131
+    },
+    {
+      "epoch": 46.74,
+      "learning_rate": 0.0002428421052631579,
+      "loss": 0.1327,
+      "step": 3132
+    },
+    {
+      "epoch": 46.76,
+      "learning_rate": 0.00024280701754385962,
+      "loss": 0.1199,
+      "step": 3133
+    },
+    {
+      "epoch": 46.77,
+      "learning_rate": 0.00024277192982456136,
+      "loss": 0.358,
+      "step": 3134
+    },
+    {
+      "epoch": 46.79,
+      "learning_rate": 0.00024273684210526314,
+      "loss": 0.08,
+      "step": 3135
+    },
+    {
+      "epoch": 46.8,
+      "learning_rate": 0.0002427017543859649,
+      "loss": 0.3817,
+      "step": 3136
+    },
+    {
+      "epoch": 46.82,
+      "learning_rate": 0.00024266666666666664,
+      "loss": 0.0779,
+      "step": 3137
+    },
+    {
+      "epoch": 46.83,
+      "learning_rate": 0.0002426315789473684,
+      "loss": 0.1329,
+      "step": 3138
+    },
+    {
+      "epoch": 46.85,
+      "learning_rate": 0.00024259649122807016,
+      "loss": 0.1074,
+      "step": 3139
+    },
+    {
+      "epoch": 46.86,
+      "learning_rate": 0.0002425614035087719,
+      "loss": 0.1433,
+      "step": 3140
+    },
+    {
+      "epoch": 46.88,
+      "learning_rate": 0.00024252631578947366,
+      "loss": 0.1577,
+      "step": 3141
+    },
+    {
+      "epoch": 46.89,
+      "learning_rate": 0.0002424912280701754,
+      "loss": 0.3007,
+      "step": 3142
+    },
+    {
+      "epoch": 46.91,
+      "learning_rate": 0.00024245614035087719,
+      "loss": 0.2814,
+      "step": 3143
+    },
+    {
+      "epoch": 46.92,
+      "learning_rate": 0.00024242105263157893,
+      "loss": 0.3594,
+      "step": 3144
+    },
+    {
+      "epoch": 46.94,
+      "learning_rate": 0.00024238596491228068,
+      "loss": 0.1507,
+      "step": 3145
+    },
+    {
+      "epoch": 46.95,
+      "learning_rate": 0.00024235087719298246,
+      "loss": 0.1063,
+      "step": 3146
+    },
+    {
+      "epoch": 46.97,
+      "learning_rate": 0.0002423157894736842,
+      "loss": 0.2693,
+      "step": 3147
+    },
+    {
+      "epoch": 46.98,
+      "learning_rate": 0.00024228070175438593,
+      "loss": 0.112,
+      "step": 3148
+    },
+    {
+      "epoch": 47.0,
+      "learning_rate": 0.00024224561403508768,
+      "loss": 0.3829,
+      "step": 3149
+    },
+    {
+      "epoch": 47.01,
+      "learning_rate": 0.00024221052631578945,
+      "loss": 0.1906,
+      "step": 3150
+    },
+    {
+      "epoch": 47.03,
+      "learning_rate": 0.0002421754385964912,
+      "loss": 0.1587,
+      "step": 3151
+    },
+    {
+      "epoch": 47.04,
+      "learning_rate": 0.00024214035087719295,
+      "loss": 0.1186,
+      "step": 3152
+    },
+    {
+      "epoch": 47.06,
+      "learning_rate": 0.0002421052631578947,
+      "loss": 0.1939,
+      "step": 3153
+    },
+    {
+      "epoch": 47.07,
+      "learning_rate": 0.00024207017543859648,
+      "loss": 0.0562,
+      "step": 3154
+    },
+    {
+      "epoch": 47.09,
+      "learning_rate": 0.00024203508771929822,
+      "loss": 0.2518,
+      "step": 3155
+    },
+    {
+      "epoch": 47.1,
+      "learning_rate": 0.00024199999999999997,
+      "loss": 0.255,
+      "step": 3156
+    },
+    {
+      "epoch": 47.12,
+      "learning_rate": 0.00024196491228070172,
+      "loss": 0.0486,
+      "step": 3157
+    },
+    {
+      "epoch": 47.13,
+      "learning_rate": 0.0002419298245614035,
+      "loss": 0.2381,
+      "step": 3158
+    },
+    {
+      "epoch": 47.15,
+      "learning_rate": 0.00024189473684210525,
+      "loss": 0.1361,
+      "step": 3159
+    },
+    {
+      "epoch": 47.16,
+      "learning_rate": 0.000241859649122807,
+      "loss": 0.157,
+      "step": 3160
+    },
+    {
+      "epoch": 47.18,
+      "learning_rate": 0.00024182456140350877,
+      "loss": 0.5923,
+      "step": 3161
+    },
+    {
+      "epoch": 47.19,
+      "learning_rate": 0.00024178947368421052,
+      "loss": 0.1404,
+      "step": 3162
+    },
+    {
+      "epoch": 47.21,
+      "learning_rate": 0.00024175438596491227,
+      "loss": 0.013,
+      "step": 3163
+    },
+    {
+      "epoch": 47.22,
+      "learning_rate": 0.000241719298245614,
+      "loss": 0.0112,
+      "step": 3164
+    },
+    {
+      "epoch": 47.24,
+      "learning_rate": 0.0002416842105263158,
+      "loss": 0.2291,
+      "step": 3165
+    },
+    {
+      "epoch": 47.25,
+      "learning_rate": 0.00024164912280701752,
+      "loss": 0.1353,
+      "step": 3166
+    },
+    {
+      "epoch": 47.27,
+      "learning_rate": 0.00024161403508771926,
+      "loss": 0.0393,
+      "step": 3167
+    },
+    {
+      "epoch": 47.28,
+      "learning_rate": 0.000241578947368421,
+      "loss": 0.0808,
+      "step": 3168
+    },
+    {
+      "epoch": 47.3,
+      "learning_rate": 0.0002415438596491228,
+      "loss": 0.0112,
+      "step": 3169
+    },
+    {
+      "epoch": 47.31,
+      "learning_rate": 0.00024150877192982454,
+      "loss": 0.0361,
+      "step": 3170
+    },
+    {
+      "epoch": 47.33,
+      "learning_rate": 0.00024147368421052629,
+      "loss": 0.1137,
+      "step": 3171
+    },
+    {
+      "epoch": 47.34,
+      "learning_rate": 0.00024143859649122806,
+      "loss": 0.0466,
+      "step": 3172
+    },
+    {
+      "epoch": 47.36,
+      "learning_rate": 0.0002414035087719298,
+      "loss": 0.0181,
+      "step": 3173
+    },
+    {
+      "epoch": 47.37,
+      "learning_rate": 0.00024136842105263156,
+      "loss": 0.0182,
+      "step": 3174
+    },
+    {
+      "epoch": 47.39,
+      "learning_rate": 0.0002413333333333333,
+      "loss": 0.0145,
+      "step": 3175
+    },
+    {
+      "epoch": 47.4,
+      "learning_rate": 0.00024129824561403508,
+      "loss": 0.2759,
+      "step": 3176
+    },
+    {
+      "epoch": 47.42,
+      "learning_rate": 0.00024126315789473683,
+      "loss": 0.1418,
+      "step": 3177
+    },
+    {
+      "epoch": 47.43,
+      "learning_rate": 0.00024122807017543858,
+      "loss": 0.1672,
+      "step": 3178
+    },
+    {
+      "epoch": 47.45,
+      "learning_rate": 0.0002411929824561403,
+      "loss": 0.1402,
+      "step": 3179
+    },
+    {
+      "epoch": 47.46,
+      "learning_rate": 0.0002411578947368421,
+      "loss": 0.1563,
+      "step": 3180
+    },
+    {
+      "epoch": 47.48,
+      "learning_rate": 0.00024112280701754383,
+      "loss": 0.0914,
+      "step": 3181
+    },
+    {
+      "epoch": 47.49,
+      "learning_rate": 0.00024108771929824558,
+      "loss": 0.0067,
+      "step": 3182
+    },
+    {
+      "epoch": 47.51,
+      "learning_rate": 0.00024105263157894735,
+      "loss": 0.1717,
+      "step": 3183
+    },
+    {
+      "epoch": 47.52,
+      "learning_rate": 0.0002410175438596491,
+      "loss": 0.2388,
+      "step": 3184
+    },
+    {
+      "epoch": 47.54,
+      "learning_rate": 0.00024098245614035085,
+      "loss": 0.561,
+      "step": 3185
+    },
+    {
+      "epoch": 47.55,
+      "learning_rate": 0.0002409473684210526,
+      "loss": 0.2598,
+      "step": 3186
+    },
+    {
+      "epoch": 47.57,
+      "learning_rate": 0.00024091228070175438,
+      "loss": 0.0856,
+      "step": 3187
+    },
+    {
+      "epoch": 47.58,
+      "learning_rate": 0.00024087719298245612,
+      "loss": 0.2389,
+      "step": 3188
+    },
+    {
+      "epoch": 47.59,
+      "learning_rate": 0.00024084210526315787,
+      "loss": 0.0659,
+      "step": 3189
+    },
+    {
+      "epoch": 47.61,
+      "learning_rate": 0.00024080701754385962,
+      "loss": 0.1238,
+      "step": 3190
+    },
+    {
+      "epoch": 47.62,
+      "learning_rate": 0.0002407719298245614,
+      "loss": 0.0759,
+      "step": 3191
+    },
+    {
+      "epoch": 47.64,
+      "learning_rate": 0.00024073684210526315,
+      "loss": 0.1634,
+      "step": 3192
+    },
+    {
+      "epoch": 47.65,
+      "learning_rate": 0.0002407017543859649,
+      "loss": 0.184,
+      "step": 3193
+    },
+    {
+      "epoch": 47.67,
+      "learning_rate": 0.00024066666666666662,
+      "loss": 0.2784,
+      "step": 3194
+    },
+    {
+      "epoch": 47.68,
+      "learning_rate": 0.00024063157894736842,
+      "loss": 0.1188,
+      "step": 3195
+    },
+    {
+      "epoch": 47.7,
+      "learning_rate": 0.00024059649122807014,
+      "loss": 0.0494,
+      "step": 3196
+    },
+    {
+      "epoch": 47.71,
+      "learning_rate": 0.0002405614035087719,
+      "loss": 0.0575,
+      "step": 3197
+    },
+    {
+      "epoch": 47.73,
+      "learning_rate": 0.00024052631578947367,
+      "loss": 0.2455,
+      "step": 3198
+    },
+    {
+      "epoch": 47.74,
+      "learning_rate": 0.00024049122807017541,
+      "loss": 0.1662,
+      "step": 3199
+    },
+    {
+      "epoch": 47.76,
+      "learning_rate": 0.00024045614035087716,
+      "loss": 0.2306,
+      "step": 3200
+    },
+    {
+      "epoch": 47.76,
+      "eval_accuracy": 0.8252569750367107,
+      "eval_f1": 0.8263631777806041,
+      "eval_loss": 0.6788427233695984,
+      "eval_runtime": 344.1209,
+      "eval_samples_per_second": 11.874,
+      "eval_steps_per_second": 0.744,
+      "step": 3200
+    },
+    {
+      "epoch": 47.77,
+      "learning_rate": 0.0002404210526315789,
+      "loss": 0.0692,
+      "step": 3201
+    },
+    {
+      "epoch": 47.79,
+      "learning_rate": 0.0002403859649122807,
+      "loss": 0.0733,
+      "step": 3202
+    },
+    {
+      "epoch": 47.8,
+      "learning_rate": 0.00024035087719298244,
+      "loss": 0.2759,
+      "step": 3203
+    },
+    {
+      "epoch": 47.82,
+      "learning_rate": 0.00024031578947368419,
+      "loss": 0.0444,
+      "step": 3204
+    },
+    {
+      "epoch": 47.83,
+      "learning_rate": 0.00024028070175438593,
+      "loss": 0.5526,
+      "step": 3205
+    },
+    {
+      "epoch": 47.85,
+      "learning_rate": 0.0002402456140350877,
+      "loss": 0.1555,
+      "step": 3206
+    },
+    {
+      "epoch": 47.86,
+      "learning_rate": 0.00024021052631578946,
+      "loss": 0.1475,
+      "step": 3207
+    },
+    {
+      "epoch": 47.88,
+      "learning_rate": 0.0002401754385964912,
+      "loss": 0.1136,
+      "step": 3208
+    },
+    {
+      "epoch": 47.89,
+      "learning_rate": 0.00024014035087719298,
+      "loss": 0.0093,
+      "step": 3209
+    },
+    {
+      "epoch": 47.91,
+      "learning_rate": 0.00024010526315789473,
+      "loss": 0.0347,
+      "step": 3210
+    },
+    {
+      "epoch": 47.92,
+      "learning_rate": 0.00024007017543859648,
+      "loss": 0.1611,
+      "step": 3211
+    },
+    {
+      "epoch": 47.94,
+      "learning_rate": 0.0002400350877192982,
+      "loss": 0.103,
+      "step": 3212
+    },
+    {
+      "epoch": 47.95,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 0.1416,
+      "step": 3213
+    },
+    {
+      "epoch": 47.97,
+      "learning_rate": 0.00023996491228070173,
+      "loss": 0.3555,
+      "step": 3214
+    },
+    {
+      "epoch": 47.98,
+      "learning_rate": 0.00023992982456140348,
+      "loss": 0.3259,
+      "step": 3215
+    },
+    {
+      "epoch": 48.0,
+      "learning_rate": 0.00023989473684210523,
+      "loss": 0.0298,
+      "step": 3216
+    },
+    {
+      "epoch": 48.01,
+      "learning_rate": 0.000239859649122807,
+      "loss": 0.3933,
+      "step": 3217
+    },
+    {
+      "epoch": 48.03,
+      "learning_rate": 0.00023982456140350875,
+      "loss": 0.0306,
+      "step": 3218
+    },
+    {
+      "epoch": 48.04,
+      "learning_rate": 0.0002397894736842105,
+      "loss": 0.242,
+      "step": 3219
+    },
+    {
+      "epoch": 48.06,
+      "learning_rate": 0.00023975438596491225,
+      "loss": 0.0583,
+      "step": 3220
+    },
+    {
+      "epoch": 48.07,
+      "learning_rate": 0.00023971929824561402,
+      "loss": 0.0489,
+      "step": 3221
+    },
+    {
+      "epoch": 48.09,
+      "learning_rate": 0.00023968421052631577,
+      "loss": 0.2804,
+      "step": 3222
+    },
+    {
+      "epoch": 48.1,
+      "learning_rate": 0.00023964912280701752,
+      "loss": 0.2151,
+      "step": 3223
+    },
+    {
+      "epoch": 48.12,
+      "learning_rate": 0.0002396140350877193,
+      "loss": 0.1241,
+      "step": 3224
+    },
+    {
+      "epoch": 48.13,
+      "learning_rate": 0.00023957894736842105,
+      "loss": 0.0905,
+      "step": 3225
+    },
+    {
+      "epoch": 48.15,
+      "learning_rate": 0.0002395438596491228,
+      "loss": 0.0301,
+      "step": 3226
+    },
+    {
+      "epoch": 48.16,
+      "learning_rate": 0.00023950877192982452,
+      "loss": 0.045,
+      "step": 3227
+    },
+    {
+      "epoch": 48.18,
+      "learning_rate": 0.00023947368421052632,
+      "loss": 0.1225,
+      "step": 3228
+    },
+    {
+      "epoch": 48.19,
+      "learning_rate": 0.00023943859649122804,
+      "loss": 0.0339,
+      "step": 3229
+    },
+    {
+      "epoch": 48.21,
+      "learning_rate": 0.0002394035087719298,
+      "loss": 0.0699,
+      "step": 3230
+    },
+    {
+      "epoch": 48.22,
+      "learning_rate": 0.00023936842105263154,
+      "loss": 0.0094,
+      "step": 3231
+    },
+    {
+      "epoch": 48.24,
+      "learning_rate": 0.00023933333333333331,
+      "loss": 0.3309,
+      "step": 3232
+    },
+    {
+      "epoch": 48.25,
+      "learning_rate": 0.00023929824561403506,
+      "loss": 0.0171,
+      "step": 3233
+    },
+    {
+      "epoch": 48.27,
+      "learning_rate": 0.0002392631578947368,
+      "loss": 0.1391,
+      "step": 3234
+    },
+    {
+      "epoch": 48.28,
+      "learning_rate": 0.0002392280701754386,
+      "loss": 0.3318,
+      "step": 3235
+    },
+    {
+      "epoch": 48.3,
+      "learning_rate": 0.00023919298245614034,
+      "loss": 0.0258,
+      "step": 3236
+    },
+    {
+      "epoch": 48.31,
+      "learning_rate": 0.00023915789473684209,
+      "loss": 0.0946,
+      "step": 3237
+    },
+    {
+      "epoch": 48.33,
+      "learning_rate": 0.00023912280701754383,
+      "loss": 0.2643,
+      "step": 3238
+    },
+    {
+      "epoch": 48.34,
+      "learning_rate": 0.0002390877192982456,
+      "loss": 0.3604,
+      "step": 3239
+    },
+    {
+      "epoch": 48.36,
+      "learning_rate": 0.00023905263157894736,
+      "loss": 0.1013,
+      "step": 3240
+    },
+    {
+      "epoch": 48.37,
+      "learning_rate": 0.0002390175438596491,
+      "loss": 0.3569,
+      "step": 3241
+    },
+    {
+      "epoch": 48.39,
+      "learning_rate": 0.00023898245614035083,
+      "loss": 0.0471,
+      "step": 3242
+    },
+    {
+      "epoch": 48.4,
+      "learning_rate": 0.00023894736842105263,
+      "loss": 0.1155,
+      "step": 3243
+    },
+    {
+      "epoch": 48.42,
+      "learning_rate": 0.00023891228070175435,
+      "loss": 0.0065,
+      "step": 3244
+    },
+    {
+      "epoch": 48.43,
+      "learning_rate": 0.0002388771929824561,
+      "loss": 0.0241,
+      "step": 3245
+    },
+    {
+      "epoch": 48.45,
+      "learning_rate": 0.00023884210526315788,
+      "loss": 0.0109,
+      "step": 3246
+    },
+    {
+      "epoch": 48.46,
+      "learning_rate": 0.00023880701754385963,
+      "loss": 0.0422,
+      "step": 3247
+    },
+    {
+      "epoch": 48.48,
+      "learning_rate": 0.00023877192982456138,
+      "loss": 0.1441,
+      "step": 3248
+    },
+    {
+      "epoch": 48.49,
+      "learning_rate": 0.00023873684210526312,
+      "loss": 0.0596,
+      "step": 3249
+    },
+    {
+      "epoch": 48.51,
+      "learning_rate": 0.0002387017543859649,
+      "loss": 0.0554,
+      "step": 3250
+    },
+    {
+      "epoch": 48.52,
+      "learning_rate": 0.00023866666666666665,
+      "loss": 0.1545,
+      "step": 3251
+    },
+    {
+      "epoch": 48.54,
+      "learning_rate": 0.0002386315789473684,
+      "loss": 0.235,
+      "step": 3252
+    },
+    {
+      "epoch": 48.55,
+      "learning_rate": 0.00023859649122807015,
+      "loss": 0.2796,
+      "step": 3253
+    },
+    {
+      "epoch": 48.57,
+      "learning_rate": 0.00023856140350877192,
+      "loss": 0.1655,
+      "step": 3254
+    },
+    {
+      "epoch": 48.58,
+      "learning_rate": 0.00023852631578947367,
+      "loss": 0.0283,
+      "step": 3255
+    },
+    {
+      "epoch": 48.59,
+      "learning_rate": 0.00023849122807017542,
+      "loss": 0.15,
+      "step": 3256
+    },
+    {
+      "epoch": 48.61,
+      "learning_rate": 0.00023845614035087717,
+      "loss": 0.1791,
+      "step": 3257
+    },
+    {
+      "epoch": 48.62,
+      "learning_rate": 0.00023842105263157895,
+      "loss": 0.167,
+      "step": 3258
+    },
+    {
+      "epoch": 48.64,
+      "learning_rate": 0.0002383859649122807,
+      "loss": 0.1388,
+      "step": 3259
+    },
+    {
+      "epoch": 48.65,
+      "learning_rate": 0.00023835087719298242,
+      "loss": 0.1566,
+      "step": 3260
+    },
+    {
+      "epoch": 48.67,
+      "learning_rate": 0.0002383157894736842,
+      "loss": 0.0098,
+      "step": 3261
+    },
+    {
+      "epoch": 48.68,
+      "learning_rate": 0.00023828070175438594,
+      "loss": 0.06,
+      "step": 3262
+    },
+    {
+      "epoch": 48.7,
+      "learning_rate": 0.0002382456140350877,
+      "loss": 0.2902,
+      "step": 3263
+    },
+    {
+      "epoch": 48.71,
+      "learning_rate": 0.00023821052631578944,
+      "loss": 0.0044,
+      "step": 3264
+    },
+    {
+      "epoch": 48.73,
+      "learning_rate": 0.00023817543859649121,
+      "loss": 0.3811,
+      "step": 3265
+    },
+    {
+      "epoch": 48.74,
+      "learning_rate": 0.00023814035087719296,
+      "loss": 0.5167,
+      "step": 3266
+    },
+    {
+      "epoch": 48.76,
+      "learning_rate": 0.0002381052631578947,
+      "loss": 0.1743,
+      "step": 3267
+    },
+    {
+      "epoch": 48.77,
+      "learning_rate": 0.00023807017543859646,
+      "loss": 0.0538,
+      "step": 3268
+    },
+    {
+      "epoch": 48.79,
+      "learning_rate": 0.00023803508771929824,
+      "loss": 0.4199,
+      "step": 3269
+    },
+    {
+      "epoch": 48.8,
+      "learning_rate": 0.00023799999999999998,
+      "loss": 0.2232,
+      "step": 3270
+    },
+    {
+      "epoch": 48.82,
+      "learning_rate": 0.00023796491228070173,
+      "loss": 0.1179,
+      "step": 3271
+    },
+    {
+      "epoch": 48.83,
+      "learning_rate": 0.0002379298245614035,
+      "loss": 0.2499,
+      "step": 3272
+    },
+    {
+      "epoch": 48.85,
+      "learning_rate": 0.00023789473684210526,
+      "loss": 0.2151,
+      "step": 3273
+    },
+    {
+      "epoch": 48.86,
+      "learning_rate": 0.000237859649122807,
+      "loss": 0.1367,
+      "step": 3274
+    },
+    {
+      "epoch": 48.88,
+      "learning_rate": 0.00023782456140350873,
+      "loss": 0.0687,
+      "step": 3275
+    },
+    {
+      "epoch": 48.89,
+      "learning_rate": 0.00023778947368421053,
+      "loss": 0.0195,
+      "step": 3276
+    },
+    {
+      "epoch": 48.91,
+      "learning_rate": 0.00023775438596491225,
+      "loss": 0.2018,
+      "step": 3277
+    },
+    {
+      "epoch": 48.92,
+      "learning_rate": 0.000237719298245614,
+      "loss": 0.187,
+      "step": 3278
+    },
+    {
+      "epoch": 48.94,
+      "learning_rate": 0.00023768421052631575,
+      "loss": 0.1456,
+      "step": 3279
+    },
+    {
+      "epoch": 48.95,
+      "learning_rate": 0.00023764912280701753,
+      "loss": 0.152,
+      "step": 3280
+    },
+    {
+      "epoch": 48.97,
+      "learning_rate": 0.00023761403508771928,
+      "loss": 0.0065,
+      "step": 3281
+    },
+    {
+      "epoch": 48.98,
+      "learning_rate": 0.00023757894736842102,
+      "loss": 0.9299,
+      "step": 3282
+    },
+    {
+      "epoch": 49.0,
+      "learning_rate": 0.00023754385964912277,
+      "loss": 0.2614,
+      "step": 3283
+    },
+    {
+      "epoch": 49.01,
+      "learning_rate": 0.00023750877192982455,
+      "loss": 0.3121,
+      "step": 3284
+    },
+    {
+      "epoch": 49.03,
+      "learning_rate": 0.0002374736842105263,
+      "loss": 0.2731,
+      "step": 3285
+    },
+    {
+      "epoch": 49.04,
+      "learning_rate": 0.00023743859649122805,
+      "loss": 0.5073,
+      "step": 3286
+    },
+    {
+      "epoch": 49.06,
+      "learning_rate": 0.00023740350877192982,
+      "loss": 0.1915,
+      "step": 3287
+    },
+    {
+      "epoch": 49.07,
+      "learning_rate": 0.00023736842105263157,
+      "loss": 0.1298,
+      "step": 3288
+    },
+    {
+      "epoch": 49.09,
+      "learning_rate": 0.00023733333333333332,
+      "loss": 0.1686,
+      "step": 3289
+    },
+    {
+      "epoch": 49.1,
+      "learning_rate": 0.00023729824561403504,
+      "loss": 0.0814,
+      "step": 3290
+    },
+    {
+      "epoch": 49.12,
+      "learning_rate": 0.00023726315789473684,
+      "loss": 0.221,
+      "step": 3291
+    },
+    {
+      "epoch": 49.13,
+      "learning_rate": 0.00023722807017543857,
+      "loss": 0.1197,
+      "step": 3292
+    },
+    {
+      "epoch": 49.15,
+      "learning_rate": 0.00023719298245614031,
+      "loss": 0.4632,
+      "step": 3293
+    },
+    {
+      "epoch": 49.16,
+      "learning_rate": 0.00023715789473684206,
+      "loss": 0.4504,
+      "step": 3294
+    },
+    {
+      "epoch": 49.18,
+      "learning_rate": 0.00023712280701754384,
+      "loss": 0.0194,
+      "step": 3295
+    },
+    {
+      "epoch": 49.19,
+      "learning_rate": 0.0002370877192982456,
+      "loss": 0.1115,
+      "step": 3296
+    },
+    {
+      "epoch": 49.21,
+      "learning_rate": 0.00023705263157894734,
+      "loss": 0.194,
+      "step": 3297
+    },
+    {
+      "epoch": 49.22,
+      "learning_rate": 0.0002370175438596491,
+      "loss": 0.1706,
+      "step": 3298
+    },
+    {
+      "epoch": 49.24,
+      "learning_rate": 0.00023698245614035086,
+      "loss": 0.2053,
+      "step": 3299
+    },
+    {
+      "epoch": 49.25,
+      "learning_rate": 0.0002369473684210526,
+      "loss": 0.0102,
+      "step": 3300
+    },
+    {
+      "epoch": 49.27,
+      "learning_rate": 0.00023691228070175436,
+      "loss": 0.1797,
+      "step": 3301
+    },
+    {
+      "epoch": 49.28,
+      "learning_rate": 0.00023687719298245614,
+      "loss": 0.1465,
+      "step": 3302
+    },
+    {
+      "epoch": 49.3,
+      "learning_rate": 0.00023684210526315788,
+      "loss": 0.2035,
+      "step": 3303
+    },
+    {
+      "epoch": 49.31,
+      "learning_rate": 0.00023680701754385963,
+      "loss": 0.266,
+      "step": 3304
+    },
+    {
+      "epoch": 49.33,
+      "learning_rate": 0.00023677192982456138,
+      "loss": 0.2117,
+      "step": 3305
+    },
+    {
+      "epoch": 49.34,
+      "learning_rate": 0.00023673684210526316,
+      "loss": 0.1285,
+      "step": 3306
+    },
+    {
+      "epoch": 49.36,
+      "learning_rate": 0.00023670175438596488,
+      "loss": 0.156,
+      "step": 3307
+    },
+    {
+      "epoch": 49.37,
+      "learning_rate": 0.00023666666666666663,
+      "loss": 0.2169,
+      "step": 3308
+    },
+    {
+      "epoch": 49.39,
+      "learning_rate": 0.0002366315789473684,
+      "loss": 0.0067,
+      "step": 3309
+    },
+    {
+      "epoch": 49.4,
+      "learning_rate": 0.00023659649122807015,
+      "loss": 0.0784,
+      "step": 3310
+    },
+    {
+      "epoch": 49.42,
+      "learning_rate": 0.0002365614035087719,
+      "loss": 0.1451,
+      "step": 3311
+    },
+    {
+      "epoch": 49.43,
+      "learning_rate": 0.00023652631578947365,
+      "loss": 0.0804,
+      "step": 3312
+    },
+    {
+      "epoch": 49.45,
+      "learning_rate": 0.00023649122807017543,
+      "loss": 0.155,
+      "step": 3313
+    },
+    {
+      "epoch": 49.46,
+      "learning_rate": 0.00023645614035087717,
+      "loss": 0.1286,
+      "step": 3314
+    },
+    {
+      "epoch": 49.48,
+      "learning_rate": 0.00023642105263157892,
+      "loss": 0.0605,
+      "step": 3315
+    },
+    {
+      "epoch": 49.49,
+      "learning_rate": 0.00023638596491228067,
+      "loss": 0.0368,
+      "step": 3316
+    },
+    {
+      "epoch": 49.51,
+      "learning_rate": 0.00023635087719298245,
+      "loss": 0.3509,
+      "step": 3317
+    },
+    {
+      "epoch": 49.52,
+      "learning_rate": 0.0002363157894736842,
+      "loss": 0.226,
+      "step": 3318
+    },
+    {
+      "epoch": 49.54,
+      "learning_rate": 0.00023628070175438595,
+      "loss": 0.2064,
+      "step": 3319
+    },
+    {
+      "epoch": 49.55,
+      "learning_rate": 0.0002362456140350877,
+      "loss": 0.0506,
+      "step": 3320
+    },
+    {
+      "epoch": 49.57,
+      "learning_rate": 0.00023621052631578947,
+      "loss": 0.0928,
+      "step": 3321
+    },
+    {
+      "epoch": 49.58,
+      "learning_rate": 0.00023617543859649122,
+      "loss": 0.1743,
+      "step": 3322
+    },
+    {
+      "epoch": 49.59,
+      "learning_rate": 0.00023614035087719294,
+      "loss": 0.0956,
+      "step": 3323
+    },
+    {
+      "epoch": 49.61,
+      "learning_rate": 0.00023610526315789474,
+      "loss": 0.2717,
+      "step": 3324
+    },
+    {
+      "epoch": 49.62,
+      "learning_rate": 0.00023607017543859647,
+      "loss": 0.0131,
+      "step": 3325
+    },
+    {
+      "epoch": 49.64,
+      "learning_rate": 0.00023603508771929821,
+      "loss": 0.1284,
+      "step": 3326
+    },
+    {
+      "epoch": 49.65,
+      "learning_rate": 0.00023599999999999996,
+      "loss": 0.0224,
+      "step": 3327
+    },
+    {
+      "epoch": 49.67,
+      "learning_rate": 0.00023596491228070174,
+      "loss": 0.2971,
+      "step": 3328
+    },
+    {
+      "epoch": 49.68,
+      "learning_rate": 0.0002359298245614035,
+      "loss": 0.0894,
+      "step": 3329
+    },
+    {
+      "epoch": 49.7,
+      "learning_rate": 0.00023589473684210524,
+      "loss": 0.0096,
+      "step": 3330
+    },
+    {
+      "epoch": 49.71,
+      "learning_rate": 0.00023585964912280699,
+      "loss": 0.1581,
+      "step": 3331
+    },
+    {
+      "epoch": 49.73,
+      "learning_rate": 0.00023582456140350876,
+      "loss": 0.0086,
+      "step": 3332
+    },
+    {
+      "epoch": 49.74,
+      "learning_rate": 0.0002357894736842105,
+      "loss": 0.2042,
+      "step": 3333
+    },
+    {
+      "epoch": 49.76,
+      "learning_rate": 0.00023575438596491226,
+      "loss": 0.0336,
+      "step": 3334
+    },
+    {
+      "epoch": 49.77,
+      "learning_rate": 0.00023571929824561403,
+      "loss": 0.047,
+      "step": 3335
+    },
+    {
+      "epoch": 49.79,
+      "learning_rate": 0.00023568421052631578,
+      "loss": 0.114,
+      "step": 3336
+    },
+    {
+      "epoch": 49.8,
+      "learning_rate": 0.00023564912280701753,
+      "loss": 0.0184,
+      "step": 3337
+    },
+    {
+      "epoch": 49.82,
+      "learning_rate": 0.00023561403508771925,
+      "loss": 0.2492,
+      "step": 3338
+    },
+    {
+      "epoch": 49.83,
+      "learning_rate": 0.00023557894736842106,
+      "loss": 0.0591,
+      "step": 3339
+    },
+    {
+      "epoch": 49.85,
+      "learning_rate": 0.00023554385964912278,
+      "loss": 0.0539,
+      "step": 3340
+    },
+    {
+      "epoch": 49.86,
+      "learning_rate": 0.00023550877192982453,
+      "loss": 0.0631,
+      "step": 3341
+    },
+    {
+      "epoch": 49.88,
+      "learning_rate": 0.00023547368421052628,
+      "loss": 0.0913,
+      "step": 3342
+    },
+    {
+      "epoch": 49.89,
+      "learning_rate": 0.00023543859649122805,
+      "loss": 0.1166,
+      "step": 3343
+    },
+    {
+      "epoch": 49.91,
+      "learning_rate": 0.0002354035087719298,
+      "loss": 0.0261,
+      "step": 3344
+    },
+    {
+      "epoch": 49.92,
+      "learning_rate": 0.00023536842105263155,
+      "loss": 0.4387,
+      "step": 3345
+    },
+    {
+      "epoch": 49.94,
+      "learning_rate": 0.00023533333333333333,
+      "loss": 0.1136,
+      "step": 3346
+    },
+    {
+      "epoch": 49.95,
+      "learning_rate": 0.00023529824561403507,
+      "loss": 0.0142,
+      "step": 3347
+    },
+    {
+      "epoch": 49.97,
+      "learning_rate": 0.00023526315789473682,
+      "loss": 0.1371,
+      "step": 3348
+    },
+    {
+      "epoch": 49.98,
+      "learning_rate": 0.00023522807017543857,
+      "loss": 0.0146,
+      "step": 3349
+    },
+    {
+      "epoch": 50.0,
+      "learning_rate": 0.00023519298245614035,
+      "loss": 0.006,
+      "step": 3350
+    },
+    {
+      "epoch": 50.01,
+      "learning_rate": 0.0002351578947368421,
+      "loss": 0.013,
+      "step": 3351
+    },
+    {
+      "epoch": 50.03,
+      "learning_rate": 0.00023512280701754385,
+      "loss": 0.0484,
+      "step": 3352
+    },
+    {
+      "epoch": 50.04,
+      "learning_rate": 0.0002350877192982456,
+      "loss": 0.1242,
+      "step": 3353
+    },
+    {
+      "epoch": 50.06,
+      "learning_rate": 0.00023505263157894737,
+      "loss": 0.158,
+      "step": 3354
+    },
+    {
+      "epoch": 50.07,
+      "learning_rate": 0.0002350175438596491,
+      "loss": 0.2993,
+      "step": 3355
+    },
+    {
+      "epoch": 50.09,
+      "learning_rate": 0.00023498245614035084,
+      "loss": 0.035,
+      "step": 3356
+    },
+    {
+      "epoch": 50.1,
+      "learning_rate": 0.0002349473684210526,
+      "loss": 0.0617,
+      "step": 3357
+    },
+    {
+      "epoch": 50.12,
+      "learning_rate": 0.00023491228070175436,
+      "loss": 0.2287,
+      "step": 3358
+    },
+    {
+      "epoch": 50.13,
+      "learning_rate": 0.00023487719298245611,
+      "loss": 0.1619,
+      "step": 3359
+    },
+    {
+      "epoch": 50.15,
+      "learning_rate": 0.00023484210526315786,
+      "loss": 0.0349,
+      "step": 3360
+    },
+    {
+      "epoch": 50.16,
+      "learning_rate": 0.00023480701754385964,
+      "loss": 0.0262,
+      "step": 3361
+    },
+    {
+      "epoch": 50.18,
+      "learning_rate": 0.0002347719298245614,
+      "loss": 0.0156,
+      "step": 3362
+    },
+    {
+      "epoch": 50.19,
+      "learning_rate": 0.00023473684210526314,
+      "loss": 0.0146,
+      "step": 3363
+    },
+    {
+      "epoch": 50.21,
+      "learning_rate": 0.00023470175438596488,
+      "loss": 0.0343,
+      "step": 3364
+    },
+    {
+      "epoch": 50.22,
+      "learning_rate": 0.00023466666666666666,
+      "loss": 0.0789,
+      "step": 3365
+    },
+    {
+      "epoch": 50.24,
+      "learning_rate": 0.0002346315789473684,
+      "loss": 0.0122,
+      "step": 3366
+    },
+    {
+      "epoch": 50.25,
+      "learning_rate": 0.00023459649122807016,
+      "loss": 0.0105,
+      "step": 3367
+    },
+    {
+      "epoch": 50.27,
+      "learning_rate": 0.0002345614035087719,
+      "loss": 0.1957,
+      "step": 3368
+    },
+    {
+      "epoch": 50.28,
+      "learning_rate": 0.00023452631578947368,
+      "loss": 0.2115,
+      "step": 3369
+    },
+    {
+      "epoch": 50.3,
+      "learning_rate": 0.00023449122807017543,
+      "loss": 0.242,
+      "step": 3370
+    },
+    {
+      "epoch": 50.31,
+      "learning_rate": 0.00023445614035087715,
+      "loss": 0.3058,
+      "step": 3371
+    },
+    {
+      "epoch": 50.33,
+      "learning_rate": 0.00023442105263157896,
+      "loss": 0.0748,
+      "step": 3372
+    },
+    {
+      "epoch": 50.34,
+      "learning_rate": 0.00023438596491228068,
+      "loss": 0.0155,
+      "step": 3373
+    },
+    {
+      "epoch": 50.36,
+      "learning_rate": 0.00023435087719298243,
+      "loss": 0.3034,
+      "step": 3374
+    },
+    {
+      "epoch": 50.37,
+      "learning_rate": 0.00023431578947368418,
+      "loss": 0.0136,
+      "step": 3375
+    },
+    {
+      "epoch": 50.39,
+      "learning_rate": 0.00023428070175438595,
+      "loss": 0.1864,
+      "step": 3376
+    },
+    {
+      "epoch": 50.4,
+      "learning_rate": 0.0002342456140350877,
+      "loss": 0.1253,
+      "step": 3377
+    },
+    {
+      "epoch": 50.42,
+      "learning_rate": 0.00023421052631578945,
+      "loss": 0.1835,
+      "step": 3378
+    },
+    {
+      "epoch": 50.43,
+      "learning_rate": 0.0002341754385964912,
+      "loss": 0.1046,
+      "step": 3379
+    },
+    {
+      "epoch": 50.45,
+      "learning_rate": 0.00023414035087719297,
+      "loss": 0.036,
+      "step": 3380
+    },
+    {
+      "epoch": 50.46,
+      "learning_rate": 0.00023410526315789472,
+      "loss": 0.0947,
+      "step": 3381
+    },
+    {
+      "epoch": 50.48,
+      "learning_rate": 0.00023407017543859647,
+      "loss": 0.0058,
+      "step": 3382
+    },
+    {
+      "epoch": 50.49,
+      "learning_rate": 0.00023403508771929822,
+      "loss": 0.1306,
+      "step": 3383
+    },
+    {
+      "epoch": 50.51,
+      "learning_rate": 0.000234,
+      "loss": 0.1747,
+      "step": 3384
+    },
+    {
+      "epoch": 50.52,
+      "learning_rate": 0.00023396491228070174,
+      "loss": 0.037,
+      "step": 3385
+    },
+    {
+      "epoch": 50.54,
+      "learning_rate": 0.00023392982456140347,
+      "loss": 0.2375,
+      "step": 3386
+    },
+    {
+      "epoch": 50.55,
+      "learning_rate": 0.00023389473684210527,
+      "loss": 0.4548,
+      "step": 3387
+    },
+    {
+      "epoch": 50.57,
+      "learning_rate": 0.000233859649122807,
+      "loss": 0.0223,
+      "step": 3388
+    },
+    {
+      "epoch": 50.58,
+      "learning_rate": 0.00023382456140350874,
+      "loss": 0.0359,
+      "step": 3389
+    },
+    {
+      "epoch": 50.59,
+      "learning_rate": 0.0002337894736842105,
+      "loss": 0.1929,
+      "step": 3390
+    },
+    {
+      "epoch": 50.61,
+      "learning_rate": 0.00023375438596491226,
+      "loss": 0.0979,
+      "step": 3391
+    },
+    {
+      "epoch": 50.62,
+      "learning_rate": 0.000233719298245614,
+      "loss": 0.0213,
+      "step": 3392
+    },
+    {
+      "epoch": 50.64,
+      "learning_rate": 0.00023368421052631576,
+      "loss": 0.05,
+      "step": 3393
+    },
+    {
+      "epoch": 50.65,
+      "learning_rate": 0.0002336491228070175,
+      "loss": 0.2083,
+      "step": 3394
+    },
+    {
+      "epoch": 50.67,
+      "learning_rate": 0.0002336140350877193,
+      "loss": 0.5703,
+      "step": 3395
+    },
+    {
+      "epoch": 50.68,
+      "learning_rate": 0.00023357894736842104,
+      "loss": 0.0331,
+      "step": 3396
+    },
+    {
+      "epoch": 50.7,
+      "learning_rate": 0.00023354385964912278,
+      "loss": 0.0684,
+      "step": 3397
+    },
+    {
+      "epoch": 50.71,
+      "learning_rate": 0.00023350877192982456,
+      "loss": 0.1066,
+      "step": 3398
+    },
+    {
+      "epoch": 50.73,
+      "learning_rate": 0.0002334736842105263,
+      "loss": 0.1049,
+      "step": 3399
+    },
+    {
+      "epoch": 50.74,
+      "learning_rate": 0.00023343859649122806,
+      "loss": 0.0975,
+      "step": 3400
+    },
+    {
+      "epoch": 50.74,
+      "eval_accuracy": 0.828928046989721,
+      "eval_f1": 0.8295199623127844,
+      "eval_loss": 0.723612368106842,
+      "eval_runtime": 345.274,
+      "eval_samples_per_second": 11.834,
+      "eval_steps_per_second": 0.741,
+      "step": 3400
+    },
+    {
+      "epoch": 50.76,
+      "learning_rate": 0.00023340350877192978,
+      "loss": 0.0765,
+      "step": 3401
+    },
+    {
+      "epoch": 50.77,
+      "learning_rate": 0.00023336842105263158,
+      "loss": 0.0704,
+      "step": 3402
+    },
+    {
+      "epoch": 50.79,
+      "learning_rate": 0.0002333333333333333,
+      "loss": 0.0318,
+      "step": 3403
+    },
+    {
+      "epoch": 50.8,
+      "learning_rate": 0.00023329824561403505,
+      "loss": 0.2962,
+      "step": 3404
+    },
+    {
+      "epoch": 50.82,
+      "learning_rate": 0.0002332631578947368,
+      "loss": 0.0897,
+      "step": 3405
+    },
+    {
+      "epoch": 50.83,
+      "learning_rate": 0.00023322807017543858,
+      "loss": 0.3142,
+      "step": 3406
+    },
+    {
+      "epoch": 50.85,
+      "learning_rate": 0.00023319298245614033,
+      "loss": 0.2082,
+      "step": 3407
+    },
+    {
+      "epoch": 50.86,
+      "learning_rate": 0.00023315789473684207,
+      "loss": 0.0214,
+      "step": 3408
+    },
+    {
+      "epoch": 50.88,
+      "learning_rate": 0.00023312280701754385,
+      "loss": 0.0919,
+      "step": 3409
+    },
+    {
+      "epoch": 50.89,
+      "learning_rate": 0.0002330877192982456,
+      "loss": 0.1899,
+      "step": 3410
+    },
+    {
+      "epoch": 50.91,
+      "learning_rate": 0.00023305263157894735,
+      "loss": 0.1403,
+      "step": 3411
+    },
+    {
+      "epoch": 50.92,
+      "learning_rate": 0.0002330175438596491,
+      "loss": 0.0434,
+      "step": 3412
+    },
+    {
+      "epoch": 50.94,
+      "learning_rate": 0.00023298245614035087,
+      "loss": 0.0165,
+      "step": 3413
+    },
+    {
+      "epoch": 50.95,
+      "learning_rate": 0.00023294736842105262,
+      "loss": 0.1556,
+      "step": 3414
+    },
+    {
+      "epoch": 50.97,
+      "learning_rate": 0.00023291228070175437,
+      "loss": 0.1885,
+      "step": 3415
+    },
+    {
+      "epoch": 50.98,
+      "learning_rate": 0.00023287719298245612,
+      "loss": 0.0331,
+      "step": 3416
+    },
+    {
+      "epoch": 51.0,
+      "learning_rate": 0.0002328421052631579,
+      "loss": 0.1439,
+      "step": 3417
+    },
+    {
+      "epoch": 51.01,
+      "learning_rate": 0.00023280701754385964,
+      "loss": 0.1991,
+      "step": 3418
+    },
+    {
+      "epoch": 51.03,
+      "learning_rate": 0.00023277192982456137,
+      "loss": 0.2616,
+      "step": 3419
+    },
+    {
+      "epoch": 51.04,
+      "learning_rate": 0.00023273684210526311,
+      "loss": 0.0325,
+      "step": 3420
+    },
+    {
+      "epoch": 51.06,
+      "learning_rate": 0.0002327017543859649,
+      "loss": 0.1062,
+      "step": 3421
+    },
+    {
+      "epoch": 51.07,
+      "learning_rate": 0.00023266666666666664,
+      "loss": 0.0539,
+      "step": 3422
+    },
+    {
+      "epoch": 51.09,
+      "learning_rate": 0.0002326315789473684,
+      "loss": 0.3147,
+      "step": 3423
+    },
+    {
+      "epoch": 51.1,
+      "learning_rate": 0.00023259649122807016,
+      "loss": 0.2051,
+      "step": 3424
+    },
+    {
+      "epoch": 51.12,
+      "learning_rate": 0.0002325614035087719,
+      "loss": 0.0063,
+      "step": 3425
+    },
+    {
+      "epoch": 51.13,
+      "learning_rate": 0.00023252631578947366,
+      "loss": 0.1891,
+      "step": 3426
+    },
+    {
+      "epoch": 51.15,
+      "learning_rate": 0.0002324912280701754,
+      "loss": 0.2393,
+      "step": 3427
+    },
+    {
+      "epoch": 51.16,
+      "learning_rate": 0.00023245614035087719,
+      "loss": 0.2608,
+      "step": 3428
+    },
+    {
+      "epoch": 51.18,
+      "learning_rate": 0.00023242105263157893,
+      "loss": 0.155,
+      "step": 3429
+    },
+    {
+      "epoch": 51.19,
+      "learning_rate": 0.00023238596491228068,
+      "loss": 0.2852,
+      "step": 3430
+    },
+    {
+      "epoch": 51.21,
+      "learning_rate": 0.00023235087719298243,
+      "loss": 0.0219,
+      "step": 3431
+    },
+    {
+      "epoch": 51.22,
+      "learning_rate": 0.0002323157894736842,
+      "loss": 0.0044,
+      "step": 3432
+    },
+    {
+      "epoch": 51.24,
+      "learning_rate": 0.00023228070175438596,
+      "loss": 0.2911,
+      "step": 3433
+    },
+    {
+      "epoch": 51.25,
+      "learning_rate": 0.00023224561403508768,
+      "loss": 0.1041,
+      "step": 3434
+    },
+    {
+      "epoch": 51.27,
+      "learning_rate": 0.00023221052631578948,
+      "loss": 0.1581,
+      "step": 3435
+    },
+    {
+      "epoch": 51.28,
+      "learning_rate": 0.0002321754385964912,
+      "loss": 0.0615,
+      "step": 3436
+    },
+    {
+      "epoch": 51.3,
+      "learning_rate": 0.00023214035087719295,
+      "loss": 0.0482,
+      "step": 3437
+    },
+    {
+      "epoch": 51.31,
+      "learning_rate": 0.0002321052631578947,
+      "loss": 0.1814,
+      "step": 3438
+    },
+    {
+      "epoch": 51.33,
+      "learning_rate": 0.00023207017543859648,
+      "loss": 0.0087,
+      "step": 3439
+    },
+    {
+      "epoch": 51.34,
+      "learning_rate": 0.00023203508771929823,
+      "loss": 0.0068,
+      "step": 3440
+    },
+    {
+      "epoch": 51.36,
+      "learning_rate": 0.00023199999999999997,
+      "loss": 0.4871,
+      "step": 3441
+    },
+    {
+      "epoch": 51.37,
+      "learning_rate": 0.00023196491228070172,
+      "loss": 0.0375,
+      "step": 3442
+    },
+    {
+      "epoch": 51.39,
+      "learning_rate": 0.0002319298245614035,
+      "loss": 0.0656,
+      "step": 3443
+    },
+    {
+      "epoch": 51.4,
+      "learning_rate": 0.00023189473684210525,
+      "loss": 0.0113,
+      "step": 3444
+    },
+    {
+      "epoch": 51.42,
+      "learning_rate": 0.000231859649122807,
+      "loss": 0.0957,
+      "step": 3445
+    },
+    {
+      "epoch": 51.43,
+      "learning_rate": 0.00023182456140350875,
+      "loss": 0.0936,
+      "step": 3446
+    },
+    {
+      "epoch": 51.45,
+      "learning_rate": 0.00023178947368421052,
+      "loss": 0.2175,
+      "step": 3447
+    },
+    {
+      "epoch": 51.46,
+      "learning_rate": 0.00023175438596491227,
+      "loss": 0.1612,
+      "step": 3448
+    },
+    {
+      "epoch": 51.48,
+      "learning_rate": 0.000231719298245614,
+      "loss": 0.0313,
+      "step": 3449
+    },
+    {
+      "epoch": 51.49,
+      "learning_rate": 0.0002316842105263158,
+      "loss": 0.0096,
+      "step": 3450
+    },
+    {
+      "epoch": 51.51,
+      "learning_rate": 0.00023164912280701752,
+      "loss": 0.0376,
+      "step": 3451
+    },
+    {
+      "epoch": 51.52,
+      "learning_rate": 0.00023161403508771927,
+      "loss": 0.0071,
+      "step": 3452
+    },
+    {
+      "epoch": 51.54,
+      "learning_rate": 0.00023157894736842101,
+      "loss": 0.2736,
+      "step": 3453
+    },
+    {
+      "epoch": 51.55,
+      "learning_rate": 0.0002315438596491228,
+      "loss": 0.1893,
+      "step": 3454
+    },
+    {
+      "epoch": 51.57,
+      "learning_rate": 0.00023150877192982454,
+      "loss": 0.0625,
+      "step": 3455
+    },
+    {
+      "epoch": 51.58,
+      "learning_rate": 0.0002314736842105263,
+      "loss": 0.0211,
+      "step": 3456
+    },
+    {
+      "epoch": 51.59,
+      "learning_rate": 0.00023143859649122804,
+      "loss": 0.026,
+      "step": 3457
+    },
+    {
+      "epoch": 51.61,
+      "learning_rate": 0.0002314035087719298,
+      "loss": 0.0453,
+      "step": 3458
+    },
+    {
+      "epoch": 51.62,
+      "learning_rate": 0.00023136842105263156,
+      "loss": 0.0545,
+      "step": 3459
+    },
+    {
+      "epoch": 51.64,
+      "learning_rate": 0.0002313333333333333,
+      "loss": 0.0336,
+      "step": 3460
+    },
+    {
+      "epoch": 51.65,
+      "learning_rate": 0.00023129824561403509,
+      "loss": 0.0492,
+      "step": 3461
+    },
+    {
+      "epoch": 51.67,
+      "learning_rate": 0.00023126315789473683,
+      "loss": 0.2222,
+      "step": 3462
+    },
+    {
+      "epoch": 51.68,
+      "learning_rate": 0.00023122807017543858,
+      "loss": 0.006,
+      "step": 3463
+    },
+    {
+      "epoch": 51.7,
+      "learning_rate": 0.00023119298245614033,
+      "loss": 0.0502,
+      "step": 3464
+    },
+    {
+      "epoch": 51.71,
+      "learning_rate": 0.0002311578947368421,
+      "loss": 0.0217,
+      "step": 3465
+    },
+    {
+      "epoch": 51.73,
+      "learning_rate": 0.00023112280701754386,
+      "loss": 0.4115,
+      "step": 3466
+    },
+    {
+      "epoch": 51.74,
+      "learning_rate": 0.00023108771929824558,
+      "loss": 0.0063,
+      "step": 3467
+    },
+    {
+      "epoch": 51.76,
+      "learning_rate": 0.00023105263157894733,
+      "loss": 0.386,
+      "step": 3468
+    },
+    {
+      "epoch": 51.77,
+      "learning_rate": 0.0002310175438596491,
+      "loss": 0.0952,
+      "step": 3469
+    },
+    {
+      "epoch": 51.79,
+      "learning_rate": 0.00023098245614035085,
+      "loss": 0.075,
+      "step": 3470
+    },
+    {
+      "epoch": 51.8,
+      "learning_rate": 0.0002309473684210526,
+      "loss": 0.2486,
+      "step": 3471
+    },
+    {
+      "epoch": 51.82,
+      "learning_rate": 0.00023091228070175438,
+      "loss": 0.5256,
+      "step": 3472
+    },
+    {
+      "epoch": 51.83,
+      "learning_rate": 0.00023087719298245612,
+      "loss": 0.0866,
+      "step": 3473
+    },
+    {
+      "epoch": 51.85,
+      "learning_rate": 0.00023084210526315787,
+      "loss": 0.1161,
+      "step": 3474
+    },
+    {
+      "epoch": 51.86,
+      "learning_rate": 0.00023080701754385962,
+      "loss": 0.035,
+      "step": 3475
+    },
+    {
+      "epoch": 51.88,
+      "learning_rate": 0.0002307719298245614,
+      "loss": 0.0224,
+      "step": 3476
+    },
+    {
+      "epoch": 51.89,
+      "learning_rate": 0.00023073684210526315,
+      "loss": 0.0111,
+      "step": 3477
+    },
+    {
+      "epoch": 51.91,
+      "learning_rate": 0.0002307017543859649,
+      "loss": 0.0374,
+      "step": 3478
+    },
+    {
+      "epoch": 51.92,
+      "learning_rate": 0.00023066666666666664,
+      "loss": 0.0163,
+      "step": 3479
+    },
+    {
+      "epoch": 51.94,
+      "learning_rate": 0.00023063157894736842,
+      "loss": 0.1842,
+      "step": 3480
+    },
+    {
+      "epoch": 51.95,
+      "learning_rate": 0.00023059649122807017,
+      "loss": 0.2779,
+      "step": 3481
+    },
+    {
+      "epoch": 51.97,
+      "learning_rate": 0.0002305614035087719,
+      "loss": 0.0568,
+      "step": 3482
+    },
+    {
+      "epoch": 51.98,
+      "learning_rate": 0.00023052631578947364,
+      "loss": 0.2603,
+      "step": 3483
+    },
+    {
+      "epoch": 52.0,
+      "learning_rate": 0.00023049122807017542,
+      "loss": 0.126,
+      "step": 3484
+    },
+    {
+      "epoch": 52.01,
+      "learning_rate": 0.00023045614035087716,
+      "loss": 0.7035,
+      "step": 3485
+    },
+    {
+      "epoch": 52.03,
+      "learning_rate": 0.0002304210526315789,
+      "loss": 0.0054,
+      "step": 3486
+    },
+    {
+      "epoch": 52.04,
+      "learning_rate": 0.0002303859649122807,
+      "loss": 0.0937,
+      "step": 3487
+    },
+    {
+      "epoch": 52.06,
+      "learning_rate": 0.00023035087719298244,
+      "loss": 0.1501,
+      "step": 3488
+    },
+    {
+      "epoch": 52.07,
+      "learning_rate": 0.0002303157894736842,
+      "loss": 0.0795,
+      "step": 3489
+    },
+    {
+      "epoch": 52.09,
+      "learning_rate": 0.00023028070175438594,
+      "loss": 0.1251,
+      "step": 3490
+    },
+    {
+      "epoch": 52.1,
+      "learning_rate": 0.0002302456140350877,
+      "loss": 0.01,
+      "step": 3491
+    },
+    {
+      "epoch": 52.12,
+      "learning_rate": 0.00023021052631578946,
+      "loss": 0.3727,
+      "step": 3492
+    },
+    {
+      "epoch": 52.13,
+      "learning_rate": 0.0002301754385964912,
+      "loss": 0.0147,
+      "step": 3493
+    },
+    {
+      "epoch": 52.15,
+      "learning_rate": 0.00023014035087719296,
+      "loss": 0.2178,
+      "step": 3494
+    },
+    {
+      "epoch": 52.16,
+      "learning_rate": 0.00023010526315789473,
+      "loss": 0.0409,
+      "step": 3495
+    },
+    {
+      "epoch": 52.18,
+      "learning_rate": 0.00023007017543859648,
+      "loss": 0.1585,
+      "step": 3496
+    },
+    {
+      "epoch": 52.19,
+      "learning_rate": 0.0002300350877192982,
+      "loss": 0.0069,
+      "step": 3497
+    },
+    {
+      "epoch": 52.21,
+      "learning_rate": 0.00023,
+      "loss": 0.0145,
+      "step": 3498
+    },
+    {
+      "epoch": 52.22,
+      "learning_rate": 0.00022996491228070173,
+      "loss": 0.0131,
+      "step": 3499
+    },
+    {
+      "epoch": 52.24,
+      "learning_rate": 0.00022992982456140348,
+      "loss": 0.2495,
+      "step": 3500
+    },
+    {
+      "epoch": 52.25,
+      "learning_rate": 0.00022989473684210523,
+      "loss": 0.1867,
+      "step": 3501
+    },
+    {
+      "epoch": 52.27,
+      "learning_rate": 0.000229859649122807,
+      "loss": 0.0064,
+      "step": 3502
+    },
+    {
+      "epoch": 52.28,
+      "learning_rate": 0.00022982456140350875,
+      "loss": 0.0257,
+      "step": 3503
+    },
+    {
+      "epoch": 52.3,
+      "learning_rate": 0.0002297894736842105,
+      "loss": 0.0654,
+      "step": 3504
+    },
+    {
+      "epoch": 52.31,
+      "learning_rate": 0.00022975438596491225,
+      "loss": 0.0885,
+      "step": 3505
+    },
+    {
+      "epoch": 52.33,
+      "learning_rate": 0.00022971929824561402,
+      "loss": 0.0556,
+      "step": 3506
+    },
+    {
+      "epoch": 52.34,
+      "learning_rate": 0.00022968421052631577,
+      "loss": 0.0429,
+      "step": 3507
+    },
+    {
+      "epoch": 52.36,
+      "learning_rate": 0.00022964912280701752,
+      "loss": 0.0837,
+      "step": 3508
+    },
+    {
+      "epoch": 52.37,
+      "learning_rate": 0.0002296140350877193,
+      "loss": 0.1373,
+      "step": 3509
+    },
+    {
+      "epoch": 52.39,
+      "learning_rate": 0.00022957894736842105,
+      "loss": 0.022,
+      "step": 3510
+    },
+    {
+      "epoch": 52.4,
+      "learning_rate": 0.0002295438596491228,
+      "loss": 0.356,
+      "step": 3511
+    },
+    {
+      "epoch": 52.42,
+      "learning_rate": 0.00022950877192982454,
+      "loss": 0.0119,
+      "step": 3512
+    },
+    {
+      "epoch": 52.43,
+      "learning_rate": 0.00022947368421052632,
+      "loss": 0.0376,
+      "step": 3513
+    },
+    {
+      "epoch": 52.45,
+      "learning_rate": 0.00022943859649122804,
+      "loss": 0.0049,
+      "step": 3514
+    },
+    {
+      "epoch": 52.46,
+      "learning_rate": 0.0002294035087719298,
+      "loss": 0.1981,
+      "step": 3515
+    },
+    {
+      "epoch": 52.48,
+      "learning_rate": 0.00022936842105263154,
+      "loss": 0.2462,
+      "step": 3516
+    },
+    {
+      "epoch": 52.49,
+      "learning_rate": 0.00022933333333333332,
+      "loss": 0.0036,
+      "step": 3517
+    },
+    {
+      "epoch": 52.51,
+      "learning_rate": 0.00022929824561403506,
+      "loss": 0.2211,
+      "step": 3518
+    },
+    {
+      "epoch": 52.52,
+      "learning_rate": 0.0002292631578947368,
+      "loss": 0.1351,
+      "step": 3519
+    },
+    {
+      "epoch": 52.54,
+      "learning_rate": 0.00022922807017543856,
+      "loss": 0.1374,
+      "step": 3520
+    },
+    {
+      "epoch": 52.55,
+      "learning_rate": 0.00022919298245614034,
+      "loss": 0.2577,
+      "step": 3521
+    },
+    {
+      "epoch": 52.57,
+      "learning_rate": 0.00022915789473684209,
+      "loss": 0.3107,
+      "step": 3522
+    },
+    {
+      "epoch": 52.58,
+      "learning_rate": 0.00022912280701754383,
+      "loss": 0.3881,
+      "step": 3523
+    },
+    {
+      "epoch": 52.59,
+      "learning_rate": 0.0002290877192982456,
+      "loss": 0.0885,
+      "step": 3524
+    },
+    {
+      "epoch": 52.61,
+      "learning_rate": 0.00022905263157894736,
+      "loss": 0.0183,
+      "step": 3525
+    },
+    {
+      "epoch": 52.62,
+      "learning_rate": 0.0002290175438596491,
+      "loss": 0.006,
+      "step": 3526
+    },
+    {
+      "epoch": 52.64,
+      "learning_rate": 0.00022898245614035086,
+      "loss": 0.1872,
+      "step": 3527
+    },
+    {
+      "epoch": 52.65,
+      "learning_rate": 0.00022894736842105263,
+      "loss": 0.1191,
+      "step": 3528
+    },
+    {
+      "epoch": 52.67,
+      "learning_rate": 0.00022891228070175438,
+      "loss": 0.1761,
+      "step": 3529
+    },
+    {
+      "epoch": 52.68,
+      "learning_rate": 0.0002288771929824561,
+      "loss": 0.0031,
+      "step": 3530
+    },
+    {
+      "epoch": 52.7,
+      "learning_rate": 0.00022884210526315785,
+      "loss": 0.0431,
+      "step": 3531
+    },
+    {
+      "epoch": 52.71,
+      "learning_rate": 0.00022880701754385963,
+      "loss": 0.0034,
+      "step": 3532
+    },
+    {
+      "epoch": 52.73,
+      "learning_rate": 0.00022877192982456138,
+      "loss": 0.1489,
+      "step": 3533
+    },
+    {
+      "epoch": 52.74,
+      "learning_rate": 0.00022873684210526313,
+      "loss": 0.0079,
+      "step": 3534
+    },
+    {
+      "epoch": 52.76,
+      "learning_rate": 0.0002287017543859649,
+      "loss": 0.2767,
+      "step": 3535
+    },
+    {
+      "epoch": 52.77,
+      "learning_rate": 0.00022866666666666665,
+      "loss": 0.4586,
+      "step": 3536
+    },
+    {
+      "epoch": 52.79,
+      "learning_rate": 0.0002286315789473684,
+      "loss": 0.0055,
+      "step": 3537
+    },
+    {
+      "epoch": 52.8,
+      "learning_rate": 0.00022859649122807015,
+      "loss": 0.2642,
+      "step": 3538
+    },
+    {
+      "epoch": 52.82,
+      "learning_rate": 0.00022856140350877192,
+      "loss": 0.0057,
+      "step": 3539
+    },
+    {
+      "epoch": 52.83,
+      "learning_rate": 0.00022852631578947367,
+      "loss": 0.0305,
+      "step": 3540
+    },
+    {
+      "epoch": 52.85,
+      "learning_rate": 0.00022849122807017542,
+      "loss": 0.1817,
+      "step": 3541
+    },
+    {
+      "epoch": 52.86,
+      "learning_rate": 0.00022845614035087717,
+      "loss": 0.1287,
+      "step": 3542
+    },
+    {
+      "epoch": 52.88,
+      "learning_rate": 0.00022842105263157895,
+      "loss": 0.2467,
+      "step": 3543
+    },
+    {
+      "epoch": 52.89,
+      "learning_rate": 0.0002283859649122807,
+      "loss": 0.0116,
+      "step": 3544
+    },
+    {
+      "epoch": 52.91,
+      "learning_rate": 0.00022835087719298242,
+      "loss": 0.4328,
+      "step": 3545
+    },
+    {
+      "epoch": 52.92,
+      "learning_rate": 0.00022831578947368417,
+      "loss": 0.0037,
+      "step": 3546
+    },
+    {
+      "epoch": 52.94,
+      "learning_rate": 0.00022828070175438594,
+      "loss": 0.2324,
+      "step": 3547
+    },
+    {
+      "epoch": 52.95,
+      "learning_rate": 0.0002282456140350877,
+      "loss": 0.0079,
+      "step": 3548
+    },
+    {
+      "epoch": 52.97,
+      "learning_rate": 0.00022821052631578944,
+      "loss": 0.2288,
+      "step": 3549
+    },
+    {
+      "epoch": 52.98,
+      "learning_rate": 0.00022817543859649121,
+      "loss": 0.0985,
+      "step": 3550
+    },
+    {
+      "epoch": 53.0,
+      "learning_rate": 0.00022814035087719296,
+      "loss": 0.3461,
+      "step": 3551
+    },
+    {
+      "epoch": 53.01,
+      "learning_rate": 0.0002281052631578947,
+      "loss": 0.115,
+      "step": 3552
+    },
+    {
+      "epoch": 53.03,
+      "learning_rate": 0.00022807017543859646,
+      "loss": 0.1899,
+      "step": 3553
+    },
+    {
+      "epoch": 53.04,
+      "learning_rate": 0.00022803508771929824,
+      "loss": 0.0037,
+      "step": 3554
+    },
+    {
+      "epoch": 53.06,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 0.0942,
+      "step": 3555
+    },
+    {
+      "epoch": 53.07,
+      "learning_rate": 0.00022796491228070173,
+      "loss": 0.1492,
+      "step": 3556
+    },
+    {
+      "epoch": 53.09,
+      "learning_rate": 0.00022792982456140348,
+      "loss": 0.014,
+      "step": 3557
+    },
+    {
+      "epoch": 53.1,
+      "learning_rate": 0.00022789473684210526,
+      "loss": 0.015,
+      "step": 3558
+    },
+    {
+      "epoch": 53.12,
+      "learning_rate": 0.000227859649122807,
+      "loss": 0.0556,
+      "step": 3559
+    },
+    {
+      "epoch": 53.13,
+      "learning_rate": 0.00022782456140350876,
+      "loss": 0.0535,
+      "step": 3560
+    },
+    {
+      "epoch": 53.15,
+      "learning_rate": 0.00022778947368421053,
+      "loss": 0.0459,
+      "step": 3561
+    },
+    {
+      "epoch": 53.16,
+      "learning_rate": 0.00022775438596491225,
+      "loss": 0.5427,
+      "step": 3562
+    },
+    {
+      "epoch": 53.18,
+      "learning_rate": 0.000227719298245614,
+      "loss": 0.0204,
+      "step": 3563
+    },
+    {
+      "epoch": 53.19,
+      "learning_rate": 0.00022768421052631575,
+      "loss": 0.0051,
+      "step": 3564
+    },
+    {
+      "epoch": 53.21,
+      "learning_rate": 0.00022764912280701753,
+      "loss": 0.1251,
+      "step": 3565
+    },
+    {
+      "epoch": 53.22,
+      "learning_rate": 0.00022761403508771928,
+      "loss": 0.1806,
+      "step": 3566
+    },
+    {
+      "epoch": 53.24,
+      "learning_rate": 0.00022757894736842102,
+      "loss": 0.0308,
+      "step": 3567
+    },
+    {
+      "epoch": 53.25,
+      "learning_rate": 0.00022754385964912277,
+      "loss": 0.2411,
+      "step": 3568
+    },
+    {
+      "epoch": 53.27,
+      "learning_rate": 0.00022750877192982455,
+      "loss": 0.2055,
+      "step": 3569
+    },
+    {
+      "epoch": 53.28,
+      "learning_rate": 0.0002274736842105263,
+      "loss": 0.0065,
+      "step": 3570
+    },
+    {
+      "epoch": 53.3,
+      "learning_rate": 0.00022743859649122805,
+      "loss": 0.0063,
+      "step": 3571
+    },
+    {
+      "epoch": 53.31,
+      "learning_rate": 0.00022740350877192982,
+      "loss": 0.1127,
+      "step": 3572
+    },
+    {
+      "epoch": 53.33,
+      "learning_rate": 0.00022736842105263157,
+      "loss": 0.15,
+      "step": 3573
+    },
+    {
+      "epoch": 53.34,
+      "learning_rate": 0.00022733333333333332,
+      "loss": 0.1083,
+      "step": 3574
+    },
+    {
+      "epoch": 53.36,
+      "learning_rate": 0.00022729824561403507,
+      "loss": 0.2448,
+      "step": 3575
+    },
+    {
+      "epoch": 53.37,
+      "learning_rate": 0.00022726315789473685,
+      "loss": 0.0562,
+      "step": 3576
+    },
+    {
+      "epoch": 53.39,
+      "learning_rate": 0.0002272280701754386,
+      "loss": 0.0156,
+      "step": 3577
+    },
+    {
+      "epoch": 53.4,
+      "learning_rate": 0.00022719298245614032,
+      "loss": 0.1385,
+      "step": 3578
+    },
+    {
+      "epoch": 53.42,
+      "learning_rate": 0.00022715789473684206,
+      "loss": 0.2497,
+      "step": 3579
+    },
+    {
+      "epoch": 53.43,
+      "learning_rate": 0.00022712280701754384,
+      "loss": 0.1978,
+      "step": 3580
+    },
+    {
+      "epoch": 53.45,
+      "learning_rate": 0.0002270877192982456,
+      "loss": 0.0094,
+      "step": 3581
+    },
+    {
+      "epoch": 53.46,
+      "learning_rate": 0.00022705263157894734,
+      "loss": 0.0449,
+      "step": 3582
+    },
+    {
+      "epoch": 53.48,
+      "learning_rate": 0.0002270175438596491,
+      "loss": 0.092,
+      "step": 3583
+    },
+    {
+      "epoch": 53.49,
+      "learning_rate": 0.00022698245614035086,
+      "loss": 0.0045,
+      "step": 3584
+    },
+    {
+      "epoch": 53.51,
+      "learning_rate": 0.0002269473684210526,
+      "loss": 0.034,
+      "step": 3585
+    },
+    {
+      "epoch": 53.52,
+      "learning_rate": 0.00022691228070175436,
+      "loss": 0.1851,
+      "step": 3586
+    },
+    {
+      "epoch": 53.54,
+      "learning_rate": 0.00022687719298245614,
+      "loss": 0.2542,
+      "step": 3587
+    },
+    {
+      "epoch": 53.55,
+      "learning_rate": 0.00022684210526315788,
+      "loss": 0.0296,
+      "step": 3588
+    },
+    {
+      "epoch": 53.57,
+      "learning_rate": 0.00022680701754385963,
+      "loss": 0.0841,
+      "step": 3589
+    },
+    {
+      "epoch": 53.58,
+      "learning_rate": 0.00022677192982456138,
+      "loss": 0.1696,
+      "step": 3590
+    },
+    {
+      "epoch": 53.59,
+      "learning_rate": 0.00022673684210526316,
+      "loss": 0.021,
+      "step": 3591
+    },
+    {
+      "epoch": 53.61,
+      "learning_rate": 0.0002267017543859649,
+      "loss": 0.0085,
+      "step": 3592
+    },
+    {
+      "epoch": 53.62,
+      "learning_rate": 0.00022666666666666663,
+      "loss": 0.0084,
+      "step": 3593
+    },
+    {
+      "epoch": 53.64,
+      "learning_rate": 0.00022663157894736838,
+      "loss": 0.1819,
+      "step": 3594
+    },
+    {
+      "epoch": 53.65,
+      "learning_rate": 0.00022659649122807015,
+      "loss": 0.0062,
+      "step": 3595
+    },
+    {
+      "epoch": 53.67,
+      "learning_rate": 0.0002265614035087719,
+      "loss": 0.2001,
+      "step": 3596
+    },
+    {
+      "epoch": 53.68,
+      "learning_rate": 0.00022652631578947365,
+      "loss": 0.0464,
+      "step": 3597
+    },
+    {
+      "epoch": 53.7,
+      "learning_rate": 0.00022649122807017543,
+      "loss": 0.1304,
+      "step": 3598
+    },
+    {
+      "epoch": 53.71,
+      "learning_rate": 0.00022645614035087718,
+      "loss": 0.0339,
+      "step": 3599
+    },
+    {
+      "epoch": 53.73,
+      "learning_rate": 0.00022642105263157892,
+      "loss": 0.0062,
+      "step": 3600
+    },
+    {
+      "epoch": 53.73,
+      "eval_accuracy": 0.8277043563387175,
+      "eval_f1": 0.8286378990021197,
+      "eval_loss": 0.6871868968009949,
+      "eval_runtime": 343.9702,
+      "eval_samples_per_second": 11.879,
+      "eval_steps_per_second": 0.744,
+      "step": 3600
+    },
+    {
+      "epoch": 53.74,
+      "learning_rate": 0.00022638596491228067,
+      "loss": 0.1994,
+      "step": 3601
+    },
+    {
+      "epoch": 53.76,
+      "learning_rate": 0.00022635087719298245,
+      "loss": 0.0094,
+      "step": 3602
+    },
+    {
+      "epoch": 53.77,
+      "learning_rate": 0.0002263157894736842,
+      "loss": 0.0321,
+      "step": 3603
+    },
+    {
+      "epoch": 53.79,
+      "learning_rate": 0.00022628070175438595,
+      "loss": 0.2154,
+      "step": 3604
+    },
+    {
+      "epoch": 53.8,
+      "learning_rate": 0.0002262456140350877,
+      "loss": 0.0209,
+      "step": 3605
+    },
+    {
+      "epoch": 53.82,
+      "learning_rate": 0.00022621052631578947,
+      "loss": 0.159,
+      "step": 3606
+    },
+    {
+      "epoch": 53.83,
+      "learning_rate": 0.00022617543859649122,
+      "loss": 0.0349,
+      "step": 3607
+    },
+    {
+      "epoch": 53.85,
+      "learning_rate": 0.00022614035087719294,
+      "loss": 0.0212,
+      "step": 3608
+    },
+    {
+      "epoch": 53.86,
+      "learning_rate": 0.0002261052631578947,
+      "loss": 0.1698,
+      "step": 3609
+    },
+    {
+      "epoch": 53.88,
+      "learning_rate": 0.00022607017543859647,
+      "loss": 0.0174,
+      "step": 3610
+    },
+    {
+      "epoch": 53.89,
+      "learning_rate": 0.00022603508771929822,
+      "loss": 0.1267,
+      "step": 3611
+    },
+    {
+      "epoch": 53.91,
+      "learning_rate": 0.00022599999999999996,
+      "loss": 0.0366,
+      "step": 3612
+    },
+    {
+      "epoch": 53.92,
+      "learning_rate": 0.00022596491228070174,
+      "loss": 0.1204,
+      "step": 3613
+    },
+    {
+      "epoch": 53.94,
+      "learning_rate": 0.0002259298245614035,
+      "loss": 0.071,
+      "step": 3614
+    },
+    {
+      "epoch": 53.95,
+      "learning_rate": 0.00022589473684210524,
+      "loss": 0.0097,
+      "step": 3615
+    },
+    {
+      "epoch": 53.97,
+      "learning_rate": 0.00022585964912280699,
+      "loss": 0.0107,
+      "step": 3616
+    },
+    {
+      "epoch": 53.98,
+      "learning_rate": 0.00022582456140350876,
+      "loss": 0.1535,
+      "step": 3617
+    },
+    {
+      "epoch": 54.0,
+      "learning_rate": 0.0002257894736842105,
+      "loss": 0.009,
+      "step": 3618
+    },
+    {
+      "epoch": 54.01,
+      "learning_rate": 0.00022575438596491226,
+      "loss": 0.0093,
+      "step": 3619
+    },
+    {
+      "epoch": 54.03,
+      "learning_rate": 0.000225719298245614,
+      "loss": 0.3425,
+      "step": 3620
+    },
+    {
+      "epoch": 54.04,
+      "learning_rate": 0.00022568421052631578,
+      "loss": 0.035,
+      "step": 3621
+    },
+    {
+      "epoch": 54.06,
+      "learning_rate": 0.00022564912280701753,
+      "loss": 0.0755,
+      "step": 3622
+    },
+    {
+      "epoch": 54.07,
+      "learning_rate": 0.00022561403508771928,
+      "loss": 0.0045,
+      "step": 3623
+    },
+    {
+      "epoch": 54.09,
+      "learning_rate": 0.00022557894736842106,
+      "loss": 0.2203,
+      "step": 3624
+    },
+    {
+      "epoch": 54.1,
+      "learning_rate": 0.0002255438596491228,
+      "loss": 0.326,
+      "step": 3625
+    },
+    {
+      "epoch": 54.12,
+      "learning_rate": 0.00022550877192982453,
+      "loss": 0.0032,
+      "step": 3626
+    },
+    {
+      "epoch": 54.13,
+      "learning_rate": 0.00022547368421052628,
+      "loss": 0.1305,
+      "step": 3627
+    },
+    {
+      "epoch": 54.15,
+      "learning_rate": 0.00022543859649122805,
+      "loss": 0.2871,
+      "step": 3628
+    },
+    {
+      "epoch": 54.16,
+      "learning_rate": 0.0002254035087719298,
+      "loss": 0.0593,
+      "step": 3629
+    },
+    {
+      "epoch": 54.18,
+      "learning_rate": 0.00022536842105263155,
+      "loss": 0.022,
+      "step": 3630
+    },
+    {
+      "epoch": 54.19,
+      "learning_rate": 0.0002253333333333333,
+      "loss": 0.0904,
+      "step": 3631
+    },
+    {
+      "epoch": 54.21,
+      "learning_rate": 0.00022529824561403508,
+      "loss": 0.0036,
+      "step": 3632
+    },
+    {
+      "epoch": 54.22,
+      "learning_rate": 0.00022526315789473682,
+      "loss": 0.0114,
+      "step": 3633
+    },
+    {
+      "epoch": 54.24,
+      "learning_rate": 0.00022522807017543857,
+      "loss": 0.1628,
+      "step": 3634
+    },
+    {
+      "epoch": 54.25,
+      "learning_rate": 0.00022519298245614035,
+      "loss": 0.2939,
+      "step": 3635
+    },
+    {
+      "epoch": 54.27,
+      "learning_rate": 0.0002251578947368421,
+      "loss": 0.0156,
+      "step": 3636
+    },
+    {
+      "epoch": 54.28,
+      "learning_rate": 0.00022512280701754385,
+      "loss": 0.0028,
+      "step": 3637
+    },
+    {
+      "epoch": 54.3,
+      "learning_rate": 0.0002250877192982456,
+      "loss": 0.0056,
+      "step": 3638
+    },
+    {
+      "epoch": 54.31,
+      "learning_rate": 0.00022505263157894737,
+      "loss": 0.0847,
+      "step": 3639
+    },
+    {
+      "epoch": 54.33,
+      "learning_rate": 0.00022501754385964912,
+      "loss": 0.3336,
+      "step": 3640
+    },
+    {
+      "epoch": 54.34,
+      "learning_rate": 0.00022498245614035084,
+      "loss": 0.1036,
+      "step": 3641
+    },
+    {
+      "epoch": 54.36,
+      "learning_rate": 0.0002249473684210526,
+      "loss": 0.0032,
+      "step": 3642
+    },
+    {
+      "epoch": 54.37,
+      "learning_rate": 0.00022491228070175437,
+      "loss": 0.0345,
+      "step": 3643
+    },
+    {
+      "epoch": 54.39,
+      "learning_rate": 0.00022487719298245611,
+      "loss": 0.1888,
+      "step": 3644
+    },
+    {
+      "epoch": 54.4,
+      "learning_rate": 0.00022484210526315786,
+      "loss": 0.0094,
+      "step": 3645
+    },
+    {
+      "epoch": 54.42,
+      "learning_rate": 0.0002248070175438596,
+      "loss": 0.0244,
+      "step": 3646
+    },
+    {
+      "epoch": 54.43,
+      "learning_rate": 0.0002247719298245614,
+      "loss": 0.0999,
+      "step": 3647
+    },
+    {
+      "epoch": 54.45,
+      "learning_rate": 0.00022473684210526314,
+      "loss": 0.0718,
+      "step": 3648
+    },
+    {
+      "epoch": 54.46,
+      "learning_rate": 0.00022470175438596489,
+      "loss": 0.2473,
+      "step": 3649
+    },
+    {
+      "epoch": 54.48,
+      "learning_rate": 0.00022466666666666666,
+      "loss": 0.164,
+      "step": 3650
+    },
+    {
+      "epoch": 54.49,
+      "learning_rate": 0.0002246315789473684,
+      "loss": 0.2189,
+      "step": 3651
+    },
+    {
+      "epoch": 54.51,
+      "learning_rate": 0.00022459649122807016,
+      "loss": 0.1645,
+      "step": 3652
+    },
+    {
+      "epoch": 54.52,
+      "learning_rate": 0.0002245614035087719,
+      "loss": 0.0121,
+      "step": 3653
+    },
+    {
+      "epoch": 54.54,
+      "learning_rate": 0.00022452631578947368,
+      "loss": 0.0113,
+      "step": 3654
+    },
+    {
+      "epoch": 54.55,
+      "learning_rate": 0.00022449122807017543,
+      "loss": 0.056,
+      "step": 3655
+    },
+    {
+      "epoch": 54.57,
+      "learning_rate": 0.00022445614035087715,
+      "loss": 0.1074,
+      "step": 3656
+    },
+    {
+      "epoch": 54.58,
+      "learning_rate": 0.0002244210526315789,
+      "loss": 0.2388,
+      "step": 3657
+    },
+    {
+      "epoch": 54.59,
+      "learning_rate": 0.00022438596491228068,
+      "loss": 0.0244,
+      "step": 3658
+    },
+    {
+      "epoch": 54.61,
+      "learning_rate": 0.00022435087719298243,
+      "loss": 0.0135,
+      "step": 3659
+    },
+    {
+      "epoch": 54.62,
+      "learning_rate": 0.00022431578947368418,
+      "loss": 0.004,
+      "step": 3660
+    },
+    {
+      "epoch": 54.64,
+      "learning_rate": 0.00022428070175438595,
+      "loss": 0.1477,
+      "step": 3661
+    },
+    {
+      "epoch": 54.65,
+      "learning_rate": 0.0002242456140350877,
+      "loss": 0.012,
+      "step": 3662
+    },
+    {
+      "epoch": 54.67,
+      "learning_rate": 0.00022421052631578945,
+      "loss": 0.2786,
+      "step": 3663
+    },
+    {
+      "epoch": 54.68,
+      "learning_rate": 0.0002241754385964912,
+      "loss": 0.0527,
+      "step": 3664
+    },
+    {
+      "epoch": 54.7,
+      "learning_rate": 0.00022414035087719297,
+      "loss": 0.0731,
+      "step": 3665
+    },
+    {
+      "epoch": 54.71,
+      "learning_rate": 0.00022410526315789472,
+      "loss": 0.2216,
+      "step": 3666
+    },
+    {
+      "epoch": 54.73,
+      "learning_rate": 0.00022407017543859647,
+      "loss": 0.0036,
+      "step": 3667
+    },
+    {
+      "epoch": 54.74,
+      "learning_rate": 0.00022403508771929822,
+      "loss": 0.0137,
+      "step": 3668
+    },
+    {
+      "epoch": 54.76,
+      "learning_rate": 0.000224,
+      "loss": 0.3671,
+      "step": 3669
+    },
+    {
+      "epoch": 54.77,
+      "learning_rate": 0.00022396491228070175,
+      "loss": 0.3146,
+      "step": 3670
+    },
+    {
+      "epoch": 54.79,
+      "learning_rate": 0.0002239298245614035,
+      "loss": 0.2034,
+      "step": 3671
+    },
+    {
+      "epoch": 54.8,
+      "learning_rate": 0.00022389473684210527,
+      "loss": 0.1474,
+      "step": 3672
+    },
+    {
+      "epoch": 54.82,
+      "learning_rate": 0.00022385964912280702,
+      "loss": 0.0591,
+      "step": 3673
+    },
+    {
+      "epoch": 54.83,
+      "learning_rate": 0.00022382456140350874,
+      "loss": 0.0043,
+      "step": 3674
+    },
+    {
+      "epoch": 54.85,
+      "learning_rate": 0.0002237894736842105,
+      "loss": 0.0173,
+      "step": 3675
+    },
+    {
+      "epoch": 54.86,
+      "learning_rate": 0.00022375438596491227,
+      "loss": 0.1988,
+      "step": 3676
+    },
+    {
+      "epoch": 54.88,
+      "learning_rate": 0.00022371929824561401,
+      "loss": 0.4382,
+      "step": 3677
+    },
+    {
+      "epoch": 54.89,
+      "learning_rate": 0.00022368421052631576,
+      "loss": 0.0258,
+      "step": 3678
+    },
+    {
+      "epoch": 54.91,
+      "learning_rate": 0.0002236491228070175,
+      "loss": 0.0638,
+      "step": 3679
+    },
+    {
+      "epoch": 54.92,
+      "learning_rate": 0.0002236140350877193,
+      "loss": 0.1948,
+      "step": 3680
+    },
+    {
+      "epoch": 54.94,
+      "learning_rate": 0.00022357894736842104,
+      "loss": 0.2229,
+      "step": 3681
+    },
+    {
+      "epoch": 54.95,
+      "learning_rate": 0.00022354385964912278,
+      "loss": 0.0038,
+      "step": 3682
+    },
+    {
+      "epoch": 54.97,
+      "learning_rate": 0.00022350877192982453,
+      "loss": 0.451,
+      "step": 3683
+    },
+    {
+      "epoch": 54.98,
+      "learning_rate": 0.0002234736842105263,
+      "loss": 0.0037,
+      "step": 3684
+    },
+    {
+      "epoch": 55.0,
+      "learning_rate": 0.00022343859649122806,
+      "loss": 0.1133,
+      "step": 3685
+    },
+    {
+      "epoch": 55.01,
+      "learning_rate": 0.0002234035087719298,
+      "loss": 0.1625,
+      "step": 3686
+    },
+    {
+      "epoch": 55.03,
+      "learning_rate": 0.00022336842105263158,
+      "loss": 0.0319,
+      "step": 3687
+    },
+    {
+      "epoch": 55.04,
+      "learning_rate": 0.00022333333333333333,
+      "loss": 0.0071,
+      "step": 3688
+    },
+    {
+      "epoch": 55.06,
+      "learning_rate": 0.00022329824561403505,
+      "loss": 0.0307,
+      "step": 3689
+    },
+    {
+      "epoch": 55.07,
+      "learning_rate": 0.0002232631578947368,
+      "loss": 0.0852,
+      "step": 3690
+    },
+    {
+      "epoch": 55.09,
+      "learning_rate": 0.00022322807017543858,
+      "loss": 0.0048,
+      "step": 3691
+    },
+    {
+      "epoch": 55.1,
+      "learning_rate": 0.00022319298245614033,
+      "loss": 0.019,
+      "step": 3692
+    },
+    {
+      "epoch": 55.12,
+      "learning_rate": 0.00022315789473684208,
+      "loss": 0.0731,
+      "step": 3693
+    },
+    {
+      "epoch": 55.13,
+      "learning_rate": 0.00022312280701754382,
+      "loss": 0.129,
+      "step": 3694
+    },
+    {
+      "epoch": 55.15,
+      "learning_rate": 0.0002230877192982456,
+      "loss": 0.037,
+      "step": 3695
+    },
+    {
+      "epoch": 55.16,
+      "learning_rate": 0.00022305263157894735,
+      "loss": 0.0068,
+      "step": 3696
+    },
+    {
+      "epoch": 55.18,
+      "learning_rate": 0.0002230175438596491,
+      "loss": 0.0246,
+      "step": 3697
+    },
+    {
+      "epoch": 55.19,
+      "learning_rate": 0.00022298245614035087,
+      "loss": 0.1615,
+      "step": 3698
+    },
+    {
+      "epoch": 55.21,
+      "learning_rate": 0.00022294736842105262,
+      "loss": 0.0029,
+      "step": 3699
+    },
+    {
+      "epoch": 55.22,
+      "learning_rate": 0.00022291228070175437,
+      "loss": 0.0246,
+      "step": 3700
+    },
+    {
+      "epoch": 55.24,
+      "learning_rate": 0.00022287719298245612,
+      "loss": 0.0065,
+      "step": 3701
+    },
+    {
+      "epoch": 55.25,
+      "learning_rate": 0.0002228421052631579,
+      "loss": 0.0155,
+      "step": 3702
+    },
+    {
+      "epoch": 55.27,
+      "learning_rate": 0.00022280701754385964,
+      "loss": 0.0479,
+      "step": 3703
+    },
+    {
+      "epoch": 55.28,
+      "learning_rate": 0.00022277192982456137,
+      "loss": 0.0084,
+      "step": 3704
+    },
+    {
+      "epoch": 55.3,
+      "learning_rate": 0.00022273684210526312,
+      "loss": 0.3419,
+      "step": 3705
+    },
+    {
+      "epoch": 55.31,
+      "learning_rate": 0.0002227017543859649,
+      "loss": 0.2642,
+      "step": 3706
+    },
+    {
+      "epoch": 55.33,
+      "learning_rate": 0.00022266666666666664,
+      "loss": 0.0043,
+      "step": 3707
+    },
+    {
+      "epoch": 55.34,
+      "learning_rate": 0.0002226315789473684,
+      "loss": 0.0485,
+      "step": 3708
+    },
+    {
+      "epoch": 55.36,
+      "learning_rate": 0.00022259649122807014,
+      "loss": 0.0498,
+      "step": 3709
+    },
+    {
+      "epoch": 55.37,
+      "learning_rate": 0.0002225614035087719,
+      "loss": 0.0062,
+      "step": 3710
+    },
+    {
+      "epoch": 55.39,
+      "learning_rate": 0.00022252631578947366,
+      "loss": 0.0073,
+      "step": 3711
+    },
+    {
+      "epoch": 55.4,
+      "learning_rate": 0.0002224912280701754,
+      "loss": 0.1017,
+      "step": 3712
+    },
+    {
+      "epoch": 55.42,
+      "learning_rate": 0.0002224561403508772,
+      "loss": 0.2038,
+      "step": 3713
+    },
+    {
+      "epoch": 55.43,
+      "learning_rate": 0.00022242105263157894,
+      "loss": 0.1198,
+      "step": 3714
+    },
+    {
+      "epoch": 55.45,
+      "learning_rate": 0.00022238596491228068,
+      "loss": 0.0198,
+      "step": 3715
+    },
+    {
+      "epoch": 55.46,
+      "learning_rate": 0.00022235087719298243,
+      "loss": 0.1242,
+      "step": 3716
+    },
+    {
+      "epoch": 55.48,
+      "learning_rate": 0.0002223157894736842,
+      "loss": 0.0246,
+      "step": 3717
+    },
+    {
+      "epoch": 55.49,
+      "learning_rate": 0.00022228070175438596,
+      "loss": 0.0113,
+      "step": 3718
+    },
+    {
+      "epoch": 55.51,
+      "learning_rate": 0.0002222456140350877,
+      "loss": 0.0202,
+      "step": 3719
+    },
+    {
+      "epoch": 55.52,
+      "learning_rate": 0.00022221052631578943,
+      "loss": 0.0141,
+      "step": 3720
+    },
+    {
+      "epoch": 55.54,
+      "learning_rate": 0.0002221754385964912,
+      "loss": 0.0651,
+      "step": 3721
+    },
+    {
+      "epoch": 55.55,
+      "learning_rate": 0.00022214035087719295,
+      "loss": 0.0952,
+      "step": 3722
+    },
+    {
+      "epoch": 55.57,
+      "learning_rate": 0.0002221052631578947,
+      "loss": 0.1741,
+      "step": 3723
+    },
+    {
+      "epoch": 55.58,
+      "learning_rate": 0.00022207017543859648,
+      "loss": 0.5287,
+      "step": 3724
+    },
+    {
+      "epoch": 55.59,
+      "learning_rate": 0.00022203508771929823,
+      "loss": 0.1499,
+      "step": 3725
+    },
+    {
+      "epoch": 55.61,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 0.2446,
+      "step": 3726
+    },
+    {
+      "epoch": 55.62,
+      "learning_rate": 0.00022196491228070172,
+      "loss": 0.0443,
+      "step": 3727
+    },
+    {
+      "epoch": 55.64,
+      "learning_rate": 0.0002219298245614035,
+      "loss": 0.0117,
+      "step": 3728
+    },
+    {
+      "epoch": 55.65,
+      "learning_rate": 0.00022189473684210525,
+      "loss": 0.0446,
+      "step": 3729
+    },
+    {
+      "epoch": 55.67,
+      "learning_rate": 0.000221859649122807,
+      "loss": 0.1981,
+      "step": 3730
+    },
+    {
+      "epoch": 55.68,
+      "learning_rate": 0.00022182456140350875,
+      "loss": 0.0033,
+      "step": 3731
+    },
+    {
+      "epoch": 55.7,
+      "learning_rate": 0.00022178947368421052,
+      "loss": 0.1273,
+      "step": 3732
+    },
+    {
+      "epoch": 55.71,
+      "learning_rate": 0.00022175438596491227,
+      "loss": 0.0741,
+      "step": 3733
+    },
+    {
+      "epoch": 55.73,
+      "learning_rate": 0.00022171929824561402,
+      "loss": 0.0265,
+      "step": 3734
+    },
+    {
+      "epoch": 55.74,
+      "learning_rate": 0.0002216842105263158,
+      "loss": 0.0115,
+      "step": 3735
+    },
+    {
+      "epoch": 55.76,
+      "learning_rate": 0.00022164912280701754,
+      "loss": 0.0161,
+      "step": 3736
+    },
+    {
+      "epoch": 55.77,
+      "learning_rate": 0.00022161403508771927,
+      "loss": 0.0085,
+      "step": 3737
+    },
+    {
+      "epoch": 55.79,
+      "learning_rate": 0.00022157894736842101,
+      "loss": 0.2448,
+      "step": 3738
+    },
+    {
+      "epoch": 55.8,
+      "learning_rate": 0.0002215438596491228,
+      "loss": 0.0145,
+      "step": 3739
+    },
+    {
+      "epoch": 55.82,
+      "learning_rate": 0.00022150877192982454,
+      "loss": 0.0153,
+      "step": 3740
+    },
+    {
+      "epoch": 55.83,
+      "learning_rate": 0.0002214736842105263,
+      "loss": 0.018,
+      "step": 3741
+    },
+    {
+      "epoch": 55.85,
+      "learning_rate": 0.00022143859649122804,
+      "loss": 0.2306,
+      "step": 3742
+    },
+    {
+      "epoch": 55.86,
+      "learning_rate": 0.0002214035087719298,
+      "loss": 0.0034,
+      "step": 3743
+    },
+    {
+      "epoch": 55.88,
+      "learning_rate": 0.00022136842105263156,
+      "loss": 0.0116,
+      "step": 3744
+    },
+    {
+      "epoch": 55.89,
+      "learning_rate": 0.0002213333333333333,
+      "loss": 0.0537,
+      "step": 3745
+    },
+    {
+      "epoch": 55.91,
+      "learning_rate": 0.00022129824561403506,
+      "loss": 0.1457,
+      "step": 3746
+    },
+    {
+      "epoch": 55.92,
+      "learning_rate": 0.00022126315789473683,
+      "loss": 0.0771,
+      "step": 3747
+    },
+    {
+      "epoch": 55.94,
+      "learning_rate": 0.00022122807017543858,
+      "loss": 0.0036,
+      "step": 3748
+    },
+    {
+      "epoch": 55.95,
+      "learning_rate": 0.00022119298245614033,
+      "loss": 0.1992,
+      "step": 3749
+    },
+    {
+      "epoch": 55.97,
+      "learning_rate": 0.0002211578947368421,
+      "loss": 0.1269,
+      "step": 3750
+    },
+    {
+      "epoch": 55.98,
+      "learning_rate": 0.00022112280701754386,
+      "loss": 0.209,
+      "step": 3751
+    },
+    {
+      "epoch": 56.0,
+      "learning_rate": 0.00022108771929824558,
+      "loss": 0.0806,
+      "step": 3752
+    },
+    {
+      "epoch": 56.01,
+      "learning_rate": 0.00022105263157894733,
+      "loss": 0.3064,
+      "step": 3753
+    },
+    {
+      "epoch": 56.03,
+      "learning_rate": 0.0002210175438596491,
+      "loss": 0.1929,
+      "step": 3754
+    },
+    {
+      "epoch": 56.04,
+      "learning_rate": 0.00022098245614035085,
+      "loss": 0.2673,
+      "step": 3755
+    },
+    {
+      "epoch": 56.06,
+      "learning_rate": 0.0002209473684210526,
+      "loss": 0.2803,
+      "step": 3756
+    },
+    {
+      "epoch": 56.07,
+      "learning_rate": 0.00022091228070175435,
+      "loss": 0.0192,
+      "step": 3757
+    },
+    {
+      "epoch": 56.09,
+      "learning_rate": 0.00022087719298245613,
+      "loss": 0.1947,
+      "step": 3758
+    },
+    {
+      "epoch": 56.1,
+      "learning_rate": 0.00022084210526315787,
+      "loss": 0.0723,
+      "step": 3759
+    },
+    {
+      "epoch": 56.12,
+      "learning_rate": 0.00022080701754385962,
+      "loss": 0.3244,
+      "step": 3760
+    },
+    {
+      "epoch": 56.13,
+      "learning_rate": 0.0002207719298245614,
+      "loss": 0.0238,
+      "step": 3761
+    },
+    {
+      "epoch": 56.15,
+      "learning_rate": 0.00022073684210526315,
+      "loss": 0.0308,
+      "step": 3762
+    },
+    {
+      "epoch": 56.16,
+      "learning_rate": 0.0002207017543859649,
+      "loss": 0.1321,
+      "step": 3763
+    },
+    {
+      "epoch": 56.18,
+      "learning_rate": 0.00022066666666666665,
+      "loss": 0.0051,
+      "step": 3764
+    },
+    {
+      "epoch": 56.19,
+      "learning_rate": 0.00022063157894736842,
+      "loss": 0.0057,
+      "step": 3765
+    },
+    {
+      "epoch": 56.21,
+      "learning_rate": 0.00022059649122807017,
+      "loss": 0.0935,
+      "step": 3766
+    },
+    {
+      "epoch": 56.22,
+      "learning_rate": 0.00022056140350877192,
+      "loss": 0.009,
+      "step": 3767
+    },
+    {
+      "epoch": 56.24,
+      "learning_rate": 0.00022052631578947364,
+      "loss": 0.0077,
+      "step": 3768
+    },
+    {
+      "epoch": 56.25,
+      "learning_rate": 0.00022049122807017542,
+      "loss": 0.0445,
+      "step": 3769
+    },
+    {
+      "epoch": 56.27,
+      "learning_rate": 0.00022045614035087717,
+      "loss": 0.1608,
+      "step": 3770
+    },
+    {
+      "epoch": 56.28,
+      "learning_rate": 0.00022042105263157891,
+      "loss": 0.1767,
+      "step": 3771
+    },
+    {
+      "epoch": 56.3,
+      "learning_rate": 0.00022038596491228066,
+      "loss": 0.0302,
+      "step": 3772
+    },
+    {
+      "epoch": 56.31,
+      "learning_rate": 0.00022035087719298244,
+      "loss": 0.1324,
+      "step": 3773
+    },
+    {
+      "epoch": 56.33,
+      "learning_rate": 0.0002203157894736842,
+      "loss": 0.0097,
+      "step": 3774
+    },
+    {
+      "epoch": 56.34,
+      "learning_rate": 0.00022028070175438594,
+      "loss": 0.3323,
+      "step": 3775
+    },
+    {
+      "epoch": 56.36,
+      "learning_rate": 0.0002202456140350877,
+      "loss": 0.1146,
+      "step": 3776
+    },
+    {
+      "epoch": 56.37,
+      "learning_rate": 0.00022021052631578946,
+      "loss": 0.1479,
+      "step": 3777
+    },
+    {
+      "epoch": 56.39,
+      "learning_rate": 0.0002201754385964912,
+      "loss": 0.1966,
+      "step": 3778
+    },
+    {
+      "epoch": 56.4,
+      "learning_rate": 0.00022014035087719296,
+      "loss": 0.0405,
+      "step": 3779
+    },
+    {
+      "epoch": 56.42,
+      "learning_rate": 0.00022010526315789473,
+      "loss": 0.3266,
+      "step": 3780
+    },
+    {
+      "epoch": 56.43,
+      "learning_rate": 0.00022007017543859648,
+      "loss": 0.0037,
+      "step": 3781
+    },
+    {
+      "epoch": 56.45,
+      "learning_rate": 0.00022003508771929823,
+      "loss": 0.089,
+      "step": 3782
+    },
+    {
+      "epoch": 56.46,
+      "learning_rate": 0.00021999999999999995,
+      "loss": 0.2181,
+      "step": 3783
+    },
+    {
+      "epoch": 56.48,
+      "learning_rate": 0.00021996491228070176,
+      "loss": 0.0876,
+      "step": 3784
+    },
+    {
+      "epoch": 56.49,
+      "learning_rate": 0.00021992982456140348,
+      "loss": 0.2992,
+      "step": 3785
+    },
+    {
+      "epoch": 56.51,
+      "learning_rate": 0.00021989473684210523,
+      "loss": 0.0213,
+      "step": 3786
+    },
+    {
+      "epoch": 56.52,
+      "learning_rate": 0.000219859649122807,
+      "loss": 0.1602,
+      "step": 3787
+    },
+    {
+      "epoch": 56.54,
+      "learning_rate": 0.00021982456140350875,
+      "loss": 0.2594,
+      "step": 3788
+    },
+    {
+      "epoch": 56.55,
+      "learning_rate": 0.0002197894736842105,
+      "loss": 0.013,
+      "step": 3789
+    },
+    {
+      "epoch": 56.57,
+      "learning_rate": 0.00021975438596491225,
+      "loss": 0.041,
+      "step": 3790
+    },
+    {
+      "epoch": 56.58,
+      "learning_rate": 0.00021971929824561403,
+      "loss": 0.1021,
+      "step": 3791
+    },
+    {
+      "epoch": 56.59,
+      "learning_rate": 0.00021968421052631577,
+      "loss": 0.1141,
+      "step": 3792
+    },
+    {
+      "epoch": 56.61,
+      "learning_rate": 0.00021964912280701752,
+      "loss": 0.0101,
+      "step": 3793
+    },
+    {
+      "epoch": 56.62,
+      "learning_rate": 0.00021961403508771927,
+      "loss": 0.0539,
+      "step": 3794
+    },
+    {
+      "epoch": 56.64,
+      "learning_rate": 0.00021957894736842105,
+      "loss": 0.0914,
+      "step": 3795
+    },
+    {
+      "epoch": 56.65,
+      "learning_rate": 0.0002195438596491228,
+      "loss": 0.0729,
+      "step": 3796
+    },
+    {
+      "epoch": 56.67,
+      "learning_rate": 0.00021950877192982454,
+      "loss": 0.0327,
+      "step": 3797
+    },
+    {
+      "epoch": 56.68,
+      "learning_rate": 0.00021947368421052632,
+      "loss": 0.0187,
+      "step": 3798
+    },
+    {
+      "epoch": 56.7,
+      "learning_rate": 0.00021943859649122807,
+      "loss": 0.0103,
+      "step": 3799
+    },
+    {
+      "epoch": 56.71,
+      "learning_rate": 0.0002194035087719298,
+      "loss": 0.1781,
+      "step": 3800
+    },
+    {
+      "epoch": 56.71,
+      "eval_accuracy": 0.8389623103279491,
+      "eval_f1": 0.8392926046674013,
+      "eval_loss": 0.6989664435386658,
+      "eval_runtime": 344.8439,
+      "eval_samples_per_second": 11.849,
+      "eval_steps_per_second": 0.742,
+      "step": 3800
+    },
+    {
+      "epoch": 56.73,
+      "learning_rate": 0.00021936842105263154,
+      "loss": 0.0225,
+      "step": 3801
+    },
+    {
+      "epoch": 56.74,
+      "learning_rate": 0.00021933333333333332,
+      "loss": 0.1503,
+      "step": 3802
+    },
+    {
+      "epoch": 56.76,
+      "learning_rate": 0.00021929824561403506,
+      "loss": 0.0784,
+      "step": 3803
+    },
+    {
+      "epoch": 56.77,
+      "learning_rate": 0.0002192631578947368,
+      "loss": 0.2565,
+      "step": 3804
+    },
+    {
+      "epoch": 56.79,
+      "learning_rate": 0.00021922807017543856,
+      "loss": 0.1283,
+      "step": 3805
+    },
+    {
+      "epoch": 56.8,
+      "learning_rate": 0.00021919298245614034,
+      "loss": 0.2075,
+      "step": 3806
+    },
+    {
+      "epoch": 56.82,
+      "learning_rate": 0.0002191578947368421,
+      "loss": 0.1317,
+      "step": 3807
+    },
+    {
+      "epoch": 56.83,
+      "learning_rate": 0.00021912280701754384,
+      "loss": 0.0164,
+      "step": 3808
+    },
+    {
+      "epoch": 56.85,
+      "learning_rate": 0.00021908771929824558,
+      "loss": 0.1049,
+      "step": 3809
+    },
+    {
+      "epoch": 56.86,
+      "learning_rate": 0.00021905263157894736,
+      "loss": 0.0052,
+      "step": 3810
+    },
+    {
+      "epoch": 56.88,
+      "learning_rate": 0.0002190175438596491,
+      "loss": 0.1653,
+      "step": 3811
+    },
+    {
+      "epoch": 56.89,
+      "learning_rate": 0.00021898245614035086,
+      "loss": 0.0032,
+      "step": 3812
+    },
+    {
+      "epoch": 56.91,
+      "learning_rate": 0.00021894736842105263,
+      "loss": 0.0823,
+      "step": 3813
+    },
+    {
+      "epoch": 56.92,
+      "learning_rate": 0.00021891228070175438,
+      "loss": 0.007,
+      "step": 3814
+    },
+    {
+      "epoch": 56.94,
+      "learning_rate": 0.0002188771929824561,
+      "loss": 0.0183,
+      "step": 3815
+    },
+    {
+      "epoch": 56.95,
+      "learning_rate": 0.00021884210526315785,
+      "loss": 0.0204,
+      "step": 3816
+    },
+    {
+      "epoch": 56.97,
+      "learning_rate": 0.00021880701754385963,
+      "loss": 0.0054,
+      "step": 3817
+    },
+    {
+      "epoch": 56.98,
+      "learning_rate": 0.00021877192982456138,
+      "loss": 0.0143,
+      "step": 3818
+    },
+    {
+      "epoch": 57.0,
+      "learning_rate": 0.00021873684210526313,
+      "loss": 0.2526,
+      "step": 3819
+    },
+    {
+      "epoch": 57.01,
+      "learning_rate": 0.00021870175438596488,
+      "loss": 0.4135,
+      "step": 3820
+    },
+    {
+      "epoch": 57.03,
+      "learning_rate": 0.00021866666666666665,
+      "loss": 0.1704,
+      "step": 3821
+    },
+    {
+      "epoch": 57.04,
+      "learning_rate": 0.0002186315789473684,
+      "loss": 0.1849,
+      "step": 3822
+    },
+    {
+      "epoch": 57.06,
+      "learning_rate": 0.00021859649122807015,
+      "loss": 0.1036,
+      "step": 3823
+    },
+    {
+      "epoch": 57.07,
+      "learning_rate": 0.00021856140350877192,
+      "loss": 0.0054,
+      "step": 3824
+    },
+    {
+      "epoch": 57.09,
+      "learning_rate": 0.00021852631578947367,
+      "loss": 0.0035,
+      "step": 3825
+    },
+    {
+      "epoch": 57.1,
+      "learning_rate": 0.00021849122807017542,
+      "loss": 0.1784,
+      "step": 3826
+    },
+    {
+      "epoch": 57.12,
+      "learning_rate": 0.00021845614035087717,
+      "loss": 0.0086,
+      "step": 3827
+    },
+    {
+      "epoch": 57.13,
+      "learning_rate": 0.00021842105263157895,
+      "loss": 0.2458,
+      "step": 3828
+    },
+    {
+      "epoch": 57.15,
+      "learning_rate": 0.0002183859649122807,
+      "loss": 0.1473,
+      "step": 3829
+    },
+    {
+      "epoch": 57.16,
+      "learning_rate": 0.00021835087719298244,
+      "loss": 0.0109,
+      "step": 3830
+    },
+    {
+      "epoch": 57.18,
+      "learning_rate": 0.00021831578947368417,
+      "loss": 0.0234,
+      "step": 3831
+    },
+    {
+      "epoch": 57.19,
+      "learning_rate": 0.00021828070175438597,
+      "loss": 0.1154,
+      "step": 3832
+    },
+    {
+      "epoch": 57.21,
+      "learning_rate": 0.0002182456140350877,
+      "loss": 0.0863,
+      "step": 3833
+    },
+    {
+      "epoch": 57.22,
+      "learning_rate": 0.00021821052631578944,
+      "loss": 0.0019,
+      "step": 3834
+    },
+    {
+      "epoch": 57.24,
+      "learning_rate": 0.00021817543859649122,
+      "loss": 0.0907,
+      "step": 3835
+    },
+    {
+      "epoch": 57.25,
+      "learning_rate": 0.00021814035087719296,
+      "loss": 0.0025,
+      "step": 3836
+    },
+    {
+      "epoch": 57.27,
+      "learning_rate": 0.0002181052631578947,
+      "loss": 0.16,
+      "step": 3837
+    },
+    {
+      "epoch": 57.28,
+      "learning_rate": 0.00021807017543859646,
+      "loss": 0.2068,
+      "step": 3838
+    },
+    {
+      "epoch": 57.3,
+      "learning_rate": 0.00021803508771929824,
+      "loss": 0.0291,
+      "step": 3839
+    },
+    {
+      "epoch": 57.31,
+      "learning_rate": 0.00021799999999999999,
+      "loss": 0.0851,
+      "step": 3840
+    },
+    {
+      "epoch": 57.33,
+      "learning_rate": 0.00021796491228070174,
+      "loss": 0.0864,
+      "step": 3841
+    },
+    {
+      "epoch": 57.34,
+      "learning_rate": 0.00021792982456140348,
+      "loss": 0.0652,
+      "step": 3842
+    },
+    {
+      "epoch": 57.36,
+      "learning_rate": 0.00021789473684210526,
+      "loss": 0.0539,
+      "step": 3843
+    },
+    {
+      "epoch": 57.37,
+      "learning_rate": 0.000217859649122807,
+      "loss": 0.0076,
+      "step": 3844
+    },
+    {
+      "epoch": 57.39,
+      "learning_rate": 0.00021782456140350876,
+      "loss": 0.0818,
+      "step": 3845
+    },
+    {
+      "epoch": 57.4,
+      "learning_rate": 0.00021778947368421048,
+      "loss": 0.0047,
+      "step": 3846
+    },
+    {
+      "epoch": 57.42,
+      "learning_rate": 0.00021775438596491228,
+      "loss": 0.0698,
+      "step": 3847
+    },
+    {
+      "epoch": 57.43,
+      "learning_rate": 0.000217719298245614,
+      "loss": 0.002,
+      "step": 3848
+    },
+    {
+      "epoch": 57.45,
+      "learning_rate": 0.00021768421052631575,
+      "loss": 0.0138,
+      "step": 3849
+    },
+    {
+      "epoch": 57.46,
+      "learning_rate": 0.00021764912280701753,
+      "loss": 0.0896,
+      "step": 3850
+    },
+    {
+      "epoch": 57.48,
+      "learning_rate": 0.00021761403508771928,
+      "loss": 0.0029,
+      "step": 3851
+    },
+    {
+      "epoch": 57.49,
+      "learning_rate": 0.00021757894736842103,
+      "loss": 0.2649,
+      "step": 3852
+    },
+    {
+      "epoch": 57.51,
+      "learning_rate": 0.00021754385964912277,
+      "loss": 0.0521,
+      "step": 3853
+    },
+    {
+      "epoch": 57.52,
+      "learning_rate": 0.00021750877192982455,
+      "loss": 0.3472,
+      "step": 3854
+    },
+    {
+      "epoch": 57.54,
+      "learning_rate": 0.0002174736842105263,
+      "loss": 0.1763,
+      "step": 3855
+    },
+    {
+      "epoch": 57.55,
+      "learning_rate": 0.00021743859649122805,
+      "loss": 0.0379,
+      "step": 3856
+    },
+    {
+      "epoch": 57.57,
+      "learning_rate": 0.0002174035087719298,
+      "loss": 0.1569,
+      "step": 3857
+    },
+    {
+      "epoch": 57.58,
+      "learning_rate": 0.00021736842105263157,
+      "loss": 0.0211,
+      "step": 3858
+    },
+    {
+      "epoch": 57.59,
+      "learning_rate": 0.00021733333333333332,
+      "loss": 0.0072,
+      "step": 3859
+    },
+    {
+      "epoch": 57.61,
+      "learning_rate": 0.00021729824561403507,
+      "loss": 0.0209,
+      "step": 3860
+    },
+    {
+      "epoch": 57.62,
+      "learning_rate": 0.00021726315789473685,
+      "loss": 0.0629,
+      "step": 3861
+    },
+    {
+      "epoch": 57.64,
+      "learning_rate": 0.0002172280701754386,
+      "loss": 0.0572,
+      "step": 3862
+    },
+    {
+      "epoch": 57.65,
+      "learning_rate": 0.00021719298245614032,
+      "loss": 0.0058,
+      "step": 3863
+    },
+    {
+      "epoch": 57.67,
+      "learning_rate": 0.00021715789473684207,
+      "loss": 0.192,
+      "step": 3864
+    },
+    {
+      "epoch": 57.68,
+      "learning_rate": 0.00021712280701754384,
+      "loss": 0.005,
+      "step": 3865
+    },
+    {
+      "epoch": 57.7,
+      "learning_rate": 0.0002170877192982456,
+      "loss": 0.0027,
+      "step": 3866
+    },
+    {
+      "epoch": 57.71,
+      "learning_rate": 0.00021705263157894734,
+      "loss": 0.0573,
+      "step": 3867
+    },
+    {
+      "epoch": 57.73,
+      "learning_rate": 0.0002170175438596491,
+      "loss": 0.0142,
+      "step": 3868
+    },
+    {
+      "epoch": 57.74,
+      "learning_rate": 0.00021698245614035086,
+      "loss": 0.0021,
+      "step": 3869
+    },
+    {
+      "epoch": 57.76,
+      "learning_rate": 0.0002169473684210526,
+      "loss": 0.0478,
+      "step": 3870
+    },
+    {
+      "epoch": 57.77,
+      "learning_rate": 0.00021691228070175436,
+      "loss": 0.0061,
+      "step": 3871
+    },
+    {
+      "epoch": 57.79,
+      "learning_rate": 0.0002168771929824561,
+      "loss": 0.0036,
+      "step": 3872
+    },
+    {
+      "epoch": 57.8,
+      "learning_rate": 0.00021684210526315789,
+      "loss": 0.0152,
+      "step": 3873
+    },
+    {
+      "epoch": 57.82,
+      "learning_rate": 0.00021680701754385963,
+      "loss": 0.0777,
+      "step": 3874
+    },
+    {
+      "epoch": 57.83,
+      "learning_rate": 0.00021677192982456138,
+      "loss": 0.0093,
+      "step": 3875
+    },
+    {
+      "epoch": 57.85,
+      "learning_rate": 0.00021673684210526316,
+      "loss": 0.2739,
+      "step": 3876
+    },
+    {
+      "epoch": 57.86,
+      "learning_rate": 0.0002167017543859649,
+      "loss": 0.1036,
+      "step": 3877
+    },
+    {
+      "epoch": 57.88,
+      "learning_rate": 0.00021666666666666666,
+      "loss": 0.0132,
+      "step": 3878
+    },
+    {
+      "epoch": 57.89,
+      "learning_rate": 0.00021663157894736838,
+      "loss": 0.0039,
+      "step": 3879
+    },
+    {
+      "epoch": 57.91,
+      "learning_rate": 0.00021659649122807018,
+      "loss": 0.0541,
+      "step": 3880
+    },
+    {
+      "epoch": 57.92,
+      "learning_rate": 0.0002165614035087719,
+      "loss": 0.0431,
+      "step": 3881
+    },
+    {
+      "epoch": 57.94,
+      "learning_rate": 0.00021652631578947365,
+      "loss": 0.1137,
+      "step": 3882
+    },
+    {
+      "epoch": 57.95,
+      "learning_rate": 0.0002164912280701754,
+      "loss": 0.0022,
+      "step": 3883
+    },
+    {
+      "epoch": 57.97,
+      "learning_rate": 0.00021645614035087718,
+      "loss": 0.1969,
+      "step": 3884
+    },
+    {
+      "epoch": 57.98,
+      "learning_rate": 0.00021642105263157893,
+      "loss": 0.0212,
+      "step": 3885
+    },
+    {
+      "epoch": 58.0,
+      "learning_rate": 0.00021638596491228067,
+      "loss": 0.3204,
+      "step": 3886
+    },
+    {
+      "epoch": 58.01,
+      "learning_rate": 0.00021635087719298245,
+      "loss": 0.0098,
+      "step": 3887
+    },
+    {
+      "epoch": 58.03,
+      "learning_rate": 0.0002163157894736842,
+      "loss": 0.1595,
+      "step": 3888
+    },
+    {
+      "epoch": 58.04,
+      "learning_rate": 0.00021628070175438595,
+      "loss": 0.086,
+      "step": 3889
+    },
+    {
+      "epoch": 58.06,
+      "learning_rate": 0.0002162456140350877,
+      "loss": 0.0183,
+      "step": 3890
+    },
+    {
+      "epoch": 58.07,
+      "learning_rate": 0.00021621052631578947,
+      "loss": 0.2086,
+      "step": 3891
+    },
+    {
+      "epoch": 58.09,
+      "learning_rate": 0.00021617543859649122,
+      "loss": 0.0196,
+      "step": 3892
+    },
+    {
+      "epoch": 58.1,
+      "learning_rate": 0.00021614035087719297,
+      "loss": 0.1816,
+      "step": 3893
+    },
+    {
+      "epoch": 58.12,
+      "learning_rate": 0.0002161052631578947,
+      "loss": 0.2019,
+      "step": 3894
+    },
+    {
+      "epoch": 58.13,
+      "learning_rate": 0.0002160701754385965,
+      "loss": 0.0304,
+      "step": 3895
+    },
+    {
+      "epoch": 58.15,
+      "learning_rate": 0.00021603508771929822,
+      "loss": 0.0164,
+      "step": 3896
+    },
+    {
+      "epoch": 58.16,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 0.002,
+      "step": 3897
+    },
+    {
+      "epoch": 58.18,
+      "learning_rate": 0.00021596491228070174,
+      "loss": 0.0027,
+      "step": 3898
+    },
+    {
+      "epoch": 58.19,
+      "learning_rate": 0.0002159298245614035,
+      "loss": 0.0026,
+      "step": 3899
+    },
+    {
+      "epoch": 58.21,
+      "learning_rate": 0.00021589473684210524,
+      "loss": 0.0901,
+      "step": 3900
+    },
+    {
+      "epoch": 58.22,
+      "learning_rate": 0.000215859649122807,
+      "loss": 0.0058,
+      "step": 3901
+    },
+    {
+      "epoch": 58.24,
+      "learning_rate": 0.00021582456140350876,
+      "loss": 0.0685,
+      "step": 3902
+    },
+    {
+      "epoch": 58.25,
+      "learning_rate": 0.0002157894736842105,
+      "loss": 0.0045,
+      "step": 3903
+    },
+    {
+      "epoch": 58.27,
+      "learning_rate": 0.00021575438596491226,
+      "loss": 0.2307,
+      "step": 3904
+    },
+    {
+      "epoch": 58.28,
+      "learning_rate": 0.000215719298245614,
+      "loss": 0.0077,
+      "step": 3905
+    },
+    {
+      "epoch": 58.3,
+      "learning_rate": 0.00021568421052631579,
+      "loss": 0.0029,
+      "step": 3906
+    },
+    {
+      "epoch": 58.31,
+      "learning_rate": 0.00021564912280701753,
+      "loss": 0.2945,
+      "step": 3907
+    },
+    {
+      "epoch": 58.33,
+      "learning_rate": 0.00021561403508771928,
+      "loss": 0.0025,
+      "step": 3908
+    },
+    {
+      "epoch": 58.34,
+      "learning_rate": 0.000215578947368421,
+      "loss": 0.019,
+      "step": 3909
+    },
+    {
+      "epoch": 58.36,
+      "learning_rate": 0.0002155438596491228,
+      "loss": 0.0021,
+      "step": 3910
+    },
+    {
+      "epoch": 58.37,
+      "learning_rate": 0.00021550877192982453,
+      "loss": 0.0225,
+      "step": 3911
+    },
+    {
+      "epoch": 58.39,
+      "learning_rate": 0.00021547368421052628,
+      "loss": 0.0101,
+      "step": 3912
+    },
+    {
+      "epoch": 58.4,
+      "learning_rate": 0.00021543859649122805,
+      "loss": 0.0075,
+      "step": 3913
+    },
+    {
+      "epoch": 58.42,
+      "learning_rate": 0.0002154035087719298,
+      "loss": 0.0023,
+      "step": 3914
+    },
+    {
+      "epoch": 58.43,
+      "learning_rate": 0.00021536842105263155,
+      "loss": 0.0037,
+      "step": 3915
+    },
+    {
+      "epoch": 58.45,
+      "learning_rate": 0.0002153333333333333,
+      "loss": 0.0055,
+      "step": 3916
+    },
+    {
+      "epoch": 58.46,
+      "learning_rate": 0.00021529824561403508,
+      "loss": 0.0205,
+      "step": 3917
+    },
+    {
+      "epoch": 58.48,
+      "learning_rate": 0.00021526315789473682,
+      "loss": 0.2707,
+      "step": 3918
+    },
+    {
+      "epoch": 58.49,
+      "learning_rate": 0.00021522807017543857,
+      "loss": 0.0289,
+      "step": 3919
+    },
+    {
+      "epoch": 58.51,
+      "learning_rate": 0.00021519298245614032,
+      "loss": 0.0175,
+      "step": 3920
+    },
+    {
+      "epoch": 58.52,
+      "learning_rate": 0.0002151578947368421,
+      "loss": 0.1888,
+      "step": 3921
+    },
+    {
+      "epoch": 58.54,
+      "learning_rate": 0.00021512280701754385,
+      "loss": 0.0133,
+      "step": 3922
+    },
+    {
+      "epoch": 58.55,
+      "learning_rate": 0.0002150877192982456,
+      "loss": 0.1769,
+      "step": 3923
+    },
+    {
+      "epoch": 58.57,
+      "learning_rate": 0.00021505263157894737,
+      "loss": 0.0578,
+      "step": 3924
+    },
+    {
+      "epoch": 58.58,
+      "learning_rate": 0.00021501754385964912,
+      "loss": 0.1585,
+      "step": 3925
+    },
+    {
+      "epoch": 58.59,
+      "learning_rate": 0.00021498245614035087,
+      "loss": 0.0783,
+      "step": 3926
+    },
+    {
+      "epoch": 58.61,
+      "learning_rate": 0.0002149473684210526,
+      "loss": 0.0021,
+      "step": 3927
+    },
+    {
+      "epoch": 58.62,
+      "learning_rate": 0.00021491228070175437,
+      "loss": 0.0082,
+      "step": 3928
+    },
+    {
+      "epoch": 58.64,
+      "learning_rate": 0.00021487719298245612,
+      "loss": 0.0019,
+      "step": 3929
+    },
+    {
+      "epoch": 58.65,
+      "learning_rate": 0.00021484210526315786,
+      "loss": 0.0022,
+      "step": 3930
+    },
+    {
+      "epoch": 58.67,
+      "learning_rate": 0.0002148070175438596,
+      "loss": 0.0103,
+      "step": 3931
+    },
+    {
+      "epoch": 58.68,
+      "learning_rate": 0.0002147719298245614,
+      "loss": 0.297,
+      "step": 3932
+    },
+    {
+      "epoch": 58.7,
+      "learning_rate": 0.00021473684210526314,
+      "loss": 0.0833,
+      "step": 3933
+    },
+    {
+      "epoch": 58.71,
+      "learning_rate": 0.00021470175438596489,
+      "loss": 0.0789,
+      "step": 3934
+    },
+    {
+      "epoch": 58.73,
+      "learning_rate": 0.00021466666666666664,
+      "loss": 0.0016,
+      "step": 3935
+    },
+    {
+      "epoch": 58.74,
+      "learning_rate": 0.0002146315789473684,
+      "loss": 0.0414,
+      "step": 3936
+    },
+    {
+      "epoch": 58.76,
+      "learning_rate": 0.00021459649122807016,
+      "loss": 0.0163,
+      "step": 3937
+    },
+    {
+      "epoch": 58.77,
+      "learning_rate": 0.0002145614035087719,
+      "loss": 0.1671,
+      "step": 3938
+    },
+    {
+      "epoch": 58.79,
+      "learning_rate": 0.00021452631578947368,
+      "loss": 0.1753,
+      "step": 3939
+    },
+    {
+      "epoch": 58.8,
+      "learning_rate": 0.00021449122807017543,
+      "loss": 0.2833,
+      "step": 3940
+    },
+    {
+      "epoch": 58.82,
+      "learning_rate": 0.00021445614035087718,
+      "loss": 0.0499,
+      "step": 3941
+    },
+    {
+      "epoch": 58.83,
+      "learning_rate": 0.0002144210526315789,
+      "loss": 0.0969,
+      "step": 3942
+    },
+    {
+      "epoch": 58.85,
+      "learning_rate": 0.0002143859649122807,
+      "loss": 0.4685,
+      "step": 3943
+    },
+    {
+      "epoch": 58.86,
+      "learning_rate": 0.00021435087719298243,
+      "loss": 0.0489,
+      "step": 3944
+    },
+    {
+      "epoch": 58.88,
+      "learning_rate": 0.00021431578947368418,
+      "loss": 0.1016,
+      "step": 3945
+    },
+    {
+      "epoch": 58.89,
+      "learning_rate": 0.00021428070175438593,
+      "loss": 0.161,
+      "step": 3946
+    },
+    {
+      "epoch": 58.91,
+      "learning_rate": 0.0002142456140350877,
+      "loss": 0.0239,
+      "step": 3947
+    },
+    {
+      "epoch": 58.92,
+      "learning_rate": 0.00021421052631578945,
+      "loss": 0.0382,
+      "step": 3948
+    },
+    {
+      "epoch": 58.94,
+      "learning_rate": 0.0002141754385964912,
+      "loss": 0.105,
+      "step": 3949
+    },
+    {
+      "epoch": 58.95,
+      "learning_rate": 0.00021414035087719298,
+      "loss": 0.5169,
+      "step": 3950
+    },
+    {
+      "epoch": 58.97,
+      "learning_rate": 0.00021410526315789472,
+      "loss": 0.1401,
+      "step": 3951
+    },
+    {
+      "epoch": 58.98,
+      "learning_rate": 0.00021407017543859647,
+      "loss": 0.4505,
+      "step": 3952
+    },
+    {
+      "epoch": 59.0,
+      "learning_rate": 0.00021403508771929822,
+      "loss": 0.2539,
+      "step": 3953
+    },
+    {
+      "epoch": 59.01,
+      "learning_rate": 0.000214,
+      "loss": 0.4692,
+      "step": 3954
+    },
+    {
+      "epoch": 59.03,
+      "learning_rate": 0.00021396491228070175,
+      "loss": 0.1235,
+      "step": 3955
+    },
+    {
+      "epoch": 59.04,
+      "learning_rate": 0.0002139298245614035,
+      "loss": 0.1404,
+      "step": 3956
+    },
+    {
+      "epoch": 59.06,
+      "learning_rate": 0.00021389473684210522,
+      "loss": 0.2029,
+      "step": 3957
+    },
+    {
+      "epoch": 59.07,
+      "learning_rate": 0.00021385964912280702,
+      "loss": 0.004,
+      "step": 3958
+    },
+    {
+      "epoch": 59.09,
+      "learning_rate": 0.00021382456140350874,
+      "loss": 0.2866,
+      "step": 3959
+    },
+    {
+      "epoch": 59.1,
+      "learning_rate": 0.0002137894736842105,
+      "loss": 0.1797,
+      "step": 3960
+    },
+    {
+      "epoch": 59.12,
+      "learning_rate": 0.00021375438596491227,
+      "loss": 0.0153,
+      "step": 3961
+    },
+    {
+      "epoch": 59.13,
+      "learning_rate": 0.00021371929824561401,
+      "loss": 0.0148,
+      "step": 3962
+    },
+    {
+      "epoch": 59.15,
+      "learning_rate": 0.00021368421052631576,
+      "loss": 0.0334,
+      "step": 3963
+    },
+    {
+      "epoch": 59.16,
+      "learning_rate": 0.0002136491228070175,
+      "loss": 0.1279,
+      "step": 3964
+    },
+    {
+      "epoch": 59.18,
+      "learning_rate": 0.0002136140350877193,
+      "loss": 0.1225,
+      "step": 3965
+    },
+    {
+      "epoch": 59.19,
+      "learning_rate": 0.00021357894736842104,
+      "loss": 0.2333,
+      "step": 3966
+    },
+    {
+      "epoch": 59.21,
+      "learning_rate": 0.00021354385964912279,
+      "loss": 0.0704,
+      "step": 3967
+    },
+    {
+      "epoch": 59.22,
+      "learning_rate": 0.00021350877192982453,
+      "loss": 0.0717,
+      "step": 3968
+    },
+    {
+      "epoch": 59.24,
+      "learning_rate": 0.0002134736842105263,
+      "loss": 0.026,
+      "step": 3969
+    },
+    {
+      "epoch": 59.25,
+      "learning_rate": 0.00021343859649122806,
+      "loss": 0.0112,
+      "step": 3970
+    },
+    {
+      "epoch": 59.27,
+      "learning_rate": 0.0002134035087719298,
+      "loss": 0.2346,
+      "step": 3971
+    },
+    {
+      "epoch": 59.28,
+      "learning_rate": 0.00021336842105263156,
+      "loss": 0.099,
+      "step": 3972
+    },
+    {
+      "epoch": 59.3,
+      "learning_rate": 0.00021333333333333333,
+      "loss": 0.1305,
+      "step": 3973
+    },
+    {
+      "epoch": 59.31,
+      "learning_rate": 0.00021329824561403508,
+      "loss": 0.1288,
+      "step": 3974
+    },
+    {
+      "epoch": 59.33,
+      "learning_rate": 0.0002132631578947368,
+      "loss": 0.0628,
+      "step": 3975
+    },
+    {
+      "epoch": 59.34,
+      "learning_rate": 0.00021322807017543858,
+      "loss": 0.1149,
+      "step": 3976
+    },
+    {
+      "epoch": 59.36,
+      "learning_rate": 0.00021319298245614033,
+      "loss": 0.0104,
+      "step": 3977
+    },
+    {
+      "epoch": 59.37,
+      "learning_rate": 0.00021315789473684208,
+      "loss": 0.2735,
+      "step": 3978
+    },
+    {
+      "epoch": 59.39,
+      "learning_rate": 0.00021312280701754383,
+      "loss": 0.0035,
+      "step": 3979
+    },
+    {
+      "epoch": 59.4,
+      "learning_rate": 0.0002130877192982456,
+      "loss": 0.0304,
+      "step": 3980
+    },
+    {
+      "epoch": 59.42,
+      "learning_rate": 0.00021305263157894735,
+      "loss": 0.0028,
+      "step": 3981
+    },
+    {
+      "epoch": 59.43,
+      "learning_rate": 0.0002130175438596491,
+      "loss": 0.0032,
+      "step": 3982
+    },
+    {
+      "epoch": 59.45,
+      "learning_rate": 0.00021298245614035085,
+      "loss": 0.1995,
+      "step": 3983
+    },
+    {
+      "epoch": 59.46,
+      "learning_rate": 0.00021294736842105262,
+      "loss": 0.0188,
+      "step": 3984
+    },
+    {
+      "epoch": 59.48,
+      "learning_rate": 0.00021291228070175437,
+      "loss": 0.0776,
+      "step": 3985
+    },
+    {
+      "epoch": 59.49,
+      "learning_rate": 0.00021287719298245612,
+      "loss": 0.0076,
+      "step": 3986
+    },
+    {
+      "epoch": 59.51,
+      "learning_rate": 0.0002128421052631579,
+      "loss": 0.5661,
+      "step": 3987
+    },
+    {
+      "epoch": 59.52,
+      "learning_rate": 0.00021280701754385965,
+      "loss": 0.4085,
+      "step": 3988
+    },
+    {
+      "epoch": 59.54,
+      "learning_rate": 0.0002127719298245614,
+      "loss": 0.1628,
+      "step": 3989
+    },
+    {
+      "epoch": 59.55,
+      "learning_rate": 0.00021273684210526312,
+      "loss": 0.1061,
+      "step": 3990
+    },
+    {
+      "epoch": 59.57,
+      "learning_rate": 0.00021270175438596492,
+      "loss": 0.2666,
+      "step": 3991
+    },
+    {
+      "epoch": 59.58,
+      "learning_rate": 0.00021266666666666664,
+      "loss": 0.0611,
+      "step": 3992
+    },
+    {
+      "epoch": 59.59,
+      "learning_rate": 0.0002126315789473684,
+      "loss": 0.0071,
+      "step": 3993
+    },
+    {
+      "epoch": 59.61,
+      "learning_rate": 0.00021259649122807014,
+      "loss": 0.0339,
+      "step": 3994
+    },
+    {
+      "epoch": 59.62,
+      "learning_rate": 0.00021256140350877191,
+      "loss": 0.2808,
+      "step": 3995
+    },
+    {
+      "epoch": 59.64,
+      "learning_rate": 0.00021252631578947366,
+      "loss": 0.0188,
+      "step": 3996
+    },
+    {
+      "epoch": 59.65,
+      "learning_rate": 0.0002124912280701754,
+      "loss": 0.1661,
+      "step": 3997
+    },
+    {
+      "epoch": 59.67,
+      "learning_rate": 0.0002124561403508772,
+      "loss": 0.0633,
+      "step": 3998
+    },
+    {
+      "epoch": 59.68,
+      "learning_rate": 0.00021242105263157894,
+      "loss": 0.0081,
+      "step": 3999
+    },
+    {
+      "epoch": 59.7,
+      "learning_rate": 0.00021238596491228069,
+      "loss": 0.0309,
+      "step": 4000
+    },
+    {
+      "epoch": 59.7,
+      "eval_accuracy": 0.8499755261869799,
+      "eval_f1": 0.8496479698093485,
+      "eval_loss": 0.6348404884338379,
+      "eval_runtime": 344.6603,
+      "eval_samples_per_second": 11.855,
+      "eval_steps_per_second": 0.743,
+      "step": 4000
+    },
+    {
+      "epoch": 59.71,
+      "learning_rate": 0.00021235087719298243,
+      "loss": 0.004,
+      "step": 4001
+    },
+    {
+      "epoch": 59.73,
+      "learning_rate": 0.0002123157894736842,
+      "loss": 0.1196,
+      "step": 4002
+    },
+    {
+      "epoch": 59.74,
+      "learning_rate": 0.00021228070175438596,
+      "loss": 0.2379,
+      "step": 4003
+    },
+    {
+      "epoch": 59.76,
+      "learning_rate": 0.0002122456140350877,
+      "loss": 0.1041,
+      "step": 4004
+    },
+    {
+      "epoch": 59.77,
+      "learning_rate": 0.00021221052631578943,
+      "loss": 0.0147,
+      "step": 4005
+    },
+    {
+      "epoch": 59.79,
+      "learning_rate": 0.00021217543859649123,
+      "loss": 0.238,
+      "step": 4006
+    },
+    {
+      "epoch": 59.8,
+      "learning_rate": 0.00021214035087719295,
+      "loss": 0.1002,
+      "step": 4007
+    },
+    {
+      "epoch": 59.82,
+      "learning_rate": 0.0002121052631578947,
+      "loss": 0.0141,
+      "step": 4008
+    },
+    {
+      "epoch": 59.83,
+      "learning_rate": 0.00021207017543859645,
+      "loss": 0.0981,
+      "step": 4009
+    },
+    {
+      "epoch": 59.85,
+      "learning_rate": 0.00021203508771929823,
+      "loss": 0.0131,
+      "step": 4010
+    },
+    {
+      "epoch": 59.86,
+      "learning_rate": 0.00021199999999999998,
+      "loss": 0.0039,
+      "step": 4011
+    },
+    {
+      "epoch": 59.88,
+      "learning_rate": 0.00021196491228070172,
+      "loss": 0.0125,
+      "step": 4012
+    },
+    {
+      "epoch": 59.89,
+      "learning_rate": 0.0002119298245614035,
+      "loss": 0.0485,
+      "step": 4013
+    },
+    {
+      "epoch": 59.91,
+      "learning_rate": 0.00021189473684210525,
+      "loss": 0.0308,
+      "step": 4014
+    },
+    {
+      "epoch": 59.92,
+      "learning_rate": 0.000211859649122807,
+      "loss": 0.0196,
+      "step": 4015
+    },
+    {
+      "epoch": 59.94,
+      "learning_rate": 0.00021182456140350875,
+      "loss": 0.0278,
+      "step": 4016
+    },
+    {
+      "epoch": 59.95,
+      "learning_rate": 0.00021178947368421052,
+      "loss": 0.0999,
+      "step": 4017
+    },
+    {
+      "epoch": 59.97,
+      "learning_rate": 0.00021175438596491227,
+      "loss": 0.3143,
+      "step": 4018
+    },
+    {
+      "epoch": 59.98,
+      "learning_rate": 0.00021171929824561402,
+      "loss": 0.0039,
+      "step": 4019
+    },
+    {
+      "epoch": 60.0,
+      "learning_rate": 0.00021168421052631577,
+      "loss": 0.0026,
+      "step": 4020
+    },
+    {
+      "epoch": 60.01,
+      "learning_rate": 0.00021164912280701755,
+      "loss": 0.0043,
+      "step": 4021
+    },
+    {
+      "epoch": 60.03,
+      "learning_rate": 0.00021161403508771927,
+      "loss": 0.0167,
+      "step": 4022
+    },
+    {
+      "epoch": 60.04,
+      "learning_rate": 0.00021157894736842102,
+      "loss": 0.0815,
+      "step": 4023
+    },
+    {
+      "epoch": 60.06,
+      "learning_rate": 0.0002115438596491228,
+      "loss": 0.0135,
+      "step": 4024
+    },
+    {
+      "epoch": 60.07,
+      "learning_rate": 0.00021150877192982454,
+      "loss": 0.0043,
+      "step": 4025
+    },
+    {
+      "epoch": 60.09,
+      "learning_rate": 0.0002114736842105263,
+      "loss": 0.206,
+      "step": 4026
+    },
+    {
+      "epoch": 60.1,
+      "learning_rate": 0.00021143859649122804,
+      "loss": 0.017,
+      "step": 4027
+    },
+    {
+      "epoch": 60.12,
+      "learning_rate": 0.00021140350877192981,
+      "loss": 0.0317,
+      "step": 4028
+    },
+    {
+      "epoch": 60.13,
+      "learning_rate": 0.00021136842105263156,
+      "loss": 0.0049,
+      "step": 4029
+    },
+    {
+      "epoch": 60.15,
+      "learning_rate": 0.0002113333333333333,
+      "loss": 0.0669,
+      "step": 4030
+    },
+    {
+      "epoch": 60.16,
+      "learning_rate": 0.00021129824561403506,
+      "loss": 0.0034,
+      "step": 4031
+    },
+    {
+      "epoch": 60.18,
+      "learning_rate": 0.00021126315789473684,
+      "loss": 0.0118,
+      "step": 4032
+    },
+    {
+      "epoch": 60.19,
+      "learning_rate": 0.00021122807017543858,
+      "loss": 0.1456,
+      "step": 4033
+    },
+    {
+      "epoch": 60.21,
+      "learning_rate": 0.00021119298245614033,
+      "loss": 0.0936,
+      "step": 4034
+    },
+    {
+      "epoch": 60.22,
+      "learning_rate": 0.00021115789473684208,
+      "loss": 0.002,
+      "step": 4035
+    },
+    {
+      "epoch": 60.24,
+      "learning_rate": 0.00021112280701754386,
+      "loss": 0.0019,
+      "step": 4036
+    },
+    {
+      "epoch": 60.25,
+      "learning_rate": 0.0002110877192982456,
+      "loss": 0.0154,
+      "step": 4037
+    },
+    {
+      "epoch": 60.27,
+      "learning_rate": 0.00021105263157894733,
+      "loss": 0.1007,
+      "step": 4038
+    },
+    {
+      "epoch": 60.28,
+      "learning_rate": 0.00021101754385964913,
+      "loss": 0.002,
+      "step": 4039
+    },
+    {
+      "epoch": 60.3,
+      "learning_rate": 0.00021098245614035085,
+      "loss": 0.0145,
+      "step": 4040
+    },
+    {
+      "epoch": 60.31,
+      "learning_rate": 0.0002109473684210526,
+      "loss": 0.0021,
+      "step": 4041
+    },
+    {
+      "epoch": 60.33,
+      "learning_rate": 0.00021091228070175435,
+      "loss": 0.0255,
+      "step": 4042
+    },
+    {
+      "epoch": 60.34,
+      "learning_rate": 0.00021087719298245613,
+      "loss": 0.1407,
+      "step": 4043
+    },
+    {
+      "epoch": 60.36,
+      "learning_rate": 0.00021084210526315788,
+      "loss": 0.0017,
+      "step": 4044
+    },
+    {
+      "epoch": 60.37,
+      "learning_rate": 0.00021080701754385962,
+      "loss": 0.0023,
+      "step": 4045
+    },
+    {
+      "epoch": 60.39,
+      "learning_rate": 0.00021077192982456137,
+      "loss": 0.2124,
+      "step": 4046
+    },
+    {
+      "epoch": 60.4,
+      "learning_rate": 0.00021073684210526315,
+      "loss": 0.0597,
+      "step": 4047
+    },
+    {
+      "epoch": 60.42,
+      "learning_rate": 0.0002107017543859649,
+      "loss": 0.002,
+      "step": 4048
+    },
+    {
+      "epoch": 60.43,
+      "learning_rate": 0.00021066666666666665,
+      "loss": 0.1701,
+      "step": 4049
+    },
+    {
+      "epoch": 60.45,
+      "learning_rate": 0.00021063157894736842,
+      "loss": 0.0052,
+      "step": 4050
+    },
+    {
+      "epoch": 60.46,
+      "learning_rate": 0.00021059649122807017,
+      "loss": 0.0016,
+      "step": 4051
+    },
+    {
+      "epoch": 60.48,
+      "learning_rate": 0.00021056140350877192,
+      "loss": 0.0016,
+      "step": 4052
+    },
+    {
+      "epoch": 60.49,
+      "learning_rate": 0.00021052631578947364,
+      "loss": 0.0013,
+      "step": 4053
+    },
+    {
+      "epoch": 60.51,
+      "learning_rate": 0.00021049122807017544,
+      "loss": 0.0026,
+      "step": 4054
+    },
+    {
+      "epoch": 60.52,
+      "learning_rate": 0.00021045614035087717,
+      "loss": 0.0104,
+      "step": 4055
+    },
+    {
+      "epoch": 60.54,
+      "learning_rate": 0.00021042105263157891,
+      "loss": 0.0041,
+      "step": 4056
+    },
+    {
+      "epoch": 60.55,
+      "learning_rate": 0.00021038596491228066,
+      "loss": 0.022,
+      "step": 4057
+    },
+    {
+      "epoch": 60.57,
+      "learning_rate": 0.00021035087719298244,
+      "loss": 0.3154,
+      "step": 4058
+    },
+    {
+      "epoch": 60.58,
+      "learning_rate": 0.0002103157894736842,
+      "loss": 0.0074,
+      "step": 4059
+    },
+    {
+      "epoch": 60.59,
+      "learning_rate": 0.00021028070175438594,
+      "loss": 0.0018,
+      "step": 4060
+    },
+    {
+      "epoch": 60.61,
+      "learning_rate": 0.0002102456140350877,
+      "loss": 0.1199,
+      "step": 4061
+    },
+    {
+      "epoch": 60.62,
+      "learning_rate": 0.00021021052631578946,
+      "loss": 0.0353,
+      "step": 4062
+    },
+    {
+      "epoch": 60.64,
+      "learning_rate": 0.0002101754385964912,
+      "loss": 0.011,
+      "step": 4063
+    },
+    {
+      "epoch": 60.65,
+      "learning_rate": 0.00021014035087719296,
+      "loss": 0.0122,
+      "step": 4064
+    },
+    {
+      "epoch": 60.67,
+      "learning_rate": 0.00021010526315789474,
+      "loss": 0.0019,
+      "step": 4065
+    },
+    {
+      "epoch": 60.68,
+      "learning_rate": 0.00021007017543859648,
+      "loss": 0.2162,
+      "step": 4066
+    },
+    {
+      "epoch": 60.7,
+      "learning_rate": 0.00021003508771929823,
+      "loss": 0.0015,
+      "step": 4067
+    },
+    {
+      "epoch": 60.71,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 0.0029,
+      "step": 4068
+    },
+    {
+      "epoch": 60.73,
+      "learning_rate": 0.00020996491228070176,
+      "loss": 0.0676,
+      "step": 4069
+    },
+    {
+      "epoch": 60.74,
+      "learning_rate": 0.00020992982456140348,
+      "loss": 0.0236,
+      "step": 4070
+    },
+    {
+      "epoch": 60.76,
+      "learning_rate": 0.00020989473684210523,
+      "loss": 0.2489,
+      "step": 4071
+    },
+    {
+      "epoch": 60.77,
+      "learning_rate": 0.00020985964912280698,
+      "loss": 0.0363,
+      "step": 4072
+    },
+    {
+      "epoch": 60.79,
+      "learning_rate": 0.00020982456140350875,
+      "loss": 0.0253,
+      "step": 4073
+    },
+    {
+      "epoch": 60.8,
+      "learning_rate": 0.0002097894736842105,
+      "loss": 0.111,
+      "step": 4074
+    },
+    {
+      "epoch": 60.82,
+      "learning_rate": 0.00020975438596491225,
+      "loss": 0.4341,
+      "step": 4075
+    },
+    {
+      "epoch": 60.83,
+      "learning_rate": 0.00020971929824561403,
+      "loss": 0.157,
+      "step": 4076
+    },
+    {
+      "epoch": 60.85,
+      "learning_rate": 0.00020968421052631577,
+      "loss": 0.0185,
+      "step": 4077
+    },
+    {
+      "epoch": 60.86,
+      "learning_rate": 0.00020964912280701752,
+      "loss": 0.1298,
+      "step": 4078
+    },
+    {
+      "epoch": 60.88,
+      "learning_rate": 0.00020961403508771927,
+      "loss": 0.1157,
+      "step": 4079
+    },
+    {
+      "epoch": 60.89,
+      "learning_rate": 0.00020957894736842105,
+      "loss": 0.0145,
+      "step": 4080
+    },
+    {
+      "epoch": 60.91,
+      "learning_rate": 0.0002095438596491228,
+      "loss": 0.0021,
+      "step": 4081
+    },
+    {
+      "epoch": 60.92,
+      "learning_rate": 0.00020950877192982455,
+      "loss": 0.0042,
+      "step": 4082
+    },
+    {
+      "epoch": 60.94,
+      "learning_rate": 0.0002094736842105263,
+      "loss": 0.1683,
+      "step": 4083
+    },
+    {
+      "epoch": 60.95,
+      "learning_rate": 0.00020943859649122807,
+      "loss": 0.3351,
+      "step": 4084
+    },
+    {
+      "epoch": 60.97,
+      "learning_rate": 0.00020940350877192982,
+      "loss": 0.2813,
+      "step": 4085
+    },
+    {
+      "epoch": 60.98,
+      "learning_rate": 0.00020936842105263154,
+      "loss": 0.0018,
+      "step": 4086
+    },
+    {
+      "epoch": 61.0,
+      "learning_rate": 0.00020933333333333334,
+      "loss": 0.1741,
+      "step": 4087
+    },
+    {
+      "epoch": 61.01,
+      "learning_rate": 0.00020929824561403507,
+      "loss": 0.0651,
+      "step": 4088
+    },
+    {
+      "epoch": 61.03,
+      "learning_rate": 0.00020926315789473681,
+      "loss": 0.2118,
+      "step": 4089
+    },
+    {
+      "epoch": 61.04,
+      "learning_rate": 0.00020922807017543856,
+      "loss": 0.003,
+      "step": 4090
+    },
+    {
+      "epoch": 61.06,
+      "learning_rate": 0.00020919298245614034,
+      "loss": 0.1208,
+      "step": 4091
+    },
+    {
+      "epoch": 61.07,
+      "learning_rate": 0.0002091578947368421,
+      "loss": 0.0036,
+      "step": 4092
+    },
+    {
+      "epoch": 61.09,
+      "learning_rate": 0.00020912280701754384,
+      "loss": 0.1213,
+      "step": 4093
+    },
+    {
+      "epoch": 61.1,
+      "learning_rate": 0.00020908771929824559,
+      "loss": 0.0017,
+      "step": 4094
+    },
+    {
+      "epoch": 61.12,
+      "learning_rate": 0.00020905263157894736,
+      "loss": 0.0023,
+      "step": 4095
+    },
+    {
+      "epoch": 61.13,
+      "learning_rate": 0.0002090175438596491,
+      "loss": 0.0016,
+      "step": 4096
+    },
+    {
+      "epoch": 61.15,
+      "learning_rate": 0.00020898245614035086,
+      "loss": 0.1565,
+      "step": 4097
+    },
+    {
+      "epoch": 61.16,
+      "learning_rate": 0.0002089473684210526,
+      "loss": 0.0819,
+      "step": 4098
+    },
+    {
+      "epoch": 61.18,
+      "learning_rate": 0.00020891228070175438,
+      "loss": 0.0196,
+      "step": 4099
+    },
+    {
+      "epoch": 61.19,
+      "learning_rate": 0.00020887719298245613,
+      "loss": 0.0111,
+      "step": 4100
+    },
+    {
+      "epoch": 61.21,
+      "learning_rate": 0.00020884210526315785,
+      "loss": 0.0028,
+      "step": 4101
+    },
+    {
+      "epoch": 61.22,
+      "learning_rate": 0.00020880701754385966,
+      "loss": 0.0065,
+      "step": 4102
+    },
+    {
+      "epoch": 61.24,
+      "learning_rate": 0.00020877192982456138,
+      "loss": 0.0015,
+      "step": 4103
+    },
+    {
+      "epoch": 61.25,
+      "learning_rate": 0.00020873684210526313,
+      "loss": 0.3828,
+      "step": 4104
+    },
+    {
+      "epoch": 61.27,
+      "learning_rate": 0.00020870175438596488,
+      "loss": 0.295,
+      "step": 4105
+    },
+    {
+      "epoch": 61.28,
+      "learning_rate": 0.00020866666666666665,
+      "loss": 0.3663,
+      "step": 4106
+    },
+    {
+      "epoch": 61.3,
+      "learning_rate": 0.0002086315789473684,
+      "loss": 0.0094,
+      "step": 4107
+    },
+    {
+      "epoch": 61.31,
+      "learning_rate": 0.00020859649122807015,
+      "loss": 0.0764,
+      "step": 4108
+    },
+    {
+      "epoch": 61.33,
+      "learning_rate": 0.0002085614035087719,
+      "loss": 0.2064,
+      "step": 4109
+    },
+    {
+      "epoch": 61.34,
+      "learning_rate": 0.00020852631578947367,
+      "loss": 0.2032,
+      "step": 4110
+    },
+    {
+      "epoch": 61.36,
+      "learning_rate": 0.00020849122807017542,
+      "loss": 0.0028,
+      "step": 4111
+    },
+    {
+      "epoch": 61.37,
+      "learning_rate": 0.00020845614035087717,
+      "loss": 0.247,
+      "step": 4112
+    },
+    {
+      "epoch": 61.39,
+      "learning_rate": 0.00020842105263157895,
+      "loss": 0.0929,
+      "step": 4113
+    },
+    {
+      "epoch": 61.4,
+      "learning_rate": 0.0002083859649122807,
+      "loss": 0.1304,
+      "step": 4114
+    },
+    {
+      "epoch": 61.42,
+      "learning_rate": 0.00020835087719298245,
+      "loss": 0.0107,
+      "step": 4115
+    },
+    {
+      "epoch": 61.43,
+      "learning_rate": 0.00020831578947368417,
+      "loss": 0.0028,
+      "step": 4116
+    },
+    {
+      "epoch": 61.45,
+      "learning_rate": 0.00020828070175438597,
+      "loss": 0.0404,
+      "step": 4117
+    },
+    {
+      "epoch": 61.46,
+      "learning_rate": 0.0002082456140350877,
+      "loss": 0.0027,
+      "step": 4118
+    },
+    {
+      "epoch": 61.48,
+      "learning_rate": 0.00020821052631578944,
+      "loss": 0.0116,
+      "step": 4119
+    },
+    {
+      "epoch": 61.49,
+      "learning_rate": 0.0002081754385964912,
+      "loss": 0.1348,
+      "step": 4120
+    },
+    {
+      "epoch": 61.51,
+      "learning_rate": 0.00020814035087719296,
+      "loss": 0.2591,
+      "step": 4121
+    },
+    {
+      "epoch": 61.52,
+      "learning_rate": 0.00020810526315789471,
+      "loss": 0.0318,
+      "step": 4122
+    },
+    {
+      "epoch": 61.54,
+      "learning_rate": 0.00020807017543859646,
+      "loss": 0.004,
+      "step": 4123
+    },
+    {
+      "epoch": 61.55,
+      "learning_rate": 0.00020803508771929824,
+      "loss": 0.1941,
+      "step": 4124
+    },
+    {
+      "epoch": 61.57,
+      "learning_rate": 0.000208,
+      "loss": 0.315,
+      "step": 4125
+    },
+    {
+      "epoch": 61.58,
+      "learning_rate": 0.00020796491228070174,
+      "loss": 0.5173,
+      "step": 4126
+    },
+    {
+      "epoch": 61.59,
+      "learning_rate": 0.00020792982456140348,
+      "loss": 0.0103,
+      "step": 4127
+    },
+    {
+      "epoch": 61.61,
+      "learning_rate": 0.00020789473684210526,
+      "loss": 0.0171,
+      "step": 4128
+    },
+    {
+      "epoch": 61.62,
+      "learning_rate": 0.000207859649122807,
+      "loss": 0.044,
+      "step": 4129
+    },
+    {
+      "epoch": 61.64,
+      "learning_rate": 0.00020782456140350876,
+      "loss": 0.1884,
+      "step": 4130
+    },
+    {
+      "epoch": 61.65,
+      "learning_rate": 0.0002077894736842105,
+      "loss": 0.0724,
+      "step": 4131
+    },
+    {
+      "epoch": 61.67,
+      "learning_rate": 0.00020775438596491228,
+      "loss": 0.1644,
+      "step": 4132
+    },
+    {
+      "epoch": 61.68,
+      "learning_rate": 0.00020771929824561403,
+      "loss": 0.0115,
+      "step": 4133
+    },
+    {
+      "epoch": 61.7,
+      "learning_rate": 0.00020768421052631575,
+      "loss": 0.0602,
+      "step": 4134
+    },
+    {
+      "epoch": 61.71,
+      "learning_rate": 0.0002076491228070175,
+      "loss": 0.1779,
+      "step": 4135
+    },
+    {
+      "epoch": 61.73,
+      "learning_rate": 0.00020761403508771928,
+      "loss": 0.3983,
+      "step": 4136
+    },
+    {
+      "epoch": 61.74,
+      "learning_rate": 0.00020757894736842103,
+      "loss": 0.034,
+      "step": 4137
+    },
+    {
+      "epoch": 61.76,
+      "learning_rate": 0.00020754385964912278,
+      "loss": 0.016,
+      "step": 4138
+    },
+    {
+      "epoch": 61.77,
+      "learning_rate": 0.00020750877192982455,
+      "loss": 0.0098,
+      "step": 4139
+    },
+    {
+      "epoch": 61.79,
+      "learning_rate": 0.0002074736842105263,
+      "loss": 0.1986,
+      "step": 4140
+    },
+    {
+      "epoch": 61.8,
+      "learning_rate": 0.00020743859649122805,
+      "loss": 0.2418,
+      "step": 4141
+    },
+    {
+      "epoch": 61.82,
+      "learning_rate": 0.0002074035087719298,
+      "loss": 0.156,
+      "step": 4142
+    },
+    {
+      "epoch": 61.83,
+      "learning_rate": 0.00020736842105263157,
+      "loss": 0.0444,
+      "step": 4143
+    },
+    {
+      "epoch": 61.85,
+      "learning_rate": 0.00020733333333333332,
+      "loss": 0.2965,
+      "step": 4144
+    },
+    {
+      "epoch": 61.86,
+      "learning_rate": 0.00020729824561403507,
+      "loss": 0.0129,
+      "step": 4145
+    },
+    {
+      "epoch": 61.88,
+      "learning_rate": 0.00020726315789473682,
+      "loss": 0.1533,
+      "step": 4146
+    },
+    {
+      "epoch": 61.89,
+      "learning_rate": 0.0002072280701754386,
+      "loss": 0.0443,
+      "step": 4147
+    },
+    {
+      "epoch": 61.91,
+      "learning_rate": 0.00020719298245614034,
+      "loss": 0.2175,
+      "step": 4148
+    },
+    {
+      "epoch": 61.92,
+      "learning_rate": 0.00020715789473684207,
+      "loss": 0.0256,
+      "step": 4149
+    },
+    {
+      "epoch": 61.94,
+      "learning_rate": 0.00020712280701754387,
+      "loss": 0.2934,
+      "step": 4150
+    },
+    {
+      "epoch": 61.95,
+      "learning_rate": 0.0002070877192982456,
+      "loss": 0.0082,
+      "step": 4151
+    },
+    {
+      "epoch": 61.97,
+      "learning_rate": 0.00020705263157894734,
+      "loss": 0.2396,
+      "step": 4152
+    },
+    {
+      "epoch": 61.98,
+      "learning_rate": 0.0002070175438596491,
+      "loss": 0.0225,
+      "step": 4153
+    },
+    {
+      "epoch": 62.0,
+      "learning_rate": 0.00020698245614035086,
+      "loss": 0.0113,
+      "step": 4154
+    },
+    {
+      "epoch": 62.01,
+      "learning_rate": 0.0002069473684210526,
+      "loss": 0.0102,
+      "step": 4155
+    },
+    {
+      "epoch": 62.03,
+      "learning_rate": 0.00020691228070175436,
+      "loss": 0.1027,
+      "step": 4156
+    },
+    {
+      "epoch": 62.04,
+      "learning_rate": 0.0002068771929824561,
+      "loss": 0.0548,
+      "step": 4157
+    },
+    {
+      "epoch": 62.06,
+      "learning_rate": 0.0002068421052631579,
+      "loss": 0.0603,
+      "step": 4158
+    },
+    {
+      "epoch": 62.07,
+      "learning_rate": 0.00020680701754385964,
+      "loss": 0.1559,
+      "step": 4159
+    },
+    {
+      "epoch": 62.09,
+      "learning_rate": 0.00020677192982456138,
+      "loss": 0.0041,
+      "step": 4160
+    },
+    {
+      "epoch": 62.1,
+      "learning_rate": 0.00020673684210526316,
+      "loss": 0.2439,
+      "step": 4161
+    },
+    {
+      "epoch": 62.12,
+      "learning_rate": 0.0002067017543859649,
+      "loss": 0.055,
+      "step": 4162
+    },
+    {
+      "epoch": 62.13,
+      "learning_rate": 0.00020666666666666666,
+      "loss": 0.009,
+      "step": 4163
+    },
+    {
+      "epoch": 62.15,
+      "learning_rate": 0.00020663157894736838,
+      "loss": 0.0038,
+      "step": 4164
+    },
+    {
+      "epoch": 62.16,
+      "learning_rate": 0.00020659649122807018,
+      "loss": 0.0186,
+      "step": 4165
+    },
+    {
+      "epoch": 62.18,
+      "learning_rate": 0.0002065614035087719,
+      "loss": 0.1261,
+      "step": 4166
+    },
+    {
+      "epoch": 62.19,
+      "learning_rate": 0.00020652631578947365,
+      "loss": 0.0824,
+      "step": 4167
+    },
+    {
+      "epoch": 62.21,
+      "learning_rate": 0.0002064912280701754,
+      "loss": 0.0174,
+      "step": 4168
+    },
+    {
+      "epoch": 62.22,
+      "learning_rate": 0.00020645614035087718,
+      "loss": 0.0025,
+      "step": 4169
+    },
+    {
+      "epoch": 62.24,
+      "learning_rate": 0.00020642105263157893,
+      "loss": 0.0024,
+      "step": 4170
+    },
+    {
+      "epoch": 62.25,
+      "learning_rate": 0.00020638596491228067,
+      "loss": 0.1987,
+      "step": 4171
+    },
+    {
+      "epoch": 62.27,
+      "learning_rate": 0.00020635087719298242,
+      "loss": 0.2195,
+      "step": 4172
+    },
+    {
+      "epoch": 62.28,
+      "learning_rate": 0.0002063157894736842,
+      "loss": 0.2643,
+      "step": 4173
+    },
+    {
+      "epoch": 62.3,
+      "learning_rate": 0.00020628070175438595,
+      "loss": 0.1246,
+      "step": 4174
+    },
+    {
+      "epoch": 62.31,
+      "learning_rate": 0.0002062456140350877,
+      "loss": 0.0093,
+      "step": 4175
+    },
+    {
+      "epoch": 62.33,
+      "learning_rate": 0.00020621052631578947,
+      "loss": 0.0032,
+      "step": 4176
+    },
+    {
+      "epoch": 62.34,
+      "learning_rate": 0.00020617543859649122,
+      "loss": 0.0156,
+      "step": 4177
+    },
+    {
+      "epoch": 62.36,
+      "learning_rate": 0.00020614035087719297,
+      "loss": 0.0021,
+      "step": 4178
+    },
+    {
+      "epoch": 62.37,
+      "learning_rate": 0.00020610526315789472,
+      "loss": 0.011,
+      "step": 4179
+    },
+    {
+      "epoch": 62.39,
+      "learning_rate": 0.0002060701754385965,
+      "loss": 0.0111,
+      "step": 4180
+    },
+    {
+      "epoch": 62.4,
+      "learning_rate": 0.00020603508771929824,
+      "loss": 0.0531,
+      "step": 4181
+    },
+    {
+      "epoch": 62.42,
+      "learning_rate": 0.00020599999999999997,
+      "loss": 0.0025,
+      "step": 4182
+    },
+    {
+      "epoch": 62.43,
+      "learning_rate": 0.00020596491228070171,
+      "loss": 0.0019,
+      "step": 4183
+    },
+    {
+      "epoch": 62.45,
+      "learning_rate": 0.0002059298245614035,
+      "loss": 0.1355,
+      "step": 4184
+    },
+    {
+      "epoch": 62.46,
+      "learning_rate": 0.00020589473684210524,
+      "loss": 0.1004,
+      "step": 4185
+    },
+    {
+      "epoch": 62.48,
+      "learning_rate": 0.000205859649122807,
+      "loss": 0.2385,
+      "step": 4186
+    },
+    {
+      "epoch": 62.49,
+      "learning_rate": 0.00020582456140350876,
+      "loss": 0.0027,
+      "step": 4187
+    },
+    {
+      "epoch": 62.51,
+      "learning_rate": 0.0002057894736842105,
+      "loss": 0.0032,
+      "step": 4188
+    },
+    {
+      "epoch": 62.52,
+      "learning_rate": 0.00020575438596491226,
+      "loss": 0.0206,
+      "step": 4189
+    },
+    {
+      "epoch": 62.54,
+      "learning_rate": 0.000205719298245614,
+      "loss": 0.0684,
+      "step": 4190
+    },
+    {
+      "epoch": 62.55,
+      "learning_rate": 0.00020568421052631579,
+      "loss": 0.0065,
+      "step": 4191
+    },
+    {
+      "epoch": 62.57,
+      "learning_rate": 0.00020564912280701753,
+      "loss": 0.0411,
+      "step": 4192
+    },
+    {
+      "epoch": 62.58,
+      "learning_rate": 0.00020561403508771928,
+      "loss": 0.1708,
+      "step": 4193
+    },
+    {
+      "epoch": 62.59,
+      "learning_rate": 0.00020557894736842103,
+      "loss": 0.0696,
+      "step": 4194
+    },
+    {
+      "epoch": 62.61,
+      "learning_rate": 0.0002055438596491228,
+      "loss": 0.111,
+      "step": 4195
+    },
+    {
+      "epoch": 62.62,
+      "learning_rate": 0.00020550877192982456,
+      "loss": 0.2224,
+      "step": 4196
+    },
+    {
+      "epoch": 62.64,
+      "learning_rate": 0.00020547368421052628,
+      "loss": 0.0619,
+      "step": 4197
+    },
+    {
+      "epoch": 62.65,
+      "learning_rate": 0.00020543859649122803,
+      "loss": 0.1194,
+      "step": 4198
+    },
+    {
+      "epoch": 62.67,
+      "learning_rate": 0.0002054035087719298,
+      "loss": 0.1641,
+      "step": 4199
+    },
+    {
+      "epoch": 62.68,
+      "learning_rate": 0.00020536842105263155,
+      "loss": 0.0026,
+      "step": 4200
+    },
+    {
+      "epoch": 62.68,
+      "eval_accuracy": 0.8565834557023985,
+      "eval_f1": 0.8584555081928308,
+      "eval_loss": 0.6736618876457214,
+      "eval_runtime": 345.5411,
+      "eval_samples_per_second": 11.825,
+      "eval_steps_per_second": 0.741,
+      "step": 4200
+    },
+    {
+      "epoch": 62.7,
+      "learning_rate": 0.0002053333333333333,
+      "loss": 0.004,
+      "step": 4201
+    },
+    {
+      "epoch": 62.71,
+      "learning_rate": 0.00020529824561403508,
+      "loss": 0.2727,
+      "step": 4202
+    },
+    {
+      "epoch": 62.73,
+      "learning_rate": 0.00020526315789473683,
+      "loss": 0.0279,
+      "step": 4203
+    },
+    {
+      "epoch": 62.74,
+      "learning_rate": 0.00020522807017543857,
+      "loss": 0.0092,
+      "step": 4204
+    },
+    {
+      "epoch": 62.76,
+      "learning_rate": 0.00020519298245614032,
+      "loss": 0.1556,
+      "step": 4205
+    },
+    {
+      "epoch": 62.77,
+      "learning_rate": 0.0002051578947368421,
+      "loss": 0.1234,
+      "step": 4206
+    },
+    {
+      "epoch": 62.79,
+      "learning_rate": 0.00020512280701754385,
+      "loss": 0.0567,
+      "step": 4207
+    },
+    {
+      "epoch": 62.8,
+      "learning_rate": 0.0002050877192982456,
+      "loss": 0.0048,
+      "step": 4208
+    },
+    {
+      "epoch": 62.82,
+      "learning_rate": 0.00020505263157894735,
+      "loss": 0.1576,
+      "step": 4209
+    },
+    {
+      "epoch": 62.83,
+      "learning_rate": 0.00020501754385964912,
+      "loss": 0.0604,
+      "step": 4210
+    },
+    {
+      "epoch": 62.85,
+      "learning_rate": 0.00020498245614035087,
+      "loss": 0.0071,
+      "step": 4211
+    },
+    {
+      "epoch": 62.86,
+      "learning_rate": 0.0002049473684210526,
+      "loss": 0.189,
+      "step": 4212
+    },
+    {
+      "epoch": 62.88,
+      "learning_rate": 0.0002049122807017544,
+      "loss": 0.0091,
+      "step": 4213
+    },
+    {
+      "epoch": 62.89,
+      "learning_rate": 0.00020487719298245612,
+      "loss": 0.1073,
+      "step": 4214
+    },
+    {
+      "epoch": 62.91,
+      "learning_rate": 0.00020484210526315787,
+      "loss": 0.0408,
+      "step": 4215
+    },
+    {
+      "epoch": 62.92,
+      "learning_rate": 0.00020480701754385961,
+      "loss": 0.1842,
+      "step": 4216
+    },
+    {
+      "epoch": 62.94,
+      "learning_rate": 0.0002047719298245614,
+      "loss": 0.0559,
+      "step": 4217
+    },
+    {
+      "epoch": 62.95,
+      "learning_rate": 0.00020473684210526314,
+      "loss": 0.0073,
+      "step": 4218
+    },
+    {
+      "epoch": 62.97,
+      "learning_rate": 0.0002047017543859649,
+      "loss": 0.071,
+      "step": 4219
+    },
+    {
+      "epoch": 62.98,
+      "learning_rate": 0.00020466666666666664,
+      "loss": 0.0027,
+      "step": 4220
+    },
+    {
+      "epoch": 63.0,
+      "learning_rate": 0.0002046315789473684,
+      "loss": 0.0257,
+      "step": 4221
+    },
+    {
+      "epoch": 63.01,
+      "learning_rate": 0.00020459649122807016,
+      "loss": 0.0126,
+      "step": 4222
+    },
+    {
+      "epoch": 63.03,
+      "learning_rate": 0.0002045614035087719,
+      "loss": 0.1263,
+      "step": 4223
+    },
+    {
+      "epoch": 63.04,
+      "learning_rate": 0.00020452631578947369,
+      "loss": 0.0039,
+      "step": 4224
+    },
+    {
+      "epoch": 63.06,
+      "learning_rate": 0.00020449122807017543,
+      "loss": 0.0579,
+      "step": 4225
+    },
+    {
+      "epoch": 63.07,
+      "learning_rate": 0.00020445614035087718,
+      "loss": 0.0764,
+      "step": 4226
+    },
+    {
+      "epoch": 63.09,
+      "learning_rate": 0.00020442105263157893,
+      "loss": 0.1427,
+      "step": 4227
+    },
+    {
+      "epoch": 63.1,
+      "learning_rate": 0.0002043859649122807,
+      "loss": 0.0591,
+      "step": 4228
+    },
+    {
+      "epoch": 63.12,
+      "learning_rate": 0.00020435087719298246,
+      "loss": 0.2312,
+      "step": 4229
+    },
+    {
+      "epoch": 63.13,
+      "learning_rate": 0.00020431578947368418,
+      "loss": 0.0047,
+      "step": 4230
+    },
+    {
+      "epoch": 63.15,
+      "learning_rate": 0.00020428070175438593,
+      "loss": 0.1699,
+      "step": 4231
+    },
+    {
+      "epoch": 63.16,
+      "learning_rate": 0.0002042456140350877,
+      "loss": 0.0909,
+      "step": 4232
+    },
+    {
+      "epoch": 63.18,
+      "learning_rate": 0.00020421052631578945,
+      "loss": 0.0058,
+      "step": 4233
+    },
+    {
+      "epoch": 63.19,
+      "learning_rate": 0.0002041754385964912,
+      "loss": 0.0023,
+      "step": 4234
+    },
+    {
+      "epoch": 63.21,
+      "learning_rate": 0.00020414035087719295,
+      "loss": 0.0038,
+      "step": 4235
+    },
+    {
+      "epoch": 63.22,
+      "learning_rate": 0.00020410526315789472,
+      "loss": 0.0051,
+      "step": 4236
+    },
+    {
+      "epoch": 63.24,
+      "learning_rate": 0.00020407017543859647,
+      "loss": 0.0402,
+      "step": 4237
+    },
+    {
+      "epoch": 63.25,
+      "learning_rate": 0.00020403508771929822,
+      "loss": 0.0693,
+      "step": 4238
+    },
+    {
+      "epoch": 63.27,
+      "learning_rate": 0.000204,
+      "loss": 0.0375,
+      "step": 4239
+    },
+    {
+      "epoch": 63.28,
+      "learning_rate": 0.00020396491228070175,
+      "loss": 0.0099,
+      "step": 4240
+    },
+    {
+      "epoch": 63.3,
+      "learning_rate": 0.0002039298245614035,
+      "loss": 0.194,
+      "step": 4241
+    },
+    {
+      "epoch": 63.31,
+      "learning_rate": 0.00020389473684210524,
+      "loss": 0.0048,
+      "step": 4242
+    },
+    {
+      "epoch": 63.33,
+      "learning_rate": 0.00020385964912280702,
+      "loss": 0.2181,
+      "step": 4243
+    },
+    {
+      "epoch": 63.34,
+      "learning_rate": 0.00020382456140350877,
+      "loss": 0.0991,
+      "step": 4244
+    },
+    {
+      "epoch": 63.36,
+      "learning_rate": 0.0002037894736842105,
+      "loss": 0.1155,
+      "step": 4245
+    },
+    {
+      "epoch": 63.37,
+      "learning_rate": 0.00020375438596491224,
+      "loss": 0.0227,
+      "step": 4246
+    },
+    {
+      "epoch": 63.39,
+      "learning_rate": 0.00020371929824561402,
+      "loss": 0.2667,
+      "step": 4247
+    },
+    {
+      "epoch": 63.4,
+      "learning_rate": 0.00020368421052631576,
+      "loss": 0.0338,
+      "step": 4248
+    },
+    {
+      "epoch": 63.42,
+      "learning_rate": 0.0002036491228070175,
+      "loss": 0.0102,
+      "step": 4249
+    },
+    {
+      "epoch": 63.43,
+      "learning_rate": 0.0002036140350877193,
+      "loss": 0.0171,
+      "step": 4250
+    },
+    {
+      "epoch": 63.45,
+      "learning_rate": 0.00020357894736842104,
+      "loss": 0.0062,
+      "step": 4251
+    },
+    {
+      "epoch": 63.46,
+      "learning_rate": 0.0002035438596491228,
+      "loss": 0.0692,
+      "step": 4252
+    },
+    {
+      "epoch": 63.48,
+      "learning_rate": 0.00020350877192982454,
+      "loss": 0.0126,
+      "step": 4253
+    },
+    {
+      "epoch": 63.49,
+      "learning_rate": 0.0002034736842105263,
+      "loss": 0.1374,
+      "step": 4254
+    },
+    {
+      "epoch": 63.51,
+      "learning_rate": 0.00020343859649122806,
+      "loss": 0.0052,
+      "step": 4255
+    },
+    {
+      "epoch": 63.52,
+      "learning_rate": 0.0002034035087719298,
+      "loss": 0.0199,
+      "step": 4256
+    },
+    {
+      "epoch": 63.54,
+      "learning_rate": 0.00020336842105263156,
+      "loss": 0.0057,
+      "step": 4257
+    },
+    {
+      "epoch": 63.55,
+      "learning_rate": 0.00020333333333333333,
+      "loss": 0.0557,
+      "step": 4258
+    },
+    {
+      "epoch": 63.57,
+      "learning_rate": 0.00020329824561403508,
+      "loss": 0.2364,
+      "step": 4259
+    },
+    {
+      "epoch": 63.58,
+      "learning_rate": 0.0002032631578947368,
+      "loss": 0.0175,
+      "step": 4260
+    },
+    {
+      "epoch": 63.59,
+      "learning_rate": 0.00020322807017543855,
+      "loss": 0.0065,
+      "step": 4261
+    },
+    {
+      "epoch": 63.61,
+      "learning_rate": 0.00020319298245614033,
+      "loss": 0.0044,
+      "step": 4262
+    },
+    {
+      "epoch": 63.62,
+      "learning_rate": 0.00020315789473684208,
+      "loss": 0.0074,
+      "step": 4263
+    },
+    {
+      "epoch": 63.64,
+      "learning_rate": 0.00020312280701754383,
+      "loss": 0.0929,
+      "step": 4264
+    },
+    {
+      "epoch": 63.65,
+      "learning_rate": 0.0002030877192982456,
+      "loss": 0.3596,
+      "step": 4265
+    },
+    {
+      "epoch": 63.67,
+      "learning_rate": 0.00020305263157894735,
+      "loss": 0.051,
+      "step": 4266
+    },
+    {
+      "epoch": 63.68,
+      "learning_rate": 0.0002030175438596491,
+      "loss": 0.038,
+      "step": 4267
+    },
+    {
+      "epoch": 63.7,
+      "learning_rate": 0.00020298245614035085,
+      "loss": 0.0597,
+      "step": 4268
+    },
+    {
+      "epoch": 63.71,
+      "learning_rate": 0.00020294736842105262,
+      "loss": 0.1669,
+      "step": 4269
+    },
+    {
+      "epoch": 63.73,
+      "learning_rate": 0.00020291228070175437,
+      "loss": 0.0057,
+      "step": 4270
+    },
+    {
+      "epoch": 63.74,
+      "learning_rate": 0.00020287719298245612,
+      "loss": 0.0087,
+      "step": 4271
+    },
+    {
+      "epoch": 63.76,
+      "learning_rate": 0.00020284210526315787,
+      "loss": 0.2221,
+      "step": 4272
+    },
+    {
+      "epoch": 63.77,
+      "learning_rate": 0.00020280701754385965,
+      "loss": 0.0157,
+      "step": 4273
+    },
+    {
+      "epoch": 63.79,
+      "learning_rate": 0.0002027719298245614,
+      "loss": 0.2365,
+      "step": 4274
+    },
+    {
+      "epoch": 63.8,
+      "learning_rate": 0.00020273684210526314,
+      "loss": 0.01,
+      "step": 4275
+    },
+    {
+      "epoch": 63.82,
+      "learning_rate": 0.00020270175438596492,
+      "loss": 0.0646,
+      "step": 4276
+    },
+    {
+      "epoch": 63.83,
+      "learning_rate": 0.00020266666666666664,
+      "loss": 0.0485,
+      "step": 4277
+    },
+    {
+      "epoch": 63.85,
+      "learning_rate": 0.0002026315789473684,
+      "loss": 0.0316,
+      "step": 4278
+    },
+    {
+      "epoch": 63.86,
+      "learning_rate": 0.00020259649122807014,
+      "loss": 0.0029,
+      "step": 4279
+    },
+    {
+      "epoch": 63.88,
+      "learning_rate": 0.00020256140350877192,
+      "loss": 0.002,
+      "step": 4280
+    },
+    {
+      "epoch": 63.89,
+      "learning_rate": 0.00020252631578947366,
+      "loss": 0.0292,
+      "step": 4281
+    },
+    {
+      "epoch": 63.91,
+      "learning_rate": 0.0002024912280701754,
+      "loss": 0.0059,
+      "step": 4282
+    },
+    {
+      "epoch": 63.92,
+      "learning_rate": 0.00020245614035087716,
+      "loss": 0.0683,
+      "step": 4283
+    },
+    {
+      "epoch": 63.94,
+      "learning_rate": 0.00020242105263157894,
+      "loss": 0.0021,
+      "step": 4284
+    },
+    {
+      "epoch": 63.95,
+      "learning_rate": 0.00020238596491228069,
+      "loss": 0.0025,
+      "step": 4285
+    },
+    {
+      "epoch": 63.97,
+      "learning_rate": 0.00020235087719298243,
+      "loss": 0.2327,
+      "step": 4286
+    },
+    {
+      "epoch": 63.98,
+      "learning_rate": 0.0002023157894736842,
+      "loss": 0.2651,
+      "step": 4287
+    },
+    {
+      "epoch": 64.0,
+      "learning_rate": 0.00020228070175438596,
+      "loss": 0.0564,
+      "step": 4288
+    },
+    {
+      "epoch": 64.01,
+      "learning_rate": 0.0002022456140350877,
+      "loss": 0.1241,
+      "step": 4289
+    },
+    {
+      "epoch": 64.03,
+      "learning_rate": 0.00020221052631578946,
+      "loss": 0.1026,
+      "step": 4290
+    },
+    {
+      "epoch": 64.04,
+      "learning_rate": 0.00020217543859649123,
+      "loss": 0.0335,
+      "step": 4291
+    },
+    {
+      "epoch": 64.06,
+      "learning_rate": 0.00020214035087719298,
+      "loss": 0.0337,
+      "step": 4292
+    },
+    {
+      "epoch": 64.07,
+      "learning_rate": 0.0002021052631578947,
+      "loss": 0.0045,
+      "step": 4293
+    },
+    {
+      "epoch": 64.09,
+      "learning_rate": 0.00020207017543859645,
+      "loss": 0.0046,
+      "step": 4294
+    },
+    {
+      "epoch": 64.1,
+      "learning_rate": 0.00020203508771929823,
+      "loss": 0.0291,
+      "step": 4295
+    },
+    {
+      "epoch": 64.12,
+      "learning_rate": 0.00020199999999999998,
+      "loss": 0.0063,
+      "step": 4296
+    },
+    {
+      "epoch": 64.13,
+      "learning_rate": 0.00020196491228070173,
+      "loss": 0.0019,
+      "step": 4297
+    },
+    {
+      "epoch": 64.15,
+      "learning_rate": 0.00020192982456140347,
+      "loss": 0.0518,
+      "step": 4298
+    },
+    {
+      "epoch": 64.16,
+      "learning_rate": 0.00020189473684210525,
+      "loss": 0.0032,
+      "step": 4299
+    },
+    {
+      "epoch": 64.18,
+      "learning_rate": 0.000201859649122807,
+      "loss": 0.0083,
+      "step": 4300
+    },
+    {
+      "epoch": 64.19,
+      "learning_rate": 0.00020182456140350875,
+      "loss": 0.0019,
+      "step": 4301
+    },
+    {
+      "epoch": 64.21,
+      "learning_rate": 0.00020178947368421052,
+      "loss": 0.0107,
+      "step": 4302
+    },
+    {
+      "epoch": 64.22,
+      "learning_rate": 0.00020175438596491227,
+      "loss": 0.0465,
+      "step": 4303
+    },
+    {
+      "epoch": 64.24,
+      "learning_rate": 0.00020171929824561402,
+      "loss": 0.0032,
+      "step": 4304
+    },
+    {
+      "epoch": 64.25,
+      "learning_rate": 0.00020168421052631577,
+      "loss": 0.0136,
+      "step": 4305
+    },
+    {
+      "epoch": 64.27,
+      "learning_rate": 0.00020164912280701755,
+      "loss": 0.1822,
+      "step": 4306
+    },
+    {
+      "epoch": 64.28,
+      "learning_rate": 0.0002016140350877193,
+      "loss": 0.022,
+      "step": 4307
+    },
+    {
+      "epoch": 64.3,
+      "learning_rate": 0.00020157894736842102,
+      "loss": 0.004,
+      "step": 4308
+    },
+    {
+      "epoch": 64.31,
+      "learning_rate": 0.00020154385964912277,
+      "loss": 0.2661,
+      "step": 4309
+    },
+    {
+      "epoch": 64.33,
+      "learning_rate": 0.00020150877192982454,
+      "loss": 0.632,
+      "step": 4310
+    },
+    {
+      "epoch": 64.34,
+      "learning_rate": 0.0002014736842105263,
+      "loss": 0.0027,
+      "step": 4311
+    },
+    {
+      "epoch": 64.36,
+      "learning_rate": 0.00020143859649122804,
+      "loss": 0.0271,
+      "step": 4312
+    },
+    {
+      "epoch": 64.37,
+      "learning_rate": 0.00020140350877192981,
+      "loss": 0.0028,
+      "step": 4313
+    },
+    {
+      "epoch": 64.39,
+      "learning_rate": 0.00020136842105263156,
+      "loss": 0.0025,
+      "step": 4314
+    },
+    {
+      "epoch": 64.4,
+      "learning_rate": 0.0002013333333333333,
+      "loss": 0.0028,
+      "step": 4315
+    },
+    {
+      "epoch": 64.42,
+      "learning_rate": 0.00020129824561403506,
+      "loss": 0.029,
+      "step": 4316
+    },
+    {
+      "epoch": 64.43,
+      "learning_rate": 0.00020126315789473684,
+      "loss": 0.0315,
+      "step": 4317
+    },
+    {
+      "epoch": 64.45,
+      "learning_rate": 0.00020122807017543859,
+      "loss": 0.2121,
+      "step": 4318
+    },
+    {
+      "epoch": 64.46,
+      "learning_rate": 0.00020119298245614033,
+      "loss": 0.0234,
+      "step": 4319
+    },
+    {
+      "epoch": 64.48,
+      "learning_rate": 0.00020115789473684208,
+      "loss": 0.0018,
+      "step": 4320
+    },
+    {
+      "epoch": 64.49,
+      "learning_rate": 0.00020112280701754386,
+      "loss": 0.3627,
+      "step": 4321
+    },
+    {
+      "epoch": 64.51,
+      "learning_rate": 0.0002010877192982456,
+      "loss": 0.0109,
+      "step": 4322
+    },
+    {
+      "epoch": 64.52,
+      "learning_rate": 0.00020105263157894736,
+      "loss": 0.2671,
+      "step": 4323
+    },
+    {
+      "epoch": 64.54,
+      "learning_rate": 0.00020101754385964913,
+      "loss": 0.2699,
+      "step": 4324
+    },
+    {
+      "epoch": 64.55,
+      "learning_rate": 0.00020098245614035085,
+      "loss": 0.0016,
+      "step": 4325
+    },
+    {
+      "epoch": 64.57,
+      "learning_rate": 0.0002009473684210526,
+      "loss": 0.1817,
+      "step": 4326
+    },
+    {
+      "epoch": 64.58,
+      "learning_rate": 0.00020091228070175435,
+      "loss": 0.1634,
+      "step": 4327
+    },
+    {
+      "epoch": 64.59,
+      "learning_rate": 0.00020087719298245613,
+      "loss": 0.0018,
+      "step": 4328
+    },
+    {
+      "epoch": 64.61,
+      "learning_rate": 0.00020084210526315788,
+      "loss": 0.0023,
+      "step": 4329
+    },
+    {
+      "epoch": 64.62,
+      "learning_rate": 0.00020080701754385963,
+      "loss": 0.0434,
+      "step": 4330
+    },
+    {
+      "epoch": 64.64,
+      "learning_rate": 0.00020077192982456137,
+      "loss": 0.1258,
+      "step": 4331
+    },
+    {
+      "epoch": 64.65,
+      "learning_rate": 0.00020073684210526315,
+      "loss": 0.0021,
+      "step": 4332
+    },
+    {
+      "epoch": 64.67,
+      "learning_rate": 0.0002007017543859649,
+      "loss": 0.0041,
+      "step": 4333
+    },
+    {
+      "epoch": 64.68,
+      "learning_rate": 0.00020066666666666665,
+      "loss": 0.0288,
+      "step": 4334
+    },
+    {
+      "epoch": 64.7,
+      "learning_rate": 0.0002006315789473684,
+      "loss": 0.0029,
+      "step": 4335
+    },
+    {
+      "epoch": 64.71,
+      "learning_rate": 0.00020059649122807017,
+      "loss": 0.0031,
+      "step": 4336
+    },
+    {
+      "epoch": 64.73,
+      "learning_rate": 0.00020056140350877192,
+      "loss": 0.0202,
+      "step": 4337
+    },
+    {
+      "epoch": 64.74,
+      "learning_rate": 0.00020052631578947367,
+      "loss": 0.007,
+      "step": 4338
+    },
+    {
+      "epoch": 64.76,
+      "learning_rate": 0.00020049122807017545,
+      "loss": 0.0825,
+      "step": 4339
+    },
+    {
+      "epoch": 64.77,
+      "learning_rate": 0.0002004561403508772,
+      "loss": 0.073,
+      "step": 4340
+    },
+    {
+      "epoch": 64.79,
+      "learning_rate": 0.00020042105263157892,
+      "loss": 0.0015,
+      "step": 4341
+    },
+    {
+      "epoch": 64.8,
+      "learning_rate": 0.00020038596491228066,
+      "loss": 0.1713,
+      "step": 4342
+    },
+    {
+      "epoch": 64.82,
+      "learning_rate": 0.00020035087719298244,
+      "loss": 0.1075,
+      "step": 4343
+    },
+    {
+      "epoch": 64.83,
+      "learning_rate": 0.0002003157894736842,
+      "loss": 0.0029,
+      "step": 4344
+    },
+    {
+      "epoch": 64.85,
+      "learning_rate": 0.00020028070175438594,
+      "loss": 0.0301,
+      "step": 4345
+    },
+    {
+      "epoch": 64.86,
+      "learning_rate": 0.0002002456140350877,
+      "loss": 0.0017,
+      "step": 4346
+    },
+    {
+      "epoch": 64.88,
+      "learning_rate": 0.00020021052631578946,
+      "loss": 0.0022,
+      "step": 4347
+    },
+    {
+      "epoch": 64.89,
+      "learning_rate": 0.0002001754385964912,
+      "loss": 0.3129,
+      "step": 4348
+    },
+    {
+      "epoch": 64.91,
+      "learning_rate": 0.00020014035087719296,
+      "loss": 0.101,
+      "step": 4349
+    },
+    {
+      "epoch": 64.92,
+      "learning_rate": 0.00020010526315789474,
+      "loss": 0.0029,
+      "step": 4350
+    },
+    {
+      "epoch": 64.94,
+      "learning_rate": 0.00020007017543859648,
+      "loss": 0.008,
+      "step": 4351
+    },
+    {
+      "epoch": 64.95,
+      "learning_rate": 0.00020003508771929823,
+      "loss": 0.1807,
+      "step": 4352
+    },
+    {
+      "epoch": 64.97,
+      "learning_rate": 0.00019999999999999998,
+      "loss": 0.053,
+      "step": 4353
+    },
+    {
+      "epoch": 64.98,
+      "learning_rate": 0.00019996491228070176,
+      "loss": 0.1483,
+      "step": 4354
+    },
+    {
+      "epoch": 65.0,
+      "learning_rate": 0.0001999298245614035,
+      "loss": 0.0238,
+      "step": 4355
+    },
+    {
+      "epoch": 65.01,
+      "learning_rate": 0.00019989473684210523,
+      "loss": 0.0031,
+      "step": 4356
+    },
+    {
+      "epoch": 65.03,
+      "learning_rate": 0.00019985964912280698,
+      "loss": 0.0186,
+      "step": 4357
+    },
+    {
+      "epoch": 65.04,
+      "learning_rate": 0.00019982456140350875,
+      "loss": 0.04,
+      "step": 4358
+    },
+    {
+      "epoch": 65.06,
+      "learning_rate": 0.0001997894736842105,
+      "loss": 0.0081,
+      "step": 4359
+    },
+    {
+      "epoch": 65.07,
+      "learning_rate": 0.00019975438596491225,
+      "loss": 0.0024,
+      "step": 4360
+    },
+    {
+      "epoch": 65.09,
+      "learning_rate": 0.000199719298245614,
+      "loss": 0.0114,
+      "step": 4361
+    },
+    {
+      "epoch": 65.1,
+      "learning_rate": 0.00019968421052631578,
+      "loss": 0.0039,
+      "step": 4362
+    },
+    {
+      "epoch": 65.12,
+      "learning_rate": 0.00019964912280701752,
+      "loss": 0.0028,
+      "step": 4363
+    },
+    {
+      "epoch": 65.13,
+      "learning_rate": 0.00019961403508771927,
+      "loss": 0.0313,
+      "step": 4364
+    },
+    {
+      "epoch": 65.15,
+      "learning_rate": 0.00019957894736842105,
+      "loss": 0.2049,
+      "step": 4365
+    },
+    {
+      "epoch": 65.16,
+      "learning_rate": 0.0001995438596491228,
+      "loss": 0.3819,
+      "step": 4366
+    },
+    {
+      "epoch": 65.18,
+      "learning_rate": 0.00019950877192982455,
+      "loss": 0.0066,
+      "step": 4367
+    },
+    {
+      "epoch": 65.19,
+      "learning_rate": 0.0001994736842105263,
+      "loss": 0.0154,
+      "step": 4368
+    },
+    {
+      "epoch": 65.21,
+      "learning_rate": 0.00019943859649122807,
+      "loss": 0.039,
+      "step": 4369
+    },
+    {
+      "epoch": 65.22,
+      "learning_rate": 0.00019940350877192982,
+      "loss": 0.0363,
+      "step": 4370
+    },
+    {
+      "epoch": 65.24,
+      "learning_rate": 0.00019936842105263154,
+      "loss": 0.1389,
+      "step": 4371
+    },
+    {
+      "epoch": 65.25,
+      "learning_rate": 0.0001993333333333333,
+      "loss": 0.0658,
+      "step": 4372
+    },
+    {
+      "epoch": 65.27,
+      "learning_rate": 0.00019929824561403507,
+      "loss": 0.2018,
+      "step": 4373
+    },
+    {
+      "epoch": 65.28,
+      "learning_rate": 0.00019926315789473682,
+      "loss": 0.0042,
+      "step": 4374
+    },
+    {
+      "epoch": 65.3,
+      "learning_rate": 0.00019922807017543856,
+      "loss": 0.0155,
+      "step": 4375
+    },
+    {
+      "epoch": 65.31,
+      "learning_rate": 0.00019919298245614034,
+      "loss": 0.1579,
+      "step": 4376
+    },
+    {
+      "epoch": 65.33,
+      "learning_rate": 0.0001991578947368421,
+      "loss": 0.4092,
+      "step": 4377
+    },
+    {
+      "epoch": 65.34,
+      "learning_rate": 0.00019912280701754384,
+      "loss": 0.0113,
+      "step": 4378
+    },
+    {
+      "epoch": 65.36,
+      "learning_rate": 0.00019908771929824559,
+      "loss": 0.3527,
+      "step": 4379
+    },
+    {
+      "epoch": 65.37,
+      "learning_rate": 0.00019905263157894736,
+      "loss": 0.0811,
+      "step": 4380
+    },
+    {
+      "epoch": 65.39,
+      "learning_rate": 0.0001990175438596491,
+      "loss": 0.0931,
+      "step": 4381
+    },
+    {
+      "epoch": 65.4,
+      "learning_rate": 0.00019898245614035086,
+      "loss": 0.1252,
+      "step": 4382
+    },
+    {
+      "epoch": 65.42,
+      "learning_rate": 0.0001989473684210526,
+      "loss": 0.263,
+      "step": 4383
+    },
+    {
+      "epoch": 65.43,
+      "learning_rate": 0.00019891228070175438,
+      "loss": 0.0018,
+      "step": 4384
+    },
+    {
+      "epoch": 65.45,
+      "learning_rate": 0.00019887719298245613,
+      "loss": 0.0345,
+      "step": 4385
+    },
+    {
+      "epoch": 65.46,
+      "learning_rate": 0.00019884210526315788,
+      "loss": 0.2768,
+      "step": 4386
+    },
+    {
+      "epoch": 65.48,
+      "learning_rate": 0.00019880701754385966,
+      "loss": 0.0059,
+      "step": 4387
+    },
+    {
+      "epoch": 65.49,
+      "learning_rate": 0.0001987719298245614,
+      "loss": 0.2509,
+      "step": 4388
+    },
+    {
+      "epoch": 65.51,
+      "learning_rate": 0.00019873684210526313,
+      "loss": 0.02,
+      "step": 4389
+    },
+    {
+      "epoch": 65.52,
+      "learning_rate": 0.00019870175438596488,
+      "loss": 0.0966,
+      "step": 4390
+    },
+    {
+      "epoch": 65.54,
+      "learning_rate": 0.00019866666666666665,
+      "loss": 0.0251,
+      "step": 4391
+    },
+    {
+      "epoch": 65.55,
+      "learning_rate": 0.0001986315789473684,
+      "loss": 0.0032,
+      "step": 4392
+    },
+    {
+      "epoch": 65.57,
+      "learning_rate": 0.00019859649122807015,
+      "loss": 0.2893,
+      "step": 4393
+    },
+    {
+      "epoch": 65.58,
+      "learning_rate": 0.0001985614035087719,
+      "loss": 0.1985,
+      "step": 4394
+    },
+    {
+      "epoch": 65.59,
+      "learning_rate": 0.00019852631578947368,
+      "loss": 0.0492,
+      "step": 4395
+    },
+    {
+      "epoch": 65.61,
+      "learning_rate": 0.00019849122807017542,
+      "loss": 0.0081,
+      "step": 4396
+    },
+    {
+      "epoch": 65.62,
+      "learning_rate": 0.00019845614035087717,
+      "loss": 0.3234,
+      "step": 4397
+    },
+    {
+      "epoch": 65.64,
+      "learning_rate": 0.00019842105263157892,
+      "loss": 0.0271,
+      "step": 4398
+    },
+    {
+      "epoch": 65.65,
+      "learning_rate": 0.0001983859649122807,
+      "loss": 0.1399,
+      "step": 4399
+    },
+    {
+      "epoch": 65.67,
+      "learning_rate": 0.00019835087719298245,
+      "loss": 0.0043,
+      "step": 4400
+    },
+    {
+      "epoch": 65.67,
+      "eval_accuracy": 0.8387175721977485,
+      "eval_f1": 0.8415862353144246,
+      "eval_loss": 0.7780405879020691,
+      "eval_runtime": 345.1415,
+      "eval_samples_per_second": 11.839,
+      "eval_steps_per_second": 0.742,
+      "step": 4400
+    },
+    {
+      "epoch": 65.68,
+      "learning_rate": 0.0001983157894736842,
+      "loss": 0.1963,
+      "step": 4401
+    },
+    {
+      "epoch": 65.7,
+      "learning_rate": 0.00019828070175438597,
+      "loss": 0.3146,
+      "step": 4402
+    },
+    {
+      "epoch": 65.71,
+      "learning_rate": 0.00019824561403508772,
+      "loss": 0.0092,
+      "step": 4403
+    },
+    {
+      "epoch": 65.73,
+      "learning_rate": 0.00019821052631578944,
+      "loss": 0.0416,
+      "step": 4404
+    },
+    {
+      "epoch": 65.74,
+      "learning_rate": 0.0001981754385964912,
+      "loss": 0.2324,
+      "step": 4405
+    },
+    {
+      "epoch": 65.76,
+      "learning_rate": 0.00019814035087719297,
+      "loss": 0.1401,
+      "step": 4406
+    },
+    {
+      "epoch": 65.77,
+      "learning_rate": 0.00019810526315789471,
+      "loss": 0.0429,
+      "step": 4407
+    },
+    {
+      "epoch": 65.79,
+      "learning_rate": 0.00019807017543859646,
+      "loss": 0.1155,
+      "step": 4408
+    },
+    {
+      "epoch": 65.8,
+      "learning_rate": 0.0001980350877192982,
+      "loss": 0.1608,
+      "step": 4409
+    },
+    {
+      "epoch": 65.82,
+      "learning_rate": 0.000198,
+      "loss": 0.3258,
+      "step": 4410
+    },
+    {
+      "epoch": 65.83,
+      "learning_rate": 0.00019796491228070174,
+      "loss": 0.0427,
+      "step": 4411
+    },
+    {
+      "epoch": 65.85,
+      "learning_rate": 0.00019792982456140349,
+      "loss": 0.2753,
+      "step": 4412
+    },
+    {
+      "epoch": 65.86,
+      "learning_rate": 0.00019789473684210526,
+      "loss": 0.0221,
+      "step": 4413
+    },
+    {
+      "epoch": 65.88,
+      "learning_rate": 0.000197859649122807,
+      "loss": 0.0376,
+      "step": 4414
+    },
+    {
+      "epoch": 65.89,
+      "learning_rate": 0.00019782456140350876,
+      "loss": 0.0378,
+      "step": 4415
+    },
+    {
+      "epoch": 65.91,
+      "learning_rate": 0.0001977894736842105,
+      "loss": 0.0455,
+      "step": 4416
+    },
+    {
+      "epoch": 65.92,
+      "learning_rate": 0.00019775438596491228,
+      "loss": 0.0532,
+      "step": 4417
+    },
+    {
+      "epoch": 65.94,
+      "learning_rate": 0.00019771929824561403,
+      "loss": 0.1445,
+      "step": 4418
+    },
+    {
+      "epoch": 65.95,
+      "learning_rate": 0.00019768421052631575,
+      "loss": 0.035,
+      "step": 4419
+    },
+    {
+      "epoch": 65.97,
+      "learning_rate": 0.0001976491228070175,
+      "loss": 0.4898,
+      "step": 4420
+    },
+    {
+      "epoch": 65.98,
+      "learning_rate": 0.00019761403508771928,
+      "loss": 0.0462,
+      "step": 4421
+    },
+    {
+      "epoch": 66.0,
+      "learning_rate": 0.00019757894736842103,
+      "loss": 0.018,
+      "step": 4422
+    },
+    {
+      "epoch": 66.01,
+      "learning_rate": 0.00019754385964912278,
+      "loss": 0.021,
+      "step": 4423
+    },
+    {
+      "epoch": 66.03,
+      "learning_rate": 0.00019750877192982453,
+      "loss": 0.1062,
+      "step": 4424
+    },
+    {
+      "epoch": 66.04,
+      "learning_rate": 0.0001974736842105263,
+      "loss": 0.2795,
+      "step": 4425
+    },
+    {
+      "epoch": 66.06,
+      "learning_rate": 0.00019743859649122805,
+      "loss": 0.0288,
+      "step": 4426
+    },
+    {
+      "epoch": 66.07,
+      "learning_rate": 0.0001974035087719298,
+      "loss": 0.0331,
+      "step": 4427
+    },
+    {
+      "epoch": 66.09,
+      "learning_rate": 0.00019736842105263157,
+      "loss": 0.0093,
+      "step": 4428
+    },
+    {
+      "epoch": 66.1,
+      "learning_rate": 0.00019733333333333332,
+      "loss": 0.0161,
+      "step": 4429
+    },
+    {
+      "epoch": 66.12,
+      "learning_rate": 0.00019729824561403507,
+      "loss": 0.0112,
+      "step": 4430
+    },
+    {
+      "epoch": 66.13,
+      "learning_rate": 0.00019726315789473682,
+      "loss": 0.0462,
+      "step": 4431
+    },
+    {
+      "epoch": 66.15,
+      "learning_rate": 0.0001972280701754386,
+      "loss": 0.0103,
+      "step": 4432
+    },
+    {
+      "epoch": 66.16,
+      "learning_rate": 0.00019719298245614035,
+      "loss": 0.0626,
+      "step": 4433
+    },
+    {
+      "epoch": 66.18,
+      "learning_rate": 0.0001971578947368421,
+      "loss": 0.0874,
+      "step": 4434
+    },
+    {
+      "epoch": 66.19,
+      "learning_rate": 0.00019712280701754382,
+      "loss": 0.0121,
+      "step": 4435
+    },
+    {
+      "epoch": 66.21,
+      "learning_rate": 0.00019708771929824562,
+      "loss": 0.2007,
+      "step": 4436
+    },
+    {
+      "epoch": 66.22,
+      "learning_rate": 0.00019705263157894734,
+      "loss": 0.0053,
+      "step": 4437
+    },
+    {
+      "epoch": 66.24,
+      "learning_rate": 0.0001970175438596491,
+      "loss": 0.0047,
+      "step": 4438
+    },
+    {
+      "epoch": 66.25,
+      "learning_rate": 0.00019698245614035087,
+      "loss": 0.0076,
+      "step": 4439
+    },
+    {
+      "epoch": 66.27,
+      "learning_rate": 0.00019694736842105261,
+      "loss": 0.1006,
+      "step": 4440
+    },
+    {
+      "epoch": 66.28,
+      "learning_rate": 0.00019691228070175436,
+      "loss": 0.0038,
+      "step": 4441
+    },
+    {
+      "epoch": 66.3,
+      "learning_rate": 0.0001968771929824561,
+      "loss": 0.0139,
+      "step": 4442
+    },
+    {
+      "epoch": 66.31,
+      "learning_rate": 0.0001968421052631579,
+      "loss": 0.0153,
+      "step": 4443
+    },
+    {
+      "epoch": 66.33,
+      "learning_rate": 0.00019680701754385964,
+      "loss": 0.087,
+      "step": 4444
+    },
+    {
+      "epoch": 66.34,
+      "learning_rate": 0.00019677192982456139,
+      "loss": 0.1041,
+      "step": 4445
+    },
+    {
+      "epoch": 66.36,
+      "learning_rate": 0.00019673684210526313,
+      "loss": 0.0709,
+      "step": 4446
+    },
+    {
+      "epoch": 66.37,
+      "learning_rate": 0.0001967017543859649,
+      "loss": 0.0218,
+      "step": 4447
+    },
+    {
+      "epoch": 66.39,
+      "learning_rate": 0.00019666666666666666,
+      "loss": 0.3569,
+      "step": 4448
+    },
+    {
+      "epoch": 66.4,
+      "learning_rate": 0.0001966315789473684,
+      "loss": 0.0036,
+      "step": 4449
+    },
+    {
+      "epoch": 66.42,
+      "learning_rate": 0.00019659649122807018,
+      "loss": 0.0239,
+      "step": 4450
+    },
+    {
+      "epoch": 66.43,
+      "learning_rate": 0.00019656140350877193,
+      "loss": 0.1154,
+      "step": 4451
+    },
+    {
+      "epoch": 66.45,
+      "learning_rate": 0.00019652631578947365,
+      "loss": 0.0184,
+      "step": 4452
+    },
+    {
+      "epoch": 66.46,
+      "learning_rate": 0.0001964912280701754,
+      "loss": 0.1561,
+      "step": 4453
+    },
+    {
+      "epoch": 66.48,
+      "learning_rate": 0.00019645614035087718,
+      "loss": 0.0059,
+      "step": 4454
+    },
+    {
+      "epoch": 66.49,
+      "learning_rate": 0.00019642105263157893,
+      "loss": 0.0033,
+      "step": 4455
+    },
+    {
+      "epoch": 66.51,
+      "learning_rate": 0.00019638596491228068,
+      "loss": 0.0037,
+      "step": 4456
+    },
+    {
+      "epoch": 66.52,
+      "learning_rate": 0.00019635087719298242,
+      "loss": 0.4539,
+      "step": 4457
+    },
+    {
+      "epoch": 66.54,
+      "learning_rate": 0.0001963157894736842,
+      "loss": 0.0782,
+      "step": 4458
+    },
+    {
+      "epoch": 66.55,
+      "learning_rate": 0.00019628070175438595,
+      "loss": 0.0233,
+      "step": 4459
+    },
+    {
+      "epoch": 66.57,
+      "learning_rate": 0.0001962456140350877,
+      "loss": 0.0077,
+      "step": 4460
+    },
+    {
+      "epoch": 66.58,
+      "learning_rate": 0.00019621052631578945,
+      "loss": 0.0778,
+      "step": 4461
+    },
+    {
+      "epoch": 66.59,
+      "learning_rate": 0.00019617543859649122,
+      "loss": 0.0108,
+      "step": 4462
+    },
+    {
+      "epoch": 66.61,
+      "learning_rate": 0.00019614035087719297,
+      "loss": 0.0633,
+      "step": 4463
+    },
+    {
+      "epoch": 66.62,
+      "learning_rate": 0.00019610526315789472,
+      "loss": 0.0154,
+      "step": 4464
+    },
+    {
+      "epoch": 66.64,
+      "learning_rate": 0.0001960701754385965,
+      "loss": 0.2093,
+      "step": 4465
+    },
+    {
+      "epoch": 66.65,
+      "learning_rate": 0.00019603508771929824,
+      "loss": 0.0022,
+      "step": 4466
+    },
+    {
+      "epoch": 66.67,
+      "learning_rate": 0.00019599999999999997,
+      "loss": 0.0282,
+      "step": 4467
+    },
+    {
+      "epoch": 66.68,
+      "learning_rate": 0.00019596491228070172,
+      "loss": 0.1057,
+      "step": 4468
+    },
+    {
+      "epoch": 66.7,
+      "learning_rate": 0.0001959298245614035,
+      "loss": 0.2186,
+      "step": 4469
+    },
+    {
+      "epoch": 66.71,
+      "learning_rate": 0.00019589473684210524,
+      "loss": 0.0053,
+      "step": 4470
+    },
+    {
+      "epoch": 66.73,
+      "learning_rate": 0.000195859649122807,
+      "loss": 0.002,
+      "step": 4471
+    },
+    {
+      "epoch": 66.74,
+      "learning_rate": 0.00019582456140350874,
+      "loss": 0.0124,
+      "step": 4472
+    },
+    {
+      "epoch": 66.76,
+      "learning_rate": 0.0001957894736842105,
+      "loss": 0.0051,
+      "step": 4473
+    },
+    {
+      "epoch": 66.77,
+      "learning_rate": 0.00019575438596491226,
+      "loss": 0.0029,
+      "step": 4474
+    },
+    {
+      "epoch": 66.79,
+      "learning_rate": 0.000195719298245614,
+      "loss": 0.1849,
+      "step": 4475
+    },
+    {
+      "epoch": 66.8,
+      "learning_rate": 0.0001956842105263158,
+      "loss": 0.0463,
+      "step": 4476
+    },
+    {
+      "epoch": 66.82,
+      "learning_rate": 0.00019564912280701754,
+      "loss": 0.0748,
+      "step": 4477
+    },
+    {
+      "epoch": 66.83,
+      "learning_rate": 0.00019561403508771928,
+      "loss": 0.1394,
+      "step": 4478
+    },
+    {
+      "epoch": 66.85,
+      "learning_rate": 0.00019557894736842103,
+      "loss": 0.0029,
+      "step": 4479
+    },
+    {
+      "epoch": 66.86,
+      "learning_rate": 0.0001955438596491228,
+      "loss": 0.0054,
+      "step": 4480
+    },
+    {
+      "epoch": 66.88,
+      "learning_rate": 0.00019550877192982456,
+      "loss": 0.1783,
+      "step": 4481
+    },
+    {
+      "epoch": 66.89,
+      "learning_rate": 0.0001954736842105263,
+      "loss": 0.1708,
+      "step": 4482
+    },
+    {
+      "epoch": 66.91,
+      "learning_rate": 0.00019543859649122803,
+      "loss": 0.0019,
+      "step": 4483
+    },
+    {
+      "epoch": 66.92,
+      "learning_rate": 0.0001954035087719298,
+      "loss": 0.0016,
+      "step": 4484
+    },
+    {
+      "epoch": 66.94,
+      "learning_rate": 0.00019536842105263155,
+      "loss": 0.0361,
+      "step": 4485
+    },
+    {
+      "epoch": 66.95,
+      "learning_rate": 0.0001953333333333333,
+      "loss": 0.0388,
+      "step": 4486
+    },
+    {
+      "epoch": 66.97,
+      "learning_rate": 0.00019529824561403508,
+      "loss": 0.1492,
+      "step": 4487
+    },
+    {
+      "epoch": 66.98,
+      "learning_rate": 0.00019526315789473683,
+      "loss": 0.246,
+      "step": 4488
+    },
+    {
+      "epoch": 67.0,
+      "learning_rate": 0.00019522807017543858,
+      "loss": 0.0024,
+      "step": 4489
+    },
+    {
+      "epoch": 67.01,
+      "learning_rate": 0.00019519298245614032,
+      "loss": 0.0028,
+      "step": 4490
+    },
+    {
+      "epoch": 67.03,
+      "learning_rate": 0.0001951578947368421,
+      "loss": 0.009,
+      "step": 4491
+    },
+    {
+      "epoch": 67.04,
+      "learning_rate": 0.00019512280701754385,
+      "loss": 0.0023,
+      "step": 4492
+    },
+    {
+      "epoch": 67.06,
+      "learning_rate": 0.0001950877192982456,
+      "loss": 0.0056,
+      "step": 4493
+    },
+    {
+      "epoch": 67.07,
+      "learning_rate": 0.00019505263157894735,
+      "loss": 0.0018,
+      "step": 4494
+    },
+    {
+      "epoch": 67.09,
+      "learning_rate": 0.00019501754385964912,
+      "loss": 0.0128,
+      "step": 4495
+    },
+    {
+      "epoch": 67.1,
+      "learning_rate": 0.00019498245614035087,
+      "loss": 0.0055,
+      "step": 4496
+    },
+    {
+      "epoch": 67.12,
+      "learning_rate": 0.00019494736842105262,
+      "loss": 0.0201,
+      "step": 4497
+    },
+    {
+      "epoch": 67.13,
+      "learning_rate": 0.00019491228070175434,
+      "loss": 0.0064,
+      "step": 4498
+    },
+    {
+      "epoch": 67.15,
+      "learning_rate": 0.00019487719298245614,
+      "loss": 0.0973,
+      "step": 4499
+    },
+    {
+      "epoch": 67.16,
+      "learning_rate": 0.00019484210526315787,
+      "loss": 0.0027,
+      "step": 4500
+    },
+    {
+      "epoch": 67.18,
+      "learning_rate": 0.00019480701754385961,
+      "loss": 0.1812,
+      "step": 4501
+    },
+    {
+      "epoch": 67.19,
+      "learning_rate": 0.0001947719298245614,
+      "loss": 0.0089,
+      "step": 4502
+    },
+    {
+      "epoch": 67.21,
+      "learning_rate": 0.00019473684210526314,
+      "loss": 0.0528,
+      "step": 4503
+    },
+    {
+      "epoch": 67.22,
+      "learning_rate": 0.0001947017543859649,
+      "loss": 0.002,
+      "step": 4504
+    },
+    {
+      "epoch": 67.24,
+      "learning_rate": 0.00019466666666666664,
+      "loss": 0.0464,
+      "step": 4505
+    },
+    {
+      "epoch": 67.25,
+      "learning_rate": 0.0001946315789473684,
+      "loss": 0.0053,
+      "step": 4506
+    },
+    {
+      "epoch": 67.27,
+      "learning_rate": 0.00019459649122807016,
+      "loss": 0.0055,
+      "step": 4507
+    },
+    {
+      "epoch": 67.28,
+      "learning_rate": 0.0001945614035087719,
+      "loss": 0.0269,
+      "step": 4508
+    },
+    {
+      "epoch": 67.3,
+      "learning_rate": 0.00019452631578947366,
+      "loss": 0.0141,
+      "step": 4509
+    },
+    {
+      "epoch": 67.31,
+      "learning_rate": 0.00019449122807017544,
+      "loss": 0.0025,
+      "step": 4510
+    },
+    {
+      "epoch": 67.33,
+      "learning_rate": 0.00019445614035087718,
+      "loss": 0.192,
+      "step": 4511
+    },
+    {
+      "epoch": 67.34,
+      "learning_rate": 0.00019442105263157893,
+      "loss": 0.1234,
+      "step": 4512
+    },
+    {
+      "epoch": 67.36,
+      "learning_rate": 0.0001943859649122807,
+      "loss": 0.0036,
+      "step": 4513
+    },
+    {
+      "epoch": 67.37,
+      "learning_rate": 0.00019435087719298246,
+      "loss": 0.1498,
+      "step": 4514
+    },
+    {
+      "epoch": 67.39,
+      "learning_rate": 0.00019431578947368418,
+      "loss": 0.004,
+      "step": 4515
+    },
+    {
+      "epoch": 67.4,
+      "learning_rate": 0.00019428070175438593,
+      "loss": 0.0039,
+      "step": 4516
+    },
+    {
+      "epoch": 67.42,
+      "learning_rate": 0.0001942456140350877,
+      "loss": 0.1355,
+      "step": 4517
+    },
+    {
+      "epoch": 67.43,
+      "learning_rate": 0.00019421052631578945,
+      "loss": 0.0039,
+      "step": 4518
+    },
+    {
+      "epoch": 67.45,
+      "learning_rate": 0.0001941754385964912,
+      "loss": 0.0032,
+      "step": 4519
+    },
+    {
+      "epoch": 67.46,
+      "learning_rate": 0.00019414035087719295,
+      "loss": 0.0042,
+      "step": 4520
+    },
+    {
+      "epoch": 67.48,
+      "learning_rate": 0.00019410526315789473,
+      "loss": 0.0115,
+      "step": 4521
+    },
+    {
+      "epoch": 67.49,
+      "learning_rate": 0.00019407017543859647,
+      "loss": 0.2845,
+      "step": 4522
+    },
+    {
+      "epoch": 67.51,
+      "learning_rate": 0.00019403508771929822,
+      "loss": 0.0287,
+      "step": 4523
+    },
+    {
+      "epoch": 67.52,
+      "learning_rate": 0.00019399999999999997,
+      "loss": 0.0677,
+      "step": 4524
+    },
+    {
+      "epoch": 67.54,
+      "learning_rate": 0.00019396491228070175,
+      "loss": 0.1119,
+      "step": 4525
+    },
+    {
+      "epoch": 67.55,
+      "learning_rate": 0.0001939298245614035,
+      "loss": 0.0017,
+      "step": 4526
+    },
+    {
+      "epoch": 67.57,
+      "learning_rate": 0.00019389473684210525,
+      "loss": 0.5221,
+      "step": 4527
+    },
+    {
+      "epoch": 67.58,
+      "learning_rate": 0.00019385964912280702,
+      "loss": 0.2991,
+      "step": 4528
+    },
+    {
+      "epoch": 67.59,
+      "learning_rate": 0.00019382456140350877,
+      "loss": 0.0145,
+      "step": 4529
+    },
+    {
+      "epoch": 67.61,
+      "learning_rate": 0.00019378947368421052,
+      "loss": 0.0054,
+      "step": 4530
+    },
+    {
+      "epoch": 67.62,
+      "learning_rate": 0.00019375438596491224,
+      "loss": 0.0019,
+      "step": 4531
+    },
+    {
+      "epoch": 67.64,
+      "learning_rate": 0.00019371929824561402,
+      "loss": 0.002,
+      "step": 4532
+    },
+    {
+      "epoch": 67.65,
+      "learning_rate": 0.00019368421052631577,
+      "loss": 0.1062,
+      "step": 4533
+    },
+    {
+      "epoch": 67.67,
+      "learning_rate": 0.00019364912280701751,
+      "loss": 0.0275,
+      "step": 4534
+    },
+    {
+      "epoch": 67.68,
+      "learning_rate": 0.00019361403508771926,
+      "loss": 0.0017,
+      "step": 4535
+    },
+    {
+      "epoch": 67.7,
+      "learning_rate": 0.00019357894736842104,
+      "loss": 0.2389,
+      "step": 4536
+    },
+    {
+      "epoch": 67.71,
+      "learning_rate": 0.0001935438596491228,
+      "loss": 0.0711,
+      "step": 4537
+    },
+    {
+      "epoch": 67.73,
+      "learning_rate": 0.00019350877192982454,
+      "loss": 0.0034,
+      "step": 4538
+    },
+    {
+      "epoch": 67.74,
+      "learning_rate": 0.0001934736842105263,
+      "loss": 0.0016,
+      "step": 4539
+    },
+    {
+      "epoch": 67.76,
+      "learning_rate": 0.00019343859649122806,
+      "loss": 0.0042,
+      "step": 4540
+    },
+    {
+      "epoch": 67.77,
+      "learning_rate": 0.0001934035087719298,
+      "loss": 0.0104,
+      "step": 4541
+    },
+    {
+      "epoch": 67.79,
+      "learning_rate": 0.00019336842105263156,
+      "loss": 0.0176,
+      "step": 4542
+    },
+    {
+      "epoch": 67.8,
+      "learning_rate": 0.00019333333333333333,
+      "loss": 0.0078,
+      "step": 4543
+    },
+    {
+      "epoch": 67.82,
+      "learning_rate": 0.00019329824561403508,
+      "loss": 0.0097,
+      "step": 4544
+    },
+    {
+      "epoch": 67.83,
+      "learning_rate": 0.00019326315789473683,
+      "loss": 0.0031,
+      "step": 4545
+    },
+    {
+      "epoch": 67.85,
+      "learning_rate": 0.00019322807017543855,
+      "loss": 0.0199,
+      "step": 4546
+    },
+    {
+      "epoch": 67.86,
+      "learning_rate": 0.00019319298245614036,
+      "loss": 0.0096,
+      "step": 4547
+    },
+    {
+      "epoch": 67.88,
+      "learning_rate": 0.00019315789473684208,
+      "loss": 0.0373,
+      "step": 4548
+    },
+    {
+      "epoch": 67.89,
+      "learning_rate": 0.00019312280701754383,
+      "loss": 0.0832,
+      "step": 4549
+    },
+    {
+      "epoch": 67.91,
+      "learning_rate": 0.0001930877192982456,
+      "loss": 0.0047,
+      "step": 4550
+    },
+    {
+      "epoch": 67.92,
+      "learning_rate": 0.00019305263157894735,
+      "loss": 0.003,
+      "step": 4551
+    },
+    {
+      "epoch": 67.94,
+      "learning_rate": 0.0001930175438596491,
+      "loss": 0.0063,
+      "step": 4552
+    },
+    {
+      "epoch": 67.95,
+      "learning_rate": 0.00019298245614035085,
+      "loss": 0.0092,
+      "step": 4553
+    },
+    {
+      "epoch": 67.97,
+      "learning_rate": 0.00019294736842105263,
+      "loss": 0.0266,
+      "step": 4554
+    },
+    {
+      "epoch": 67.98,
+      "learning_rate": 0.00019291228070175437,
+      "loss": 0.0016,
+      "step": 4555
+    },
+    {
+      "epoch": 68.0,
+      "learning_rate": 0.00019287719298245612,
+      "loss": 0.0127,
+      "step": 4556
+    },
+    {
+      "epoch": 68.01,
+      "learning_rate": 0.00019284210526315787,
+      "loss": 0.0633,
+      "step": 4557
+    },
+    {
+      "epoch": 68.03,
+      "learning_rate": 0.00019280701754385965,
+      "loss": 0.0054,
+      "step": 4558
+    },
+    {
+      "epoch": 68.04,
+      "learning_rate": 0.0001927719298245614,
+      "loss": 0.003,
+      "step": 4559
+    },
+    {
+      "epoch": 68.06,
+      "learning_rate": 0.00019273684210526315,
+      "loss": 0.1727,
+      "step": 4560
+    },
+    {
+      "epoch": 68.07,
+      "learning_rate": 0.00019270175438596487,
+      "loss": 0.0038,
+      "step": 4561
+    },
+    {
+      "epoch": 68.09,
+      "learning_rate": 0.00019266666666666667,
+      "loss": 0.0344,
+      "step": 4562
+    },
+    {
+      "epoch": 68.1,
+      "learning_rate": 0.0001926315789473684,
+      "loss": 0.0019,
+      "step": 4563
+    },
+    {
+      "epoch": 68.12,
+      "learning_rate": 0.00019259649122807014,
+      "loss": 0.0056,
+      "step": 4564
+    },
+    {
+      "epoch": 68.13,
+      "learning_rate": 0.00019256140350877192,
+      "loss": 0.0031,
+      "step": 4565
+    },
+    {
+      "epoch": 68.15,
+      "learning_rate": 0.00019252631578947366,
+      "loss": 0.002,
+      "step": 4566
+    },
+    {
+      "epoch": 68.16,
+      "learning_rate": 0.00019249122807017541,
+      "loss": 0.01,
+      "step": 4567
+    },
+    {
+      "epoch": 68.18,
+      "learning_rate": 0.00019245614035087716,
+      "loss": 0.0028,
+      "step": 4568
+    },
+    {
+      "epoch": 68.19,
+      "learning_rate": 0.00019242105263157894,
+      "loss": 0.1223,
+      "step": 4569
+    },
+    {
+      "epoch": 68.21,
+      "learning_rate": 0.0001923859649122807,
+      "loss": 0.0052,
+      "step": 4570
+    },
+    {
+      "epoch": 68.22,
+      "learning_rate": 0.00019235087719298244,
+      "loss": 0.3806,
+      "step": 4571
+    },
+    {
+      "epoch": 68.24,
+      "learning_rate": 0.00019231578947368418,
+      "loss": 0.0019,
+      "step": 4572
+    },
+    {
+      "epoch": 68.25,
+      "learning_rate": 0.00019228070175438596,
+      "loss": 0.0223,
+      "step": 4573
+    },
+    {
+      "epoch": 68.27,
+      "learning_rate": 0.0001922456140350877,
+      "loss": 0.0017,
+      "step": 4574
+    },
+    {
+      "epoch": 68.28,
+      "learning_rate": 0.00019221052631578946,
+      "loss": 0.1442,
+      "step": 4575
+    },
+    {
+      "epoch": 68.3,
+      "learning_rate": 0.00019217543859649123,
+      "loss": 0.0333,
+      "step": 4576
+    },
+    {
+      "epoch": 68.31,
+      "learning_rate": 0.00019214035087719298,
+      "loss": 0.1573,
+      "step": 4577
+    },
+    {
+      "epoch": 68.33,
+      "learning_rate": 0.0001921052631578947,
+      "loss": 0.1816,
+      "step": 4578
+    },
+    {
+      "epoch": 68.34,
+      "learning_rate": 0.00019207017543859645,
+      "loss": 0.0019,
+      "step": 4579
+    },
+    {
+      "epoch": 68.36,
+      "learning_rate": 0.00019203508771929823,
+      "loss": 0.0022,
+      "step": 4580
+    },
+    {
+      "epoch": 68.37,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 0.0021,
+      "step": 4581
+    },
+    {
+      "epoch": 68.39,
+      "learning_rate": 0.00019196491228070173,
+      "loss": 0.0018,
+      "step": 4582
+    },
+    {
+      "epoch": 68.4,
+      "learning_rate": 0.00019192982456140348,
+      "loss": 0.0197,
+      "step": 4583
+    },
+    {
+      "epoch": 68.42,
+      "learning_rate": 0.00019189473684210525,
+      "loss": 0.1204,
+      "step": 4584
+    },
+    {
+      "epoch": 68.43,
+      "learning_rate": 0.000191859649122807,
+      "loss": 0.0027,
+      "step": 4585
+    },
+    {
+      "epoch": 68.45,
+      "learning_rate": 0.00019182456140350875,
+      "loss": 0.194,
+      "step": 4586
+    },
+    {
+      "epoch": 68.46,
+      "learning_rate": 0.0001917894736842105,
+      "loss": 0.0022,
+      "step": 4587
+    },
+    {
+      "epoch": 68.48,
+      "learning_rate": 0.00019175438596491227,
+      "loss": 0.003,
+      "step": 4588
+    },
+    {
+      "epoch": 68.49,
+      "learning_rate": 0.00019171929824561402,
+      "loss": 0.1746,
+      "step": 4589
+    },
+    {
+      "epoch": 68.51,
+      "learning_rate": 0.00019168421052631577,
+      "loss": 0.2832,
+      "step": 4590
+    },
+    {
+      "epoch": 68.52,
+      "learning_rate": 0.00019164912280701755,
+      "loss": 0.0399,
+      "step": 4591
+    },
+    {
+      "epoch": 68.54,
+      "learning_rate": 0.0001916140350877193,
+      "loss": 0.0209,
+      "step": 4592
+    },
+    {
+      "epoch": 68.55,
+      "learning_rate": 0.00019157894736842104,
+      "loss": 0.0055,
+      "step": 4593
+    },
+    {
+      "epoch": 68.57,
+      "learning_rate": 0.00019154385964912277,
+      "loss": 0.0084,
+      "step": 4594
+    },
+    {
+      "epoch": 68.58,
+      "learning_rate": 0.00019150877192982457,
+      "loss": 0.0051,
+      "step": 4595
+    },
+    {
+      "epoch": 68.59,
+      "learning_rate": 0.0001914736842105263,
+      "loss": 0.1722,
+      "step": 4596
+    },
+    {
+      "epoch": 68.61,
+      "learning_rate": 0.00019143859649122804,
+      "loss": 0.0606,
+      "step": 4597
+    },
+    {
+      "epoch": 68.62,
+      "learning_rate": 0.0001914035087719298,
+      "loss": 0.0611,
+      "step": 4598
+    },
+    {
+      "epoch": 68.64,
+      "learning_rate": 0.00019136842105263156,
+      "loss": 0.0172,
+      "step": 4599
+    },
+    {
+      "epoch": 68.65,
+      "learning_rate": 0.0001913333333333333,
+      "loss": 0.0032,
+      "step": 4600
+    },
+    {
+      "epoch": 68.65,
+      "eval_accuracy": 0.846059716103769,
+      "eval_f1": 0.8482206471494375,
+      "eval_loss": 0.689896821975708,
+      "eval_runtime": 344.0746,
+      "eval_samples_per_second": 11.875,
+      "eval_steps_per_second": 0.744,
+      "step": 4600
+    },
+    {
+      "epoch": 68.67,
+      "learning_rate": 0.00019129824561403506,
+      "loss": 0.0392,
+      "step": 4601
+    },
+    {
+      "epoch": 68.68,
+      "learning_rate": 0.00019126315789473684,
+      "loss": 0.0019,
+      "step": 4602
+    },
+    {
+      "epoch": 68.7,
+      "learning_rate": 0.00019122807017543859,
+      "loss": 0.0025,
+      "step": 4603
+    },
+    {
+      "epoch": 68.71,
+      "learning_rate": 0.00019119298245614034,
+      "loss": 0.0398,
+      "step": 4604
+    },
+    {
+      "epoch": 68.73,
+      "learning_rate": 0.00019115789473684208,
+      "loss": 0.1833,
+      "step": 4605
+    },
+    {
+      "epoch": 68.74,
+      "learning_rate": 0.00019112280701754386,
+      "loss": 0.0022,
+      "step": 4606
+    },
+    {
+      "epoch": 68.76,
+      "learning_rate": 0.0001910877192982456,
+      "loss": 0.0574,
+      "step": 4607
+    },
+    {
+      "epoch": 68.77,
+      "learning_rate": 0.00019105263157894736,
+      "loss": 0.0379,
+      "step": 4608
+    },
+    {
+      "epoch": 68.79,
+      "learning_rate": 0.00019101754385964908,
+      "loss": 0.0053,
+      "step": 4609
+    },
+    {
+      "epoch": 68.8,
+      "learning_rate": 0.00019098245614035088,
+      "loss": 0.0207,
+      "step": 4610
+    },
+    {
+      "epoch": 68.82,
+      "learning_rate": 0.0001909473684210526,
+      "loss": 0.0028,
+      "step": 4611
+    },
+    {
+      "epoch": 68.83,
+      "learning_rate": 0.00019091228070175435,
+      "loss": 0.0186,
+      "step": 4612
+    },
+    {
+      "epoch": 68.85,
+      "learning_rate": 0.00019087719298245613,
+      "loss": 0.1757,
+      "step": 4613
+    },
+    {
+      "epoch": 68.86,
+      "learning_rate": 0.00019084210526315788,
+      "loss": 0.0196,
+      "step": 4614
+    },
+    {
+      "epoch": 68.88,
+      "learning_rate": 0.00019080701754385963,
+      "loss": 0.1241,
+      "step": 4615
+    },
+    {
+      "epoch": 68.89,
+      "learning_rate": 0.00019077192982456137,
+      "loss": 0.0025,
+      "step": 4616
+    },
+    {
+      "epoch": 68.91,
+      "learning_rate": 0.00019073684210526315,
+      "loss": 0.2688,
+      "step": 4617
+    },
+    {
+      "epoch": 68.92,
+      "learning_rate": 0.0001907017543859649,
+      "loss": 0.0955,
+      "step": 4618
+    },
+    {
+      "epoch": 68.94,
+      "learning_rate": 0.00019066666666666665,
+      "loss": 0.1777,
+      "step": 4619
+    },
+    {
+      "epoch": 68.95,
+      "learning_rate": 0.0001906315789473684,
+      "loss": 0.0022,
+      "step": 4620
+    },
+    {
+      "epoch": 68.97,
+      "learning_rate": 0.00019059649122807017,
+      "loss": 0.0059,
+      "step": 4621
+    },
+    {
+      "epoch": 68.98,
+      "learning_rate": 0.00019056140350877192,
+      "loss": 0.0378,
+      "step": 4622
+    },
+    {
+      "epoch": 69.0,
+      "learning_rate": 0.00019052631578947367,
+      "loss": 0.1003,
+      "step": 4623
+    },
+    {
+      "epoch": 69.01,
+      "learning_rate": 0.00019049122807017542,
+      "loss": 0.2193,
+      "step": 4624
+    },
+    {
+      "epoch": 69.03,
+      "learning_rate": 0.0001904561403508772,
+      "loss": 0.211,
+      "step": 4625
+    },
+    {
+      "epoch": 69.04,
+      "learning_rate": 0.00019042105263157892,
+      "loss": 0.1146,
+      "step": 4626
+    },
+    {
+      "epoch": 69.06,
+      "learning_rate": 0.00019038596491228067,
+      "loss": 0.0106,
+      "step": 4627
+    },
+    {
+      "epoch": 69.07,
+      "learning_rate": 0.00019035087719298244,
+      "loss": 0.2057,
+      "step": 4628
+    },
+    {
+      "epoch": 69.09,
+      "learning_rate": 0.0001903157894736842,
+      "loss": 0.011,
+      "step": 4629
+    },
+    {
+      "epoch": 69.1,
+      "learning_rate": 0.00019028070175438594,
+      "loss": 0.1871,
+      "step": 4630
+    },
+    {
+      "epoch": 69.12,
+      "learning_rate": 0.0001902456140350877,
+      "loss": 0.2694,
+      "step": 4631
+    },
+    {
+      "epoch": 69.13,
+      "learning_rate": 0.00019021052631578946,
+      "loss": 0.0786,
+      "step": 4632
+    },
+    {
+      "epoch": 69.15,
+      "learning_rate": 0.0001901754385964912,
+      "loss": 0.0609,
+      "step": 4633
+    },
+    {
+      "epoch": 69.16,
+      "learning_rate": 0.00019014035087719296,
+      "loss": 0.1266,
+      "step": 4634
+    },
+    {
+      "epoch": 69.18,
+      "learning_rate": 0.0001901052631578947,
+      "loss": 0.0047,
+      "step": 4635
+    },
+    {
+      "epoch": 69.19,
+      "learning_rate": 0.00019007017543859649,
+      "loss": 0.0186,
+      "step": 4636
+    },
+    {
+      "epoch": 69.21,
+      "learning_rate": 0.00019003508771929823,
+      "loss": 0.1044,
+      "step": 4637
+    },
+    {
+      "epoch": 69.22,
+      "learning_rate": 0.00018999999999999998,
+      "loss": 0.0125,
+      "step": 4638
+    },
+    {
+      "epoch": 69.24,
+      "learning_rate": 0.00018996491228070176,
+      "loss": 0.0122,
+      "step": 4639
+    },
+    {
+      "epoch": 69.25,
+      "learning_rate": 0.0001899298245614035,
+      "loss": 0.0034,
+      "step": 4640
+    },
+    {
+      "epoch": 69.27,
+      "learning_rate": 0.00018989473684210526,
+      "loss": 0.2012,
+      "step": 4641
+    },
+    {
+      "epoch": 69.28,
+      "learning_rate": 0.00018985964912280698,
+      "loss": 0.1219,
+      "step": 4642
+    },
+    {
+      "epoch": 69.3,
+      "learning_rate": 0.00018982456140350878,
+      "loss": 0.128,
+      "step": 4643
+    },
+    {
+      "epoch": 69.31,
+      "learning_rate": 0.0001897894736842105,
+      "loss": 0.2125,
+      "step": 4644
+    },
+    {
+      "epoch": 69.33,
+      "learning_rate": 0.00018975438596491225,
+      "loss": 0.0541,
+      "step": 4645
+    },
+    {
+      "epoch": 69.34,
+      "learning_rate": 0.000189719298245614,
+      "loss": 0.0034,
+      "step": 4646
+    },
+    {
+      "epoch": 69.36,
+      "learning_rate": 0.00018968421052631578,
+      "loss": 0.1952,
+      "step": 4647
+    },
+    {
+      "epoch": 69.37,
+      "learning_rate": 0.00018964912280701753,
+      "loss": 0.1402,
+      "step": 4648
+    },
+    {
+      "epoch": 69.39,
+      "learning_rate": 0.00018961403508771927,
+      "loss": 0.0026,
+      "step": 4649
+    },
+    {
+      "epoch": 69.4,
+      "learning_rate": 0.00018957894736842105,
+      "loss": 0.0261,
+      "step": 4650
+    },
+    {
+      "epoch": 69.42,
+      "learning_rate": 0.0001895438596491228,
+      "loss": 0.0645,
+      "step": 4651
+    },
+    {
+      "epoch": 69.43,
+      "learning_rate": 0.00018950877192982455,
+      "loss": 0.0089,
+      "step": 4652
+    },
+    {
+      "epoch": 69.45,
+      "learning_rate": 0.0001894736842105263,
+      "loss": 0.0456,
+      "step": 4653
+    },
+    {
+      "epoch": 69.46,
+      "learning_rate": 0.00018943859649122807,
+      "loss": 0.0451,
+      "step": 4654
+    },
+    {
+      "epoch": 69.48,
+      "learning_rate": 0.00018940350877192982,
+      "loss": 0.1816,
+      "step": 4655
+    },
+    {
+      "epoch": 69.49,
+      "learning_rate": 0.00018936842105263157,
+      "loss": 0.0079,
+      "step": 4656
+    },
+    {
+      "epoch": 69.51,
+      "learning_rate": 0.0001893333333333333,
+      "loss": 0.0054,
+      "step": 4657
+    },
+    {
+      "epoch": 69.52,
+      "learning_rate": 0.0001892982456140351,
+      "loss": 0.0098,
+      "step": 4658
+    },
+    {
+      "epoch": 69.54,
+      "learning_rate": 0.00018926315789473682,
+      "loss": 0.1181,
+      "step": 4659
+    },
+    {
+      "epoch": 69.55,
+      "learning_rate": 0.00018922807017543856,
+      "loss": 0.0056,
+      "step": 4660
+    },
+    {
+      "epoch": 69.57,
+      "learning_rate": 0.00018919298245614031,
+      "loss": 0.0632,
+      "step": 4661
+    },
+    {
+      "epoch": 69.58,
+      "learning_rate": 0.0001891578947368421,
+      "loss": 0.0759,
+      "step": 4662
+    },
+    {
+      "epoch": 69.59,
+      "learning_rate": 0.00018912280701754384,
+      "loss": 0.0149,
+      "step": 4663
+    },
+    {
+      "epoch": 69.61,
+      "learning_rate": 0.0001890877192982456,
+      "loss": 0.0175,
+      "step": 4664
+    },
+    {
+      "epoch": 69.62,
+      "learning_rate": 0.00018905263157894736,
+      "loss": 0.0583,
+      "step": 4665
+    },
+    {
+      "epoch": 69.64,
+      "learning_rate": 0.0001890175438596491,
+      "loss": 0.0057,
+      "step": 4666
+    },
+    {
+      "epoch": 69.65,
+      "learning_rate": 0.00018898245614035086,
+      "loss": 0.023,
+      "step": 4667
+    },
+    {
+      "epoch": 69.67,
+      "learning_rate": 0.0001889473684210526,
+      "loss": 0.0054,
+      "step": 4668
+    },
+    {
+      "epoch": 69.68,
+      "learning_rate": 0.00018891228070175439,
+      "loss": 0.0018,
+      "step": 4669
+    },
+    {
+      "epoch": 69.7,
+      "learning_rate": 0.00018887719298245613,
+      "loss": 0.0502,
+      "step": 4670
+    },
+    {
+      "epoch": 69.71,
+      "learning_rate": 0.00018884210526315788,
+      "loss": 0.0018,
+      "step": 4671
+    },
+    {
+      "epoch": 69.73,
+      "learning_rate": 0.0001888070175438596,
+      "loss": 0.005,
+      "step": 4672
+    },
+    {
+      "epoch": 69.74,
+      "learning_rate": 0.0001887719298245614,
+      "loss": 0.3566,
+      "step": 4673
+    },
+    {
+      "epoch": 69.76,
+      "learning_rate": 0.00018873684210526313,
+      "loss": 0.0016,
+      "step": 4674
+    },
+    {
+      "epoch": 69.77,
+      "learning_rate": 0.00018870175438596488,
+      "loss": 0.1829,
+      "step": 4675
+    },
+    {
+      "epoch": 69.79,
+      "learning_rate": 0.00018866666666666665,
+      "loss": 0.0085,
+      "step": 4676
+    },
+    {
+      "epoch": 69.8,
+      "learning_rate": 0.0001886315789473684,
+      "loss": 0.0765,
+      "step": 4677
+    },
+    {
+      "epoch": 69.82,
+      "learning_rate": 0.00018859649122807015,
+      "loss": 0.0022,
+      "step": 4678
+    },
+    {
+      "epoch": 69.83,
+      "learning_rate": 0.0001885614035087719,
+      "loss": 0.0162,
+      "step": 4679
+    },
+    {
+      "epoch": 69.85,
+      "learning_rate": 0.00018852631578947368,
+      "loss": 0.2237,
+      "step": 4680
+    },
+    {
+      "epoch": 69.86,
+      "learning_rate": 0.00018849122807017542,
+      "loss": 0.2743,
+      "step": 4681
+    },
+    {
+      "epoch": 69.88,
+      "learning_rate": 0.00018845614035087717,
+      "loss": 0.0056,
+      "step": 4682
+    },
+    {
+      "epoch": 69.89,
+      "learning_rate": 0.00018842105263157892,
+      "loss": 0.0636,
+      "step": 4683
+    },
+    {
+      "epoch": 69.91,
+      "learning_rate": 0.0001883859649122807,
+      "loss": 0.0069,
+      "step": 4684
+    },
+    {
+      "epoch": 69.92,
+      "learning_rate": 0.00018835087719298245,
+      "loss": 0.1107,
+      "step": 4685
+    },
+    {
+      "epoch": 69.94,
+      "learning_rate": 0.0001883157894736842,
+      "loss": 0.0033,
+      "step": 4686
+    },
+    {
+      "epoch": 69.95,
+      "learning_rate": 0.00018828070175438594,
+      "loss": 0.0026,
+      "step": 4687
+    },
+    {
+      "epoch": 69.97,
+      "learning_rate": 0.00018824561403508772,
+      "loss": 0.188,
+      "step": 4688
+    },
+    {
+      "epoch": 69.98,
+      "learning_rate": 0.00018821052631578947,
+      "loss": 0.1188,
+      "step": 4689
+    },
+    {
+      "epoch": 70.0,
+      "learning_rate": 0.0001881754385964912,
+      "loss": 0.0255,
+      "step": 4690
+    },
+    {
+      "epoch": 70.01,
+      "learning_rate": 0.00018814035087719297,
+      "loss": 0.0019,
+      "step": 4691
+    },
+    {
+      "epoch": 70.03,
+      "learning_rate": 0.00018810526315789472,
+      "loss": 0.2015,
+      "step": 4692
+    },
+    {
+      "epoch": 70.04,
+      "learning_rate": 0.00018807017543859646,
+      "loss": 0.0018,
+      "step": 4693
+    },
+    {
+      "epoch": 70.06,
+      "learning_rate": 0.0001880350877192982,
+      "loss": 0.0096,
+      "step": 4694
+    },
+    {
+      "epoch": 70.07,
+      "learning_rate": 0.000188,
+      "loss": 0.0097,
+      "step": 4695
+    },
+    {
+      "epoch": 70.09,
+      "learning_rate": 0.00018796491228070174,
+      "loss": 0.1639,
+      "step": 4696
+    },
+    {
+      "epoch": 70.1,
+      "learning_rate": 0.0001879298245614035,
+      "loss": 0.0949,
+      "step": 4697
+    },
+    {
+      "epoch": 70.12,
+      "learning_rate": 0.00018789473684210524,
+      "loss": 0.0125,
+      "step": 4698
+    },
+    {
+      "epoch": 70.13,
+      "learning_rate": 0.000187859649122807,
+      "loss": 0.0019,
+      "step": 4699
+    },
+    {
+      "epoch": 70.15,
+      "learning_rate": 0.00018782456140350876,
+      "loss": 0.0016,
+      "step": 4700
+    },
+    {
+      "epoch": 70.16,
+      "learning_rate": 0.0001877894736842105,
+      "loss": 0.0095,
+      "step": 4701
+    },
+    {
+      "epoch": 70.18,
+      "learning_rate": 0.00018775438596491228,
+      "loss": 0.0103,
+      "step": 4702
+    },
+    {
+      "epoch": 70.19,
+      "learning_rate": 0.00018771929824561403,
+      "loss": 0.0036,
+      "step": 4703
+    },
+    {
+      "epoch": 70.21,
+      "learning_rate": 0.00018768421052631578,
+      "loss": 0.0878,
+      "step": 4704
+    },
+    {
+      "epoch": 70.22,
+      "learning_rate": 0.0001876491228070175,
+      "loss": 0.1278,
+      "step": 4705
+    },
+    {
+      "epoch": 70.24,
+      "learning_rate": 0.0001876140350877193,
+      "loss": 0.0028,
+      "step": 4706
+    },
+    {
+      "epoch": 70.25,
+      "learning_rate": 0.00018757894736842103,
+      "loss": 0.0033,
+      "step": 4707
+    },
+    {
+      "epoch": 70.27,
+      "learning_rate": 0.00018754385964912278,
+      "loss": 0.0084,
+      "step": 4708
+    },
+    {
+      "epoch": 70.28,
+      "learning_rate": 0.00018750877192982453,
+      "loss": 0.002,
+      "step": 4709
+    },
+    {
+      "epoch": 70.3,
+      "learning_rate": 0.0001874736842105263,
+      "loss": 0.2291,
+      "step": 4710
+    },
+    {
+      "epoch": 70.31,
+      "learning_rate": 0.00018743859649122805,
+      "loss": 0.005,
+      "step": 4711
+    },
+    {
+      "epoch": 70.33,
+      "learning_rate": 0.0001874035087719298,
+      "loss": 0.133,
+      "step": 4712
+    },
+    {
+      "epoch": 70.34,
+      "learning_rate": 0.00018736842105263158,
+      "loss": 0.0036,
+      "step": 4713
+    },
+    {
+      "epoch": 70.36,
+      "learning_rate": 0.00018733333333333332,
+      "loss": 0.1317,
+      "step": 4714
+    },
+    {
+      "epoch": 70.37,
+      "learning_rate": 0.00018729824561403507,
+      "loss": 0.0028,
+      "step": 4715
+    },
+    {
+      "epoch": 70.39,
+      "learning_rate": 0.00018726315789473682,
+      "loss": 0.02,
+      "step": 4716
+    },
+    {
+      "epoch": 70.4,
+      "learning_rate": 0.0001872280701754386,
+      "loss": 0.0013,
+      "step": 4717
+    },
+    {
+      "epoch": 70.42,
+      "learning_rate": 0.00018719298245614035,
+      "loss": 0.0013,
+      "step": 4718
+    },
+    {
+      "epoch": 70.43,
+      "learning_rate": 0.0001871578947368421,
+      "loss": 0.0015,
+      "step": 4719
+    },
+    {
+      "epoch": 70.45,
+      "learning_rate": 0.00018712280701754382,
+      "loss": 0.0062,
+      "step": 4720
+    },
+    {
+      "epoch": 70.46,
+      "learning_rate": 0.00018708771929824562,
+      "loss": 0.3317,
+      "step": 4721
+    },
+    {
+      "epoch": 70.48,
+      "learning_rate": 0.00018705263157894734,
+      "loss": 0.0073,
+      "step": 4722
+    },
+    {
+      "epoch": 70.49,
+      "learning_rate": 0.0001870175438596491,
+      "loss": 0.0042,
+      "step": 4723
+    },
+    {
+      "epoch": 70.51,
+      "learning_rate": 0.00018698245614035084,
+      "loss": 0.0138,
+      "step": 4724
+    },
+    {
+      "epoch": 70.52,
+      "learning_rate": 0.00018694736842105261,
+      "loss": 0.1278,
+      "step": 4725
+    },
+    {
+      "epoch": 70.54,
+      "learning_rate": 0.00018691228070175436,
+      "loss": 0.0046,
+      "step": 4726
+    },
+    {
+      "epoch": 70.55,
+      "learning_rate": 0.0001868771929824561,
+      "loss": 0.003,
+      "step": 4727
+    },
+    {
+      "epoch": 70.57,
+      "learning_rate": 0.0001868421052631579,
+      "loss": 0.0073,
+      "step": 4728
+    },
+    {
+      "epoch": 70.58,
+      "learning_rate": 0.00018680701754385964,
+      "loss": 0.0081,
+      "step": 4729
+    },
+    {
+      "epoch": 70.59,
+      "learning_rate": 0.00018677192982456139,
+      "loss": 0.0057,
+      "step": 4730
+    },
+    {
+      "epoch": 70.61,
+      "learning_rate": 0.00018673684210526313,
+      "loss": 0.0133,
+      "step": 4731
+    },
+    {
+      "epoch": 70.62,
+      "learning_rate": 0.0001867017543859649,
+      "loss": 0.0017,
+      "step": 4732
+    },
+    {
+      "epoch": 70.64,
+      "learning_rate": 0.00018666666666666666,
+      "loss": 0.0448,
+      "step": 4733
+    },
+    {
+      "epoch": 70.65,
+      "learning_rate": 0.0001866315789473684,
+      "loss": 0.0026,
+      "step": 4734
+    },
+    {
+      "epoch": 70.67,
+      "learning_rate": 0.00018659649122807016,
+      "loss": 0.2009,
+      "step": 4735
+    },
+    {
+      "epoch": 70.68,
+      "learning_rate": 0.00018656140350877193,
+      "loss": 0.0104,
+      "step": 4736
+    },
+    {
+      "epoch": 70.7,
+      "learning_rate": 0.00018652631578947368,
+      "loss": 0.008,
+      "step": 4737
+    },
+    {
+      "epoch": 70.71,
+      "learning_rate": 0.0001864912280701754,
+      "loss": 0.0966,
+      "step": 4738
+    },
+    {
+      "epoch": 70.73,
+      "learning_rate": 0.00018645614035087718,
+      "loss": 0.0021,
+      "step": 4739
+    },
+    {
+      "epoch": 70.74,
+      "learning_rate": 0.00018642105263157893,
+      "loss": 0.0016,
+      "step": 4740
+    },
+    {
+      "epoch": 70.76,
+      "learning_rate": 0.00018638596491228068,
+      "loss": 0.1067,
+      "step": 4741
+    },
+    {
+      "epoch": 70.77,
+      "learning_rate": 0.00018635087719298243,
+      "loss": 0.0963,
+      "step": 4742
+    },
+    {
+      "epoch": 70.79,
+      "learning_rate": 0.0001863157894736842,
+      "loss": 0.0917,
+      "step": 4743
+    },
+    {
+      "epoch": 70.8,
+      "learning_rate": 0.00018628070175438595,
+      "loss": 0.0018,
+      "step": 4744
+    },
+    {
+      "epoch": 70.82,
+      "learning_rate": 0.0001862456140350877,
+      "loss": 0.183,
+      "step": 4745
+    },
+    {
+      "epoch": 70.83,
+      "learning_rate": 0.00018621052631578945,
+      "loss": 0.0012,
+      "step": 4746
+    },
+    {
+      "epoch": 70.85,
+      "learning_rate": 0.00018617543859649122,
+      "loss": 0.0013,
+      "step": 4747
+    },
+    {
+      "epoch": 70.86,
+      "learning_rate": 0.00018614035087719297,
+      "loss": 0.002,
+      "step": 4748
+    },
+    {
+      "epoch": 70.88,
+      "learning_rate": 0.00018610526315789472,
+      "loss": 0.0035,
+      "step": 4749
+    },
+    {
+      "epoch": 70.89,
+      "learning_rate": 0.0001860701754385965,
+      "loss": 0.0024,
+      "step": 4750
+    },
+    {
+      "epoch": 70.91,
+      "learning_rate": 0.00018603508771929825,
+      "loss": 0.0031,
+      "step": 4751
+    },
+    {
+      "epoch": 70.92,
+      "learning_rate": 0.000186,
+      "loss": 0.0021,
+      "step": 4752
+    },
+    {
+      "epoch": 70.94,
+      "learning_rate": 0.00018596491228070172,
+      "loss": 0.0054,
+      "step": 4753
+    },
+    {
+      "epoch": 70.95,
+      "learning_rate": 0.00018592982456140352,
+      "loss": 0.0037,
+      "step": 4754
+    },
+    {
+      "epoch": 70.97,
+      "learning_rate": 0.00018589473684210524,
+      "loss": 0.002,
+      "step": 4755
+    },
+    {
+      "epoch": 70.98,
+      "learning_rate": 0.000185859649122807,
+      "loss": 0.0839,
+      "step": 4756
+    },
+    {
+      "epoch": 71.0,
+      "learning_rate": 0.00018582456140350874,
+      "loss": 0.1267,
+      "step": 4757
+    },
+    {
+      "epoch": 71.01,
+      "learning_rate": 0.00018578947368421051,
+      "loss": 0.0022,
+      "step": 4758
+    },
+    {
+      "epoch": 71.03,
+      "learning_rate": 0.00018575438596491226,
+      "loss": 0.001,
+      "step": 4759
+    },
+    {
+      "epoch": 71.04,
+      "learning_rate": 0.000185719298245614,
+      "loss": 0.0012,
+      "step": 4760
+    },
+    {
+      "epoch": 71.06,
+      "learning_rate": 0.00018568421052631576,
+      "loss": 0.0075,
+      "step": 4761
+    },
+    {
+      "epoch": 71.07,
+      "learning_rate": 0.00018564912280701754,
+      "loss": 0.0519,
+      "step": 4762
+    },
+    {
+      "epoch": 71.09,
+      "learning_rate": 0.00018561403508771929,
+      "loss": 0.0016,
+      "step": 4763
+    },
+    {
+      "epoch": 71.1,
+      "learning_rate": 0.00018557894736842103,
+      "loss": 0.0017,
+      "step": 4764
+    },
+    {
+      "epoch": 71.12,
+      "learning_rate": 0.0001855438596491228,
+      "loss": 0.0289,
+      "step": 4765
+    },
+    {
+      "epoch": 71.13,
+      "learning_rate": 0.00018550877192982456,
+      "loss": 0.0028,
+      "step": 4766
+    },
+    {
+      "epoch": 71.15,
+      "learning_rate": 0.0001854736842105263,
+      "loss": 0.0536,
+      "step": 4767
+    },
+    {
+      "epoch": 71.16,
+      "learning_rate": 0.00018543859649122803,
+      "loss": 0.1368,
+      "step": 4768
+    },
+    {
+      "epoch": 71.18,
+      "learning_rate": 0.00018540350877192983,
+      "loss": 0.0013,
+      "step": 4769
+    },
+    {
+      "epoch": 71.19,
+      "learning_rate": 0.00018536842105263155,
+      "loss": 0.2852,
+      "step": 4770
+    },
+    {
+      "epoch": 71.21,
+      "learning_rate": 0.0001853333333333333,
+      "loss": 0.0144,
+      "step": 4771
+    },
+    {
+      "epoch": 71.22,
+      "learning_rate": 0.00018529824561403505,
+      "loss": 0.0072,
+      "step": 4772
+    },
+    {
+      "epoch": 71.24,
+      "learning_rate": 0.00018526315789473683,
+      "loss": 0.0088,
+      "step": 4773
+    },
+    {
+      "epoch": 71.25,
+      "learning_rate": 0.00018522807017543858,
+      "loss": 0.3189,
+      "step": 4774
+    },
+    {
+      "epoch": 71.27,
+      "learning_rate": 0.00018519298245614032,
+      "loss": 0.1993,
+      "step": 4775
+    },
+    {
+      "epoch": 71.28,
+      "learning_rate": 0.0001851578947368421,
+      "loss": 0.255,
+      "step": 4776
+    },
+    {
+      "epoch": 71.3,
+      "learning_rate": 0.00018512280701754385,
+      "loss": 0.0594,
+      "step": 4777
+    },
+    {
+      "epoch": 71.31,
+      "learning_rate": 0.0001850877192982456,
+      "loss": 0.0034,
+      "step": 4778
+    },
+    {
+      "epoch": 71.33,
+      "learning_rate": 0.00018505263157894735,
+      "loss": 0.014,
+      "step": 4779
+    },
+    {
+      "epoch": 71.34,
+      "learning_rate": 0.00018501754385964912,
+      "loss": 0.0016,
+      "step": 4780
+    },
+    {
+      "epoch": 71.36,
+      "learning_rate": 0.00018498245614035087,
+      "loss": 0.1977,
+      "step": 4781
+    },
+    {
+      "epoch": 71.37,
+      "learning_rate": 0.00018494736842105262,
+      "loss": 0.2658,
+      "step": 4782
+    },
+    {
+      "epoch": 71.39,
+      "learning_rate": 0.00018491228070175437,
+      "loss": 0.005,
+      "step": 4783
+    },
+    {
+      "epoch": 71.4,
+      "learning_rate": 0.00018487719298245615,
+      "loss": 0.0766,
+      "step": 4784
+    },
+    {
+      "epoch": 71.42,
+      "learning_rate": 0.00018484210526315787,
+      "loss": 0.084,
+      "step": 4785
+    },
+    {
+      "epoch": 71.43,
+      "learning_rate": 0.00018480701754385962,
+      "loss": 0.0876,
+      "step": 4786
+    },
+    {
+      "epoch": 71.45,
+      "learning_rate": 0.00018477192982456136,
+      "loss": 0.002,
+      "step": 4787
+    },
+    {
+      "epoch": 71.46,
+      "learning_rate": 0.00018473684210526314,
+      "loss": 0.0182,
+      "step": 4788
+    },
+    {
+      "epoch": 71.48,
+      "learning_rate": 0.0001847017543859649,
+      "loss": 0.0067,
+      "step": 4789
+    },
+    {
+      "epoch": 71.49,
+      "learning_rate": 0.00018466666666666664,
+      "loss": 0.0028,
+      "step": 4790
+    },
+    {
+      "epoch": 71.51,
+      "learning_rate": 0.00018463157894736841,
+      "loss": 0.0689,
+      "step": 4791
+    },
+    {
+      "epoch": 71.52,
+      "learning_rate": 0.00018459649122807016,
+      "loss": 0.3814,
+      "step": 4792
+    },
+    {
+      "epoch": 71.54,
+      "learning_rate": 0.0001845614035087719,
+      "loss": 0.0075,
+      "step": 4793
+    },
+    {
+      "epoch": 71.55,
+      "learning_rate": 0.00018452631578947366,
+      "loss": 0.0132,
+      "step": 4794
+    },
+    {
+      "epoch": 71.57,
+      "learning_rate": 0.00018449122807017544,
+      "loss": 0.1036,
+      "step": 4795
+    },
+    {
+      "epoch": 71.58,
+      "learning_rate": 0.00018445614035087718,
+      "loss": 0.0515,
+      "step": 4796
+    },
+    {
+      "epoch": 71.59,
+      "learning_rate": 0.00018442105263157893,
+      "loss": 0.0664,
+      "step": 4797
+    },
+    {
+      "epoch": 71.61,
+      "learning_rate": 0.00018438596491228068,
+      "loss": 0.0349,
+      "step": 4798
+    },
+    {
+      "epoch": 71.62,
+      "learning_rate": 0.00018435087719298246,
+      "loss": 0.1953,
+      "step": 4799
+    },
+    {
+      "epoch": 71.64,
+      "learning_rate": 0.0001843157894736842,
+      "loss": 0.0302,
+      "step": 4800
+    },
+    {
+      "epoch": 71.64,
+      "eval_accuracy": 0.8494860499265786,
+      "eval_f1": 0.8515287106120767,
+      "eval_loss": 0.6812880635261536,
+      "eval_runtime": 344.4321,
+      "eval_samples_per_second": 11.863,
+      "eval_steps_per_second": 0.743,
+      "step": 4800
+    },
+    {
+      "epoch": 71.65,
+      "learning_rate": 0.00018428070175438593,
+      "loss": 0.0782,
+      "step": 4801
+    },
+    {
+      "epoch": 71.67,
+      "learning_rate": 0.00018424561403508773,
+      "loss": 0.0039,
+      "step": 4802
+    },
+    {
+      "epoch": 71.68,
+      "learning_rate": 0.00018421052631578945,
+      "loss": 0.0041,
+      "step": 4803
+    },
+    {
+      "epoch": 71.7,
+      "learning_rate": 0.0001841754385964912,
+      "loss": 0.0041,
+      "step": 4804
+    },
+    {
+      "epoch": 71.71,
+      "learning_rate": 0.00018414035087719295,
+      "loss": 0.0499,
+      "step": 4805
+    },
+    {
+      "epoch": 71.73,
+      "learning_rate": 0.00018410526315789473,
+      "loss": 0.0392,
+      "step": 4806
+    },
+    {
+      "epoch": 71.74,
+      "learning_rate": 0.00018407017543859648,
+      "loss": 0.0955,
+      "step": 4807
+    },
+    {
+      "epoch": 71.76,
+      "learning_rate": 0.00018403508771929822,
+      "loss": 0.2134,
+      "step": 4808
+    },
+    {
+      "epoch": 71.77,
+      "learning_rate": 0.00018399999999999997,
+      "loss": 0.2199,
+      "step": 4809
+    },
+    {
+      "epoch": 71.79,
+      "learning_rate": 0.00018396491228070175,
+      "loss": 0.0032,
+      "step": 4810
+    },
+    {
+      "epoch": 71.8,
+      "learning_rate": 0.0001839298245614035,
+      "loss": 0.0542,
+      "step": 4811
+    },
+    {
+      "epoch": 71.82,
+      "learning_rate": 0.00018389473684210525,
+      "loss": 0.1352,
+      "step": 4812
+    },
+    {
+      "epoch": 71.83,
+      "learning_rate": 0.00018385964912280702,
+      "loss": 0.1094,
+      "step": 4813
+    },
+    {
+      "epoch": 71.85,
+      "learning_rate": 0.00018382456140350877,
+      "loss": 0.0961,
+      "step": 4814
+    },
+    {
+      "epoch": 71.86,
+      "learning_rate": 0.00018378947368421052,
+      "loss": 0.0033,
+      "step": 4815
+    },
+    {
+      "epoch": 71.88,
+      "learning_rate": 0.00018375438596491224,
+      "loss": 0.3464,
+      "step": 4816
+    },
+    {
+      "epoch": 71.89,
+      "learning_rate": 0.00018371929824561404,
+      "loss": 0.0021,
+      "step": 4817
+    },
+    {
+      "epoch": 71.91,
+      "learning_rate": 0.00018368421052631577,
+      "loss": 0.0212,
+      "step": 4818
+    },
+    {
+      "epoch": 71.92,
+      "learning_rate": 0.00018364912280701752,
+      "loss": 0.09,
+      "step": 4819
+    },
+    {
+      "epoch": 71.94,
+      "learning_rate": 0.00018361403508771926,
+      "loss": 0.1911,
+      "step": 4820
+    },
+    {
+      "epoch": 71.95,
+      "learning_rate": 0.00018357894736842104,
+      "loss": 0.0022,
+      "step": 4821
+    },
+    {
+      "epoch": 71.97,
+      "learning_rate": 0.0001835438596491228,
+      "loss": 0.0026,
+      "step": 4822
+    },
+    {
+      "epoch": 71.98,
+      "learning_rate": 0.00018350877192982454,
+      "loss": 0.003,
+      "step": 4823
+    },
+    {
+      "epoch": 72.0,
+      "learning_rate": 0.00018347368421052629,
+      "loss": 0.0518,
+      "step": 4824
+    },
+    {
+      "epoch": 72.01,
+      "learning_rate": 0.00018343859649122806,
+      "loss": 0.1218,
+      "step": 4825
+    },
+    {
+      "epoch": 72.03,
+      "learning_rate": 0.0001834035087719298,
+      "loss": 0.2565,
+      "step": 4826
+    },
+    {
+      "epoch": 72.04,
+      "learning_rate": 0.00018336842105263156,
+      "loss": 0.0015,
+      "step": 4827
+    },
+    {
+      "epoch": 72.06,
+      "learning_rate": 0.00018333333333333334,
+      "loss": 0.1193,
+      "step": 4828
+    },
+    {
+      "epoch": 72.07,
+      "learning_rate": 0.00018329824561403508,
+      "loss": 0.0566,
+      "step": 4829
+    },
+    {
+      "epoch": 72.09,
+      "learning_rate": 0.00018326315789473683,
+      "loss": 0.0022,
+      "step": 4830
+    },
+    {
+      "epoch": 72.1,
+      "learning_rate": 0.00018322807017543858,
+      "loss": 0.0022,
+      "step": 4831
+    },
+    {
+      "epoch": 72.12,
+      "learning_rate": 0.00018319298245614036,
+      "loss": 0.1068,
+      "step": 4832
+    },
+    {
+      "epoch": 72.13,
+      "learning_rate": 0.00018315789473684208,
+      "loss": 0.062,
+      "step": 4833
+    },
+    {
+      "epoch": 72.15,
+      "learning_rate": 0.00018312280701754383,
+      "loss": 0.0677,
+      "step": 4834
+    },
+    {
+      "epoch": 72.16,
+      "learning_rate": 0.00018308771929824558,
+      "loss": 0.0032,
+      "step": 4835
+    },
+    {
+      "epoch": 72.18,
+      "learning_rate": 0.00018305263157894735,
+      "loss": 0.0027,
+      "step": 4836
+    },
+    {
+      "epoch": 72.19,
+      "learning_rate": 0.0001830175438596491,
+      "loss": 0.0203,
+      "step": 4837
+    },
+    {
+      "epoch": 72.21,
+      "learning_rate": 0.00018298245614035085,
+      "loss": 0.0049,
+      "step": 4838
+    },
+    {
+      "epoch": 72.22,
+      "learning_rate": 0.00018294736842105263,
+      "loss": 0.0329,
+      "step": 4839
+    },
+    {
+      "epoch": 72.24,
+      "learning_rate": 0.00018291228070175437,
+      "loss": 0.0059,
+      "step": 4840
+    },
+    {
+      "epoch": 72.25,
+      "learning_rate": 0.00018287719298245612,
+      "loss": 0.0042,
+      "step": 4841
+    },
+    {
+      "epoch": 72.27,
+      "learning_rate": 0.00018284210526315787,
+      "loss": 0.1367,
+      "step": 4842
+    },
+    {
+      "epoch": 72.28,
+      "learning_rate": 0.00018280701754385965,
+      "loss": 0.0015,
+      "step": 4843
+    },
+    {
+      "epoch": 72.3,
+      "learning_rate": 0.0001827719298245614,
+      "loss": 0.1505,
+      "step": 4844
+    },
+    {
+      "epoch": 72.31,
+      "learning_rate": 0.00018273684210526315,
+      "loss": 0.0677,
+      "step": 4845
+    },
+    {
+      "epoch": 72.33,
+      "learning_rate": 0.0001827017543859649,
+      "loss": 0.0066,
+      "step": 4846
+    },
+    {
+      "epoch": 72.34,
+      "learning_rate": 0.00018266666666666667,
+      "loss": 0.0131,
+      "step": 4847
+    },
+    {
+      "epoch": 72.36,
+      "learning_rate": 0.00018263157894736842,
+      "loss": 0.095,
+      "step": 4848
+    },
+    {
+      "epoch": 72.37,
+      "learning_rate": 0.00018259649122807014,
+      "loss": 0.0308,
+      "step": 4849
+    },
+    {
+      "epoch": 72.39,
+      "learning_rate": 0.0001825614035087719,
+      "loss": 0.0013,
+      "step": 4850
+    },
+    {
+      "epoch": 72.4,
+      "learning_rate": 0.00018252631578947367,
+      "loss": 0.1971,
+      "step": 4851
+    },
+    {
+      "epoch": 72.42,
+      "learning_rate": 0.00018249122807017541,
+      "loss": 0.0014,
+      "step": 4852
+    },
+    {
+      "epoch": 72.43,
+      "learning_rate": 0.00018245614035087716,
+      "loss": 0.0013,
+      "step": 4853
+    },
+    {
+      "epoch": 72.45,
+      "learning_rate": 0.00018242105263157894,
+      "loss": 0.0355,
+      "step": 4854
+    },
+    {
+      "epoch": 72.46,
+      "learning_rate": 0.0001823859649122807,
+      "loss": 0.0038,
+      "step": 4855
+    },
+    {
+      "epoch": 72.48,
+      "learning_rate": 0.00018235087719298244,
+      "loss": 0.0012,
+      "step": 4856
+    },
+    {
+      "epoch": 72.49,
+      "learning_rate": 0.00018231578947368419,
+      "loss": 0.102,
+      "step": 4857
+    },
+    {
+      "epoch": 72.51,
+      "learning_rate": 0.00018228070175438596,
+      "loss": 0.0082,
+      "step": 4858
+    },
+    {
+      "epoch": 72.52,
+      "learning_rate": 0.0001822456140350877,
+      "loss": 0.0481,
+      "step": 4859
+    },
+    {
+      "epoch": 72.54,
+      "learning_rate": 0.00018221052631578946,
+      "loss": 0.0025,
+      "step": 4860
+    },
+    {
+      "epoch": 72.55,
+      "learning_rate": 0.0001821754385964912,
+      "loss": 0.0018,
+      "step": 4861
+    },
+    {
+      "epoch": 72.57,
+      "learning_rate": 0.00018214035087719298,
+      "loss": 0.1083,
+      "step": 4862
+    },
+    {
+      "epoch": 72.58,
+      "learning_rate": 0.00018210526315789473,
+      "loss": 0.0437,
+      "step": 4863
+    },
+    {
+      "epoch": 72.59,
+      "learning_rate": 0.00018207017543859645,
+      "loss": 0.0088,
+      "step": 4864
+    },
+    {
+      "epoch": 72.61,
+      "learning_rate": 0.00018203508771929826,
+      "loss": 0.1357,
+      "step": 4865
+    },
+    {
+      "epoch": 72.62,
+      "learning_rate": 0.00018199999999999998,
+      "loss": 0.0158,
+      "step": 4866
+    },
+    {
+      "epoch": 72.64,
+      "learning_rate": 0.00018196491228070173,
+      "loss": 0.0188,
+      "step": 4867
+    },
+    {
+      "epoch": 72.65,
+      "learning_rate": 0.00018192982456140348,
+      "loss": 0.0024,
+      "step": 4868
+    },
+    {
+      "epoch": 72.67,
+      "learning_rate": 0.00018189473684210525,
+      "loss": 0.005,
+      "step": 4869
+    },
+    {
+      "epoch": 72.68,
+      "learning_rate": 0.000181859649122807,
+      "loss": 0.0025,
+      "step": 4870
+    },
+    {
+      "epoch": 72.7,
+      "learning_rate": 0.00018182456140350875,
+      "loss": 0.0077,
+      "step": 4871
+    },
+    {
+      "epoch": 72.71,
+      "learning_rate": 0.0001817894736842105,
+      "loss": 0.0025,
+      "step": 4872
+    },
+    {
+      "epoch": 72.73,
+      "learning_rate": 0.00018175438596491227,
+      "loss": 0.058,
+      "step": 4873
+    },
+    {
+      "epoch": 72.74,
+      "learning_rate": 0.00018171929824561402,
+      "loss": 0.1914,
+      "step": 4874
+    },
+    {
+      "epoch": 72.76,
+      "learning_rate": 0.00018168421052631577,
+      "loss": 0.1722,
+      "step": 4875
+    },
+    {
+      "epoch": 72.77,
+      "learning_rate": 0.00018164912280701755,
+      "loss": 0.0266,
+      "step": 4876
+    },
+    {
+      "epoch": 72.79,
+      "learning_rate": 0.0001816140350877193,
+      "loss": 0.0278,
+      "step": 4877
+    },
+    {
+      "epoch": 72.8,
+      "learning_rate": 0.00018157894736842105,
+      "loss": 0.0495,
+      "step": 4878
+    },
+    {
+      "epoch": 72.82,
+      "learning_rate": 0.00018154385964912277,
+      "loss": 0.1768,
+      "step": 4879
+    },
+    {
+      "epoch": 72.83,
+      "learning_rate": 0.00018150877192982457,
+      "loss": 0.0619,
+      "step": 4880
+    },
+    {
+      "epoch": 72.85,
+      "learning_rate": 0.0001814736842105263,
+      "loss": 0.1561,
+      "step": 4881
+    },
+    {
+      "epoch": 72.86,
+      "learning_rate": 0.00018143859649122804,
+      "loss": 0.0668,
+      "step": 4882
+    },
+    {
+      "epoch": 72.88,
+      "learning_rate": 0.0001814035087719298,
+      "loss": 0.0052,
+      "step": 4883
+    },
+    {
+      "epoch": 72.89,
+      "learning_rate": 0.00018136842105263157,
+      "loss": 0.0246,
+      "step": 4884
+    },
+    {
+      "epoch": 72.91,
+      "learning_rate": 0.00018133333333333331,
+      "loss": 0.2206,
+      "step": 4885
+    },
+    {
+      "epoch": 72.92,
+      "learning_rate": 0.00018129824561403506,
+      "loss": 0.2973,
+      "step": 4886
+    },
+    {
+      "epoch": 72.94,
+      "learning_rate": 0.0001812631578947368,
+      "loss": 0.3662,
+      "step": 4887
+    },
+    {
+      "epoch": 72.95,
+      "learning_rate": 0.0001812280701754386,
+      "loss": 0.2203,
+      "step": 4888
+    },
+    {
+      "epoch": 72.97,
+      "learning_rate": 0.00018119298245614034,
+      "loss": 0.0179,
+      "step": 4889
+    },
+    {
+      "epoch": 72.98,
+      "learning_rate": 0.00018115789473684208,
+      "loss": 0.0022,
+      "step": 4890
+    },
+    {
+      "epoch": 73.0,
+      "learning_rate": 0.00018112280701754386,
+      "loss": 0.0622,
+      "step": 4891
+    },
+    {
+      "epoch": 73.01,
+      "learning_rate": 0.0001810877192982456,
+      "loss": 0.0017,
+      "step": 4892
+    },
+    {
+      "epoch": 73.03,
+      "learning_rate": 0.00018105263157894736,
+      "loss": 0.0018,
+      "step": 4893
+    },
+    {
+      "epoch": 73.04,
+      "learning_rate": 0.0001810175438596491,
+      "loss": 0.0528,
+      "step": 4894
+    },
+    {
+      "epoch": 73.06,
+      "learning_rate": 0.00018098245614035088,
+      "loss": 0.0023,
+      "step": 4895
+    },
+    {
+      "epoch": 73.07,
+      "learning_rate": 0.00018094736842105263,
+      "loss": 0.0043,
+      "step": 4896
+    },
+    {
+      "epoch": 73.09,
+      "learning_rate": 0.00018091228070175435,
+      "loss": 0.0218,
+      "step": 4897
+    },
+    {
+      "epoch": 73.1,
+      "learning_rate": 0.0001808771929824561,
+      "loss": 0.2016,
+      "step": 4898
+    },
+    {
+      "epoch": 73.12,
+      "learning_rate": 0.00018084210526315788,
+      "loss": 0.1458,
+      "step": 4899
+    },
+    {
+      "epoch": 73.13,
+      "learning_rate": 0.00018080701754385963,
+      "loss": 0.0088,
+      "step": 4900
+    },
+    {
+      "epoch": 73.15,
+      "learning_rate": 0.00018077192982456138,
+      "loss": 0.0022,
+      "step": 4901
+    },
+    {
+      "epoch": 73.16,
+      "learning_rate": 0.00018073684210526315,
+      "loss": 0.0228,
+      "step": 4902
+    },
+    {
+      "epoch": 73.18,
+      "learning_rate": 0.0001807017543859649,
+      "loss": 0.0058,
+      "step": 4903
+    },
+    {
+      "epoch": 73.19,
+      "learning_rate": 0.00018066666666666665,
+      "loss": 0.0074,
+      "step": 4904
+    },
+    {
+      "epoch": 73.21,
+      "learning_rate": 0.0001806315789473684,
+      "loss": 0.0103,
+      "step": 4905
+    },
+    {
+      "epoch": 73.22,
+      "learning_rate": 0.00018059649122807017,
+      "loss": 0.1562,
+      "step": 4906
+    },
+    {
+      "epoch": 73.24,
+      "learning_rate": 0.00018056140350877192,
+      "loss": 0.004,
+      "step": 4907
+    },
+    {
+      "epoch": 73.25,
+      "learning_rate": 0.00018052631578947367,
+      "loss": 0.0222,
+      "step": 4908
+    },
+    {
+      "epoch": 73.27,
+      "learning_rate": 0.00018049122807017542,
+      "loss": 0.0036,
+      "step": 4909
+    },
+    {
+      "epoch": 73.28,
+      "learning_rate": 0.0001804561403508772,
+      "loss": 0.0019,
+      "step": 4910
+    },
+    {
+      "epoch": 73.3,
+      "learning_rate": 0.00018042105263157894,
+      "loss": 0.0021,
+      "step": 4911
+    },
+    {
+      "epoch": 73.31,
+      "learning_rate": 0.00018038596491228067,
+      "loss": 0.2216,
+      "step": 4912
+    },
+    {
+      "epoch": 73.33,
+      "learning_rate": 0.00018035087719298247,
+      "loss": 0.1907,
+      "step": 4913
+    },
+    {
+      "epoch": 73.34,
+      "learning_rate": 0.0001803157894736842,
+      "loss": 0.0057,
+      "step": 4914
+    },
+    {
+      "epoch": 73.36,
+      "learning_rate": 0.00018028070175438594,
+      "loss": 0.0024,
+      "step": 4915
+    },
+    {
+      "epoch": 73.37,
+      "learning_rate": 0.0001802456140350877,
+      "loss": 0.2023,
+      "step": 4916
+    },
+    {
+      "epoch": 73.39,
+      "learning_rate": 0.00018021052631578946,
+      "loss": 0.0057,
+      "step": 4917
+    },
+    {
+      "epoch": 73.4,
+      "learning_rate": 0.0001801754385964912,
+      "loss": 0.0878,
+      "step": 4918
+    },
+    {
+      "epoch": 73.42,
+      "learning_rate": 0.00018014035087719296,
+      "loss": 0.0024,
+      "step": 4919
+    },
+    {
+      "epoch": 73.43,
+      "learning_rate": 0.0001801052631578947,
+      "loss": 0.0191,
+      "step": 4920
+    },
+    {
+      "epoch": 73.45,
+      "learning_rate": 0.0001800701754385965,
+      "loss": 0.0756,
+      "step": 4921
+    },
+    {
+      "epoch": 73.46,
+      "learning_rate": 0.00018003508771929824,
+      "loss": 0.0028,
+      "step": 4922
+    },
+    {
+      "epoch": 73.48,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 0.2512,
+      "step": 4923
+    },
+    {
+      "epoch": 73.49,
+      "learning_rate": 0.00017996491228070173,
+      "loss": 0.1202,
+      "step": 4924
+    },
+    {
+      "epoch": 73.51,
+      "learning_rate": 0.0001799298245614035,
+      "loss": 0.1339,
+      "step": 4925
+    },
+    {
+      "epoch": 73.52,
+      "learning_rate": 0.00017989473684210526,
+      "loss": 0.0635,
+      "step": 4926
+    },
+    {
+      "epoch": 73.54,
+      "learning_rate": 0.00017985964912280698,
+      "loss": 0.0029,
+      "step": 4927
+    },
+    {
+      "epoch": 73.55,
+      "learning_rate": 0.00017982456140350878,
+      "loss": 0.0032,
+      "step": 4928
+    },
+    {
+      "epoch": 73.57,
+      "learning_rate": 0.0001797894736842105,
+      "loss": 0.0057,
+      "step": 4929
+    },
+    {
+      "epoch": 73.58,
+      "learning_rate": 0.00017975438596491225,
+      "loss": 0.267,
+      "step": 4930
+    },
+    {
+      "epoch": 73.59,
+      "learning_rate": 0.000179719298245614,
+      "loss": 0.1825,
+      "step": 4931
+    },
+    {
+      "epoch": 73.61,
+      "learning_rate": 0.00017968421052631578,
+      "loss": 0.1034,
+      "step": 4932
+    },
+    {
+      "epoch": 73.62,
+      "learning_rate": 0.00017964912280701753,
+      "loss": 0.0088,
+      "step": 4933
+    },
+    {
+      "epoch": 73.64,
+      "learning_rate": 0.00017961403508771927,
+      "loss": 0.27,
+      "step": 4934
+    },
+    {
+      "epoch": 73.65,
+      "learning_rate": 0.00017957894736842102,
+      "loss": 0.0016,
+      "step": 4935
+    },
+    {
+      "epoch": 73.67,
+      "learning_rate": 0.0001795438596491228,
+      "loss": 0.0059,
+      "step": 4936
+    },
+    {
+      "epoch": 73.68,
+      "learning_rate": 0.00017950877192982455,
+      "loss": 0.3191,
+      "step": 4937
+    },
+    {
+      "epoch": 73.7,
+      "learning_rate": 0.0001794736842105263,
+      "loss": 0.0091,
+      "step": 4938
+    },
+    {
+      "epoch": 73.71,
+      "learning_rate": 0.00017943859649122807,
+      "loss": 0.037,
+      "step": 4939
+    },
+    {
+      "epoch": 73.73,
+      "learning_rate": 0.00017940350877192982,
+      "loss": 0.0173,
+      "step": 4940
+    },
+    {
+      "epoch": 73.74,
+      "learning_rate": 0.00017936842105263157,
+      "loss": 0.0028,
+      "step": 4941
+    },
+    {
+      "epoch": 73.76,
+      "learning_rate": 0.00017933333333333332,
+      "loss": 0.0125,
+      "step": 4942
+    },
+    {
+      "epoch": 73.77,
+      "learning_rate": 0.0001792982456140351,
+      "loss": 0.0519,
+      "step": 4943
+    },
+    {
+      "epoch": 73.79,
+      "learning_rate": 0.00017926315789473684,
+      "loss": 0.3839,
+      "step": 4944
+    },
+    {
+      "epoch": 73.8,
+      "learning_rate": 0.00017922807017543857,
+      "loss": 0.0054,
+      "step": 4945
+    },
+    {
+      "epoch": 73.82,
+      "learning_rate": 0.00017919298245614031,
+      "loss": 0.036,
+      "step": 4946
+    },
+    {
+      "epoch": 73.83,
+      "learning_rate": 0.0001791578947368421,
+      "loss": 0.0084,
+      "step": 4947
+    },
+    {
+      "epoch": 73.85,
+      "learning_rate": 0.00017912280701754384,
+      "loss": 0.0497,
+      "step": 4948
+    },
+    {
+      "epoch": 73.86,
+      "learning_rate": 0.0001790877192982456,
+      "loss": 0.0131,
+      "step": 4949
+    },
+    {
+      "epoch": 73.88,
+      "learning_rate": 0.00017905263157894734,
+      "loss": 0.2431,
+      "step": 4950
+    },
+    {
+      "epoch": 73.89,
+      "learning_rate": 0.0001790175438596491,
+      "loss": 0.215,
+      "step": 4951
+    },
+    {
+      "epoch": 73.91,
+      "learning_rate": 0.00017898245614035086,
+      "loss": 0.011,
+      "step": 4952
+    },
+    {
+      "epoch": 73.92,
+      "learning_rate": 0.0001789473684210526,
+      "loss": 0.0145,
+      "step": 4953
+    },
+    {
+      "epoch": 73.94,
+      "learning_rate": 0.00017891228070175439,
+      "loss": 0.0019,
+      "step": 4954
+    },
+    {
+      "epoch": 73.95,
+      "learning_rate": 0.00017887719298245613,
+      "loss": 0.0811,
+      "step": 4955
+    },
+    {
+      "epoch": 73.97,
+      "learning_rate": 0.00017884210526315788,
+      "loss": 0.008,
+      "step": 4956
+    },
+    {
+      "epoch": 73.98,
+      "learning_rate": 0.00017880701754385963,
+      "loss": 0.2649,
+      "step": 4957
+    },
+    {
+      "epoch": 74.0,
+      "learning_rate": 0.0001787719298245614,
+      "loss": 0.1351,
+      "step": 4958
+    },
+    {
+      "epoch": 74.01,
+      "learning_rate": 0.00017873684210526316,
+      "loss": 0.0034,
+      "step": 4959
+    },
+    {
+      "epoch": 74.03,
+      "learning_rate": 0.00017870175438596488,
+      "loss": 0.012,
+      "step": 4960
+    },
+    {
+      "epoch": 74.04,
+      "learning_rate": 0.00017866666666666663,
+      "loss": 0.0023,
+      "step": 4961
+    },
+    {
+      "epoch": 74.06,
+      "learning_rate": 0.0001786315789473684,
+      "loss": 0.1564,
+      "step": 4962
+    },
+    {
+      "epoch": 74.07,
+      "learning_rate": 0.00017859649122807015,
+      "loss": 0.1394,
+      "step": 4963
+    },
+    {
+      "epoch": 74.09,
+      "learning_rate": 0.0001785614035087719,
+      "loss": 0.0147,
+      "step": 4964
+    },
+    {
+      "epoch": 74.1,
+      "learning_rate": 0.00017852631578947368,
+      "loss": 0.0139,
+      "step": 4965
+    },
+    {
+      "epoch": 74.12,
+      "learning_rate": 0.00017849122807017543,
+      "loss": 0.028,
+      "step": 4966
+    },
+    {
+      "epoch": 74.13,
+      "learning_rate": 0.00017845614035087717,
+      "loss": 0.0709,
+      "step": 4967
+    },
+    {
+      "epoch": 74.15,
+      "learning_rate": 0.00017842105263157892,
+      "loss": 0.0028,
+      "step": 4968
+    },
+    {
+      "epoch": 74.16,
+      "learning_rate": 0.0001783859649122807,
+      "loss": 0.0064,
+      "step": 4969
+    },
+    {
+      "epoch": 74.18,
+      "learning_rate": 0.00017835087719298245,
+      "loss": 0.005,
+      "step": 4970
+    },
+    {
+      "epoch": 74.19,
+      "learning_rate": 0.0001783157894736842,
+      "loss": 0.0279,
+      "step": 4971
+    },
+    {
+      "epoch": 74.21,
+      "learning_rate": 0.00017828070175438595,
+      "loss": 0.0143,
+      "step": 4972
+    },
+    {
+      "epoch": 74.22,
+      "learning_rate": 0.00017824561403508772,
+      "loss": 0.0258,
+      "step": 4973
+    },
+    {
+      "epoch": 74.24,
+      "learning_rate": 0.00017821052631578947,
+      "loss": 0.0602,
+      "step": 4974
+    },
+    {
+      "epoch": 74.25,
+      "learning_rate": 0.0001781754385964912,
+      "loss": 0.1954,
+      "step": 4975
+    },
+    {
+      "epoch": 74.27,
+      "learning_rate": 0.000178140350877193,
+      "loss": 0.0018,
+      "step": 4976
+    },
+    {
+      "epoch": 74.28,
+      "learning_rate": 0.00017810526315789472,
+      "loss": 0.0082,
+      "step": 4977
+    },
+    {
+      "epoch": 74.3,
+      "learning_rate": 0.00017807017543859647,
+      "loss": 0.1672,
+      "step": 4978
+    },
+    {
+      "epoch": 74.31,
+      "learning_rate": 0.00017803508771929821,
+      "loss": 0.0016,
+      "step": 4979
+    },
+    {
+      "epoch": 74.33,
+      "learning_rate": 0.000178,
+      "loss": 0.0091,
+      "step": 4980
+    },
+    {
+      "epoch": 74.34,
+      "learning_rate": 0.00017796491228070174,
+      "loss": 0.0312,
+      "step": 4981
+    },
+    {
+      "epoch": 74.36,
+      "learning_rate": 0.0001779298245614035,
+      "loss": 0.2499,
+      "step": 4982
+    },
+    {
+      "epoch": 74.37,
+      "learning_rate": 0.00017789473684210524,
+      "loss": 0.1338,
+      "step": 4983
+    },
+    {
+      "epoch": 74.39,
+      "learning_rate": 0.000177859649122807,
+      "loss": 0.011,
+      "step": 4984
+    },
+    {
+      "epoch": 74.4,
+      "learning_rate": 0.00017782456140350876,
+      "loss": 0.0024,
+      "step": 4985
+    },
+    {
+      "epoch": 74.42,
+      "learning_rate": 0.0001777894736842105,
+      "loss": 0.0041,
+      "step": 4986
+    },
+    {
+      "epoch": 74.43,
+      "learning_rate": 0.00017775438596491226,
+      "loss": 0.0016,
+      "step": 4987
+    },
+    {
+      "epoch": 74.45,
+      "learning_rate": 0.00017771929824561403,
+      "loss": 0.0135,
+      "step": 4988
+    },
+    {
+      "epoch": 74.46,
+      "learning_rate": 0.00017768421052631578,
+      "loss": 0.0148,
+      "step": 4989
+    },
+    {
+      "epoch": 74.48,
+      "learning_rate": 0.00017764912280701753,
+      "loss": 0.0708,
+      "step": 4990
+    },
+    {
+      "epoch": 74.49,
+      "learning_rate": 0.0001776140350877193,
+      "loss": 0.083,
+      "step": 4991
+    },
+    {
+      "epoch": 74.51,
+      "learning_rate": 0.00017757894736842103,
+      "loss": 0.0161,
+      "step": 4992
+    },
+    {
+      "epoch": 74.52,
+      "learning_rate": 0.00017754385964912278,
+      "loss": 0.0177,
+      "step": 4993
+    },
+    {
+      "epoch": 74.54,
+      "learning_rate": 0.00017750877192982453,
+      "loss": 0.0022,
+      "step": 4994
+    },
+    {
+      "epoch": 74.55,
+      "learning_rate": 0.0001774736842105263,
+      "loss": 0.0189,
+      "step": 4995
+    },
+    {
+      "epoch": 74.57,
+      "learning_rate": 0.00017743859649122805,
+      "loss": 0.0014,
+      "step": 4996
+    },
+    {
+      "epoch": 74.58,
+      "learning_rate": 0.0001774035087719298,
+      "loss": 0.0093,
+      "step": 4997
+    },
+    {
+      "epoch": 74.59,
+      "learning_rate": 0.00017736842105263155,
+      "loss": 0.0283,
+      "step": 4998
+    },
+    {
+      "epoch": 74.61,
+      "learning_rate": 0.00017733333333333333,
+      "loss": 0.139,
+      "step": 4999
+    },
+    {
+      "epoch": 74.62,
+      "learning_rate": 0.00017729824561403507,
+      "loss": 0.0027,
+      "step": 5000
+    },
+    {
+      "epoch": 74.62,
+      "eval_accuracy": 0.8529123837493882,
+      "eval_f1": 0.8530140782623041,
+      "eval_loss": 0.7162572145462036,
+      "eval_runtime": 343.7711,
+      "eval_samples_per_second": 11.886,
+      "eval_steps_per_second": 0.745,
+      "step": 5000
+    },
+    {
+      "epoch": 74.64,
+      "learning_rate": 0.00017726315789473682,
+      "loss": 0.0074,
+      "step": 5001
+    },
+    {
+      "epoch": 74.65,
+      "learning_rate": 0.0001772280701754386,
+      "loss": 0.0027,
+      "step": 5002
+    },
+    {
+      "epoch": 74.67,
+      "learning_rate": 0.00017719298245614035,
+      "loss": 0.0679,
+      "step": 5003
+    },
+    {
+      "epoch": 74.68,
+      "learning_rate": 0.0001771578947368421,
+      "loss": 0.0013,
+      "step": 5004
+    },
+    {
+      "epoch": 74.7,
+      "learning_rate": 0.00017712280701754384,
+      "loss": 0.0307,
+      "step": 5005
+    },
+    {
+      "epoch": 74.71,
+      "learning_rate": 0.00017708771929824562,
+      "loss": 0.0014,
+      "step": 5006
+    },
+    {
+      "epoch": 74.73,
+      "learning_rate": 0.00017705263157894737,
+      "loss": 0.0016,
+      "step": 5007
+    },
+    {
+      "epoch": 74.74,
+      "learning_rate": 0.00017705263157894737,
+      "loss": 0.3708,
+      "step": 5008
+    },
+    {
+      "epoch": 74.76,
+      "learning_rate": 0.0001770175438596491,
+      "loss": 0.0718,
+      "step": 5009
+    },
+    {
+      "epoch": 74.77,
+      "learning_rate": 0.00017698245614035084,
+      "loss": 0.0015,
+      "step": 5010
+    },
+    {
+      "epoch": 74.79,
+      "learning_rate": 0.00017694736842105262,
+      "loss": 0.0336,
+      "step": 5011
+    },
+    {
+      "epoch": 74.8,
+      "learning_rate": 0.00017691228070175436,
+      "loss": 0.0311,
+      "step": 5012
+    },
+    {
+      "epoch": 74.82,
+      "learning_rate": 0.0001768771929824561,
+      "loss": 0.0123,
+      "step": 5013
+    },
+    {
+      "epoch": 74.83,
+      "learning_rate": 0.00017684210526315786,
+      "loss": 0.0049,
+      "step": 5014
+    },
+    {
+      "epoch": 74.85,
+      "learning_rate": 0.00017680701754385964,
+      "loss": 0.0137,
+      "step": 5015
+    },
+    {
+      "epoch": 74.86,
+      "learning_rate": 0.0001767719298245614,
+      "loss": 0.0619,
+      "step": 5016
+    },
+    {
+      "epoch": 74.88,
+      "learning_rate": 0.00017673684210526314,
+      "loss": 0.1754,
+      "step": 5017
+    },
+    {
+      "epoch": 74.89,
+      "learning_rate": 0.0001767017543859649,
+      "loss": 0.0045,
+      "step": 5018
+    },
+    {
+      "epoch": 74.91,
+      "learning_rate": 0.00017666666666666666,
+      "loss": 0.0024,
+      "step": 5019
+    },
+    {
+      "epoch": 74.92,
+      "learning_rate": 0.0001766315789473684,
+      "loss": 0.0026,
+      "step": 5020
+    },
+    {
+      "epoch": 74.94,
+      "learning_rate": 0.00017659649122807016,
+      "loss": 0.1055,
+      "step": 5021
+    },
+    {
+      "epoch": 74.95,
+      "learning_rate": 0.00017656140350877193,
+      "loss": 0.004,
+      "step": 5022
+    },
+    {
+      "epoch": 74.97,
+      "learning_rate": 0.00017652631578947368,
+      "loss": 0.0023,
+      "step": 5023
+    },
+    {
+      "epoch": 74.98,
+      "learning_rate": 0.0001764912280701754,
+      "loss": 0.0035,
+      "step": 5024
+    },
+    {
+      "epoch": 75.0,
+      "learning_rate": 0.00017645614035087715,
+      "loss": 0.0064,
+      "step": 5025
+    },
+    {
+      "epoch": 75.01,
+      "learning_rate": 0.00017642105263157893,
+      "loss": 0.0107,
+      "step": 5026
+    },
+    {
+      "epoch": 75.03,
+      "learning_rate": 0.00017638596491228068,
+      "loss": 0.0037,
+      "step": 5027
+    },
+    {
+      "epoch": 75.04,
+      "learning_rate": 0.00017635087719298243,
+      "loss": 0.0015,
+      "step": 5028
+    },
+    {
+      "epoch": 75.06,
+      "learning_rate": 0.0001763157894736842,
+      "loss": 0.0016,
+      "step": 5029
+    },
+    {
+      "epoch": 75.07,
+      "learning_rate": 0.00017628070175438595,
+      "loss": 0.1804,
+      "step": 5030
+    },
+    {
+      "epoch": 75.09,
+      "learning_rate": 0.0001762456140350877,
+      "loss": 0.0013,
+      "step": 5031
+    },
+    {
+      "epoch": 75.1,
+      "learning_rate": 0.00017621052631578945,
+      "loss": 0.0021,
+      "step": 5032
+    },
+    {
+      "epoch": 75.12,
+      "learning_rate": 0.00017617543859649122,
+      "loss": 0.0022,
+      "step": 5033
+    },
+    {
+      "epoch": 75.13,
+      "learning_rate": 0.00017614035087719297,
+      "loss": 0.0011,
+      "step": 5034
+    },
+    {
+      "epoch": 75.15,
+      "learning_rate": 0.00017610526315789472,
+      "loss": 0.2328,
+      "step": 5035
+    },
+    {
+      "epoch": 75.16,
+      "learning_rate": 0.00017607017543859647,
+      "loss": 0.2112,
+      "step": 5036
+    },
+    {
+      "epoch": 75.18,
+      "learning_rate": 0.00017603508771929825,
+      "loss": 0.0083,
+      "step": 5037
+    },
+    {
+      "epoch": 75.19,
+      "learning_rate": 0.000176,
+      "loss": 0.2025,
+      "step": 5038
+    },
+    {
+      "epoch": 75.21,
+      "learning_rate": 0.00017596491228070174,
+      "loss": 0.0229,
+      "step": 5039
+    },
+    {
+      "epoch": 75.22,
+      "learning_rate": 0.00017592982456140352,
+      "loss": 0.0018,
+      "step": 5040
+    },
+    {
+      "epoch": 75.24,
+      "learning_rate": 0.00017589473684210524,
+      "loss": 0.0022,
+      "step": 5041
+    },
+    {
+      "epoch": 75.25,
+      "learning_rate": 0.000175859649122807,
+      "loss": 0.0017,
+      "step": 5042
+    },
+    {
+      "epoch": 75.27,
+      "learning_rate": 0.00017582456140350874,
+      "loss": 0.0017,
+      "step": 5043
+    },
+    {
+      "epoch": 75.28,
+      "learning_rate": 0.00017578947368421052,
+      "loss": 0.1587,
+      "step": 5044
+    },
+    {
+      "epoch": 75.3,
+      "learning_rate": 0.00017575438596491226,
+      "loss": 0.0028,
+      "step": 5045
+    },
+    {
+      "epoch": 75.31,
+      "learning_rate": 0.000175719298245614,
+      "loss": 0.0041,
+      "step": 5046
+    },
+    {
+      "epoch": 75.33,
+      "learning_rate": 0.00017568421052631576,
+      "loss": 0.0012,
+      "step": 5047
+    },
+    {
+      "epoch": 75.34,
+      "learning_rate": 0.00017564912280701754,
+      "loss": 0.0014,
+      "step": 5048
+    },
+    {
+      "epoch": 75.36,
+      "learning_rate": 0.00017561403508771929,
+      "loss": 0.1035,
+      "step": 5049
+    },
+    {
+      "epoch": 75.37,
+      "learning_rate": 0.00017557894736842103,
+      "loss": 0.0893,
+      "step": 5050
+    },
+    {
+      "epoch": 75.39,
+      "learning_rate": 0.00017554385964912278,
+      "loss": 0.162,
+      "step": 5051
+    },
+    {
+      "epoch": 75.4,
+      "learning_rate": 0.00017550877192982456,
+      "loss": 0.0009,
+      "step": 5052
+    },
+    {
+      "epoch": 75.42,
+      "learning_rate": 0.0001754736842105263,
+      "loss": 0.001,
+      "step": 5053
+    },
+    {
+      "epoch": 75.43,
+      "learning_rate": 0.00017543859649122806,
+      "loss": 0.001,
+      "step": 5054
+    },
+    {
+      "epoch": 75.45,
+      "learning_rate": 0.00017540350877192983,
+      "loss": 0.0011,
+      "step": 5055
+    },
+    {
+      "epoch": 75.46,
+      "learning_rate": 0.00017536842105263158,
+      "loss": 0.1141,
+      "step": 5056
+    },
+    {
+      "epoch": 75.48,
+      "learning_rate": 0.0001753333333333333,
+      "loss": 0.0036,
+      "step": 5057
+    },
+    {
+      "epoch": 75.49,
+      "learning_rate": 0.00017529824561403505,
+      "loss": 0.001,
+      "step": 5058
+    },
+    {
+      "epoch": 75.51,
+      "learning_rate": 0.00017526315789473683,
+      "loss": 0.16,
+      "step": 5059
+    },
+    {
+      "epoch": 75.52,
+      "learning_rate": 0.00017522807017543858,
+      "loss": 0.2575,
+      "step": 5060
+    },
+    {
+      "epoch": 75.54,
+      "learning_rate": 0.00017519298245614033,
+      "loss": 0.0013,
+      "step": 5061
+    },
+    {
+      "epoch": 75.55,
+      "learning_rate": 0.00017515789473684207,
+      "loss": 0.1232,
+      "step": 5062
+    },
+    {
+      "epoch": 75.57,
+      "learning_rate": 0.00017512280701754385,
+      "loss": 0.0014,
+      "step": 5063
+    },
+    {
+      "epoch": 75.58,
+      "learning_rate": 0.0001750877192982456,
+      "loss": 0.0102,
+      "step": 5064
+    },
+    {
+      "epoch": 75.59,
+      "learning_rate": 0.00017505263157894735,
+      "loss": 0.1149,
+      "step": 5065
+    },
+    {
+      "epoch": 75.61,
+      "learning_rate": 0.00017501754385964912,
+      "loss": 0.1935,
+      "step": 5066
+    },
+    {
+      "epoch": 75.62,
+      "learning_rate": 0.00017498245614035087,
+      "loss": 0.0018,
+      "step": 5067
+    },
+    {
+      "epoch": 75.64,
+      "learning_rate": 0.00017494736842105262,
+      "loss": 0.1331,
+      "step": 5068
+    },
+    {
+      "epoch": 75.65,
+      "learning_rate": 0.00017491228070175437,
+      "loss": 0.1212,
+      "step": 5069
+    },
+    {
+      "epoch": 75.67,
+      "learning_rate": 0.00017487719298245615,
+      "loss": 0.1023,
+      "step": 5070
+    },
+    {
+      "epoch": 75.68,
+      "learning_rate": 0.0001748421052631579,
+      "loss": 0.1183,
+      "step": 5071
+    },
+    {
+      "epoch": 75.7,
+      "learning_rate": 0.00017480701754385962,
+      "loss": 0.145,
+      "step": 5072
+    },
+    {
+      "epoch": 75.71,
+      "learning_rate": 0.00017477192982456137,
+      "loss": 0.0044,
+      "step": 5073
+    },
+    {
+      "epoch": 75.73,
+      "learning_rate": 0.00017473684210526314,
+      "loss": 0.0016,
+      "step": 5074
+    },
+    {
+      "epoch": 75.74,
+      "learning_rate": 0.0001747017543859649,
+      "loss": 0.0024,
+      "step": 5075
+    },
+    {
+      "epoch": 75.76,
+      "learning_rate": 0.00017466666666666664,
+      "loss": 0.0338,
+      "step": 5076
+    },
+    {
+      "epoch": 75.77,
+      "learning_rate": 0.00017463157894736841,
+      "loss": 0.0053,
+      "step": 5077
+    },
+    {
+      "epoch": 75.79,
+      "learning_rate": 0.00017459649122807016,
+      "loss": 0.0742,
+      "step": 5078
+    },
+    {
+      "epoch": 75.8,
+      "learning_rate": 0.0001745614035087719,
+      "loss": 0.0017,
+      "step": 5079
+    },
+    {
+      "epoch": 75.82,
+      "learning_rate": 0.00017452631578947366,
+      "loss": 0.0229,
+      "step": 5080
+    },
+    {
+      "epoch": 75.83,
+      "learning_rate": 0.00017449122807017544,
+      "loss": 0.0323,
+      "step": 5081
+    },
+    {
+      "epoch": 75.85,
+      "learning_rate": 0.00017445614035087719,
+      "loss": 0.009,
+      "step": 5082
+    },
+    {
+      "epoch": 75.86,
+      "learning_rate": 0.00017442105263157893,
+      "loss": 0.0308,
+      "step": 5083
+    },
+    {
+      "epoch": 75.88,
+      "learning_rate": 0.00017438596491228068,
+      "loss": 0.1732,
+      "step": 5084
+    },
+    {
+      "epoch": 75.89,
+      "learning_rate": 0.00017435087719298246,
+      "loss": 0.0183,
+      "step": 5085
+    },
+    {
+      "epoch": 75.91,
+      "learning_rate": 0.0001743157894736842,
+      "loss": 0.0027,
+      "step": 5086
+    },
+    {
+      "epoch": 75.92,
+      "learning_rate": 0.00017428070175438593,
+      "loss": 0.0059,
+      "step": 5087
+    },
+    {
+      "epoch": 75.94,
+      "learning_rate": 0.00017424561403508768,
+      "loss": 0.0011,
+      "step": 5088
+    },
+    {
+      "epoch": 75.95,
+      "learning_rate": 0.00017421052631578945,
+      "loss": 0.0154,
+      "step": 5089
+    },
+    {
+      "epoch": 75.97,
+      "learning_rate": 0.0001741754385964912,
+      "loss": 0.2746,
+      "step": 5090
+    },
+    {
+      "epoch": 75.98,
+      "learning_rate": 0.00017414035087719295,
+      "loss": 0.009,
+      "step": 5091
+    },
+    {
+      "epoch": 76.0,
+      "learning_rate": 0.00017410526315789473,
+      "loss": 0.1161,
+      "step": 5092
+    },
+    {
+      "epoch": 76.01,
+      "learning_rate": 0.00017407017543859648,
+      "loss": 0.0044,
+      "step": 5093
+    },
+    {
+      "epoch": 76.03,
+      "learning_rate": 0.00017403508771929823,
+      "loss": 0.0122,
+      "step": 5094
+    },
+    {
+      "epoch": 76.04,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 0.0024,
+      "step": 5095
+    },
+    {
+      "epoch": 76.06,
+      "learning_rate": 0.00017396491228070175,
+      "loss": 0.017,
+      "step": 5096
+    },
+    {
+      "epoch": 76.07,
+      "learning_rate": 0.0001739298245614035,
+      "loss": 0.3844,
+      "step": 5097
+    },
+    {
+      "epoch": 76.09,
+      "learning_rate": 0.00017389473684210525,
+      "loss": 0.0014,
+      "step": 5098
+    },
+    {
+      "epoch": 76.1,
+      "learning_rate": 0.000173859649122807,
+      "loss": 0.0806,
+      "step": 5099
+    },
+    {
+      "epoch": 76.12,
+      "learning_rate": 0.00017382456140350877,
+      "loss": 0.0194,
+      "step": 5100
+    },
+    {
+      "epoch": 76.13,
+      "learning_rate": 0.00017378947368421052,
+      "loss": 0.1099,
+      "step": 5101
+    },
+    {
+      "epoch": 76.15,
+      "learning_rate": 0.00017375438596491227,
+      "loss": 0.0016,
+      "step": 5102
+    },
+    {
+      "epoch": 76.16,
+      "learning_rate": 0.00017371929824561405,
+      "loss": 0.0011,
+      "step": 5103
+    },
+    {
+      "epoch": 76.18,
+      "learning_rate": 0.0001736842105263158,
+      "loss": 0.0017,
+      "step": 5104
+    },
+    {
+      "epoch": 76.19,
+      "learning_rate": 0.00017364912280701752,
+      "loss": 0.0024,
+      "step": 5105
+    },
+    {
+      "epoch": 76.21,
+      "learning_rate": 0.00017361403508771926,
+      "loss": 0.0012,
+      "step": 5106
+    },
+    {
+      "epoch": 76.22,
+      "learning_rate": 0.00017357894736842104,
+      "loss": 0.0047,
+      "step": 5107
+    },
+    {
+      "epoch": 76.24,
+      "learning_rate": 0.0001735438596491228,
+      "loss": 0.001,
+      "step": 5108
+    },
+    {
+      "epoch": 76.25,
+      "learning_rate": 0.00017350877192982454,
+      "loss": 0.3858,
+      "step": 5109
+    },
+    {
+      "epoch": 76.27,
+      "learning_rate": 0.0001734736842105263,
+      "loss": 0.0133,
+      "step": 5110
+    },
+    {
+      "epoch": 76.28,
+      "learning_rate": 0.00017343859649122806,
+      "loss": 0.0053,
+      "step": 5111
+    },
+    {
+      "epoch": 76.3,
+      "learning_rate": 0.0001734035087719298,
+      "loss": 0.0933,
+      "step": 5112
+    },
+    {
+      "epoch": 76.31,
+      "learning_rate": 0.00017336842105263156,
+      "loss": 0.023,
+      "step": 5113
+    },
+    {
+      "epoch": 76.33,
+      "learning_rate": 0.0001733333333333333,
+      "loss": 0.0009,
+      "step": 5114
+    },
+    {
+      "epoch": 76.34,
+      "learning_rate": 0.00017329824561403508,
+      "loss": 0.0033,
+      "step": 5115
+    },
+    {
+      "epoch": 76.36,
+      "learning_rate": 0.00017326315789473683,
+      "loss": 0.0283,
+      "step": 5116
+    },
+    {
+      "epoch": 76.37,
+      "learning_rate": 0.00017322807017543858,
+      "loss": 0.0076,
+      "step": 5117
+    },
+    {
+      "epoch": 76.39,
+      "learning_rate": 0.00017319298245614036,
+      "loss": 0.0467,
+      "step": 5118
+    },
+    {
+      "epoch": 76.4,
+      "learning_rate": 0.0001731578947368421,
+      "loss": 0.0849,
+      "step": 5119
+    },
+    {
+      "epoch": 76.42,
+      "learning_rate": 0.00017312280701754383,
+      "loss": 0.0441,
+      "step": 5120
+    },
+    {
+      "epoch": 76.43,
+      "learning_rate": 0.00017308771929824558,
+      "loss": 0.0016,
+      "step": 5121
+    },
+    {
+      "epoch": 76.45,
+      "learning_rate": 0.00017305263157894735,
+      "loss": 0.0011,
+      "step": 5122
+    },
+    {
+      "epoch": 76.46,
+      "learning_rate": 0.0001730175438596491,
+      "loss": 0.1175,
+      "step": 5123
+    },
+    {
+      "epoch": 76.48,
+      "learning_rate": 0.00017298245614035085,
+      "loss": 0.0439,
+      "step": 5124
+    },
+    {
+      "epoch": 76.49,
+      "learning_rate": 0.0001729473684210526,
+      "loss": 0.002,
+      "step": 5125
+    },
+    {
+      "epoch": 76.51,
+      "learning_rate": 0.00017291228070175438,
+      "loss": 0.2141,
+      "step": 5126
+    },
+    {
+      "epoch": 76.52,
+      "learning_rate": 0.00017287719298245612,
+      "loss": 0.0033,
+      "step": 5127
+    },
+    {
+      "epoch": 76.54,
+      "learning_rate": 0.00017284210526315787,
+      "loss": 0.0052,
+      "step": 5128
+    },
+    {
+      "epoch": 76.55,
+      "learning_rate": 0.00017280701754385965,
+      "loss": 0.001,
+      "step": 5129
+    },
+    {
+      "epoch": 76.57,
+      "learning_rate": 0.0001727719298245614,
+      "loss": 0.1329,
+      "step": 5130
+    },
+    {
+      "epoch": 76.58,
+      "learning_rate": 0.00017273684210526315,
+      "loss": 0.1306,
+      "step": 5131
+    },
+    {
+      "epoch": 76.59,
+      "learning_rate": 0.0001727017543859649,
+      "loss": 0.0259,
+      "step": 5132
+    },
+    {
+      "epoch": 76.61,
+      "learning_rate": 0.00017266666666666667,
+      "loss": 0.0014,
+      "step": 5133
+    },
+    {
+      "epoch": 76.62,
+      "learning_rate": 0.00017263157894736842,
+      "loss": 0.0735,
+      "step": 5134
+    },
+    {
+      "epoch": 76.64,
+      "learning_rate": 0.00017259649122807014,
+      "loss": 0.001,
+      "step": 5135
+    },
+    {
+      "epoch": 76.65,
+      "learning_rate": 0.0001725614035087719,
+      "loss": 0.0068,
+      "step": 5136
+    },
+    {
+      "epoch": 76.67,
+      "learning_rate": 0.00017252631578947367,
+      "loss": 0.113,
+      "step": 5137
+    },
+    {
+      "epoch": 76.68,
+      "learning_rate": 0.00017249122807017542,
+      "loss": 0.1599,
+      "step": 5138
+    },
+    {
+      "epoch": 76.7,
+      "learning_rate": 0.00017245614035087716,
+      "loss": 0.0022,
+      "step": 5139
+    },
+    {
+      "epoch": 76.71,
+      "learning_rate": 0.00017242105263157894,
+      "loss": 0.0012,
+      "step": 5140
+    },
+    {
+      "epoch": 76.73,
+      "learning_rate": 0.0001723859649122807,
+      "loss": 0.0692,
+      "step": 5141
+    },
+    {
+      "epoch": 76.74,
+      "learning_rate": 0.00017235087719298244,
+      "loss": 0.0012,
+      "step": 5142
+    },
+    {
+      "epoch": 76.76,
+      "learning_rate": 0.00017231578947368419,
+      "loss": 0.0655,
+      "step": 5143
+    },
+    {
+      "epoch": 76.77,
+      "learning_rate": 0.00017228070175438596,
+      "loss": 0.1018,
+      "step": 5144
+    },
+    {
+      "epoch": 76.79,
+      "learning_rate": 0.0001722456140350877,
+      "loss": 0.0011,
+      "step": 5145
+    },
+    {
+      "epoch": 76.8,
+      "learning_rate": 0.00017221052631578946,
+      "loss": 0.0603,
+      "step": 5146
+    },
+    {
+      "epoch": 76.82,
+      "learning_rate": 0.0001721754385964912,
+      "loss": 0.1936,
+      "step": 5147
+    },
+    {
+      "epoch": 76.83,
+      "learning_rate": 0.00017214035087719298,
+      "loss": 0.0032,
+      "step": 5148
+    },
+    {
+      "epoch": 76.85,
+      "learning_rate": 0.00017210526315789473,
+      "loss": 0.003,
+      "step": 5149
+    },
+    {
+      "epoch": 76.86,
+      "learning_rate": 0.00017207017543859648,
+      "loss": 0.0038,
+      "step": 5150
+    },
+    {
+      "epoch": 76.88,
+      "learning_rate": 0.0001720350877192982,
+      "loss": 0.1508,
+      "step": 5151
+    },
+    {
+      "epoch": 76.89,
+      "learning_rate": 0.000172,
+      "loss": 0.001,
+      "step": 5152
+    },
+    {
+      "epoch": 76.91,
+      "learning_rate": 0.00017196491228070173,
+      "loss": 0.0362,
+      "step": 5153
+    },
+    {
+      "epoch": 76.92,
+      "learning_rate": 0.00017192982456140348,
+      "loss": 0.0024,
+      "step": 5154
+    },
+    {
+      "epoch": 76.94,
+      "learning_rate": 0.00017189473684210525,
+      "loss": 0.0012,
+      "step": 5155
+    },
+    {
+      "epoch": 76.95,
+      "learning_rate": 0.000171859649122807,
+      "loss": 0.0012,
+      "step": 5156
+    },
+    {
+      "epoch": 76.97,
+      "learning_rate": 0.00017182456140350875,
+      "loss": 0.0092,
+      "step": 5157
+    },
+    {
+      "epoch": 76.98,
+      "learning_rate": 0.0001717894736842105,
+      "loss": 0.0012,
+      "step": 5158
+    },
+    {
+      "epoch": 77.0,
+      "learning_rate": 0.00017175438596491228,
+      "loss": 0.0331,
+      "step": 5159
+    },
+    {
+      "epoch": 77.01,
+      "learning_rate": 0.00017171929824561402,
+      "loss": 0.0255,
+      "step": 5160
+    },
+    {
+      "epoch": 77.03,
+      "learning_rate": 0.00017168421052631577,
+      "loss": 0.0015,
+      "step": 5161
+    },
+    {
+      "epoch": 77.04,
+      "learning_rate": 0.00017164912280701752,
+      "loss": 0.1901,
+      "step": 5162
+    },
+    {
+      "epoch": 77.06,
+      "learning_rate": 0.0001716140350877193,
+      "loss": 0.0017,
+      "step": 5163
+    },
+    {
+      "epoch": 77.07,
+      "learning_rate": 0.00017157894736842105,
+      "loss": 0.013,
+      "step": 5164
+    },
+    {
+      "epoch": 77.09,
+      "learning_rate": 0.0001715438596491228,
+      "loss": 0.0037,
+      "step": 5165
+    },
+    {
+      "epoch": 77.1,
+      "learning_rate": 0.00017150877192982457,
+      "loss": 0.0039,
+      "step": 5166
+    },
+    {
+      "epoch": 77.12,
+      "learning_rate": 0.00017147368421052632,
+      "loss": 0.0437,
+      "step": 5167
+    },
+    {
+      "epoch": 77.13,
+      "learning_rate": 0.00017143859649122804,
+      "loss": 0.1028,
+      "step": 5168
+    },
+    {
+      "epoch": 77.15,
+      "learning_rate": 0.0001714035087719298,
+      "loss": 0.0809,
+      "step": 5169
+    },
+    {
+      "epoch": 77.16,
+      "learning_rate": 0.00017136842105263157,
+      "loss": 0.0092,
+      "step": 5170
+    },
+    {
+      "epoch": 77.18,
+      "learning_rate": 0.00017133333333333331,
+      "loss": 0.0021,
+      "step": 5171
+    },
+    {
+      "epoch": 77.19,
+      "learning_rate": 0.00017129824561403506,
+      "loss": 0.0011,
+      "step": 5172
+    },
+    {
+      "epoch": 77.21,
+      "learning_rate": 0.0001712631578947368,
+      "loss": 0.1418,
+      "step": 5173
+    },
+    {
+      "epoch": 77.22,
+      "learning_rate": 0.0001712280701754386,
+      "loss": 0.0012,
+      "step": 5174
+    },
+    {
+      "epoch": 77.24,
+      "learning_rate": 0.00017119298245614034,
+      "loss": 0.0011,
+      "step": 5175
+    },
+    {
+      "epoch": 77.25,
+      "learning_rate": 0.00017115789473684209,
+      "loss": 0.1556,
+      "step": 5176
+    },
+    {
+      "epoch": 77.27,
+      "learning_rate": 0.00017112280701754383,
+      "loss": 0.02,
+      "step": 5177
+    },
+    {
+      "epoch": 77.28,
+      "learning_rate": 0.0001710877192982456,
+      "loss": 0.0541,
+      "step": 5178
+    },
+    {
+      "epoch": 77.3,
+      "learning_rate": 0.00017105263157894736,
+      "loss": 0.0024,
+      "step": 5179
+    },
+    {
+      "epoch": 77.31,
+      "learning_rate": 0.0001710175438596491,
+      "loss": 0.0144,
+      "step": 5180
+    },
+    {
+      "epoch": 77.33,
+      "learning_rate": 0.00017098245614035088,
+      "loss": 0.0672,
+      "step": 5181
+    },
+    {
+      "epoch": 77.34,
+      "learning_rate": 0.00017094736842105263,
+      "loss": 0.0022,
+      "step": 5182
+    },
+    {
+      "epoch": 77.36,
+      "learning_rate": 0.00017091228070175435,
+      "loss": 0.1882,
+      "step": 5183
+    },
+    {
+      "epoch": 77.37,
+      "learning_rate": 0.0001708771929824561,
+      "loss": 0.0506,
+      "step": 5184
+    },
+    {
+      "epoch": 77.39,
+      "learning_rate": 0.00017084210526315788,
+      "loss": 0.0947,
+      "step": 5185
+    },
+    {
+      "epoch": 77.4,
+      "learning_rate": 0.00017080701754385963,
+      "loss": 0.0038,
+      "step": 5186
+    },
+    {
+      "epoch": 77.42,
+      "learning_rate": 0.00017077192982456138,
+      "loss": 0.1022,
+      "step": 5187
+    },
+    {
+      "epoch": 77.43,
+      "learning_rate": 0.00017073684210526313,
+      "loss": 0.0033,
+      "step": 5188
+    },
+    {
+      "epoch": 77.45,
+      "learning_rate": 0.0001707017543859649,
+      "loss": 0.0016,
+      "step": 5189
+    },
+    {
+      "epoch": 77.46,
+      "learning_rate": 0.00017066666666666665,
+      "loss": 0.0544,
+      "step": 5190
+    },
+    {
+      "epoch": 77.48,
+      "learning_rate": 0.0001706315789473684,
+      "loss": 0.0466,
+      "step": 5191
+    },
+    {
+      "epoch": 77.49,
+      "learning_rate": 0.00017059649122807017,
+      "loss": 0.0066,
+      "step": 5192
+    },
+    {
+      "epoch": 77.51,
+      "learning_rate": 0.00017056140350877192,
+      "loss": 0.1766,
+      "step": 5193
+    },
+    {
+      "epoch": 77.52,
+      "learning_rate": 0.00017052631578947367,
+      "loss": 0.0044,
+      "step": 5194
+    },
+    {
+      "epoch": 77.54,
+      "learning_rate": 0.00017049122807017542,
+      "loss": 0.1594,
+      "step": 5195
+    },
+    {
+      "epoch": 77.55,
+      "learning_rate": 0.0001704561403508772,
+      "loss": 0.002,
+      "step": 5196
+    },
+    {
+      "epoch": 77.57,
+      "learning_rate": 0.00017042105263157895,
+      "loss": 0.5177,
+      "step": 5197
+    },
+    {
+      "epoch": 77.58,
+      "learning_rate": 0.0001703859649122807,
+      "loss": 0.007,
+      "step": 5198
+    },
+    {
+      "epoch": 77.59,
+      "learning_rate": 0.00017035087719298242,
+      "loss": 0.0725,
+      "step": 5199
+    },
+    {
+      "epoch": 77.61,
+      "learning_rate": 0.00017031578947368422,
+      "loss": 0.1165,
+      "step": 5200
+    },
+    {
+      "epoch": 77.61,
+      "eval_accuracy": 0.8595203132648067,
+      "eval_f1": 0.860270309660957,
+      "eval_loss": 0.6248674988746643,
+      "eval_runtime": 345.784,
+      "eval_samples_per_second": 11.817,
+      "eval_steps_per_second": 0.74,
+      "step": 5200
+    },
+    {
+      "epoch": 77.62,
+      "learning_rate": 0.00017028070175438594,
+      "loss": 0.011,
+      "step": 5201
+    },
+    {
+      "epoch": 77.64,
+      "learning_rate": 0.0001702456140350877,
+      "loss": 0.0015,
+      "step": 5202
+    },
+    {
+      "epoch": 77.65,
+      "learning_rate": 0.00017021052631578947,
+      "loss": 0.0022,
+      "step": 5203
+    },
+    {
+      "epoch": 77.67,
+      "learning_rate": 0.00017017543859649121,
+      "loss": 0.0012,
+      "step": 5204
+    },
+    {
+      "epoch": 77.68,
+      "learning_rate": 0.00017014035087719296,
+      "loss": 0.0163,
+      "step": 5205
+    },
+    {
+      "epoch": 77.7,
+      "learning_rate": 0.0001701052631578947,
+      "loss": 0.0082,
+      "step": 5206
+    },
+    {
+      "epoch": 77.71,
+      "learning_rate": 0.0001700701754385965,
+      "loss": 0.0025,
+      "step": 5207
+    },
+    {
+      "epoch": 77.73,
+      "learning_rate": 0.00017003508771929824,
+      "loss": 0.0022,
+      "step": 5208
+    },
+    {
+      "epoch": 77.74,
+      "learning_rate": 0.00016999999999999999,
+      "loss": 0.0288,
+      "step": 5209
+    },
+    {
+      "epoch": 77.76,
+      "learning_rate": 0.00016996491228070173,
+      "loss": 0.1087,
+      "step": 5210
+    },
+    {
+      "epoch": 77.77,
+      "learning_rate": 0.0001699298245614035,
+      "loss": 0.0049,
+      "step": 5211
+    },
+    {
+      "epoch": 77.79,
+      "learning_rate": 0.00016989473684210526,
+      "loss": 0.0052,
+      "step": 5212
+    },
+    {
+      "epoch": 77.8,
+      "learning_rate": 0.000169859649122807,
+      "loss": 0.0126,
+      "step": 5213
+    },
+    {
+      "epoch": 77.82,
+      "learning_rate": 0.00016982456140350873,
+      "loss": 0.0048,
+      "step": 5214
+    },
+    {
+      "epoch": 77.83,
+      "learning_rate": 0.00016978947368421053,
+      "loss": 0.0034,
+      "step": 5215
+    },
+    {
+      "epoch": 77.85,
+      "learning_rate": 0.00016975438596491225,
+      "loss": 0.1069,
+      "step": 5216
+    },
+    {
+      "epoch": 77.86,
+      "learning_rate": 0.000169719298245614,
+      "loss": 0.0046,
+      "step": 5217
+    },
+    {
+      "epoch": 77.88,
+      "learning_rate": 0.00016968421052631578,
+      "loss": 0.0169,
+      "step": 5218
+    },
+    {
+      "epoch": 77.89,
+      "learning_rate": 0.00016964912280701753,
+      "loss": 0.2265,
+      "step": 5219
+    },
+    {
+      "epoch": 77.91,
+      "learning_rate": 0.00016961403508771928,
+      "loss": 0.002,
+      "step": 5220
+    },
+    {
+      "epoch": 77.92,
+      "learning_rate": 0.00016957894736842102,
+      "loss": 0.0123,
+      "step": 5221
+    },
+    {
+      "epoch": 77.94,
+      "learning_rate": 0.0001695438596491228,
+      "loss": 0.0356,
+      "step": 5222
+    },
+    {
+      "epoch": 77.95,
+      "learning_rate": 0.00016950877192982455,
+      "loss": 0.0099,
+      "step": 5223
+    },
+    {
+      "epoch": 77.97,
+      "learning_rate": 0.0001694736842105263,
+      "loss": 0.003,
+      "step": 5224
+    },
+    {
+      "epoch": 77.98,
+      "learning_rate": 0.00016943859649122805,
+      "loss": 0.0033,
+      "step": 5225
+    },
+    {
+      "epoch": 78.0,
+      "learning_rate": 0.00016940350877192982,
+      "loss": 0.1931,
+      "step": 5226
+    },
+    {
+      "epoch": 78.01,
+      "learning_rate": 0.00016936842105263157,
+      "loss": 0.2121,
+      "step": 5227
+    },
+    {
+      "epoch": 78.03,
+      "learning_rate": 0.00016933333333333332,
+      "loss": 0.0295,
+      "step": 5228
+    },
+    {
+      "epoch": 78.04,
+      "learning_rate": 0.0001692982456140351,
+      "loss": 0.0095,
+      "step": 5229
+    },
+    {
+      "epoch": 78.06,
+      "learning_rate": 0.00016926315789473684,
+      "loss": 0.1848,
+      "step": 5230
+    },
+    {
+      "epoch": 78.07,
+      "learning_rate": 0.00016922807017543857,
+      "loss": 0.179,
+      "step": 5231
+    },
+    {
+      "epoch": 78.09,
+      "learning_rate": 0.00016919298245614032,
+      "loss": 0.002,
+      "step": 5232
+    },
+    {
+      "epoch": 78.1,
+      "learning_rate": 0.0001691578947368421,
+      "loss": 0.0021,
+      "step": 5233
+    },
+    {
+      "epoch": 78.12,
+      "learning_rate": 0.00016912280701754384,
+      "loss": 0.001,
+      "step": 5234
+    },
+    {
+      "epoch": 78.13,
+      "learning_rate": 0.0001690877192982456,
+      "loss": 0.0018,
+      "step": 5235
+    },
+    {
+      "epoch": 78.15,
+      "learning_rate": 0.00016905263157894734,
+      "loss": 0.0209,
+      "step": 5236
+    },
+    {
+      "epoch": 78.16,
+      "learning_rate": 0.0001690175438596491,
+      "loss": 0.003,
+      "step": 5237
+    },
+    {
+      "epoch": 78.18,
+      "learning_rate": 0.00016898245614035086,
+      "loss": 0.0016,
+      "step": 5238
+    },
+    {
+      "epoch": 78.19,
+      "learning_rate": 0.0001689473684210526,
+      "loss": 0.0861,
+      "step": 5239
+    },
+    {
+      "epoch": 78.21,
+      "learning_rate": 0.0001689122807017544,
+      "loss": 0.3714,
+      "step": 5240
+    },
+    {
+      "epoch": 78.22,
+      "learning_rate": 0.00016887719298245614,
+      "loss": 0.0318,
+      "step": 5241
+    },
+    {
+      "epoch": 78.24,
+      "learning_rate": 0.00016884210526315788,
+      "loss": 0.0144,
+      "step": 5242
+    },
+    {
+      "epoch": 78.25,
+      "learning_rate": 0.00016880701754385963,
+      "loss": 0.0178,
+      "step": 5243
+    },
+    {
+      "epoch": 78.27,
+      "learning_rate": 0.0001687719298245614,
+      "loss": 0.0019,
+      "step": 5244
+    },
+    {
+      "epoch": 78.28,
+      "learning_rate": 0.00016873684210526316,
+      "loss": 0.0023,
+      "step": 5245
+    },
+    {
+      "epoch": 78.3,
+      "learning_rate": 0.0001687017543859649,
+      "loss": 0.1122,
+      "step": 5246
+    },
+    {
+      "epoch": 78.31,
+      "learning_rate": 0.00016866666666666663,
+      "loss": 0.0778,
+      "step": 5247
+    },
+    {
+      "epoch": 78.33,
+      "learning_rate": 0.0001686315789473684,
+      "loss": 0.0029,
+      "step": 5248
+    },
+    {
+      "epoch": 78.34,
+      "learning_rate": 0.00016859649122807015,
+      "loss": 0.0016,
+      "step": 5249
+    },
+    {
+      "epoch": 78.36,
+      "learning_rate": 0.0001685614035087719,
+      "loss": 0.1838,
+      "step": 5250
+    },
+    {
+      "epoch": 78.37,
+      "learning_rate": 0.00016852631578947365,
+      "loss": 0.0189,
+      "step": 5251
+    },
+    {
+      "epoch": 78.39,
+      "learning_rate": 0.00016849122807017543,
+      "loss": 0.0798,
+      "step": 5252
+    },
+    {
+      "epoch": 78.4,
+      "learning_rate": 0.00016845614035087718,
+      "loss": 0.0018,
+      "step": 5253
+    },
+    {
+      "epoch": 78.42,
+      "learning_rate": 0.00016842105263157892,
+      "loss": 0.0105,
+      "step": 5254
+    },
+    {
+      "epoch": 78.43,
+      "learning_rate": 0.0001683859649122807,
+      "loss": 0.0016,
+      "step": 5255
+    },
+    {
+      "epoch": 78.45,
+      "learning_rate": 0.00016835087719298245,
+      "loss": 0.0053,
+      "step": 5256
+    },
+    {
+      "epoch": 78.46,
+      "learning_rate": 0.0001683157894736842,
+      "loss": 0.0009,
+      "step": 5257
+    },
+    {
+      "epoch": 78.48,
+      "learning_rate": 0.00016828070175438595,
+      "loss": 0.0009,
+      "step": 5258
+    },
+    {
+      "epoch": 78.49,
+      "learning_rate": 0.00016824561403508772,
+      "loss": 0.0011,
+      "step": 5259
+    },
+    {
+      "epoch": 78.51,
+      "learning_rate": 0.00016821052631578947,
+      "loss": 0.1651,
+      "step": 5260
+    },
+    {
+      "epoch": 78.52,
+      "learning_rate": 0.00016817543859649122,
+      "loss": 0.0027,
+      "step": 5261
+    },
+    {
+      "epoch": 78.54,
+      "learning_rate": 0.00016814035087719294,
+      "loss": 0.0896,
+      "step": 5262
+    },
+    {
+      "epoch": 78.55,
+      "learning_rate": 0.00016810526315789474,
+      "loss": 0.1429,
+      "step": 5263
+    },
+    {
+      "epoch": 78.57,
+      "learning_rate": 0.00016807017543859647,
+      "loss": 0.0048,
+      "step": 5264
+    },
+    {
+      "epoch": 78.58,
+      "learning_rate": 0.00016803508771929821,
+      "loss": 0.0124,
+      "step": 5265
+    },
+    {
+      "epoch": 78.59,
+      "learning_rate": 0.000168,
+      "loss": 0.0183,
+      "step": 5266
+    },
+    {
+      "epoch": 78.61,
+      "learning_rate": 0.00016796491228070174,
+      "loss": 0.0027,
+      "step": 5267
+    },
+    {
+      "epoch": 78.62,
+      "learning_rate": 0.0001679298245614035,
+      "loss": 0.0021,
+      "step": 5268
+    },
+    {
+      "epoch": 78.64,
+      "learning_rate": 0.00016789473684210524,
+      "loss": 0.0305,
+      "step": 5269
+    },
+    {
+      "epoch": 78.65,
+      "learning_rate": 0.000167859649122807,
+      "loss": 0.0382,
+      "step": 5270
+    },
+    {
+      "epoch": 78.67,
+      "learning_rate": 0.00016782456140350876,
+      "loss": 0.0012,
+      "step": 5271
+    },
+    {
+      "epoch": 78.68,
+      "learning_rate": 0.0001677894736842105,
+      "loss": 0.0018,
+      "step": 5272
+    },
+    {
+      "epoch": 78.7,
+      "learning_rate": 0.00016775438596491226,
+      "loss": 0.0016,
+      "step": 5273
+    },
+    {
+      "epoch": 78.71,
+      "learning_rate": 0.00016771929824561404,
+      "loss": 0.0149,
+      "step": 5274
+    },
+    {
+      "epoch": 78.73,
+      "learning_rate": 0.00016768421052631578,
+      "loss": 0.0011,
+      "step": 5275
+    },
+    {
+      "epoch": 78.74,
+      "learning_rate": 0.00016764912280701753,
+      "loss": 0.2443,
+      "step": 5276
+    },
+    {
+      "epoch": 78.76,
+      "learning_rate": 0.00016761403508771925,
+      "loss": 0.0069,
+      "step": 5277
+    },
+    {
+      "epoch": 78.77,
+      "learning_rate": 0.00016757894736842106,
+      "loss": 0.0438,
+      "step": 5278
+    },
+    {
+      "epoch": 78.79,
+      "learning_rate": 0.00016754385964912278,
+      "loss": 0.002,
+      "step": 5279
+    },
+    {
+      "epoch": 78.8,
+      "learning_rate": 0.00016750877192982453,
+      "loss": 0.0701,
+      "step": 5280
+    },
+    {
+      "epoch": 78.82,
+      "learning_rate": 0.0001674736842105263,
+      "loss": 0.0552,
+      "step": 5281
+    },
+    {
+      "epoch": 78.83,
+      "learning_rate": 0.00016743859649122805,
+      "loss": 0.1124,
+      "step": 5282
+    },
+    {
+      "epoch": 78.85,
+      "learning_rate": 0.0001674035087719298,
+      "loss": 0.0138,
+      "step": 5283
+    },
+    {
+      "epoch": 78.86,
+      "learning_rate": 0.00016736842105263155,
+      "loss": 0.0011,
+      "step": 5284
+    },
+    {
+      "epoch": 78.88,
+      "learning_rate": 0.00016733333333333333,
+      "loss": 0.0077,
+      "step": 5285
+    },
+    {
+      "epoch": 78.89,
+      "learning_rate": 0.00016729824561403507,
+      "loss": 0.0048,
+      "step": 5286
+    },
+    {
+      "epoch": 78.91,
+      "learning_rate": 0.00016726315789473682,
+      "loss": 0.0017,
+      "step": 5287
+    },
+    {
+      "epoch": 78.92,
+      "learning_rate": 0.00016722807017543857,
+      "loss": 0.1648,
+      "step": 5288
+    },
+    {
+      "epoch": 78.94,
+      "learning_rate": 0.00016719298245614035,
+      "loss": 0.1637,
+      "step": 5289
+    },
+    {
+      "epoch": 78.95,
+      "learning_rate": 0.0001671578947368421,
+      "loss": 0.0021,
+      "step": 5290
+    },
+    {
+      "epoch": 78.97,
+      "learning_rate": 0.00016712280701754385,
+      "loss": 0.0633,
+      "step": 5291
+    },
+    {
+      "epoch": 78.98,
+      "learning_rate": 0.00016708771929824562,
+      "loss": 0.0014,
+      "step": 5292
+    },
+    {
+      "epoch": 79.0,
+      "learning_rate": 0.00016705263157894737,
+      "loss": 0.0012,
+      "step": 5293
+    },
+    {
+      "epoch": 79.01,
+      "learning_rate": 0.00016701754385964912,
+      "loss": 0.0053,
+      "step": 5294
+    },
+    {
+      "epoch": 79.03,
+      "learning_rate": 0.00016698245614035084,
+      "loss": 0.0032,
+      "step": 5295
+    },
+    {
+      "epoch": 79.04,
+      "learning_rate": 0.00016694736842105262,
+      "loss": 0.0259,
+      "step": 5296
+    },
+    {
+      "epoch": 79.06,
+      "learning_rate": 0.00016691228070175437,
+      "loss": 0.0016,
+      "step": 5297
+    },
+    {
+      "epoch": 79.07,
+      "learning_rate": 0.00016687719298245611,
+      "loss": 0.0043,
+      "step": 5298
+    },
+    {
+      "epoch": 79.09,
+      "learning_rate": 0.00016684210526315786,
+      "loss": 0.099,
+      "step": 5299
+    },
+    {
+      "epoch": 79.1,
+      "learning_rate": 0.00016680701754385964,
+      "loss": 0.0156,
+      "step": 5300
+    },
+    {
+      "epoch": 79.12,
+      "learning_rate": 0.0001667719298245614,
+      "loss": 0.0023,
+      "step": 5301
+    },
+    {
+      "epoch": 79.13,
+      "learning_rate": 0.00016673684210526314,
+      "loss": 0.0013,
+      "step": 5302
+    },
+    {
+      "epoch": 79.15,
+      "learning_rate": 0.0001667017543859649,
+      "loss": 0.147,
+      "step": 5303
+    },
+    {
+      "epoch": 79.16,
+      "learning_rate": 0.00016666666666666666,
+      "loss": 0.0026,
+      "step": 5304
+    },
+    {
+      "epoch": 79.18,
+      "learning_rate": 0.0001666315789473684,
+      "loss": 0.1411,
+      "step": 5305
+    },
+    {
+      "epoch": 79.19,
+      "learning_rate": 0.00016659649122807016,
+      "loss": 0.0008,
+      "step": 5306
+    },
+    {
+      "epoch": 79.21,
+      "learning_rate": 0.00016656140350877193,
+      "loss": 0.002,
+      "step": 5307
+    },
+    {
+      "epoch": 79.22,
+      "learning_rate": 0.00016652631578947368,
+      "loss": 0.0238,
+      "step": 5308
+    },
+    {
+      "epoch": 79.24,
+      "learning_rate": 0.00016649122807017543,
+      "loss": 0.0234,
+      "step": 5309
+    },
+    {
+      "epoch": 79.25,
+      "learning_rate": 0.00016645614035087715,
+      "loss": 0.0023,
+      "step": 5310
+    },
+    {
+      "epoch": 79.27,
+      "learning_rate": 0.00016642105263157896,
+      "loss": 0.0031,
+      "step": 5311
+    },
+    {
+      "epoch": 79.28,
+      "learning_rate": 0.00016638596491228068,
+      "loss": 0.0033,
+      "step": 5312
+    },
+    {
+      "epoch": 79.3,
+      "learning_rate": 0.00016635087719298243,
+      "loss": 0.0009,
+      "step": 5313
+    },
+    {
+      "epoch": 79.31,
+      "learning_rate": 0.00016631578947368418,
+      "loss": 0.0379,
+      "step": 5314
+    },
+    {
+      "epoch": 79.33,
+      "learning_rate": 0.00016628070175438595,
+      "loss": 0.001,
+      "step": 5315
+    },
+    {
+      "epoch": 79.34,
+      "learning_rate": 0.0001662456140350877,
+      "loss": 0.0029,
+      "step": 5316
+    },
+    {
+      "epoch": 79.36,
+      "learning_rate": 0.00016621052631578945,
+      "loss": 0.0009,
+      "step": 5317
+    },
+    {
+      "epoch": 79.37,
+      "learning_rate": 0.00016617543859649123,
+      "loss": 0.0497,
+      "step": 5318
+    },
+    {
+      "epoch": 79.39,
+      "learning_rate": 0.00016614035087719297,
+      "loss": 0.0067,
+      "step": 5319
+    },
+    {
+      "epoch": 79.4,
+      "learning_rate": 0.00016610526315789472,
+      "loss": 0.001,
+      "step": 5320
+    },
+    {
+      "epoch": 79.42,
+      "learning_rate": 0.00016607017543859647,
+      "loss": 0.0771,
+      "step": 5321
+    },
+    {
+      "epoch": 79.43,
+      "learning_rate": 0.00016603508771929825,
+      "loss": 0.0084,
+      "step": 5322
+    },
+    {
+      "epoch": 79.45,
+      "learning_rate": 0.000166,
+      "loss": 0.0012,
+      "step": 5323
+    },
+    {
+      "epoch": 79.46,
+      "learning_rate": 0.00016596491228070175,
+      "loss": 0.0063,
+      "step": 5324
+    },
+    {
+      "epoch": 79.48,
+      "learning_rate": 0.00016592982456140347,
+      "loss": 0.0082,
+      "step": 5325
+    },
+    {
+      "epoch": 79.49,
+      "learning_rate": 0.00016589473684210527,
+      "loss": 0.0008,
+      "step": 5326
+    },
+    {
+      "epoch": 79.51,
+      "learning_rate": 0.000165859649122807,
+      "loss": 0.0708,
+      "step": 5327
+    },
+    {
+      "epoch": 79.52,
+      "learning_rate": 0.00016582456140350874,
+      "loss": 0.2766,
+      "step": 5328
+    },
+    {
+      "epoch": 79.54,
+      "learning_rate": 0.00016578947368421052,
+      "loss": 0.1932,
+      "step": 5329
+    },
+    {
+      "epoch": 79.55,
+      "learning_rate": 0.00016575438596491226,
+      "loss": 0.0024,
+      "step": 5330
+    },
+    {
+      "epoch": 79.57,
+      "learning_rate": 0.00016571929824561401,
+      "loss": 0.001,
+      "step": 5331
+    },
+    {
+      "epoch": 79.58,
+      "learning_rate": 0.00016568421052631576,
+      "loss": 0.026,
+      "step": 5332
+    },
+    {
+      "epoch": 79.59,
+      "learning_rate": 0.00016564912280701754,
+      "loss": 0.0014,
+      "step": 5333
+    },
+    {
+      "epoch": 79.61,
+      "learning_rate": 0.0001656140350877193,
+      "loss": 0.0027,
+      "step": 5334
+    },
+    {
+      "epoch": 79.62,
+      "learning_rate": 0.00016557894736842104,
+      "loss": 0.0038,
+      "step": 5335
+    },
+    {
+      "epoch": 79.64,
+      "learning_rate": 0.00016554385964912278,
+      "loss": 0.0042,
+      "step": 5336
+    },
+    {
+      "epoch": 79.65,
+      "learning_rate": 0.00016550877192982456,
+      "loss": 0.0009,
+      "step": 5337
+    },
+    {
+      "epoch": 79.67,
+      "learning_rate": 0.0001654736842105263,
+      "loss": 0.0024,
+      "step": 5338
+    },
+    {
+      "epoch": 79.68,
+      "learning_rate": 0.00016543859649122806,
+      "loss": 0.0009,
+      "step": 5339
+    },
+    {
+      "epoch": 79.7,
+      "learning_rate": 0.0001654035087719298,
+      "loss": 0.2441,
+      "step": 5340
+    },
+    {
+      "epoch": 79.71,
+      "learning_rate": 0.00016536842105263158,
+      "loss": 0.0387,
+      "step": 5341
+    },
+    {
+      "epoch": 79.73,
+      "learning_rate": 0.0001653333333333333,
+      "loss": 0.001,
+      "step": 5342
+    },
+    {
+      "epoch": 79.74,
+      "learning_rate": 0.00016529824561403505,
+      "loss": 0.0024,
+      "step": 5343
+    },
+    {
+      "epoch": 79.76,
+      "learning_rate": 0.00016526315789473683,
+      "loss": 0.0043,
+      "step": 5344
+    },
+    {
+      "epoch": 79.77,
+      "learning_rate": 0.00016522807017543858,
+      "loss": 0.0826,
+      "step": 5345
+    },
+    {
+      "epoch": 79.79,
+      "learning_rate": 0.00016519298245614033,
+      "loss": 0.0019,
+      "step": 5346
+    },
+    {
+      "epoch": 79.8,
+      "learning_rate": 0.00016515789473684208,
+      "loss": 0.0067,
+      "step": 5347
+    },
+    {
+      "epoch": 79.82,
+      "learning_rate": 0.00016512280701754385,
+      "loss": 0.1593,
+      "step": 5348
+    },
+    {
+      "epoch": 79.83,
+      "learning_rate": 0.0001650877192982456,
+      "loss": 0.0102,
+      "step": 5349
+    },
+    {
+      "epoch": 79.85,
+      "learning_rate": 0.00016505263157894735,
+      "loss": 0.0048,
+      "step": 5350
+    },
+    {
+      "epoch": 79.86,
+      "learning_rate": 0.0001650175438596491,
+      "loss": 0.0009,
+      "step": 5351
+    },
+    {
+      "epoch": 79.88,
+      "learning_rate": 0.00016498245614035087,
+      "loss": 0.1238,
+      "step": 5352
+    },
+    {
+      "epoch": 79.89,
+      "learning_rate": 0.00016494736842105262,
+      "loss": 0.0311,
+      "step": 5353
+    },
+    {
+      "epoch": 79.91,
+      "learning_rate": 0.00016491228070175437,
+      "loss": 0.0013,
+      "step": 5354
+    },
+    {
+      "epoch": 79.92,
+      "learning_rate": 0.00016487719298245615,
+      "loss": 0.0785,
+      "step": 5355
+    },
+    {
+      "epoch": 79.94,
+      "learning_rate": 0.0001648421052631579,
+      "loss": 0.2155,
+      "step": 5356
+    },
+    {
+      "epoch": 79.95,
+      "learning_rate": 0.00016480701754385964,
+      "loss": 0.2146,
+      "step": 5357
+    },
+    {
+      "epoch": 79.97,
+      "learning_rate": 0.00016477192982456137,
+      "loss": 0.1953,
+      "step": 5358
+    },
+    {
+      "epoch": 79.98,
+      "learning_rate": 0.00016473684210526317,
+      "loss": 0.0021,
+      "step": 5359
+    },
+    {
+      "epoch": 80.0,
+      "learning_rate": 0.0001647017543859649,
+      "loss": 0.0479,
+      "step": 5360
+    },
+    {
+      "epoch": 80.01,
+      "learning_rate": 0.00016466666666666664,
+      "loss": 0.0028,
+      "step": 5361
+    },
+    {
+      "epoch": 80.03,
+      "learning_rate": 0.0001646315789473684,
+      "loss": 0.001,
+      "step": 5362
+    },
+    {
+      "epoch": 80.04,
+      "learning_rate": 0.00016459649122807016,
+      "loss": 0.1739,
+      "step": 5363
+    },
+    {
+      "epoch": 80.06,
+      "learning_rate": 0.0001645614035087719,
+      "loss": 0.1624,
+      "step": 5364
+    },
+    {
+      "epoch": 80.07,
+      "learning_rate": 0.00016452631578947366,
+      "loss": 0.1035,
+      "step": 5365
+    },
+    {
+      "epoch": 80.09,
+      "learning_rate": 0.00016449122807017544,
+      "loss": 0.0018,
+      "step": 5366
+    },
+    {
+      "epoch": 80.1,
+      "learning_rate": 0.0001644561403508772,
+      "loss": 0.1251,
+      "step": 5367
+    },
+    {
+      "epoch": 80.12,
+      "learning_rate": 0.00016442105263157894,
+      "loss": 0.0059,
+      "step": 5368
+    },
+    {
+      "epoch": 80.13,
+      "learning_rate": 0.00016438596491228068,
+      "loss": 0.3782,
+      "step": 5369
+    },
+    {
+      "epoch": 80.15,
+      "learning_rate": 0.00016435087719298246,
+      "loss": 0.0023,
+      "step": 5370
+    },
+    {
+      "epoch": 80.16,
+      "learning_rate": 0.0001643157894736842,
+      "loss": 0.0096,
+      "step": 5371
+    },
+    {
+      "epoch": 80.18,
+      "learning_rate": 0.00016428070175438596,
+      "loss": 0.3213,
+      "step": 5372
+    },
+    {
+      "epoch": 80.19,
+      "learning_rate": 0.00016424561403508768,
+      "loss": 0.0009,
+      "step": 5373
+    },
+    {
+      "epoch": 80.21,
+      "learning_rate": 0.00016421052631578948,
+      "loss": 0.0033,
+      "step": 5374
+    },
+    {
+      "epoch": 80.22,
+      "learning_rate": 0.0001641754385964912,
+      "loss": 0.001,
+      "step": 5375
+    },
+    {
+      "epoch": 80.24,
+      "learning_rate": 0.00016414035087719295,
+      "loss": 0.0185,
+      "step": 5376
+    },
+    {
+      "epoch": 80.25,
+      "learning_rate": 0.0001641052631578947,
+      "loss": 0.0059,
+      "step": 5377
+    },
+    {
+      "epoch": 80.27,
+      "learning_rate": 0.00016407017543859648,
+      "loss": 0.0022,
+      "step": 5378
+    },
+    {
+      "epoch": 80.28,
+      "learning_rate": 0.00016403508771929823,
+      "loss": 0.0538,
+      "step": 5379
+    },
+    {
+      "epoch": 80.3,
+      "learning_rate": 0.00016399999999999997,
+      "loss": 0.0096,
+      "step": 5380
+    },
+    {
+      "epoch": 80.31,
+      "learning_rate": 0.00016396491228070175,
+      "loss": 0.0257,
+      "step": 5381
+    },
+    {
+      "epoch": 80.33,
+      "learning_rate": 0.0001639298245614035,
+      "loss": 0.0072,
+      "step": 5382
+    },
+    {
+      "epoch": 80.34,
+      "learning_rate": 0.00016389473684210525,
+      "loss": 0.1744,
+      "step": 5383
+    },
+    {
+      "epoch": 80.36,
+      "learning_rate": 0.000163859649122807,
+      "loss": 0.0038,
+      "step": 5384
+    },
+    {
+      "epoch": 80.37,
+      "learning_rate": 0.00016382456140350877,
+      "loss": 0.0013,
+      "step": 5385
+    },
+    {
+      "epoch": 80.39,
+      "learning_rate": 0.00016378947368421052,
+      "loss": 0.032,
+      "step": 5386
+    },
+    {
+      "epoch": 80.4,
+      "learning_rate": 0.00016375438596491227,
+      "loss": 0.001,
+      "step": 5387
+    },
+    {
+      "epoch": 80.42,
+      "learning_rate": 0.000163719298245614,
+      "loss": 0.0909,
+      "step": 5388
+    },
+    {
+      "epoch": 80.43,
+      "learning_rate": 0.0001636842105263158,
+      "loss": 0.0019,
+      "step": 5389
+    },
+    {
+      "epoch": 80.45,
+      "learning_rate": 0.00016364912280701752,
+      "loss": 0.0483,
+      "step": 5390
+    },
+    {
+      "epoch": 80.46,
+      "learning_rate": 0.00016361403508771927,
+      "loss": 0.2005,
+      "step": 5391
+    },
+    {
+      "epoch": 80.48,
+      "learning_rate": 0.00016357894736842104,
+      "loss": 0.0697,
+      "step": 5392
+    },
+    {
+      "epoch": 80.49,
+      "learning_rate": 0.0001635438596491228,
+      "loss": 0.001,
+      "step": 5393
+    },
+    {
+      "epoch": 80.51,
+      "learning_rate": 0.00016350877192982454,
+      "loss": 0.0186,
+      "step": 5394
+    },
+    {
+      "epoch": 80.52,
+      "learning_rate": 0.0001634736842105263,
+      "loss": 0.1185,
+      "step": 5395
+    },
+    {
+      "epoch": 80.54,
+      "learning_rate": 0.00016343859649122806,
+      "loss": 0.243,
+      "step": 5396
+    },
+    {
+      "epoch": 80.55,
+      "learning_rate": 0.0001634035087719298,
+      "loss": 0.0291,
+      "step": 5397
+    },
+    {
+      "epoch": 80.57,
+      "learning_rate": 0.00016336842105263156,
+      "loss": 0.0045,
+      "step": 5398
+    },
+    {
+      "epoch": 80.58,
+      "learning_rate": 0.0001633333333333333,
+      "loss": 0.0055,
+      "step": 5399
+    },
+    {
+      "epoch": 80.59,
+      "learning_rate": 0.00016329824561403509,
+      "loss": 0.0021,
+      "step": 5400
+    },
+    {
+      "epoch": 80.59,
+      "eval_accuracy": 0.8578071463534018,
+      "eval_f1": 0.8587756014640304,
+      "eval_loss": 0.6747012138366699,
+      "eval_runtime": 344.6737,
+      "eval_samples_per_second": 11.855,
+      "eval_steps_per_second": 0.743,
+      "step": 5400
+    },
+    {
+      "epoch": 80.61,
+      "learning_rate": 0.00016326315789473683,
+      "loss": 0.0016,
+      "step": 5401
+    },
+    {
+      "epoch": 80.62,
+      "learning_rate": 0.00016322807017543858,
+      "loss": 0.0357,
+      "step": 5402
+    },
+    {
+      "epoch": 80.64,
+      "learning_rate": 0.00016319298245614036,
+      "loss": 0.0355,
+      "step": 5403
+    },
+    {
+      "epoch": 80.65,
+      "learning_rate": 0.0001631578947368421,
+      "loss": 0.0025,
+      "step": 5404
+    },
+    {
+      "epoch": 80.67,
+      "learning_rate": 0.00016312280701754386,
+      "loss": 0.005,
+      "step": 5405
+    },
+    {
+      "epoch": 80.68,
+      "learning_rate": 0.00016308771929824558,
+      "loss": 0.002,
+      "step": 5406
+    },
+    {
+      "epoch": 80.7,
+      "learning_rate": 0.00016305263157894738,
+      "loss": 0.0022,
+      "step": 5407
+    },
+    {
+      "epoch": 80.71,
+      "learning_rate": 0.0001630175438596491,
+      "loss": 0.2577,
+      "step": 5408
+    },
+    {
+      "epoch": 80.73,
+      "learning_rate": 0.00016298245614035085,
+      "loss": 0.0013,
+      "step": 5409
+    },
+    {
+      "epoch": 80.74,
+      "learning_rate": 0.0001629473684210526,
+      "loss": 0.007,
+      "step": 5410
+    },
+    {
+      "epoch": 80.76,
+      "learning_rate": 0.00016291228070175438,
+      "loss": 0.0017,
+      "step": 5411
+    },
+    {
+      "epoch": 80.77,
+      "learning_rate": 0.00016287719298245613,
+      "loss": 0.0038,
+      "step": 5412
+    },
+    {
+      "epoch": 80.79,
+      "learning_rate": 0.00016284210526315787,
+      "loss": 0.2622,
+      "step": 5413
+    },
+    {
+      "epoch": 80.8,
+      "learning_rate": 0.00016280701754385962,
+      "loss": 0.0013,
+      "step": 5414
+    },
+    {
+      "epoch": 80.82,
+      "learning_rate": 0.0001627719298245614,
+      "loss": 0.1833,
+      "step": 5415
+    },
+    {
+      "epoch": 80.83,
+      "learning_rate": 0.00016273684210526315,
+      "loss": 0.066,
+      "step": 5416
+    },
+    {
+      "epoch": 80.85,
+      "learning_rate": 0.0001627017543859649,
+      "loss": 0.0032,
+      "step": 5417
+    },
+    {
+      "epoch": 80.86,
+      "learning_rate": 0.00016266666666666667,
+      "loss": 0.0973,
+      "step": 5418
+    },
+    {
+      "epoch": 80.88,
+      "learning_rate": 0.00016263157894736842,
+      "loss": 0.0088,
+      "step": 5419
+    },
+    {
+      "epoch": 80.89,
+      "learning_rate": 0.00016259649122807017,
+      "loss": 0.001,
+      "step": 5420
+    },
+    {
+      "epoch": 80.91,
+      "learning_rate": 0.0001625614035087719,
+      "loss": 0.1095,
+      "step": 5421
+    },
+    {
+      "epoch": 80.92,
+      "learning_rate": 0.0001625263157894737,
+      "loss": 0.0488,
+      "step": 5422
+    },
+    {
+      "epoch": 80.94,
+      "learning_rate": 0.00016249122807017542,
+      "loss": 0.0919,
+      "step": 5423
+    },
+    {
+      "epoch": 80.95,
+      "learning_rate": 0.00016245614035087716,
+      "loss": 0.1362,
+      "step": 5424
+    },
+    {
+      "epoch": 80.97,
+      "learning_rate": 0.00016242105263157891,
+      "loss": 0.0035,
+      "step": 5425
+    },
+    {
+      "epoch": 80.98,
+      "learning_rate": 0.0001623859649122807,
+      "loss": 0.0581,
+      "step": 5426
+    },
+    {
+      "epoch": 81.0,
+      "learning_rate": 0.00016235087719298244,
+      "loss": 0.2841,
+      "step": 5427
+    },
+    {
+      "epoch": 81.01,
+      "learning_rate": 0.0001623157894736842,
+      "loss": 0.3199,
+      "step": 5428
+    },
+    {
+      "epoch": 81.03,
+      "learning_rate": 0.00016228070175438596,
+      "loss": 0.0018,
+      "step": 5429
+    },
+    {
+      "epoch": 81.04,
+      "learning_rate": 0.0001622456140350877,
+      "loss": 0.1194,
+      "step": 5430
+    },
+    {
+      "epoch": 81.06,
+      "learning_rate": 0.00016221052631578946,
+      "loss": 0.2508,
+      "step": 5431
+    },
+    {
+      "epoch": 81.07,
+      "learning_rate": 0.0001621754385964912,
+      "loss": 0.0017,
+      "step": 5432
+    },
+    {
+      "epoch": 81.09,
+      "learning_rate": 0.00016214035087719299,
+      "loss": 0.0014,
+      "step": 5433
+    },
+    {
+      "epoch": 81.1,
+      "learning_rate": 0.00016210526315789473,
+      "loss": 0.0689,
+      "step": 5434
+    },
+    {
+      "epoch": 81.12,
+      "learning_rate": 0.00016207017543859648,
+      "loss": 0.0054,
+      "step": 5435
+    },
+    {
+      "epoch": 81.13,
+      "learning_rate": 0.0001620350877192982,
+      "loss": 0.0099,
+      "step": 5436
+    },
+    {
+      "epoch": 81.15,
+      "learning_rate": 0.000162,
+      "loss": 0.122,
+      "step": 5437
+    },
+    {
+      "epoch": 81.16,
+      "learning_rate": 0.00016196491228070173,
+      "loss": 0.0013,
+      "step": 5438
+    },
+    {
+      "epoch": 81.18,
+      "learning_rate": 0.00016192982456140348,
+      "loss": 0.0066,
+      "step": 5439
+    },
+    {
+      "epoch": 81.19,
+      "learning_rate": 0.00016189473684210523,
+      "loss": 0.013,
+      "step": 5440
+    },
+    {
+      "epoch": 81.21,
+      "learning_rate": 0.000161859649122807,
+      "loss": 0.1842,
+      "step": 5441
+    },
+    {
+      "epoch": 81.22,
+      "learning_rate": 0.00016182456140350875,
+      "loss": 0.0014,
+      "step": 5442
+    },
+    {
+      "epoch": 81.24,
+      "learning_rate": 0.0001617894736842105,
+      "loss": 0.0029,
+      "step": 5443
+    },
+    {
+      "epoch": 81.25,
+      "learning_rate": 0.00016175438596491228,
+      "loss": 0.001,
+      "step": 5444
+    },
+    {
+      "epoch": 81.27,
+      "learning_rate": 0.00016171929824561402,
+      "loss": 0.0012,
+      "step": 5445
+    },
+    {
+      "epoch": 81.28,
+      "learning_rate": 0.00016168421052631577,
+      "loss": 0.002,
+      "step": 5446
+    },
+    {
+      "epoch": 81.3,
+      "learning_rate": 0.00016164912280701752,
+      "loss": 0.004,
+      "step": 5447
+    },
+    {
+      "epoch": 81.31,
+      "learning_rate": 0.0001616140350877193,
+      "loss": 0.0059,
+      "step": 5448
+    },
+    {
+      "epoch": 81.33,
+      "learning_rate": 0.00016157894736842105,
+      "loss": 0.1543,
+      "step": 5449
+    },
+    {
+      "epoch": 81.34,
+      "learning_rate": 0.0001615438596491228,
+      "loss": 0.0173,
+      "step": 5450
+    },
+    {
+      "epoch": 81.36,
+      "learning_rate": 0.00016150877192982454,
+      "loss": 0.0082,
+      "step": 5451
+    },
+    {
+      "epoch": 81.37,
+      "learning_rate": 0.00016147368421052632,
+      "loss": 0.107,
+      "step": 5452
+    },
+    {
+      "epoch": 81.39,
+      "learning_rate": 0.00016143859649122807,
+      "loss": 0.001,
+      "step": 5453
+    },
+    {
+      "epoch": 81.4,
+      "learning_rate": 0.0001614035087719298,
+      "loss": 0.0013,
+      "step": 5454
+    },
+    {
+      "epoch": 81.42,
+      "learning_rate": 0.00016136842105263157,
+      "loss": 0.0149,
+      "step": 5455
+    },
+    {
+      "epoch": 81.43,
+      "learning_rate": 0.00016133333333333332,
+      "loss": 0.0018,
+      "step": 5456
+    },
+    {
+      "epoch": 81.45,
+      "learning_rate": 0.00016129824561403506,
+      "loss": 0.0016,
+      "step": 5457
+    },
+    {
+      "epoch": 81.46,
+      "learning_rate": 0.0001612631578947368,
+      "loss": 0.0064,
+      "step": 5458
+    },
+    {
+      "epoch": 81.48,
+      "learning_rate": 0.0001612280701754386,
+      "loss": 0.0017,
+      "step": 5459
+    },
+    {
+      "epoch": 81.49,
+      "learning_rate": 0.00016119298245614034,
+      "loss": 0.0033,
+      "step": 5460
+    },
+    {
+      "epoch": 81.51,
+      "learning_rate": 0.0001611578947368421,
+      "loss": 0.0052,
+      "step": 5461
+    },
+    {
+      "epoch": 81.52,
+      "learning_rate": 0.00016112280701754384,
+      "loss": 0.2818,
+      "step": 5462
+    },
+    {
+      "epoch": 81.54,
+      "learning_rate": 0.0001610877192982456,
+      "loss": 0.0061,
+      "step": 5463
+    },
+    {
+      "epoch": 81.55,
+      "learning_rate": 0.00016105263157894736,
+      "loss": 0.0035,
+      "step": 5464
+    },
+    {
+      "epoch": 81.57,
+      "learning_rate": 0.0001610175438596491,
+      "loss": 0.0019,
+      "step": 5465
+    },
+    {
+      "epoch": 81.58,
+      "learning_rate": 0.00016098245614035088,
+      "loss": 0.1041,
+      "step": 5466
+    },
+    {
+      "epoch": 81.59,
+      "learning_rate": 0.00016094736842105263,
+      "loss": 0.0522,
+      "step": 5467
+    },
+    {
+      "epoch": 81.61,
+      "learning_rate": 0.00016091228070175438,
+      "loss": 0.0039,
+      "step": 5468
+    },
+    {
+      "epoch": 81.62,
+      "learning_rate": 0.0001608771929824561,
+      "loss": 0.0032,
+      "step": 5469
+    },
+    {
+      "epoch": 81.64,
+      "learning_rate": 0.0001608421052631579,
+      "loss": 0.001,
+      "step": 5470
+    },
+    {
+      "epoch": 81.65,
+      "learning_rate": 0.00016080701754385963,
+      "loss": 0.0013,
+      "step": 5471
+    },
+    {
+      "epoch": 81.67,
+      "learning_rate": 0.00016077192982456138,
+      "loss": 0.0008,
+      "step": 5472
+    },
+    {
+      "epoch": 81.68,
+      "learning_rate": 0.00016073684210526313,
+      "loss": 0.0009,
+      "step": 5473
+    },
+    {
+      "epoch": 81.7,
+      "learning_rate": 0.0001607017543859649,
+      "loss": 0.0013,
+      "step": 5474
+    },
+    {
+      "epoch": 81.71,
+      "learning_rate": 0.00016066666666666665,
+      "loss": 0.001,
+      "step": 5475
+    },
+    {
+      "epoch": 81.73,
+      "learning_rate": 0.0001606315789473684,
+      "loss": 0.0009,
+      "step": 5476
+    },
+    {
+      "epoch": 81.74,
+      "learning_rate": 0.00016059649122807015,
+      "loss": 0.0008,
+      "step": 5477
+    },
+    {
+      "epoch": 81.76,
+      "learning_rate": 0.00016056140350877192,
+      "loss": 0.0014,
+      "step": 5478
+    },
+    {
+      "epoch": 81.77,
+      "learning_rate": 0.00016052631578947367,
+      "loss": 0.0065,
+      "step": 5479
+    },
+    {
+      "epoch": 81.79,
+      "learning_rate": 0.00016049122807017542,
+      "loss": 0.0279,
+      "step": 5480
+    },
+    {
+      "epoch": 81.8,
+      "learning_rate": 0.0001604561403508772,
+      "loss": 0.0008,
+      "step": 5481
+    },
+    {
+      "epoch": 81.82,
+      "learning_rate": 0.00016042105263157895,
+      "loss": 0.001,
+      "step": 5482
+    },
+    {
+      "epoch": 81.83,
+      "learning_rate": 0.0001603859649122807,
+      "loss": 0.0019,
+      "step": 5483
+    },
+    {
+      "epoch": 81.85,
+      "learning_rate": 0.00016035087719298242,
+      "loss": 0.0048,
+      "step": 5484
+    },
+    {
+      "epoch": 81.86,
+      "learning_rate": 0.00016031578947368422,
+      "loss": 0.0362,
+      "step": 5485
+    },
+    {
+      "epoch": 81.88,
+      "learning_rate": 0.00016028070175438594,
+      "loss": 0.009,
+      "step": 5486
+    },
+    {
+      "epoch": 81.89,
+      "learning_rate": 0.0001602456140350877,
+      "loss": 0.0007,
+      "step": 5487
+    },
+    {
+      "epoch": 81.91,
+      "learning_rate": 0.00016021052631578944,
+      "loss": 0.0149,
+      "step": 5488
+    },
+    {
+      "epoch": 81.92,
+      "learning_rate": 0.00016017543859649121,
+      "loss": 0.0784,
+      "step": 5489
+    },
+    {
+      "epoch": 81.94,
+      "learning_rate": 0.00016014035087719296,
+      "loss": 0.0009,
+      "step": 5490
+    },
+    {
+      "epoch": 81.95,
+      "learning_rate": 0.0001601052631578947,
+      "loss": 0.005,
+      "step": 5491
+    },
+    {
+      "epoch": 81.97,
+      "learning_rate": 0.0001600701754385965,
+      "loss": 0.0018,
+      "step": 5492
+    },
+    {
+      "epoch": 81.98,
+      "learning_rate": 0.00016003508771929824,
+      "loss": 0.0299,
+      "step": 5493
+    },
+    {
+      "epoch": 82.0,
+      "learning_rate": 0.00015999999999999999,
+      "loss": 0.2323,
+      "step": 5494
+    },
+    {
+      "epoch": 82.01,
+      "learning_rate": 0.00015996491228070173,
+      "loss": 0.0014,
+      "step": 5495
+    },
+    {
+      "epoch": 82.03,
+      "learning_rate": 0.0001599298245614035,
+      "loss": 0.0014,
+      "step": 5496
+    },
+    {
+      "epoch": 82.04,
+      "learning_rate": 0.00015989473684210526,
+      "loss": 0.0885,
+      "step": 5497
+    },
+    {
+      "epoch": 82.06,
+      "learning_rate": 0.000159859649122807,
+      "loss": 0.0017,
+      "step": 5498
+    },
+    {
+      "epoch": 82.07,
+      "learning_rate": 0.00015982456140350876,
+      "loss": 0.0009,
+      "step": 5499
+    },
+    {
+      "epoch": 82.09,
+      "learning_rate": 0.00015978947368421053,
+      "loss": 0.0017,
+      "step": 5500
+    },
+    {
+      "epoch": 82.1,
+      "learning_rate": 0.00015975438596491228,
+      "loss": 0.0016,
+      "step": 5501
+    },
+    {
+      "epoch": 82.12,
+      "learning_rate": 0.000159719298245614,
+      "loss": 0.1006,
+      "step": 5502
+    },
+    {
+      "epoch": 82.13,
+      "learning_rate": 0.00015968421052631575,
+      "loss": 0.0051,
+      "step": 5503
+    },
+    {
+      "epoch": 82.15,
+      "learning_rate": 0.00015964912280701753,
+      "loss": 0.0014,
+      "step": 5504
+    },
+    {
+      "epoch": 82.16,
+      "learning_rate": 0.00015961403508771928,
+      "loss": 0.0016,
+      "step": 5505
+    },
+    {
+      "epoch": 82.18,
+      "learning_rate": 0.00015957894736842103,
+      "loss": 0.0018,
+      "step": 5506
+    },
+    {
+      "epoch": 82.19,
+      "learning_rate": 0.0001595438596491228,
+      "loss": 0.0018,
+      "step": 5507
+    },
+    {
+      "epoch": 82.21,
+      "learning_rate": 0.00015950877192982455,
+      "loss": 0.0027,
+      "step": 5508
+    },
+    {
+      "epoch": 82.22,
+      "learning_rate": 0.0001594736842105263,
+      "loss": 0.0009,
+      "step": 5509
+    },
+    {
+      "epoch": 82.24,
+      "learning_rate": 0.00015943859649122805,
+      "loss": 0.0029,
+      "step": 5510
+    },
+    {
+      "epoch": 82.25,
+      "learning_rate": 0.00015940350877192982,
+      "loss": 0.0521,
+      "step": 5511
+    },
+    {
+      "epoch": 82.27,
+      "learning_rate": 0.00015936842105263157,
+      "loss": 0.019,
+      "step": 5512
+    },
+    {
+      "epoch": 82.28,
+      "learning_rate": 0.00015933333333333332,
+      "loss": 0.0095,
+      "step": 5513
+    },
+    {
+      "epoch": 82.3,
+      "learning_rate": 0.00015929824561403507,
+      "loss": 0.0915,
+      "step": 5514
+    },
+    {
+      "epoch": 82.31,
+      "learning_rate": 0.00015926315789473685,
+      "loss": 0.0017,
+      "step": 5515
+    },
+    {
+      "epoch": 82.33,
+      "learning_rate": 0.0001592280701754386,
+      "loss": 0.0024,
+      "step": 5516
+    },
+    {
+      "epoch": 82.34,
+      "learning_rate": 0.00015919298245614032,
+      "loss": 0.0162,
+      "step": 5517
+    },
+    {
+      "epoch": 82.36,
+      "learning_rate": 0.00015915789473684212,
+      "loss": 0.0009,
+      "step": 5518
+    },
+    {
+      "epoch": 82.37,
+      "learning_rate": 0.00015912280701754384,
+      "loss": 0.0009,
+      "step": 5519
+    },
+    {
+      "epoch": 82.39,
+      "learning_rate": 0.0001590877192982456,
+      "loss": 0.0008,
+      "step": 5520
+    },
+    {
+      "epoch": 82.4,
+      "learning_rate": 0.00015905263157894734,
+      "loss": 0.0008,
+      "step": 5521
+    },
+    {
+      "epoch": 82.42,
+      "learning_rate": 0.00015901754385964911,
+      "loss": 0.0081,
+      "step": 5522
+    },
+    {
+      "epoch": 82.43,
+      "learning_rate": 0.00015898245614035086,
+      "loss": 0.001,
+      "step": 5523
+    },
+    {
+      "epoch": 82.45,
+      "learning_rate": 0.0001589473684210526,
+      "loss": 0.0013,
+      "step": 5524
+    },
+    {
+      "epoch": 82.46,
+      "learning_rate": 0.00015891228070175436,
+      "loss": 0.0824,
+      "step": 5525
+    },
+    {
+      "epoch": 82.48,
+      "learning_rate": 0.00015887719298245614,
+      "loss": 0.0008,
+      "step": 5526
+    },
+    {
+      "epoch": 82.49,
+      "learning_rate": 0.00015884210526315789,
+      "loss": 0.0013,
+      "step": 5527
+    },
+    {
+      "epoch": 82.51,
+      "learning_rate": 0.00015880701754385963,
+      "loss": 0.0014,
+      "step": 5528
+    },
+    {
+      "epoch": 82.52,
+      "learning_rate": 0.0001587719298245614,
+      "loss": 0.0571,
+      "step": 5529
+    },
+    {
+      "epoch": 82.54,
+      "learning_rate": 0.00015873684210526316,
+      "loss": 0.0397,
+      "step": 5530
+    },
+    {
+      "epoch": 82.55,
+      "learning_rate": 0.0001587017543859649,
+      "loss": 0.0006,
+      "step": 5531
+    },
+    {
+      "epoch": 82.57,
+      "learning_rate": 0.00015866666666666663,
+      "loss": 0.106,
+      "step": 5532
+    },
+    {
+      "epoch": 82.58,
+      "learning_rate": 0.00015863157894736843,
+      "loss": 0.0012,
+      "step": 5533
+    },
+    {
+      "epoch": 82.59,
+      "learning_rate": 0.00015859649122807015,
+      "loss": 0.0023,
+      "step": 5534
+    },
+    {
+      "epoch": 82.61,
+      "learning_rate": 0.0001585614035087719,
+      "loss": 0.0007,
+      "step": 5535
+    },
+    {
+      "epoch": 82.62,
+      "learning_rate": 0.00015852631578947365,
+      "loss": 0.0007,
+      "step": 5536
+    },
+    {
+      "epoch": 82.64,
+      "learning_rate": 0.00015849122807017543,
+      "loss": 0.0006,
+      "step": 5537
+    },
+    {
+      "epoch": 82.65,
+      "learning_rate": 0.00015845614035087718,
+      "loss": 0.0947,
+      "step": 5538
+    },
+    {
+      "epoch": 82.67,
+      "learning_rate": 0.00015842105263157892,
+      "loss": 0.0804,
+      "step": 5539
+    },
+    {
+      "epoch": 82.68,
+      "learning_rate": 0.00015838596491228067,
+      "loss": 0.0006,
+      "step": 5540
+    },
+    {
+      "epoch": 82.7,
+      "learning_rate": 0.00015835087719298245,
+      "loss": 0.002,
+      "step": 5541
+    },
+    {
+      "epoch": 82.71,
+      "learning_rate": 0.0001583157894736842,
+      "loss": 0.0091,
+      "step": 5542
+    },
+    {
+      "epoch": 82.73,
+      "learning_rate": 0.00015828070175438595,
+      "loss": 0.0725,
+      "step": 5543
+    },
+    {
+      "epoch": 82.74,
+      "learning_rate": 0.00015824561403508772,
+      "loss": 0.0369,
+      "step": 5544
+    },
+    {
+      "epoch": 82.76,
+      "learning_rate": 0.00015821052631578947,
+      "loss": 0.0008,
+      "step": 5545
+    },
+    {
+      "epoch": 82.77,
+      "learning_rate": 0.00015817543859649122,
+      "loss": 0.2917,
+      "step": 5546
+    },
+    {
+      "epoch": 82.79,
+      "learning_rate": 0.00015814035087719297,
+      "loss": 0.3772,
+      "step": 5547
+    },
+    {
+      "epoch": 82.8,
+      "learning_rate": 0.00015810526315789475,
+      "loss": 0.0007,
+      "step": 5548
+    },
+    {
+      "epoch": 82.82,
+      "learning_rate": 0.00015807017543859647,
+      "loss": 0.1434,
+      "step": 5549
+    },
+    {
+      "epoch": 82.83,
+      "learning_rate": 0.00015803508771929822,
+      "loss": 0.0369,
+      "step": 5550
+    },
+    {
+      "epoch": 82.85,
+      "learning_rate": 0.00015799999999999996,
+      "loss": 0.0007,
+      "step": 5551
+    },
+    {
+      "epoch": 82.86,
+      "learning_rate": 0.00015796491228070174,
+      "loss": 0.0007,
+      "step": 5552
+    },
+    {
+      "epoch": 82.88,
+      "learning_rate": 0.0001579298245614035,
+      "loss": 0.0009,
+      "step": 5553
+    },
+    {
+      "epoch": 82.89,
+      "learning_rate": 0.00015789473684210524,
+      "loss": 0.0011,
+      "step": 5554
+    },
+    {
+      "epoch": 82.91,
+      "learning_rate": 0.00015785964912280701,
+      "loss": 0.0009,
+      "step": 5555
+    },
+    {
+      "epoch": 82.92,
+      "learning_rate": 0.00015782456140350876,
+      "loss": 0.0047,
+      "step": 5556
+    },
+    {
+      "epoch": 82.94,
+      "learning_rate": 0.0001577894736842105,
+      "loss": 0.0011,
+      "step": 5557
+    },
+    {
+      "epoch": 82.95,
+      "learning_rate": 0.00015775438596491226,
+      "loss": 0.0142,
+      "step": 5558
+    },
+    {
+      "epoch": 82.97,
+      "learning_rate": 0.00015771929824561404,
+      "loss": 0.0384,
+      "step": 5559
+    },
+    {
+      "epoch": 82.98,
+      "learning_rate": 0.00015768421052631578,
+      "loss": 0.0017,
+      "step": 5560
+    },
+    {
+      "epoch": 83.0,
+      "learning_rate": 0.00015764912280701753,
+      "loss": 0.0011,
+      "step": 5561
+    },
+    {
+      "epoch": 83.01,
+      "learning_rate": 0.00015761403508771928,
+      "loss": 0.0043,
+      "step": 5562
+    },
+    {
+      "epoch": 83.03,
+      "learning_rate": 0.00015757894736842106,
+      "loss": 0.0012,
+      "step": 5563
+    },
+    {
+      "epoch": 83.04,
+      "learning_rate": 0.0001575438596491228,
+      "loss": 0.1857,
+      "step": 5564
+    },
+    {
+      "epoch": 83.06,
+      "learning_rate": 0.00015750877192982453,
+      "loss": 0.1551,
+      "step": 5565
+    },
+    {
+      "epoch": 83.07,
+      "learning_rate": 0.00015747368421052633,
+      "loss": 0.0013,
+      "step": 5566
+    },
+    {
+      "epoch": 83.09,
+      "learning_rate": 0.00015743859649122805,
+      "loss": 0.0018,
+      "step": 5567
+    },
+    {
+      "epoch": 83.1,
+      "learning_rate": 0.0001574035087719298,
+      "loss": 0.0416,
+      "step": 5568
+    },
+    {
+      "epoch": 83.12,
+      "learning_rate": 0.00015736842105263155,
+      "loss": 0.0016,
+      "step": 5569
+    },
+    {
+      "epoch": 83.13,
+      "learning_rate": 0.00015733333333333333,
+      "loss": 0.0565,
+      "step": 5570
+    },
+    {
+      "epoch": 83.15,
+      "learning_rate": 0.00015729824561403508,
+      "loss": 0.0044,
+      "step": 5571
+    },
+    {
+      "epoch": 83.16,
+      "learning_rate": 0.00015726315789473682,
+      "loss": 0.0031,
+      "step": 5572
+    },
+    {
+      "epoch": 83.18,
+      "learning_rate": 0.00015722807017543857,
+      "loss": 0.0062,
+      "step": 5573
+    },
+    {
+      "epoch": 83.19,
+      "learning_rate": 0.00015719298245614035,
+      "loss": 0.0015,
+      "step": 5574
+    },
+    {
+      "epoch": 83.21,
+      "learning_rate": 0.0001571578947368421,
+      "loss": 0.0162,
+      "step": 5575
+    },
+    {
+      "epoch": 83.22,
+      "learning_rate": 0.00015712280701754385,
+      "loss": 0.0023,
+      "step": 5576
+    },
+    {
+      "epoch": 83.24,
+      "learning_rate": 0.0001570877192982456,
+      "loss": 0.1509,
+      "step": 5577
+    },
+    {
+      "epoch": 83.25,
+      "learning_rate": 0.00015705263157894737,
+      "loss": 0.0032,
+      "step": 5578
+    },
+    {
+      "epoch": 83.27,
+      "learning_rate": 0.00015701754385964912,
+      "loss": 0.0601,
+      "step": 5579
+    },
+    {
+      "epoch": 83.28,
+      "learning_rate": 0.00015698245614035084,
+      "loss": 0.0017,
+      "step": 5580
+    },
+    {
+      "epoch": 83.3,
+      "learning_rate": 0.00015694736842105264,
+      "loss": 0.0017,
+      "step": 5581
+    },
+    {
+      "epoch": 83.31,
+      "learning_rate": 0.00015691228070175437,
+      "loss": 0.0074,
+      "step": 5582
+    },
+    {
+      "epoch": 83.33,
+      "learning_rate": 0.00015687719298245612,
+      "loss": 0.2237,
+      "step": 5583
+    },
+    {
+      "epoch": 83.34,
+      "learning_rate": 0.00015684210526315786,
+      "loss": 0.0026,
+      "step": 5584
+    },
+    {
+      "epoch": 83.36,
+      "learning_rate": 0.00015680701754385964,
+      "loss": 0.0321,
+      "step": 5585
+    },
+    {
+      "epoch": 83.37,
+      "learning_rate": 0.0001567719298245614,
+      "loss": 0.0013,
+      "step": 5586
+    },
+    {
+      "epoch": 83.39,
+      "learning_rate": 0.00015673684210526314,
+      "loss": 0.0012,
+      "step": 5587
+    },
+    {
+      "epoch": 83.4,
+      "learning_rate": 0.00015670175438596489,
+      "loss": 0.0011,
+      "step": 5588
+    },
+    {
+      "epoch": 83.42,
+      "learning_rate": 0.00015666666666666666,
+      "loss": 0.0089,
+      "step": 5589
+    },
+    {
+      "epoch": 83.43,
+      "learning_rate": 0.0001566315789473684,
+      "loss": 0.0007,
+      "step": 5590
+    },
+    {
+      "epoch": 83.45,
+      "learning_rate": 0.00015659649122807016,
+      "loss": 0.0008,
+      "step": 5591
+    },
+    {
+      "epoch": 83.46,
+      "learning_rate": 0.00015656140350877194,
+      "loss": 0.0024,
+      "step": 5592
+    },
+    {
+      "epoch": 83.48,
+      "learning_rate": 0.00015652631578947368,
+      "loss": 0.0008,
+      "step": 5593
+    },
+    {
+      "epoch": 83.49,
+      "learning_rate": 0.00015649122807017543,
+      "loss": 0.0019,
+      "step": 5594
+    },
+    {
+      "epoch": 83.51,
+      "learning_rate": 0.00015645614035087718,
+      "loss": 0.2801,
+      "step": 5595
+    },
+    {
+      "epoch": 83.52,
+      "learning_rate": 0.00015642105263157896,
+      "loss": 0.0012,
+      "step": 5596
+    },
+    {
+      "epoch": 83.54,
+      "learning_rate": 0.00015638596491228068,
+      "loss": 0.0034,
+      "step": 5597
+    },
+    {
+      "epoch": 83.55,
+      "learning_rate": 0.00015635087719298243,
+      "loss": 0.0846,
+      "step": 5598
+    },
+    {
+      "epoch": 83.57,
+      "learning_rate": 0.00015631578947368418,
+      "loss": 0.0549,
+      "step": 5599
+    },
+    {
+      "epoch": 83.58,
+      "learning_rate": 0.00015628070175438595,
+      "loss": 0.2558,
+      "step": 5600
+    },
+    {
+      "epoch": 83.58,
+      "eval_accuracy": 0.8580518844836026,
+      "eval_f1": 0.8581217984766242,
+      "eval_loss": 0.7513839602470398,
+      "eval_runtime": 345.0011,
+      "eval_samples_per_second": 11.843,
+      "eval_steps_per_second": 0.742,
+      "step": 5600
+    },
+    {
+      "epoch": 83.59,
+      "learning_rate": 0.0001562456140350877,
+      "loss": 0.0009,
+      "step": 5601
+    },
+    {
+      "epoch": 83.61,
+      "learning_rate": 0.00015621052631578945,
+      "loss": 0.0996,
+      "step": 5602
+    },
+    {
+      "epoch": 83.62,
+      "learning_rate": 0.0001561754385964912,
+      "loss": 0.0023,
+      "step": 5603
+    },
+    {
+      "epoch": 83.64,
+      "learning_rate": 0.00015614035087719297,
+      "loss": 0.0013,
+      "step": 5604
+    },
+    {
+      "epoch": 83.65,
+      "learning_rate": 0.00015610526315789472,
+      "loss": 0.0011,
+      "step": 5605
+    },
+    {
+      "epoch": 83.67,
+      "learning_rate": 0.00015607017543859647,
+      "loss": 0.0021,
+      "step": 5606
+    },
+    {
+      "epoch": 83.68,
+      "learning_rate": 0.00015603508771929825,
+      "loss": 0.267,
+      "step": 5607
+    },
+    {
+      "epoch": 83.7,
+      "learning_rate": 0.000156,
+      "loss": 0.0036,
+      "step": 5608
+    },
+    {
+      "epoch": 83.71,
+      "learning_rate": 0.00015596491228070175,
+      "loss": 0.0013,
+      "step": 5609
+    },
+    {
+      "epoch": 83.73,
+      "learning_rate": 0.0001559298245614035,
+      "loss": 0.0077,
+      "step": 5610
+    },
+    {
+      "epoch": 83.74,
+      "learning_rate": 0.00015589473684210527,
+      "loss": 0.0034,
+      "step": 5611
+    },
+    {
+      "epoch": 83.76,
+      "learning_rate": 0.00015585964912280702,
+      "loss": 0.0024,
+      "step": 5612
+    },
+    {
+      "epoch": 83.77,
+      "learning_rate": 0.00015582456140350874,
+      "loss": 0.1785,
+      "step": 5613
+    },
+    {
+      "epoch": 83.79,
+      "learning_rate": 0.0001557894736842105,
+      "loss": 0.0032,
+      "step": 5614
+    },
+    {
+      "epoch": 83.8,
+      "learning_rate": 0.00015575438596491227,
+      "loss": 0.0021,
+      "step": 5615
+    },
+    {
+      "epoch": 83.82,
+      "learning_rate": 0.00015571929824561401,
+      "loss": 0.4422,
+      "step": 5616
+    },
+    {
+      "epoch": 83.83,
+      "learning_rate": 0.00015568421052631576,
+      "loss": 0.0062,
+      "step": 5617
+    },
+    {
+      "epoch": 83.85,
+      "learning_rate": 0.00015564912280701754,
+      "loss": 0.3256,
+      "step": 5618
+    },
+    {
+      "epoch": 83.86,
+      "learning_rate": 0.0001556140350877193,
+      "loss": 0.0058,
+      "step": 5619
+    },
+    {
+      "epoch": 83.88,
+      "learning_rate": 0.00015557894736842104,
+      "loss": 0.0015,
+      "step": 5620
+    },
+    {
+      "epoch": 83.89,
+      "learning_rate": 0.00015554385964912279,
+      "loss": 0.0021,
+      "step": 5621
+    },
+    {
+      "epoch": 83.91,
+      "learning_rate": 0.00015550877192982456,
+      "loss": 0.3244,
+      "step": 5622
+    },
+    {
+      "epoch": 83.92,
+      "learning_rate": 0.0001554736842105263,
+      "loss": 0.002,
+      "step": 5623
+    },
+    {
+      "epoch": 83.94,
+      "learning_rate": 0.00015543859649122806,
+      "loss": 0.0073,
+      "step": 5624
+    },
+    {
+      "epoch": 83.95,
+      "learning_rate": 0.0001554035087719298,
+      "loss": 0.034,
+      "step": 5625
+    },
+    {
+      "epoch": 83.97,
+      "learning_rate": 0.00015536842105263158,
+      "loss": 0.0015,
+      "step": 5626
+    },
+    {
+      "epoch": 83.98,
+      "learning_rate": 0.00015533333333333333,
+      "loss": 0.0368,
+      "step": 5627
+    },
+    {
+      "epoch": 84.0,
+      "learning_rate": 0.00015529824561403505,
+      "loss": 0.0769,
+      "step": 5628
+    },
+    {
+      "epoch": 84.01,
+      "learning_rate": 0.00015526315789473686,
+      "loss": 0.0027,
+      "step": 5629
+    },
+    {
+      "epoch": 84.03,
+      "learning_rate": 0.00015522807017543858,
+      "loss": 0.1014,
+      "step": 5630
+    },
+    {
+      "epoch": 84.04,
+      "learning_rate": 0.00015519298245614033,
+      "loss": 0.0016,
+      "step": 5631
+    },
+    {
+      "epoch": 84.06,
+      "learning_rate": 0.00015515789473684208,
+      "loss": 0.0019,
+      "step": 5632
+    },
+    {
+      "epoch": 84.07,
+      "learning_rate": 0.00015512280701754385,
+      "loss": 0.0227,
+      "step": 5633
+    },
+    {
+      "epoch": 84.09,
+      "learning_rate": 0.0001550877192982456,
+      "loss": 0.0014,
+      "step": 5634
+    },
+    {
+      "epoch": 84.1,
+      "learning_rate": 0.00015505263157894735,
+      "loss": 0.0755,
+      "step": 5635
+    },
+    {
+      "epoch": 84.12,
+      "learning_rate": 0.0001550175438596491,
+      "loss": 0.0014,
+      "step": 5636
+    },
+    {
+      "epoch": 84.13,
+      "learning_rate": 0.00015498245614035087,
+      "loss": 0.0031,
+      "step": 5637
+    },
+    {
+      "epoch": 84.15,
+      "learning_rate": 0.00015494736842105262,
+      "loss": 0.107,
+      "step": 5638
+    },
+    {
+      "epoch": 84.16,
+      "learning_rate": 0.00015491228070175437,
+      "loss": 0.3357,
+      "step": 5639
+    },
+    {
+      "epoch": 84.18,
+      "learning_rate": 0.00015487719298245612,
+      "loss": 0.0045,
+      "step": 5640
+    },
+    {
+      "epoch": 84.19,
+      "learning_rate": 0.0001548421052631579,
+      "loss": 0.0018,
+      "step": 5641
+    },
+    {
+      "epoch": 84.21,
+      "learning_rate": 0.00015480701754385965,
+      "loss": 0.0163,
+      "step": 5642
+    },
+    {
+      "epoch": 84.22,
+      "learning_rate": 0.00015477192982456137,
+      "loss": 0.3467,
+      "step": 5643
+    },
+    {
+      "epoch": 84.24,
+      "learning_rate": 0.00015473684210526317,
+      "loss": 0.0168,
+      "step": 5644
+    },
+    {
+      "epoch": 84.25,
+      "learning_rate": 0.0001547017543859649,
+      "loss": 0.0943,
+      "step": 5645
+    },
+    {
+      "epoch": 84.27,
+      "learning_rate": 0.00015466666666666664,
+      "loss": 0.002,
+      "step": 5646
+    },
+    {
+      "epoch": 84.28,
+      "learning_rate": 0.0001546315789473684,
+      "loss": 0.193,
+      "step": 5647
+    },
+    {
+      "epoch": 84.3,
+      "learning_rate": 0.00015459649122807017,
+      "loss": 0.0045,
+      "step": 5648
+    },
+    {
+      "epoch": 84.31,
+      "learning_rate": 0.00015456140350877191,
+      "loss": 0.0017,
+      "step": 5649
+    },
+    {
+      "epoch": 84.33,
+      "learning_rate": 0.00015452631578947366,
+      "loss": 0.4212,
+      "step": 5650
+    },
+    {
+      "epoch": 84.34,
+      "learning_rate": 0.0001544912280701754,
+      "loss": 0.0033,
+      "step": 5651
+    },
+    {
+      "epoch": 84.36,
+      "learning_rate": 0.0001544561403508772,
+      "loss": 0.0022,
+      "step": 5652
+    },
+    {
+      "epoch": 84.37,
+      "learning_rate": 0.00015442105263157894,
+      "loss": 0.0021,
+      "step": 5653
+    },
+    {
+      "epoch": 84.39,
+      "learning_rate": 0.00015438596491228068,
+      "loss": 0.0087,
+      "step": 5654
+    },
+    {
+      "epoch": 84.4,
+      "learning_rate": 0.00015435087719298246,
+      "loss": 0.0464,
+      "step": 5655
+    },
+    {
+      "epoch": 84.42,
+      "learning_rate": 0.0001543157894736842,
+      "loss": 0.0063,
+      "step": 5656
+    },
+    {
+      "epoch": 84.43,
+      "learning_rate": 0.00015428070175438596,
+      "loss": 0.0176,
+      "step": 5657
+    },
+    {
+      "epoch": 84.45,
+      "learning_rate": 0.0001542456140350877,
+      "loss": 0.3616,
+      "step": 5658
+    },
+    {
+      "epoch": 84.46,
+      "learning_rate": 0.00015421052631578948,
+      "loss": 0.0088,
+      "step": 5659
+    },
+    {
+      "epoch": 84.48,
+      "learning_rate": 0.00015417543859649123,
+      "loss": 0.0016,
+      "step": 5660
+    },
+    {
+      "epoch": 84.49,
+      "learning_rate": 0.00015414035087719295,
+      "loss": 0.0014,
+      "step": 5661
+    },
+    {
+      "epoch": 84.51,
+      "learning_rate": 0.0001541052631578947,
+      "loss": 0.0018,
+      "step": 5662
+    },
+    {
+      "epoch": 84.52,
+      "learning_rate": 0.00015407017543859648,
+      "loss": 0.0014,
+      "step": 5663
+    },
+    {
+      "epoch": 84.54,
+      "learning_rate": 0.00015403508771929823,
+      "loss": 0.0066,
+      "step": 5664
+    },
+    {
+      "epoch": 84.55,
+      "learning_rate": 0.00015399999999999998,
+      "loss": 0.0012,
+      "step": 5665
+    },
+    {
+      "epoch": 84.57,
+      "learning_rate": 0.00015396491228070172,
+      "loss": 0.0026,
+      "step": 5666
+    },
+    {
+      "epoch": 84.58,
+      "learning_rate": 0.0001539298245614035,
+      "loss": 0.0016,
+      "step": 5667
+    },
+    {
+      "epoch": 84.59,
+      "learning_rate": 0.00015389473684210525,
+      "loss": 0.0063,
+      "step": 5668
+    },
+    {
+      "epoch": 84.61,
+      "learning_rate": 0.000153859649122807,
+      "loss": 0.0022,
+      "step": 5669
+    },
+    {
+      "epoch": 84.62,
+      "learning_rate": 0.00015382456140350877,
+      "loss": 0.0018,
+      "step": 5670
+    },
+    {
+      "epoch": 84.64,
+      "learning_rate": 0.00015378947368421052,
+      "loss": 0.1457,
+      "step": 5671
+    },
+    {
+      "epoch": 84.65,
+      "learning_rate": 0.00015375438596491227,
+      "loss": 0.0018,
+      "step": 5672
+    },
+    {
+      "epoch": 84.67,
+      "learning_rate": 0.00015371929824561402,
+      "loss": 0.0831,
+      "step": 5673
+    },
+    {
+      "epoch": 84.68,
+      "learning_rate": 0.0001536842105263158,
+      "loss": 0.0848,
+      "step": 5674
+    },
+    {
+      "epoch": 84.7,
+      "learning_rate": 0.00015364912280701754,
+      "loss": 0.1043,
+      "step": 5675
+    },
+    {
+      "epoch": 84.71,
+      "learning_rate": 0.00015361403508771927,
+      "loss": 0.01,
+      "step": 5676
+    },
+    {
+      "epoch": 84.73,
+      "learning_rate": 0.00015357894736842102,
+      "loss": 0.0043,
+      "step": 5677
+    },
+    {
+      "epoch": 84.74,
+      "learning_rate": 0.0001535438596491228,
+      "loss": 0.0017,
+      "step": 5678
+    },
+    {
+      "epoch": 84.76,
+      "learning_rate": 0.00015350877192982454,
+      "loss": 0.0034,
+      "step": 5679
+    },
+    {
+      "epoch": 84.77,
+      "learning_rate": 0.0001534736842105263,
+      "loss": 0.0042,
+      "step": 5680
+    },
+    {
+      "epoch": 84.79,
+      "learning_rate": 0.00015343859649122806,
+      "loss": 0.0725,
+      "step": 5681
+    },
+    {
+      "epoch": 84.8,
+      "learning_rate": 0.0001534035087719298,
+      "loss": 0.0264,
+      "step": 5682
+    },
+    {
+      "epoch": 84.82,
+      "learning_rate": 0.00015336842105263156,
+      "loss": 0.0028,
+      "step": 5683
+    },
+    {
+      "epoch": 84.83,
+      "learning_rate": 0.0001533333333333333,
+      "loss": 0.6024,
+      "step": 5684
+    },
+    {
+      "epoch": 84.85,
+      "learning_rate": 0.0001532982456140351,
+      "loss": 0.0025,
+      "step": 5685
+    },
+    {
+      "epoch": 84.86,
+      "learning_rate": 0.00015326315789473684,
+      "loss": 0.0011,
+      "step": 5686
+    },
+    {
+      "epoch": 84.88,
+      "learning_rate": 0.00015322807017543858,
+      "loss": 0.0021,
+      "step": 5687
+    },
+    {
+      "epoch": 84.89,
+      "learning_rate": 0.00015319298245614033,
+      "loss": 0.001,
+      "step": 5688
+    },
+    {
+      "epoch": 84.91,
+      "learning_rate": 0.0001531578947368421,
+      "loss": 0.0023,
+      "step": 5689
+    },
+    {
+      "epoch": 84.92,
+      "learning_rate": 0.00015312280701754386,
+      "loss": 0.0969,
+      "step": 5690
+    },
+    {
+      "epoch": 84.94,
+      "learning_rate": 0.00015308771929824558,
+      "loss": 0.0009,
+      "step": 5691
+    },
+    {
+      "epoch": 84.95,
+      "learning_rate": 0.00015305263157894738,
+      "loss": 0.0018,
+      "step": 5692
+    },
+    {
+      "epoch": 84.97,
+      "learning_rate": 0.0001530175438596491,
+      "loss": 0.0348,
+      "step": 5693
+    },
+    {
+      "epoch": 84.98,
+      "learning_rate": 0.00015298245614035085,
+      "loss": 0.006,
+      "step": 5694
+    },
+    {
+      "epoch": 85.0,
+      "learning_rate": 0.0001529473684210526,
+      "loss": 0.0012,
+      "step": 5695
+    },
+    {
+      "epoch": 85.01,
+      "learning_rate": 0.00015291228070175438,
+      "loss": 0.0085,
+      "step": 5696
+    },
+    {
+      "epoch": 85.03,
+      "learning_rate": 0.00015287719298245613,
+      "loss": 0.0019,
+      "step": 5697
+    },
+    {
+      "epoch": 85.04,
+      "learning_rate": 0.00015284210526315788,
+      "loss": 0.0055,
+      "step": 5698
+    },
+    {
+      "epoch": 85.06,
+      "learning_rate": 0.00015280701754385962,
+      "loss": 0.1146,
+      "step": 5699
+    },
+    {
+      "epoch": 85.07,
+      "learning_rate": 0.0001527719298245614,
+      "loss": 0.0261,
+      "step": 5700
+    },
+    {
+      "epoch": 85.09,
+      "learning_rate": 0.00015273684210526315,
+      "loss": 0.0082,
+      "step": 5701
+    },
+    {
+      "epoch": 85.1,
+      "learning_rate": 0.0001527017543859649,
+      "loss": 0.001,
+      "step": 5702
+    },
+    {
+      "epoch": 85.12,
+      "learning_rate": 0.00015266666666666665,
+      "loss": 0.0011,
+      "step": 5703
+    },
+    {
+      "epoch": 85.13,
+      "learning_rate": 0.00015263157894736842,
+      "loss": 0.0084,
+      "step": 5704
+    },
+    {
+      "epoch": 85.15,
+      "learning_rate": 0.00015259649122807017,
+      "loss": 0.005,
+      "step": 5705
+    },
+    {
+      "epoch": 85.16,
+      "learning_rate": 0.00015256140350877192,
+      "loss": 0.0011,
+      "step": 5706
+    },
+    {
+      "epoch": 85.18,
+      "learning_rate": 0.0001525263157894737,
+      "loss": 0.0007,
+      "step": 5707
+    },
+    {
+      "epoch": 85.19,
+      "learning_rate": 0.00015249122807017544,
+      "loss": 0.0018,
+      "step": 5708
+    },
+    {
+      "epoch": 85.21,
+      "learning_rate": 0.00015245614035087717,
+      "loss": 0.0009,
+      "step": 5709
+    },
+    {
+      "epoch": 85.22,
+      "learning_rate": 0.00015242105263157891,
+      "loss": 0.001,
+      "step": 5710
+    },
+    {
+      "epoch": 85.24,
+      "learning_rate": 0.0001523859649122807,
+      "loss": 0.0112,
+      "step": 5711
+    },
+    {
+      "epoch": 85.25,
+      "learning_rate": 0.00015235087719298244,
+      "loss": 0.0023,
+      "step": 5712
+    },
+    {
+      "epoch": 85.27,
+      "learning_rate": 0.0001523157894736842,
+      "loss": 0.0017,
+      "step": 5713
+    },
+    {
+      "epoch": 85.28,
+      "learning_rate": 0.00015228070175438594,
+      "loss": 0.0008,
+      "step": 5714
+    },
+    {
+      "epoch": 85.3,
+      "learning_rate": 0.0001522456140350877,
+      "loss": 0.0268,
+      "step": 5715
+    },
+    {
+      "epoch": 85.31,
+      "learning_rate": 0.00015221052631578946,
+      "loss": 0.0033,
+      "step": 5716
+    },
+    {
+      "epoch": 85.33,
+      "learning_rate": 0.0001521754385964912,
+      "loss": 0.0422,
+      "step": 5717
+    },
+    {
+      "epoch": 85.34,
+      "learning_rate": 0.00015214035087719299,
+      "loss": 0.0048,
+      "step": 5718
+    },
+    {
+      "epoch": 85.36,
+      "learning_rate": 0.00015210526315789473,
+      "loss": 0.005,
+      "step": 5719
+    },
+    {
+      "epoch": 85.37,
+      "learning_rate": 0.00015207017543859648,
+      "loss": 0.0054,
+      "step": 5720
+    },
+    {
+      "epoch": 85.39,
+      "learning_rate": 0.00015203508771929823,
+      "loss": 0.1974,
+      "step": 5721
+    },
+    {
+      "epoch": 85.4,
+      "learning_rate": 0.000152,
+      "loss": 0.0008,
+      "step": 5722
+    },
+    {
+      "epoch": 85.42,
+      "learning_rate": 0.00015196491228070176,
+      "loss": 0.0007,
+      "step": 5723
+    },
+    {
+      "epoch": 85.43,
+      "learning_rate": 0.00015192982456140348,
+      "loss": 0.0016,
+      "step": 5724
+    },
+    {
+      "epoch": 85.45,
+      "learning_rate": 0.00015189473684210523,
+      "loss": 0.0007,
+      "step": 5725
+    },
+    {
+      "epoch": 85.46,
+      "learning_rate": 0.000151859649122807,
+      "loss": 0.0115,
+      "step": 5726
+    },
+    {
+      "epoch": 85.48,
+      "learning_rate": 0.00015182456140350875,
+      "loss": 0.0009,
+      "step": 5727
+    },
+    {
+      "epoch": 85.49,
+      "learning_rate": 0.0001517894736842105,
+      "loss": 0.0008,
+      "step": 5728
+    },
+    {
+      "epoch": 85.51,
+      "learning_rate": 0.00015175438596491228,
+      "loss": 0.001,
+      "step": 5729
+    },
+    {
+      "epoch": 85.52,
+      "learning_rate": 0.00015171929824561403,
+      "loss": 0.0024,
+      "step": 5730
+    },
+    {
+      "epoch": 85.54,
+      "learning_rate": 0.00015168421052631577,
+      "loss": 0.0352,
+      "step": 5731
+    },
+    {
+      "epoch": 85.55,
+      "learning_rate": 0.00015164912280701752,
+      "loss": 0.0773,
+      "step": 5732
+    },
+    {
+      "epoch": 85.57,
+      "learning_rate": 0.0001516140350877193,
+      "loss": 0.0067,
+      "step": 5733
+    },
+    {
+      "epoch": 85.58,
+      "learning_rate": 0.00015157894736842105,
+      "loss": 0.001,
+      "step": 5734
+    },
+    {
+      "epoch": 85.59,
+      "learning_rate": 0.0001515438596491228,
+      "loss": 0.0658,
+      "step": 5735
+    },
+    {
+      "epoch": 85.61,
+      "learning_rate": 0.00015150877192982455,
+      "loss": 0.0008,
+      "step": 5736
+    },
+    {
+      "epoch": 85.62,
+      "learning_rate": 0.00015147368421052632,
+      "loss": 0.0021,
+      "step": 5737
+    },
+    {
+      "epoch": 85.64,
+      "learning_rate": 0.00015143859649122807,
+      "loss": 0.0005,
+      "step": 5738
+    },
+    {
+      "epoch": 85.65,
+      "learning_rate": 0.0001514035087719298,
+      "loss": 0.0008,
+      "step": 5739
+    },
+    {
+      "epoch": 85.67,
+      "learning_rate": 0.00015136842105263154,
+      "loss": 0.0009,
+      "step": 5740
+    },
+    {
+      "epoch": 85.68,
+      "learning_rate": 0.00015133333333333332,
+      "loss": 0.0006,
+      "step": 5741
+    },
+    {
+      "epoch": 85.7,
+      "learning_rate": 0.00015129824561403507,
+      "loss": 0.0015,
+      "step": 5742
+    },
+    {
+      "epoch": 85.71,
+      "learning_rate": 0.00015126315789473681,
+      "loss": 0.0008,
+      "step": 5743
+    },
+    {
+      "epoch": 85.73,
+      "learning_rate": 0.0001512280701754386,
+      "loss": 0.0008,
+      "step": 5744
+    },
+    {
+      "epoch": 85.74,
+      "learning_rate": 0.00015119298245614034,
+      "loss": 0.0009,
+      "step": 5745
+    },
+    {
+      "epoch": 85.76,
+      "learning_rate": 0.0001511578947368421,
+      "loss": 0.0076,
+      "step": 5746
+    },
+    {
+      "epoch": 85.77,
+      "learning_rate": 0.00015112280701754384,
+      "loss": 0.0253,
+      "step": 5747
+    },
+    {
+      "epoch": 85.79,
+      "learning_rate": 0.0001510877192982456,
+      "loss": 0.0017,
+      "step": 5748
+    },
+    {
+      "epoch": 85.8,
+      "learning_rate": 0.00015105263157894736,
+      "loss": 0.0044,
+      "step": 5749
+    },
+    {
+      "epoch": 85.82,
+      "learning_rate": 0.0001510175438596491,
+      "loss": 0.0056,
+      "step": 5750
+    },
+    {
+      "epoch": 85.83,
+      "learning_rate": 0.00015098245614035086,
+      "loss": 0.0009,
+      "step": 5751
+    },
+    {
+      "epoch": 85.85,
+      "learning_rate": 0.00015094736842105263,
+      "loss": 0.0039,
+      "step": 5752
+    },
+    {
+      "epoch": 85.86,
+      "learning_rate": 0.00015091228070175438,
+      "loss": 0.0027,
+      "step": 5753
+    },
+    {
+      "epoch": 85.88,
+      "learning_rate": 0.00015087719298245613,
+      "loss": 0.0222,
+      "step": 5754
+    },
+    {
+      "epoch": 85.89,
+      "learning_rate": 0.0001508421052631579,
+      "loss": 0.1151,
+      "step": 5755
+    },
+    {
+      "epoch": 85.91,
+      "learning_rate": 0.00015080701754385963,
+      "loss": 0.322,
+      "step": 5756
+    },
+    {
+      "epoch": 85.92,
+      "learning_rate": 0.00015077192982456138,
+      "loss": 0.0006,
+      "step": 5757
+    },
+    {
+      "epoch": 85.94,
+      "learning_rate": 0.00015073684210526313,
+      "loss": 0.0007,
+      "step": 5758
+    },
+    {
+      "epoch": 85.95,
+      "learning_rate": 0.0001507017543859649,
+      "loss": 0.0091,
+      "step": 5759
+    },
+    {
+      "epoch": 85.97,
+      "learning_rate": 0.00015066666666666665,
+      "loss": 0.1359,
+      "step": 5760
+    },
+    {
+      "epoch": 85.98,
+      "learning_rate": 0.0001506315789473684,
+      "loss": 0.0006,
+      "step": 5761
+    },
+    {
+      "epoch": 86.0,
+      "learning_rate": 0.00015059649122807015,
+      "loss": 0.0034,
+      "step": 5762
+    },
+    {
+      "epoch": 86.01,
+      "learning_rate": 0.00015056140350877193,
+      "loss": 0.0012,
+      "step": 5763
+    },
+    {
+      "epoch": 86.03,
+      "learning_rate": 0.00015052631578947367,
+      "loss": 0.0997,
+      "step": 5764
+    },
+    {
+      "epoch": 86.04,
+      "learning_rate": 0.00015049122807017542,
+      "loss": 0.2033,
+      "step": 5765
+    },
+    {
+      "epoch": 86.06,
+      "learning_rate": 0.00015045614035087717,
+      "loss": 0.0006,
+      "step": 5766
+    },
+    {
+      "epoch": 86.07,
+      "learning_rate": 0.00015042105263157895,
+      "loss": 0.4448,
+      "step": 5767
+    },
+    {
+      "epoch": 86.09,
+      "learning_rate": 0.0001503859649122807,
+      "loss": 0.0009,
+      "step": 5768
+    },
+    {
+      "epoch": 86.1,
+      "learning_rate": 0.00015035087719298244,
+      "loss": 0.0669,
+      "step": 5769
+    },
+    {
+      "epoch": 86.12,
+      "learning_rate": 0.00015031578947368422,
+      "loss": 0.055,
+      "step": 5770
+    },
+    {
+      "epoch": 86.13,
+      "learning_rate": 0.00015028070175438597,
+      "loss": 0.0017,
+      "step": 5771
+    },
+    {
+      "epoch": 86.15,
+      "learning_rate": 0.0001502456140350877,
+      "loss": 0.0013,
+      "step": 5772
+    },
+    {
+      "epoch": 86.16,
+      "learning_rate": 0.00015021052631578944,
+      "loss": 0.0008,
+      "step": 5773
+    },
+    {
+      "epoch": 86.18,
+      "learning_rate": 0.00015017543859649122,
+      "loss": 0.0009,
+      "step": 5774
+    },
+    {
+      "epoch": 86.19,
+      "learning_rate": 0.00015014035087719296,
+      "loss": 0.0008,
+      "step": 5775
+    },
+    {
+      "epoch": 86.21,
+      "learning_rate": 0.0001501052631578947,
+      "loss": 0.1733,
+      "step": 5776
+    },
+    {
+      "epoch": 86.22,
+      "learning_rate": 0.00015007017543859646,
+      "loss": 0.0011,
+      "step": 5777
+    },
+    {
+      "epoch": 86.24,
+      "learning_rate": 0.00015003508771929824,
+      "loss": 0.0011,
+      "step": 5778
+    },
+    {
+      "epoch": 86.25,
+      "learning_rate": 0.00015,
+      "loss": 0.1676,
+      "step": 5779
+    },
+    {
+      "epoch": 86.27,
+      "learning_rate": 0.00014996491228070174,
+      "loss": 0.0156,
+      "step": 5780
+    },
+    {
+      "epoch": 86.28,
+      "learning_rate": 0.00014992982456140348,
+      "loss": 0.0034,
+      "step": 5781
+    },
+    {
+      "epoch": 86.3,
+      "learning_rate": 0.00014989473684210526,
+      "loss": 0.0028,
+      "step": 5782
+    },
+    {
+      "epoch": 86.31,
+      "learning_rate": 0.000149859649122807,
+      "loss": 0.0805,
+      "step": 5783
+    },
+    {
+      "epoch": 86.33,
+      "learning_rate": 0.00014982456140350876,
+      "loss": 0.0009,
+      "step": 5784
+    },
+    {
+      "epoch": 86.34,
+      "learning_rate": 0.0001497894736842105,
+      "loss": 0.001,
+      "step": 5785
+    },
+    {
+      "epoch": 86.36,
+      "learning_rate": 0.00014975438596491228,
+      "loss": 0.0014,
+      "step": 5786
+    },
+    {
+      "epoch": 86.37,
+      "learning_rate": 0.000149719298245614,
+      "loss": 0.0013,
+      "step": 5787
+    },
+    {
+      "epoch": 86.39,
+      "learning_rate": 0.00014968421052631578,
+      "loss": 0.004,
+      "step": 5788
+    },
+    {
+      "epoch": 86.4,
+      "learning_rate": 0.00014964912280701753,
+      "loss": 0.0011,
+      "step": 5789
+    },
+    {
+      "epoch": 86.42,
+      "learning_rate": 0.00014961403508771928,
+      "loss": 0.3142,
+      "step": 5790
+    },
+    {
+      "epoch": 86.43,
+      "learning_rate": 0.00014957894736842103,
+      "loss": 0.0419,
+      "step": 5791
+    },
+    {
+      "epoch": 86.45,
+      "learning_rate": 0.0001495438596491228,
+      "loss": 0.0111,
+      "step": 5792
+    },
+    {
+      "epoch": 86.46,
+      "learning_rate": 0.00014950877192982455,
+      "loss": 0.008,
+      "step": 5793
+    },
+    {
+      "epoch": 86.48,
+      "learning_rate": 0.0001494736842105263,
+      "loss": 0.0049,
+      "step": 5794
+    },
+    {
+      "epoch": 86.49,
+      "learning_rate": 0.00014943859649122808,
+      "loss": 0.1487,
+      "step": 5795
+    },
+    {
+      "epoch": 86.51,
+      "learning_rate": 0.0001494035087719298,
+      "loss": 0.0008,
+      "step": 5796
+    },
+    {
+      "epoch": 86.52,
+      "learning_rate": 0.00014936842105263157,
+      "loss": 0.0008,
+      "step": 5797
+    },
+    {
+      "epoch": 86.54,
+      "learning_rate": 0.00014933333333333332,
+      "loss": 0.001,
+      "step": 5798
+    },
+    {
+      "epoch": 86.55,
+      "learning_rate": 0.00014929824561403507,
+      "loss": 0.0466,
+      "step": 5799
+    },
+    {
+      "epoch": 86.57,
+      "learning_rate": 0.00014926315789473682,
+      "loss": 0.0162,
+      "step": 5800
+    },
+    {
+      "epoch": 86.57,
+      "eval_accuracy": 0.8663729809104258,
+      "eval_f1": 0.8667231212452812,
+      "eval_loss": 0.6782434582710266,
+      "eval_runtime": 344.5523,
+      "eval_samples_per_second": 11.859,
+      "eval_steps_per_second": 0.743,
+      "step": 5800
+    },
+    {
+      "epoch": 86.58,
+      "learning_rate": 0.0001492280701754386,
+      "loss": 0.1697,
+      "step": 5801
+    },
+    {
+      "epoch": 86.59,
+      "learning_rate": 0.00014919298245614034,
+      "loss": 0.0398,
+      "step": 5802
+    },
+    {
+      "epoch": 86.61,
+      "learning_rate": 0.0001491578947368421,
+      "loss": 0.0209,
+      "step": 5803
+    },
+    {
+      "epoch": 86.62,
+      "learning_rate": 0.00014912280701754384,
+      "loss": 0.001,
+      "step": 5804
+    },
+    {
+      "epoch": 86.64,
+      "learning_rate": 0.0001490877192982456,
+      "loss": 0.0625,
+      "step": 5805
+    },
+    {
+      "epoch": 86.65,
+      "learning_rate": 0.00014905263157894737,
+      "loss": 0.0015,
+      "step": 5806
+    },
+    {
+      "epoch": 86.67,
+      "learning_rate": 0.00014901754385964912,
+      "loss": 0.1041,
+      "step": 5807
+    },
+    {
+      "epoch": 86.68,
+      "learning_rate": 0.00014898245614035086,
+      "loss": 0.0991,
+      "step": 5808
+    },
+    {
+      "epoch": 86.7,
+      "learning_rate": 0.0001489473684210526,
+      "loss": 0.0031,
+      "step": 5809
+    },
+    {
+      "epoch": 86.71,
+      "learning_rate": 0.0001489122807017544,
+      "loss": 0.0017,
+      "step": 5810
+    },
+    {
+      "epoch": 86.73,
+      "learning_rate": 0.0001488771929824561,
+      "loss": 0.0084,
+      "step": 5811
+    },
+    {
+      "epoch": 86.74,
+      "learning_rate": 0.00014884210526315789,
+      "loss": 0.0009,
+      "step": 5812
+    },
+    {
+      "epoch": 86.76,
+      "learning_rate": 0.00014880701754385964,
+      "loss": 0.0012,
+      "step": 5813
+    },
+    {
+      "epoch": 86.77,
+      "learning_rate": 0.00014877192982456138,
+      "loss": 0.0019,
+      "step": 5814
+    },
+    {
+      "epoch": 86.79,
+      "learning_rate": 0.00014873684210526313,
+      "loss": 0.0017,
+      "step": 5815
+    },
+    {
+      "epoch": 86.8,
+      "learning_rate": 0.0001487017543859649,
+      "loss": 0.0127,
+      "step": 5816
+    },
+    {
+      "epoch": 86.82,
+      "learning_rate": 0.00014866666666666666,
+      "loss": 0.0011,
+      "step": 5817
+    },
+    {
+      "epoch": 86.83,
+      "learning_rate": 0.0001486315789473684,
+      "loss": 0.0011,
+      "step": 5818
+    },
+    {
+      "epoch": 86.85,
+      "learning_rate": 0.00014859649122807018,
+      "loss": 0.2392,
+      "step": 5819
+    },
+    {
+      "epoch": 86.86,
+      "learning_rate": 0.0001485614035087719,
+      "loss": 0.0805,
+      "step": 5820
+    },
+    {
+      "epoch": 86.88,
+      "learning_rate": 0.00014852631578947368,
+      "loss": 0.0014,
+      "step": 5821
+    },
+    {
+      "epoch": 86.89,
+      "learning_rate": 0.00014849122807017543,
+      "loss": 0.0017,
+      "step": 5822
+    },
+    {
+      "epoch": 86.91,
+      "learning_rate": 0.00014845614035087718,
+      "loss": 0.3464,
+      "step": 5823
+    },
+    {
+      "epoch": 86.92,
+      "learning_rate": 0.00014842105263157893,
+      "loss": 0.0017,
+      "step": 5824
+    },
+    {
+      "epoch": 86.94,
+      "learning_rate": 0.0001483859649122807,
+      "loss": 0.0026,
+      "step": 5825
+    },
+    {
+      "epoch": 86.95,
+      "learning_rate": 0.00014835087719298242,
+      "loss": 0.0041,
+      "step": 5826
+    },
+    {
+      "epoch": 86.97,
+      "learning_rate": 0.0001483157894736842,
+      "loss": 0.0068,
+      "step": 5827
+    },
+    {
+      "epoch": 86.98,
+      "learning_rate": 0.00014828070175438595,
+      "loss": 0.0137,
+      "step": 5828
+    },
+    {
+      "epoch": 87.0,
+      "learning_rate": 0.0001482456140350877,
+      "loss": 0.1182,
+      "step": 5829
+    },
+    {
+      "epoch": 87.01,
+      "learning_rate": 0.00014821052631578947,
+      "loss": 0.0045,
+      "step": 5830
+    },
+    {
+      "epoch": 87.03,
+      "learning_rate": 0.00014817543859649122,
+      "loss": 0.0103,
+      "step": 5831
+    },
+    {
+      "epoch": 87.04,
+      "learning_rate": 0.00014814035087719297,
+      "loss": 0.0048,
+      "step": 5832
+    },
+    {
+      "epoch": 87.06,
+      "learning_rate": 0.00014810526315789472,
+      "loss": 0.0839,
+      "step": 5833
+    },
+    {
+      "epoch": 87.07,
+      "learning_rate": 0.0001480701754385965,
+      "loss": 0.0019,
+      "step": 5834
+    },
+    {
+      "epoch": 87.09,
+      "learning_rate": 0.00014803508771929822,
+      "loss": 0.0032,
+      "step": 5835
+    },
+    {
+      "epoch": 87.1,
+      "learning_rate": 0.000148,
+      "loss": 0.0013,
+      "step": 5836
+    },
+    {
+      "epoch": 87.12,
+      "learning_rate": 0.00014796491228070174,
+      "loss": 0.0122,
+      "step": 5837
+    },
+    {
+      "epoch": 87.13,
+      "learning_rate": 0.0001479298245614035,
+      "loss": 0.024,
+      "step": 5838
+    },
+    {
+      "epoch": 87.15,
+      "learning_rate": 0.00014789473684210524,
+      "loss": 0.1766,
+      "step": 5839
+    },
+    {
+      "epoch": 87.16,
+      "learning_rate": 0.00014785964912280701,
+      "loss": 0.0011,
+      "step": 5840
+    },
+    {
+      "epoch": 87.18,
+      "learning_rate": 0.00014782456140350876,
+      "loss": 0.1159,
+      "step": 5841
+    },
+    {
+      "epoch": 87.19,
+      "learning_rate": 0.0001477894736842105,
+      "loss": 0.0019,
+      "step": 5842
+    },
+    {
+      "epoch": 87.21,
+      "learning_rate": 0.0001477543859649123,
+      "loss": 0.0011,
+      "step": 5843
+    },
+    {
+      "epoch": 87.22,
+      "learning_rate": 0.000147719298245614,
+      "loss": 0.0013,
+      "step": 5844
+    },
+    {
+      "epoch": 87.24,
+      "learning_rate": 0.00014768421052631579,
+      "loss": 0.0013,
+      "step": 5845
+    },
+    {
+      "epoch": 87.25,
+      "learning_rate": 0.00014764912280701753,
+      "loss": 0.0241,
+      "step": 5846
+    },
+    {
+      "epoch": 87.27,
+      "learning_rate": 0.00014761403508771928,
+      "loss": 0.0051,
+      "step": 5847
+    },
+    {
+      "epoch": 87.28,
+      "learning_rate": 0.00014757894736842103,
+      "loss": 0.0034,
+      "step": 5848
+    },
+    {
+      "epoch": 87.3,
+      "learning_rate": 0.0001475438596491228,
+      "loss": 0.0059,
+      "step": 5849
+    },
+    {
+      "epoch": 87.31,
+      "learning_rate": 0.00014750877192982453,
+      "loss": 0.002,
+      "step": 5850
+    },
+    {
+      "epoch": 87.33,
+      "learning_rate": 0.0001474736842105263,
+      "loss": 0.079,
+      "step": 5851
+    },
+    {
+      "epoch": 87.34,
+      "learning_rate": 0.00014743859649122805,
+      "loss": 0.0011,
+      "step": 5852
+    },
+    {
+      "epoch": 87.36,
+      "learning_rate": 0.0001474035087719298,
+      "loss": 0.0013,
+      "step": 5853
+    },
+    {
+      "epoch": 87.37,
+      "learning_rate": 0.00014736842105263155,
+      "loss": 0.004,
+      "step": 5854
+    },
+    {
+      "epoch": 87.39,
+      "learning_rate": 0.00014733333333333333,
+      "loss": 0.23,
+      "step": 5855
+    },
+    {
+      "epoch": 87.4,
+      "learning_rate": 0.00014729824561403508,
+      "loss": 0.0115,
+      "step": 5856
+    },
+    {
+      "epoch": 87.42,
+      "learning_rate": 0.00014726315789473683,
+      "loss": 0.0328,
+      "step": 5857
+    },
+    {
+      "epoch": 87.43,
+      "learning_rate": 0.0001472280701754386,
+      "loss": 0.0018,
+      "step": 5858
+    },
+    {
+      "epoch": 87.45,
+      "learning_rate": 0.00014719298245614032,
+      "loss": 0.0015,
+      "step": 5859
+    },
+    {
+      "epoch": 87.46,
+      "learning_rate": 0.0001471578947368421,
+      "loss": 0.0015,
+      "step": 5860
+    },
+    {
+      "epoch": 87.48,
+      "learning_rate": 0.00014712280701754385,
+      "loss": 0.0011,
+      "step": 5861
+    },
+    {
+      "epoch": 87.49,
+      "learning_rate": 0.0001470877192982456,
+      "loss": 0.0487,
+      "step": 5862
+    },
+    {
+      "epoch": 87.51,
+      "learning_rate": 0.00014705263157894734,
+      "loss": 0.0065,
+      "step": 5863
+    },
+    {
+      "epoch": 87.52,
+      "learning_rate": 0.00014701754385964912,
+      "loss": 0.0877,
+      "step": 5864
+    },
+    {
+      "epoch": 87.54,
+      "learning_rate": 0.00014698245614035087,
+      "loss": 0.0022,
+      "step": 5865
+    },
+    {
+      "epoch": 87.55,
+      "learning_rate": 0.00014694736842105262,
+      "loss": 0.0031,
+      "step": 5866
+    },
+    {
+      "epoch": 87.57,
+      "learning_rate": 0.0001469122807017544,
+      "loss": 0.2098,
+      "step": 5867
+    },
+    {
+      "epoch": 87.58,
+      "learning_rate": 0.00014687719298245612,
+      "loss": 0.002,
+      "step": 5868
+    },
+    {
+      "epoch": 87.59,
+      "learning_rate": 0.0001468421052631579,
+      "loss": 0.0015,
+      "step": 5869
+    },
+    {
+      "epoch": 87.61,
+      "learning_rate": 0.00014680701754385964,
+      "loss": 0.0051,
+      "step": 5870
+    },
+    {
+      "epoch": 87.62,
+      "learning_rate": 0.0001467719298245614,
+      "loss": 0.0025,
+      "step": 5871
+    },
+    {
+      "epoch": 87.64,
+      "learning_rate": 0.00014673684210526314,
+      "loss": 0.0078,
+      "step": 5872
+    },
+    {
+      "epoch": 87.65,
+      "learning_rate": 0.00014670175438596491,
+      "loss": 0.0011,
+      "step": 5873
+    },
+    {
+      "epoch": 87.67,
+      "learning_rate": 0.00014666666666666664,
+      "loss": 0.0079,
+      "step": 5874
+    },
+    {
+      "epoch": 87.68,
+      "learning_rate": 0.0001466315789473684,
+      "loss": 0.0046,
+      "step": 5875
+    },
+    {
+      "epoch": 87.7,
+      "learning_rate": 0.00014659649122807016,
+      "loss": 0.0021,
+      "step": 5876
+    },
+    {
+      "epoch": 87.71,
+      "learning_rate": 0.0001465614035087719,
+      "loss": 0.0013,
+      "step": 5877
+    },
+    {
+      "epoch": 87.73,
+      "learning_rate": 0.00014652631578947366,
+      "loss": 0.0009,
+      "step": 5878
+    },
+    {
+      "epoch": 87.74,
+      "learning_rate": 0.00014649122807017543,
+      "loss": 0.0011,
+      "step": 5879
+    },
+    {
+      "epoch": 87.76,
+      "learning_rate": 0.00014645614035087718,
+      "loss": 0.0012,
+      "step": 5880
+    },
+    {
+      "epoch": 87.77,
+      "learning_rate": 0.00014642105263157893,
+      "loss": 0.0009,
+      "step": 5881
+    },
+    {
+      "epoch": 87.79,
+      "learning_rate": 0.0001463859649122807,
+      "loss": 0.0008,
+      "step": 5882
+    },
+    {
+      "epoch": 87.8,
+      "learning_rate": 0.00014635087719298243,
+      "loss": 0.0012,
+      "step": 5883
+    },
+    {
+      "epoch": 87.82,
+      "learning_rate": 0.0001463157894736842,
+      "loss": 0.075,
+      "step": 5884
+    },
+    {
+      "epoch": 87.83,
+      "learning_rate": 0.00014628070175438595,
+      "loss": 0.003,
+      "step": 5885
+    },
+    {
+      "epoch": 87.85,
+      "learning_rate": 0.0001462456140350877,
+      "loss": 0.0025,
+      "step": 5886
+    },
+    {
+      "epoch": 87.86,
+      "learning_rate": 0.00014621052631578945,
+      "loss": 0.001,
+      "step": 5887
+    },
+    {
+      "epoch": 87.88,
+      "learning_rate": 0.00014617543859649123,
+      "loss": 0.0007,
+      "step": 5888
+    },
+    {
+      "epoch": 87.89,
+      "learning_rate": 0.00014614035087719298,
+      "loss": 0.0009,
+      "step": 5889
+    },
+    {
+      "epoch": 87.91,
+      "learning_rate": 0.00014610526315789472,
+      "loss": 0.0394,
+      "step": 5890
+    },
+    {
+      "epoch": 87.92,
+      "learning_rate": 0.00014607017543859647,
+      "loss": 0.0007,
+      "step": 5891
+    },
+    {
+      "epoch": 87.94,
+      "learning_rate": 0.00014603508771929822,
+      "loss": 0.0369,
+      "step": 5892
+    },
+    {
+      "epoch": 87.95,
+      "learning_rate": 0.000146,
+      "loss": 0.0059,
+      "step": 5893
+    },
+    {
+      "epoch": 87.97,
+      "learning_rate": 0.00014596491228070175,
+      "loss": 0.0013,
+      "step": 5894
+    },
+    {
+      "epoch": 87.98,
+      "learning_rate": 0.0001459298245614035,
+      "loss": 0.2778,
+      "step": 5895
+    },
+    {
+      "epoch": 88.0,
+      "learning_rate": 0.00014589473684210524,
+      "loss": 0.0017,
+      "step": 5896
+    },
+    {
+      "epoch": 88.01,
+      "learning_rate": 0.00014585964912280702,
+      "loss": 0.0066,
+      "step": 5897
+    },
+    {
+      "epoch": 88.03,
+      "learning_rate": 0.00014582456140350874,
+      "loss": 0.0257,
+      "step": 5898
+    },
+    {
+      "epoch": 88.04,
+      "learning_rate": 0.00014578947368421052,
+      "loss": 0.0007,
+      "step": 5899
+    },
+    {
+      "epoch": 88.06,
+      "learning_rate": 0.00014575438596491227,
+      "loss": 0.1894,
+      "step": 5900
+    },
+    {
+      "epoch": 88.07,
+      "learning_rate": 0.00014571929824561402,
+      "loss": 0.1666,
+      "step": 5901
+    },
+    {
+      "epoch": 88.09,
+      "learning_rate": 0.00014568421052631576,
+      "loss": 0.0976,
+      "step": 5902
+    },
+    {
+      "epoch": 88.1,
+      "learning_rate": 0.00014564912280701754,
+      "loss": 0.0007,
+      "step": 5903
+    },
+    {
+      "epoch": 88.12,
+      "learning_rate": 0.0001456140350877193,
+      "loss": 0.0009,
+      "step": 5904
+    },
+    {
+      "epoch": 88.13,
+      "learning_rate": 0.00014557894736842104,
+      "loss": 0.0785,
+      "step": 5905
+    },
+    {
+      "epoch": 88.15,
+      "learning_rate": 0.0001455438596491228,
+      "loss": 0.1406,
+      "step": 5906
+    },
+    {
+      "epoch": 88.16,
+      "learning_rate": 0.00014550877192982454,
+      "loss": 0.0008,
+      "step": 5907
+    },
+    {
+      "epoch": 88.18,
+      "learning_rate": 0.0001454736842105263,
+      "loss": 0.0007,
+      "step": 5908
+    },
+    {
+      "epoch": 88.19,
+      "learning_rate": 0.00014543859649122806,
+      "loss": 0.0007,
+      "step": 5909
+    },
+    {
+      "epoch": 88.21,
+      "learning_rate": 0.0001454035087719298,
+      "loss": 0.1557,
+      "step": 5910
+    },
+    {
+      "epoch": 88.22,
+      "learning_rate": 0.00014536842105263156,
+      "loss": 0.0011,
+      "step": 5911
+    },
+    {
+      "epoch": 88.24,
+      "learning_rate": 0.00014533333333333333,
+      "loss": 0.2107,
+      "step": 5912
+    },
+    {
+      "epoch": 88.25,
+      "learning_rate": 0.00014529824561403508,
+      "loss": 0.2468,
+      "step": 5913
+    },
+    {
+      "epoch": 88.27,
+      "learning_rate": 0.00014526315789473683,
+      "loss": 0.0007,
+      "step": 5914
+    },
+    {
+      "epoch": 88.28,
+      "learning_rate": 0.00014522807017543858,
+      "loss": 0.0032,
+      "step": 5915
+    },
+    {
+      "epoch": 88.3,
+      "learning_rate": 0.00014519298245614033,
+      "loss": 0.002,
+      "step": 5916
+    },
+    {
+      "epoch": 88.31,
+      "learning_rate": 0.0001451578947368421,
+      "loss": 0.0035,
+      "step": 5917
+    },
+    {
+      "epoch": 88.33,
+      "learning_rate": 0.00014512280701754385,
+      "loss": 0.0014,
+      "step": 5918
+    },
+    {
+      "epoch": 88.34,
+      "learning_rate": 0.0001450877192982456,
+      "loss": 0.0042,
+      "step": 5919
+    },
+    {
+      "epoch": 88.36,
+      "learning_rate": 0.00014505263157894735,
+      "loss": 0.0015,
+      "step": 5920
+    },
+    {
+      "epoch": 88.37,
+      "learning_rate": 0.00014501754385964913,
+      "loss": 0.0027,
+      "step": 5921
+    },
+    {
+      "epoch": 88.39,
+      "learning_rate": 0.00014498245614035085,
+      "loss": 0.2759,
+      "step": 5922
+    },
+    {
+      "epoch": 88.4,
+      "learning_rate": 0.00014494736842105262,
+      "loss": 0.0025,
+      "step": 5923
+    },
+    {
+      "epoch": 88.42,
+      "learning_rate": 0.00014491228070175437,
+      "loss": 0.0016,
+      "step": 5924
+    },
+    {
+      "epoch": 88.43,
+      "learning_rate": 0.00014487719298245612,
+      "loss": 0.0013,
+      "step": 5925
+    },
+    {
+      "epoch": 88.45,
+      "learning_rate": 0.00014484210526315787,
+      "loss": 0.0033,
+      "step": 5926
+    },
+    {
+      "epoch": 88.46,
+      "learning_rate": 0.00014480701754385965,
+      "loss": 0.0038,
+      "step": 5927
+    },
+    {
+      "epoch": 88.48,
+      "learning_rate": 0.0001447719298245614,
+      "loss": 0.1503,
+      "step": 5928
+    },
+    {
+      "epoch": 88.49,
+      "learning_rate": 0.00014473684210526314,
+      "loss": 0.0013,
+      "step": 5929
+    },
+    {
+      "epoch": 88.51,
+      "learning_rate": 0.00014470175438596492,
+      "loss": 0.0019,
+      "step": 5930
+    },
+    {
+      "epoch": 88.52,
+      "learning_rate": 0.00014466666666666664,
+      "loss": 0.0517,
+      "step": 5931
+    },
+    {
+      "epoch": 88.54,
+      "learning_rate": 0.00014463157894736842,
+      "loss": 0.0095,
+      "step": 5932
+    },
+    {
+      "epoch": 88.55,
+      "learning_rate": 0.00014459649122807017,
+      "loss": 0.0308,
+      "step": 5933
+    },
+    {
+      "epoch": 88.57,
+      "learning_rate": 0.00014456140350877191,
+      "loss": 0.0025,
+      "step": 5934
+    },
+    {
+      "epoch": 88.58,
+      "learning_rate": 0.00014452631578947366,
+      "loss": 0.0047,
+      "step": 5935
+    },
+    {
+      "epoch": 88.59,
+      "learning_rate": 0.00014449122807017544,
+      "loss": 0.0013,
+      "step": 5936
+    },
+    {
+      "epoch": 88.61,
+      "learning_rate": 0.0001444561403508772,
+      "loss": 0.004,
+      "step": 5937
+    },
+    {
+      "epoch": 88.62,
+      "learning_rate": 0.00014442105263157894,
+      "loss": 0.0042,
+      "step": 5938
+    },
+    {
+      "epoch": 88.64,
+      "learning_rate": 0.00014438596491228069,
+      "loss": 0.2497,
+      "step": 5939
+    },
+    {
+      "epoch": 88.65,
+      "learning_rate": 0.00014435087719298243,
+      "loss": 0.0088,
+      "step": 5940
+    },
+    {
+      "epoch": 88.67,
+      "learning_rate": 0.00014431578947368418,
+      "loss": 0.0128,
+      "step": 5941
+    },
+    {
+      "epoch": 88.68,
+      "learning_rate": 0.00014428070175438596,
+      "loss": 0.0015,
+      "step": 5942
+    },
+    {
+      "epoch": 88.7,
+      "learning_rate": 0.0001442456140350877,
+      "loss": 0.0049,
+      "step": 5943
+    },
+    {
+      "epoch": 88.71,
+      "learning_rate": 0.00014421052631578946,
+      "loss": 0.0013,
+      "step": 5944
+    },
+    {
+      "epoch": 88.73,
+      "learning_rate": 0.00014417543859649123,
+      "loss": 0.0062,
+      "step": 5945
+    },
+    {
+      "epoch": 88.74,
+      "learning_rate": 0.00014414035087719295,
+      "loss": 0.0102,
+      "step": 5946
+    },
+    {
+      "epoch": 88.76,
+      "learning_rate": 0.00014410526315789473,
+      "loss": 0.1041,
+      "step": 5947
+    },
+    {
+      "epoch": 88.77,
+      "learning_rate": 0.00014407017543859648,
+      "loss": 0.0318,
+      "step": 5948
+    },
+    {
+      "epoch": 88.79,
+      "learning_rate": 0.00014403508771929823,
+      "loss": 0.1817,
+      "step": 5949
+    },
+    {
+      "epoch": 88.8,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 0.0012,
+      "step": 5950
+    },
+    {
+      "epoch": 88.82,
+      "learning_rate": 0.00014396491228070175,
+      "loss": 0.2596,
+      "step": 5951
+    },
+    {
+      "epoch": 88.83,
+      "learning_rate": 0.0001439298245614035,
+      "loss": 0.0012,
+      "step": 5952
+    },
+    {
+      "epoch": 88.85,
+      "learning_rate": 0.00014389473684210525,
+      "loss": 0.162,
+      "step": 5953
+    },
+    {
+      "epoch": 88.86,
+      "learning_rate": 0.000143859649122807,
+      "loss": 0.0988,
+      "step": 5954
+    },
+    {
+      "epoch": 88.88,
+      "learning_rate": 0.00014382456140350875,
+      "loss": 0.0008,
+      "step": 5955
+    },
+    {
+      "epoch": 88.89,
+      "learning_rate": 0.00014378947368421052,
+      "loss": 0.0078,
+      "step": 5956
+    },
+    {
+      "epoch": 88.91,
+      "learning_rate": 0.00014375438596491227,
+      "loss": 0.0186,
+      "step": 5957
+    },
+    {
+      "epoch": 88.92,
+      "learning_rate": 0.00014371929824561402,
+      "loss": 0.0009,
+      "step": 5958
+    },
+    {
+      "epoch": 88.94,
+      "learning_rate": 0.00014368421052631577,
+      "loss": 0.0046,
+      "step": 5959
+    },
+    {
+      "epoch": 88.95,
+      "learning_rate": 0.00014364912280701755,
+      "loss": 0.0011,
+      "step": 5960
+    },
+    {
+      "epoch": 88.97,
+      "learning_rate": 0.0001436140350877193,
+      "loss": 0.001,
+      "step": 5961
+    },
+    {
+      "epoch": 88.98,
+      "learning_rate": 0.00014357894736842104,
+      "loss": 0.0008,
+      "step": 5962
+    },
+    {
+      "epoch": 89.0,
+      "learning_rate": 0.0001435438596491228,
+      "loss": 0.0851,
+      "step": 5963
+    },
+    {
+      "epoch": 89.01,
+      "learning_rate": 0.00014350877192982454,
+      "loss": 0.0031,
+      "step": 5964
+    },
+    {
+      "epoch": 89.03,
+      "learning_rate": 0.0001434736842105263,
+      "loss": 0.0015,
+      "step": 5965
+    },
+    {
+      "epoch": 89.04,
+      "learning_rate": 0.00014343859649122807,
+      "loss": 0.2782,
+      "step": 5966
+    },
+    {
+      "epoch": 89.06,
+      "learning_rate": 0.00014340350877192981,
+      "loss": 0.0386,
+      "step": 5967
+    },
+    {
+      "epoch": 89.07,
+      "learning_rate": 0.00014336842105263156,
+      "loss": 0.2128,
+      "step": 5968
+    },
+    {
+      "epoch": 89.09,
+      "learning_rate": 0.00014333333333333334,
+      "loss": 0.001,
+      "step": 5969
+    },
+    {
+      "epoch": 89.1,
+      "learning_rate": 0.00014329824561403506,
+      "loss": 0.0344,
+      "step": 5970
+    },
+    {
+      "epoch": 89.12,
+      "learning_rate": 0.00014326315789473684,
+      "loss": 0.001,
+      "step": 5971
+    },
+    {
+      "epoch": 89.13,
+      "learning_rate": 0.00014322807017543859,
+      "loss": 0.0008,
+      "step": 5972
+    },
+    {
+      "epoch": 89.15,
+      "learning_rate": 0.00014319298245614033,
+      "loss": 0.0195,
+      "step": 5973
+    },
+    {
+      "epoch": 89.16,
+      "learning_rate": 0.00014315789473684208,
+      "loss": 0.0012,
+      "step": 5974
+    },
+    {
+      "epoch": 89.18,
+      "learning_rate": 0.00014312280701754386,
+      "loss": 0.0016,
+      "step": 5975
+    },
+    {
+      "epoch": 89.19,
+      "learning_rate": 0.0001430877192982456,
+      "loss": 0.0053,
+      "step": 5976
+    },
+    {
+      "epoch": 89.21,
+      "learning_rate": 0.00014305263157894736,
+      "loss": 0.0014,
+      "step": 5977
+    },
+    {
+      "epoch": 89.22,
+      "learning_rate": 0.0001430175438596491,
+      "loss": 0.1234,
+      "step": 5978
+    },
+    {
+      "epoch": 89.24,
+      "learning_rate": 0.00014298245614035085,
+      "loss": 0.0009,
+      "step": 5979
+    },
+    {
+      "epoch": 89.25,
+      "learning_rate": 0.00014294736842105263,
+      "loss": 0.0274,
+      "step": 5980
+    },
+    {
+      "epoch": 89.27,
+      "learning_rate": 0.00014291228070175438,
+      "loss": 0.0013,
+      "step": 5981
+    },
+    {
+      "epoch": 89.28,
+      "learning_rate": 0.00014287719298245613,
+      "loss": 0.0067,
+      "step": 5982
+    },
+    {
+      "epoch": 89.3,
+      "learning_rate": 0.00014284210526315788,
+      "loss": 0.0061,
+      "step": 5983
+    },
+    {
+      "epoch": 89.31,
+      "learning_rate": 0.00014280701754385965,
+      "loss": 0.1048,
+      "step": 5984
+    },
+    {
+      "epoch": 89.33,
+      "learning_rate": 0.0001427719298245614,
+      "loss": 0.2524,
+      "step": 5985
+    },
+    {
+      "epoch": 89.34,
+      "learning_rate": 0.00014273684210526315,
+      "loss": 0.0997,
+      "step": 5986
+    },
+    {
+      "epoch": 89.36,
+      "learning_rate": 0.0001427017543859649,
+      "loss": 0.0029,
+      "step": 5987
+    },
+    {
+      "epoch": 89.37,
+      "learning_rate": 0.00014266666666666665,
+      "loss": 0.1637,
+      "step": 5988
+    },
+    {
+      "epoch": 89.39,
+      "learning_rate": 0.0001426315789473684,
+      "loss": 0.0022,
+      "step": 5989
+    },
+    {
+      "epoch": 89.4,
+      "learning_rate": 0.00014259649122807017,
+      "loss": 0.0015,
+      "step": 5990
+    },
+    {
+      "epoch": 89.42,
+      "learning_rate": 0.00014256140350877192,
+      "loss": 0.0012,
+      "step": 5991
+    },
+    {
+      "epoch": 89.43,
+      "learning_rate": 0.00014252631578947367,
+      "loss": 0.1172,
+      "step": 5992
+    },
+    {
+      "epoch": 89.45,
+      "learning_rate": 0.00014249122807017545,
+      "loss": 0.1637,
+      "step": 5993
+    },
+    {
+      "epoch": 89.46,
+      "learning_rate": 0.00014245614035087717,
+      "loss": 0.0271,
+      "step": 5994
+    },
+    {
+      "epoch": 89.48,
+      "learning_rate": 0.00014242105263157894,
+      "loss": 0.0028,
+      "step": 5995
+    },
+    {
+      "epoch": 89.49,
+      "learning_rate": 0.0001423859649122807,
+      "loss": 0.2502,
+      "step": 5996
+    },
+    {
+      "epoch": 89.51,
+      "learning_rate": 0.00014235087719298244,
+      "loss": 0.0021,
+      "step": 5997
+    },
+    {
+      "epoch": 89.52,
+      "learning_rate": 0.0001423157894736842,
+      "loss": 0.1202,
+      "step": 5998
+    },
+    {
+      "epoch": 89.54,
+      "learning_rate": 0.00014228070175438596,
+      "loss": 0.1288,
+      "step": 5999
+    },
+    {
+      "epoch": 89.55,
+      "learning_rate": 0.00014224561403508771,
+      "loss": 0.1929,
+      "step": 6000
+    },
+    {
+      "epoch": 89.55,
+      "eval_accuracy": 0.860009789525208,
+      "eval_f1": 0.8615169079479937,
+      "eval_loss": 0.6371402144432068,
+      "eval_runtime": 345.6404,
+      "eval_samples_per_second": 11.822,
+      "eval_steps_per_second": 0.741,
+      "step": 6000
+    },
+    {
+      "epoch": 89.57,
+      "learning_rate": 0.00014221052631578946,
+      "loss": 0.0023,
+      "step": 6001
+    },
+    {
+      "epoch": 89.58,
+      "learning_rate": 0.0001421754385964912,
+      "loss": 0.1027,
+      "step": 6002
+    },
+    {
+      "epoch": 89.59,
+      "learning_rate": 0.00014214035087719296,
+      "loss": 0.0391,
+      "step": 6003
+    },
+    {
+      "epoch": 89.61,
+      "learning_rate": 0.0001421052631578947,
+      "loss": 0.0332,
+      "step": 6004
+    },
+    {
+      "epoch": 89.62,
+      "learning_rate": 0.00014207017543859648,
+      "loss": 0.0038,
+      "step": 6005
+    },
+    {
+      "epoch": 89.64,
+      "learning_rate": 0.00014203508771929823,
+      "loss": 0.0185,
+      "step": 6006
+    },
+    {
+      "epoch": 89.65,
+      "learning_rate": 0.00014199999999999998,
+      "loss": 0.0101,
+      "step": 6007
+    },
+    {
+      "epoch": 89.67,
+      "learning_rate": 0.00014196491228070176,
+      "loss": 0.0023,
+      "step": 6008
+    },
+    {
+      "epoch": 89.68,
+      "learning_rate": 0.0001419298245614035,
+      "loss": 0.0232,
+      "step": 6009
+    },
+    {
+      "epoch": 89.7,
+      "learning_rate": 0.00014189473684210526,
+      "loss": 0.1696,
+      "step": 6010
+    },
+    {
+      "epoch": 89.71,
+      "learning_rate": 0.000141859649122807,
+      "loss": 0.0579,
+      "step": 6011
+    },
+    {
+      "epoch": 89.73,
+      "learning_rate": 0.00014182456140350875,
+      "loss": 0.0025,
+      "step": 6012
+    },
+    {
+      "epoch": 89.74,
+      "learning_rate": 0.0001417894736842105,
+      "loss": 0.0084,
+      "step": 6013
+    },
+    {
+      "epoch": 89.76,
+      "learning_rate": 0.00014175438596491228,
+      "loss": 0.0014,
+      "step": 6014
+    },
+    {
+      "epoch": 89.77,
+      "learning_rate": 0.00014171929824561403,
+      "loss": 0.0051,
+      "step": 6015
+    },
+    {
+      "epoch": 89.79,
+      "learning_rate": 0.00014168421052631578,
+      "loss": 0.0147,
+      "step": 6016
+    },
+    {
+      "epoch": 89.8,
+      "learning_rate": 0.00014164912280701752,
+      "loss": 0.0327,
+      "step": 6017
+    },
+    {
+      "epoch": 89.82,
+      "learning_rate": 0.00014161403508771927,
+      "loss": 0.0069,
+      "step": 6018
+    },
+    {
+      "epoch": 89.83,
+      "learning_rate": 0.00014157894736842105,
+      "loss": 0.011,
+      "step": 6019
+    },
+    {
+      "epoch": 89.85,
+      "learning_rate": 0.0001415438596491228,
+      "loss": 0.016,
+      "step": 6020
+    },
+    {
+      "epoch": 89.86,
+      "learning_rate": 0.00014150877192982455,
+      "loss": 0.0099,
+      "step": 6021
+    },
+    {
+      "epoch": 89.88,
+      "learning_rate": 0.0001414736842105263,
+      "loss": 0.0036,
+      "step": 6022
+    },
+    {
+      "epoch": 89.89,
+      "learning_rate": 0.00014143859649122807,
+      "loss": 0.0856,
+      "step": 6023
+    },
+    {
+      "epoch": 89.91,
+      "learning_rate": 0.00014140350877192982,
+      "loss": 0.0012,
+      "step": 6024
+    },
+    {
+      "epoch": 89.92,
+      "learning_rate": 0.00014136842105263157,
+      "loss": 0.0026,
+      "step": 6025
+    },
+    {
+      "epoch": 89.94,
+      "learning_rate": 0.00014133333333333332,
+      "loss": 0.225,
+      "step": 6026
+    },
+    {
+      "epoch": 89.95,
+      "learning_rate": 0.00014129824561403507,
+      "loss": 0.2369,
+      "step": 6027
+    },
+    {
+      "epoch": 89.97,
+      "learning_rate": 0.00014126315789473681,
+      "loss": 0.0025,
+      "step": 6028
+    },
+    {
+      "epoch": 89.98,
+      "learning_rate": 0.0001412280701754386,
+      "loss": 0.0009,
+      "step": 6029
+    },
+    {
+      "epoch": 90.0,
+      "learning_rate": 0.00014119298245614034,
+      "loss": 0.0011,
+      "step": 6030
+    },
+    {
+      "epoch": 90.01,
+      "learning_rate": 0.0001411578947368421,
+      "loss": 0.0779,
+      "step": 6031
+    },
+    {
+      "epoch": 90.03,
+      "learning_rate": 0.00014112280701754386,
+      "loss": 0.127,
+      "step": 6032
+    },
+    {
+      "epoch": 90.04,
+      "learning_rate": 0.0001410877192982456,
+      "loss": 0.1497,
+      "step": 6033
+    },
+    {
+      "epoch": 90.06,
+      "learning_rate": 0.00014105263157894736,
+      "loss": 0.0013,
+      "step": 6034
+    },
+    {
+      "epoch": 90.07,
+      "learning_rate": 0.0001410175438596491,
+      "loss": 0.0189,
+      "step": 6035
+    },
+    {
+      "epoch": 90.09,
+      "learning_rate": 0.00014098245614035086,
+      "loss": 0.2014,
+      "step": 6036
+    },
+    {
+      "epoch": 90.1,
+      "learning_rate": 0.0001409473684210526,
+      "loss": 0.0027,
+      "step": 6037
+    },
+    {
+      "epoch": 90.12,
+      "learning_rate": 0.00014091228070175438,
+      "loss": 0.0085,
+      "step": 6038
+    },
+    {
+      "epoch": 90.13,
+      "learning_rate": 0.00014087719298245613,
+      "loss": 0.0017,
+      "step": 6039
+    },
+    {
+      "epoch": 90.15,
+      "learning_rate": 0.00014084210526315788,
+      "loss": 0.0009,
+      "step": 6040
+    },
+    {
+      "epoch": 90.16,
+      "learning_rate": 0.00014080701754385963,
+      "loss": 0.0031,
+      "step": 6041
+    },
+    {
+      "epoch": 90.18,
+      "learning_rate": 0.00014077192982456138,
+      "loss": 0.0009,
+      "step": 6042
+    },
+    {
+      "epoch": 90.19,
+      "learning_rate": 0.00014073684210526315,
+      "loss": 0.0007,
+      "step": 6043
+    },
+    {
+      "epoch": 90.21,
+      "learning_rate": 0.0001407017543859649,
+      "loss": 0.0053,
+      "step": 6044
+    },
+    {
+      "epoch": 90.22,
+      "learning_rate": 0.00014066666666666665,
+      "loss": 0.0176,
+      "step": 6045
+    },
+    {
+      "epoch": 90.24,
+      "learning_rate": 0.0001406315789473684,
+      "loss": 0.0015,
+      "step": 6046
+    },
+    {
+      "epoch": 90.25,
+      "learning_rate": 0.00014059649122807018,
+      "loss": 0.0021,
+      "step": 6047
+    },
+    {
+      "epoch": 90.27,
+      "learning_rate": 0.00014056140350877193,
+      "loss": 0.0071,
+      "step": 6048
+    },
+    {
+      "epoch": 90.28,
+      "learning_rate": 0.00014052631578947367,
+      "loss": 0.0008,
+      "step": 6049
+    },
+    {
+      "epoch": 90.3,
+      "learning_rate": 0.00014049122807017542,
+      "loss": 0.0057,
+      "step": 6050
+    },
+    {
+      "epoch": 90.31,
+      "learning_rate": 0.00014045614035087717,
+      "loss": 0.0154,
+      "step": 6051
+    },
+    {
+      "epoch": 90.33,
+      "learning_rate": 0.00014042105263157892,
+      "loss": 0.001,
+      "step": 6052
+    },
+    {
+      "epoch": 90.34,
+      "learning_rate": 0.0001403859649122807,
+      "loss": 0.001,
+      "step": 6053
+    },
+    {
+      "epoch": 90.36,
+      "learning_rate": 0.00014035087719298245,
+      "loss": 0.0235,
+      "step": 6054
+    },
+    {
+      "epoch": 90.37,
+      "learning_rate": 0.0001403157894736842,
+      "loss": 0.0128,
+      "step": 6055
+    },
+    {
+      "epoch": 90.39,
+      "learning_rate": 0.00014028070175438597,
+      "loss": 0.0008,
+      "step": 6056
+    },
+    {
+      "epoch": 90.4,
+      "learning_rate": 0.0001402456140350877,
+      "loss": 0.0027,
+      "step": 6057
+    },
+    {
+      "epoch": 90.42,
+      "learning_rate": 0.00014021052631578947,
+      "loss": 0.0024,
+      "step": 6058
+    },
+    {
+      "epoch": 90.43,
+      "learning_rate": 0.00014017543859649122,
+      "loss": 0.0008,
+      "step": 6059
+    },
+    {
+      "epoch": 90.45,
+      "learning_rate": 0.00014014035087719297,
+      "loss": 0.0984,
+      "step": 6060
+    },
+    {
+      "epoch": 90.46,
+      "learning_rate": 0.00014010526315789471,
+      "loss": 0.0008,
+      "step": 6061
+    },
+    {
+      "epoch": 90.48,
+      "learning_rate": 0.0001400701754385965,
+      "loss": 0.0008,
+      "step": 6062
+    },
+    {
+      "epoch": 90.49,
+      "learning_rate": 0.00014003508771929824,
+      "loss": 0.0664,
+      "step": 6063
+    },
+    {
+      "epoch": 90.51,
+      "learning_rate": 0.00014,
+      "loss": 0.0387,
+      "step": 6064
+    },
+    {
+      "epoch": 90.52,
+      "learning_rate": 0.00013996491228070174,
+      "loss": 0.0026,
+      "step": 6065
+    },
+    {
+      "epoch": 90.54,
+      "learning_rate": 0.00013992982456140349,
+      "loss": 0.003,
+      "step": 6066
+    },
+    {
+      "epoch": 90.55,
+      "learning_rate": 0.00013989473684210523,
+      "loss": 0.0011,
+      "step": 6067
+    },
+    {
+      "epoch": 90.57,
+      "learning_rate": 0.000139859649122807,
+      "loss": 0.0022,
+      "step": 6068
+    },
+    {
+      "epoch": 90.58,
+      "learning_rate": 0.00013982456140350876,
+      "loss": 0.001,
+      "step": 6069
+    },
+    {
+      "epoch": 90.59,
+      "learning_rate": 0.0001397894736842105,
+      "loss": 0.0246,
+      "step": 6070
+    },
+    {
+      "epoch": 90.61,
+      "learning_rate": 0.00013975438596491228,
+      "loss": 0.0157,
+      "step": 6071
+    },
+    {
+      "epoch": 90.62,
+      "learning_rate": 0.00013971929824561403,
+      "loss": 0.001,
+      "step": 6072
+    },
+    {
+      "epoch": 90.64,
+      "learning_rate": 0.00013968421052631578,
+      "loss": 0.0108,
+      "step": 6073
+    },
+    {
+      "epoch": 90.65,
+      "learning_rate": 0.00013964912280701753,
+      "loss": 0.0009,
+      "step": 6074
+    },
+    {
+      "epoch": 90.67,
+      "learning_rate": 0.00013961403508771928,
+      "loss": 0.0008,
+      "step": 6075
+    },
+    {
+      "epoch": 90.68,
+      "learning_rate": 0.00013957894736842103,
+      "loss": 0.0034,
+      "step": 6076
+    },
+    {
+      "epoch": 90.7,
+      "learning_rate": 0.0001395438596491228,
+      "loss": 0.0009,
+      "step": 6077
+    },
+    {
+      "epoch": 90.71,
+      "learning_rate": 0.00013950877192982455,
+      "loss": 0.0557,
+      "step": 6078
+    },
+    {
+      "epoch": 90.73,
+      "learning_rate": 0.0001394736842105263,
+      "loss": 0.0007,
+      "step": 6079
+    },
+    {
+      "epoch": 90.74,
+      "learning_rate": 0.00013943859649122808,
+      "loss": 0.0019,
+      "step": 6080
+    },
+    {
+      "epoch": 90.76,
+      "learning_rate": 0.0001394035087719298,
+      "loss": 0.0069,
+      "step": 6081
+    },
+    {
+      "epoch": 90.77,
+      "learning_rate": 0.00013936842105263157,
+      "loss": 0.0013,
+      "step": 6082
+    },
+    {
+      "epoch": 90.79,
+      "learning_rate": 0.00013933333333333332,
+      "loss": 0.0486,
+      "step": 6083
+    },
+    {
+      "epoch": 90.8,
+      "learning_rate": 0.00013929824561403507,
+      "loss": 0.0448,
+      "step": 6084
+    },
+    {
+      "epoch": 90.82,
+      "learning_rate": 0.00013926315789473682,
+      "loss": 0.0008,
+      "step": 6085
+    },
+    {
+      "epoch": 90.83,
+      "learning_rate": 0.0001392280701754386,
+      "loss": 0.0208,
+      "step": 6086
+    },
+    {
+      "epoch": 90.85,
+      "learning_rate": 0.00013919298245614035,
+      "loss": 0.0014,
+      "step": 6087
+    },
+    {
+      "epoch": 90.86,
+      "learning_rate": 0.0001391578947368421,
+      "loss": 0.258,
+      "step": 6088
+    },
+    {
+      "epoch": 90.88,
+      "learning_rate": 0.00013912280701754384,
+      "loss": 0.0007,
+      "step": 6089
+    },
+    {
+      "epoch": 90.89,
+      "learning_rate": 0.0001390877192982456,
+      "loss": 0.0034,
+      "step": 6090
+    },
+    {
+      "epoch": 90.91,
+      "learning_rate": 0.00013905263157894734,
+      "loss": 0.001,
+      "step": 6091
+    },
+    {
+      "epoch": 90.92,
+      "learning_rate": 0.00013901754385964912,
+      "loss": 0.0009,
+      "step": 6092
+    },
+    {
+      "epoch": 90.94,
+      "learning_rate": 0.00013898245614035086,
+      "loss": 0.0006,
+      "step": 6093
+    },
+    {
+      "epoch": 90.95,
+      "learning_rate": 0.00013894736842105261,
+      "loss": 0.0012,
+      "step": 6094
+    },
+    {
+      "epoch": 90.97,
+      "learning_rate": 0.0001389122807017544,
+      "loss": 0.001,
+      "step": 6095
+    },
+    {
+      "epoch": 90.98,
+      "learning_rate": 0.00013887719298245614,
+      "loss": 0.001,
+      "step": 6096
+    },
+    {
+      "epoch": 91.0,
+      "learning_rate": 0.0001388421052631579,
+      "loss": 0.0011,
+      "step": 6097
+    },
+    {
+      "epoch": 91.01,
+      "learning_rate": 0.00013880701754385964,
+      "loss": 0.0022,
+      "step": 6098
+    },
+    {
+      "epoch": 91.03,
+      "learning_rate": 0.00013877192982456138,
+      "loss": 0.0029,
+      "step": 6099
+    },
+    {
+      "epoch": 91.04,
+      "learning_rate": 0.00013873684210526313,
+      "loss": 0.0463,
+      "step": 6100
+    },
+    {
+      "epoch": 91.06,
+      "learning_rate": 0.0001387017543859649,
+      "loss": 0.0864,
+      "step": 6101
+    },
+    {
+      "epoch": 91.07,
+      "learning_rate": 0.00013866666666666666,
+      "loss": 0.001,
+      "step": 6102
+    },
+    {
+      "epoch": 91.09,
+      "learning_rate": 0.0001386315789473684,
+      "loss": 0.0009,
+      "step": 6103
+    },
+    {
+      "epoch": 91.1,
+      "learning_rate": 0.00013859649122807016,
+      "loss": 0.0079,
+      "step": 6104
+    },
+    {
+      "epoch": 91.12,
+      "learning_rate": 0.0001385614035087719,
+      "loss": 0.0109,
+      "step": 6105
+    },
+    {
+      "epoch": 91.13,
+      "learning_rate": 0.00013852631578947368,
+      "loss": 0.0017,
+      "step": 6106
+    },
+    {
+      "epoch": 91.15,
+      "learning_rate": 0.00013849122807017543,
+      "loss": 0.0009,
+      "step": 6107
+    },
+    {
+      "epoch": 91.16,
+      "learning_rate": 0.00013845614035087718,
+      "loss": 0.0028,
+      "step": 6108
+    },
+    {
+      "epoch": 91.18,
+      "learning_rate": 0.00013842105263157893,
+      "loss": 0.0007,
+      "step": 6109
+    },
+    {
+      "epoch": 91.19,
+      "learning_rate": 0.0001383859649122807,
+      "loss": 0.0013,
+      "step": 6110
+    },
+    {
+      "epoch": 91.21,
+      "learning_rate": 0.00013835087719298245,
+      "loss": 0.0051,
+      "step": 6111
+    },
+    {
+      "epoch": 91.22,
+      "learning_rate": 0.0001383157894736842,
+      "loss": 0.0016,
+      "step": 6112
+    },
+    {
+      "epoch": 91.24,
+      "learning_rate": 0.00013828070175438595,
+      "loss": 0.0112,
+      "step": 6113
+    },
+    {
+      "epoch": 91.25,
+      "learning_rate": 0.0001382456140350877,
+      "loss": 0.0008,
+      "step": 6114
+    },
+    {
+      "epoch": 91.27,
+      "learning_rate": 0.00013821052631578945,
+      "loss": 0.1127,
+      "step": 6115
+    },
+    {
+      "epoch": 91.28,
+      "learning_rate": 0.00013817543859649122,
+      "loss": 0.0011,
+      "step": 6116
+    },
+    {
+      "epoch": 91.3,
+      "learning_rate": 0.00013814035087719297,
+      "loss": 0.0027,
+      "step": 6117
+    },
+    {
+      "epoch": 91.31,
+      "learning_rate": 0.00013810526315789472,
+      "loss": 0.0007,
+      "step": 6118
+    },
+    {
+      "epoch": 91.33,
+      "learning_rate": 0.0001380701754385965,
+      "loss": 0.0555,
+      "step": 6119
+    },
+    {
+      "epoch": 91.34,
+      "learning_rate": 0.00013803508771929824,
+      "loss": 0.0025,
+      "step": 6120
+    },
+    {
+      "epoch": 91.36,
+      "learning_rate": 0.000138,
+      "loss": 0.0009,
+      "step": 6121
+    },
+    {
+      "epoch": 91.37,
+      "learning_rate": 0.00013796491228070174,
+      "loss": 0.0059,
+      "step": 6122
+    },
+    {
+      "epoch": 91.39,
+      "learning_rate": 0.0001379298245614035,
+      "loss": 0.0016,
+      "step": 6123
+    },
+    {
+      "epoch": 91.4,
+      "learning_rate": 0.00013789473684210524,
+      "loss": 0.0484,
+      "step": 6124
+    },
+    {
+      "epoch": 91.42,
+      "learning_rate": 0.00013785964912280702,
+      "loss": 0.0589,
+      "step": 6125
+    },
+    {
+      "epoch": 91.43,
+      "learning_rate": 0.00013782456140350876,
+      "loss": 0.0498,
+      "step": 6126
+    },
+    {
+      "epoch": 91.45,
+      "learning_rate": 0.0001377894736842105,
+      "loss": 0.0007,
+      "step": 6127
+    },
+    {
+      "epoch": 91.46,
+      "learning_rate": 0.00013775438596491226,
+      "loss": 0.0008,
+      "step": 6128
+    },
+    {
+      "epoch": 91.48,
+      "learning_rate": 0.000137719298245614,
+      "loss": 0.0011,
+      "step": 6129
+    },
+    {
+      "epoch": 91.49,
+      "learning_rate": 0.00013768421052631576,
+      "loss": 0.0008,
+      "step": 6130
+    },
+    {
+      "epoch": 91.51,
+      "learning_rate": 0.00013764912280701754,
+      "loss": 0.001,
+      "step": 6131
+    },
+    {
+      "epoch": 91.52,
+      "learning_rate": 0.00013761403508771928,
+      "loss": 0.0006,
+      "step": 6132
+    },
+    {
+      "epoch": 91.54,
+      "learning_rate": 0.00013757894736842103,
+      "loss": 0.0101,
+      "step": 6133
+    },
+    {
+      "epoch": 91.55,
+      "learning_rate": 0.0001375438596491228,
+      "loss": 0.0091,
+      "step": 6134
+    },
+    {
+      "epoch": 91.57,
+      "learning_rate": 0.00013750877192982456,
+      "loss": 0.0023,
+      "step": 6135
+    },
+    {
+      "epoch": 91.58,
+      "learning_rate": 0.0001374736842105263,
+      "loss": 0.001,
+      "step": 6136
+    },
+    {
+      "epoch": 91.59,
+      "learning_rate": 0.00013743859649122806,
+      "loss": 0.0009,
+      "step": 6137
+    },
+    {
+      "epoch": 91.61,
+      "learning_rate": 0.0001374035087719298,
+      "loss": 0.0023,
+      "step": 6138
+    },
+    {
+      "epoch": 91.62,
+      "learning_rate": 0.00013736842105263155,
+      "loss": 0.0014,
+      "step": 6139
+    },
+    {
+      "epoch": 91.64,
+      "learning_rate": 0.00013733333333333333,
+      "loss": 0.0016,
+      "step": 6140
+    },
+    {
+      "epoch": 91.65,
+      "learning_rate": 0.00013729824561403508,
+      "loss": 0.0045,
+      "step": 6141
+    },
+    {
+      "epoch": 91.67,
+      "learning_rate": 0.00013726315789473683,
+      "loss": 0.2162,
+      "step": 6142
+    },
+    {
+      "epoch": 91.68,
+      "learning_rate": 0.0001372280701754386,
+      "loss": 0.0006,
+      "step": 6143
+    },
+    {
+      "epoch": 91.7,
+      "learning_rate": 0.00013719298245614035,
+      "loss": 0.0014,
+      "step": 6144
+    },
+    {
+      "epoch": 91.71,
+      "learning_rate": 0.0001371578947368421,
+      "loss": 0.0006,
+      "step": 6145
+    },
+    {
+      "epoch": 91.73,
+      "learning_rate": 0.00013712280701754385,
+      "loss": 0.0007,
+      "step": 6146
+    },
+    {
+      "epoch": 91.74,
+      "learning_rate": 0.0001370877192982456,
+      "loss": 0.0077,
+      "step": 6147
+    },
+    {
+      "epoch": 91.76,
+      "learning_rate": 0.00013705263157894735,
+      "loss": 0.0008,
+      "step": 6148
+    },
+    {
+      "epoch": 91.77,
+      "learning_rate": 0.00013701754385964912,
+      "loss": 0.004,
+      "step": 6149
+    },
+    {
+      "epoch": 91.79,
+      "learning_rate": 0.00013698245614035087,
+      "loss": 0.0006,
+      "step": 6150
+    },
+    {
+      "epoch": 91.8,
+      "learning_rate": 0.00013694736842105262,
+      "loss": 0.1144,
+      "step": 6151
+    },
+    {
+      "epoch": 91.82,
+      "learning_rate": 0.00013691228070175437,
+      "loss": 0.0006,
+      "step": 6152
+    },
+    {
+      "epoch": 91.83,
+      "learning_rate": 0.00013687719298245612,
+      "loss": 0.0028,
+      "step": 6153
+    },
+    {
+      "epoch": 91.85,
+      "learning_rate": 0.00013684210526315787,
+      "loss": 0.0038,
+      "step": 6154
+    },
+    {
+      "epoch": 91.86,
+      "learning_rate": 0.00013680701754385964,
+      "loss": 0.0016,
+      "step": 6155
+    },
+    {
+      "epoch": 91.88,
+      "learning_rate": 0.0001367719298245614,
+      "loss": 0.0007,
+      "step": 6156
+    },
+    {
+      "epoch": 91.89,
+      "learning_rate": 0.00013673684210526314,
+      "loss": 0.0006,
+      "step": 6157
+    },
+    {
+      "epoch": 91.91,
+      "learning_rate": 0.00013670175438596491,
+      "loss": 0.001,
+      "step": 6158
+    },
+    {
+      "epoch": 91.92,
+      "learning_rate": 0.00013666666666666666,
+      "loss": 0.0008,
+      "step": 6159
+    },
+    {
+      "epoch": 91.94,
+      "learning_rate": 0.0001366315789473684,
+      "loss": 0.0173,
+      "step": 6160
+    },
+    {
+      "epoch": 91.95,
+      "learning_rate": 0.00013659649122807016,
+      "loss": 0.0082,
+      "step": 6161
+    },
+    {
+      "epoch": 91.97,
+      "learning_rate": 0.0001365614035087719,
+      "loss": 0.0005,
+      "step": 6162
+    },
+    {
+      "epoch": 91.98,
+      "learning_rate": 0.00013652631578947366,
+      "loss": 0.0015,
+      "step": 6163
+    },
+    {
+      "epoch": 92.0,
+      "learning_rate": 0.00013649122807017543,
+      "loss": 0.117,
+      "step": 6164
+    },
+    {
+      "epoch": 92.01,
+      "learning_rate": 0.00013645614035087718,
+      "loss": 0.0019,
+      "step": 6165
+    },
+    {
+      "epoch": 92.03,
+      "learning_rate": 0.00013642105263157893,
+      "loss": 0.0015,
+      "step": 6166
+    },
+    {
+      "epoch": 92.04,
+      "learning_rate": 0.00013638596491228068,
+      "loss": 0.0028,
+      "step": 6167
+    },
+    {
+      "epoch": 92.06,
+      "learning_rate": 0.00013635087719298246,
+      "loss": 0.0007,
+      "step": 6168
+    },
+    {
+      "epoch": 92.07,
+      "learning_rate": 0.0001363157894736842,
+      "loss": 0.0005,
+      "step": 6169
+    },
+    {
+      "epoch": 92.09,
+      "learning_rate": 0.00013628070175438595,
+      "loss": 0.0019,
+      "step": 6170
+    },
+    {
+      "epoch": 92.1,
+      "learning_rate": 0.0001362456140350877,
+      "loss": 0.0083,
+      "step": 6171
+    },
+    {
+      "epoch": 92.12,
+      "learning_rate": 0.00013621052631578945,
+      "loss": 0.0007,
+      "step": 6172
+    },
+    {
+      "epoch": 92.13,
+      "learning_rate": 0.00013617543859649123,
+      "loss": 0.0006,
+      "step": 6173
+    },
+    {
+      "epoch": 92.15,
+      "learning_rate": 0.00013614035087719298,
+      "loss": 0.0821,
+      "step": 6174
+    },
+    {
+      "epoch": 92.16,
+      "learning_rate": 0.00013610526315789473,
+      "loss": 0.0006,
+      "step": 6175
+    },
+    {
+      "epoch": 92.18,
+      "learning_rate": 0.00013607017543859647,
+      "loss": 0.0115,
+      "step": 6176
+    },
+    {
+      "epoch": 92.19,
+      "learning_rate": 0.00013603508771929822,
+      "loss": 0.0006,
+      "step": 6177
+    },
+    {
+      "epoch": 92.21,
+      "learning_rate": 0.00013599999999999997,
+      "loss": 0.0011,
+      "step": 6178
+    },
+    {
+      "epoch": 92.22,
+      "learning_rate": 0.00013596491228070175,
+      "loss": 0.0006,
+      "step": 6179
+    },
+    {
+      "epoch": 92.24,
+      "learning_rate": 0.0001359298245614035,
+      "loss": 0.0005,
+      "step": 6180
+    },
+    {
+      "epoch": 92.25,
+      "learning_rate": 0.00013589473684210525,
+      "loss": 0.0007,
+      "step": 6181
+    },
+    {
+      "epoch": 92.27,
+      "learning_rate": 0.00013585964912280702,
+      "loss": 0.0055,
+      "step": 6182
+    },
+    {
+      "epoch": 92.28,
+      "learning_rate": 0.00013582456140350877,
+      "loss": 0.0007,
+      "step": 6183
+    },
+    {
+      "epoch": 92.3,
+      "learning_rate": 0.00013578947368421052,
+      "loss": 0.0107,
+      "step": 6184
+    },
+    {
+      "epoch": 92.31,
+      "learning_rate": 0.00013575438596491227,
+      "loss": 0.0112,
+      "step": 6185
+    },
+    {
+      "epoch": 92.33,
+      "learning_rate": 0.00013571929824561402,
+      "loss": 0.0007,
+      "step": 6186
+    },
+    {
+      "epoch": 92.34,
+      "learning_rate": 0.00013568421052631577,
+      "loss": 0.0005,
+      "step": 6187
+    },
+    {
+      "epoch": 92.36,
+      "learning_rate": 0.00013564912280701754,
+      "loss": 0.0006,
+      "step": 6188
+    },
+    {
+      "epoch": 92.37,
+      "learning_rate": 0.0001356140350877193,
+      "loss": 0.0073,
+      "step": 6189
+    },
+    {
+      "epoch": 92.39,
+      "learning_rate": 0.00013557894736842104,
+      "loss": 0.0006,
+      "step": 6190
+    },
+    {
+      "epoch": 92.4,
+      "learning_rate": 0.0001355438596491228,
+      "loss": 0.0006,
+      "step": 6191
+    },
+    {
+      "epoch": 92.42,
+      "learning_rate": 0.00013550877192982456,
+      "loss": 0.0006,
+      "step": 6192
+    },
+    {
+      "epoch": 92.43,
+      "learning_rate": 0.0001354736842105263,
+      "loss": 0.0005,
+      "step": 6193
+    },
+    {
+      "epoch": 92.45,
+      "learning_rate": 0.00013543859649122806,
+      "loss": 0.0007,
+      "step": 6194
+    },
+    {
+      "epoch": 92.46,
+      "learning_rate": 0.0001354035087719298,
+      "loss": 0.0005,
+      "step": 6195
+    },
+    {
+      "epoch": 92.48,
+      "learning_rate": 0.00013536842105263156,
+      "loss": 0.1313,
+      "step": 6196
+    },
+    {
+      "epoch": 92.49,
+      "learning_rate": 0.00013533333333333333,
+      "loss": 0.0005,
+      "step": 6197
+    },
+    {
+      "epoch": 92.51,
+      "learning_rate": 0.00013529824561403508,
+      "loss": 0.0802,
+      "step": 6198
+    },
+    {
+      "epoch": 92.52,
+      "learning_rate": 0.00013526315789473683,
+      "loss": 0.0005,
+      "step": 6199
+    },
+    {
+      "epoch": 92.54,
+      "learning_rate": 0.00013522807017543858,
+      "loss": 0.0621,
+      "step": 6200
+    },
+    {
+      "epoch": 92.54,
+      "eval_accuracy": 0.8607440039158101,
+      "eval_f1": 0.8599750596640915,
+      "eval_loss": 0.8078528642654419,
+      "eval_runtime": 344.8499,
+      "eval_samples_per_second": 11.849,
+      "eval_steps_per_second": 0.742,
+      "step": 6200
+    },
+    {
+      "epoch": 92.55,
+      "learning_rate": 0.00013519298245614033,
+      "loss": 0.0008,
+      "step": 6201
+    },
+    {
+      "epoch": 92.57,
+      "learning_rate": 0.00013515789473684208,
+      "loss": 0.0007,
+      "step": 6202
+    },
+    {
+      "epoch": 92.58,
+      "learning_rate": 0.00013512280701754385,
+      "loss": 0.001,
+      "step": 6203
+    },
+    {
+      "epoch": 92.59,
+      "learning_rate": 0.0001350877192982456,
+      "loss": 0.0007,
+      "step": 6204
+    },
+    {
+      "epoch": 92.61,
+      "learning_rate": 0.00013505263157894735,
+      "loss": 0.0006,
+      "step": 6205
+    },
+    {
+      "epoch": 92.62,
+      "learning_rate": 0.00013501754385964913,
+      "loss": 0.0015,
+      "step": 6206
+    },
+    {
+      "epoch": 92.64,
+      "learning_rate": 0.00013498245614035088,
+      "loss": 0.0006,
+      "step": 6207
+    },
+    {
+      "epoch": 92.65,
+      "learning_rate": 0.00013494736842105262,
+      "loss": 0.0005,
+      "step": 6208
+    },
+    {
+      "epoch": 92.67,
+      "learning_rate": 0.00013491228070175437,
+      "loss": 0.0017,
+      "step": 6209
+    },
+    {
+      "epoch": 92.68,
+      "learning_rate": 0.00013487719298245612,
+      "loss": 0.0016,
+      "step": 6210
+    },
+    {
+      "epoch": 92.7,
+      "learning_rate": 0.00013484210526315787,
+      "loss": 0.0011,
+      "step": 6211
+    },
+    {
+      "epoch": 92.71,
+      "learning_rate": 0.00013480701754385965,
+      "loss": 0.0957,
+      "step": 6212
+    },
+    {
+      "epoch": 92.73,
+      "learning_rate": 0.0001347719298245614,
+      "loss": 0.0011,
+      "step": 6213
+    },
+    {
+      "epoch": 92.74,
+      "learning_rate": 0.00013473684210526314,
+      "loss": 0.0005,
+      "step": 6214
+    },
+    {
+      "epoch": 92.76,
+      "learning_rate": 0.0001347017543859649,
+      "loss": 0.0006,
+      "step": 6215
+    },
+    {
+      "epoch": 92.77,
+      "learning_rate": 0.00013466666666666667,
+      "loss": 0.0032,
+      "step": 6216
+    },
+    {
+      "epoch": 92.79,
+      "learning_rate": 0.0001346315789473684,
+      "loss": 0.0005,
+      "step": 6217
+    },
+    {
+      "epoch": 92.8,
+      "learning_rate": 0.00013459649122807017,
+      "loss": 0.0136,
+      "step": 6218
+    },
+    {
+      "epoch": 92.82,
+      "learning_rate": 0.00013456140350877192,
+      "loss": 0.0005,
+      "step": 6219
+    },
+    {
+      "epoch": 92.83,
+      "learning_rate": 0.00013452631578947366,
+      "loss": 0.0005,
+      "step": 6220
+    },
+    {
+      "epoch": 92.85,
+      "learning_rate": 0.00013449122807017544,
+      "loss": 0.0007,
+      "step": 6221
+    },
+    {
+      "epoch": 92.86,
+      "learning_rate": 0.0001344561403508772,
+      "loss": 0.0007,
+      "step": 6222
+    },
+    {
+      "epoch": 92.88,
+      "learning_rate": 0.00013442105263157894,
+      "loss": 0.0016,
+      "step": 6223
+    },
+    {
+      "epoch": 92.89,
+      "learning_rate": 0.0001343859649122807,
+      "loss": 0.0035,
+      "step": 6224
+    },
+    {
+      "epoch": 92.91,
+      "learning_rate": 0.00013435087719298244,
+      "loss": 0.0029,
+      "step": 6225
+    },
+    {
+      "epoch": 92.92,
+      "learning_rate": 0.00013431578947368418,
+      "loss": 0.0985,
+      "step": 6226
+    },
+    {
+      "epoch": 92.94,
+      "learning_rate": 0.00013428070175438596,
+      "loss": 0.0301,
+      "step": 6227
+    },
+    {
+      "epoch": 92.95,
+      "learning_rate": 0.0001342456140350877,
+      "loss": 0.0009,
+      "step": 6228
+    },
+    {
+      "epoch": 92.97,
+      "learning_rate": 0.00013421052631578946,
+      "loss": 0.0006,
+      "step": 6229
+    },
+    {
+      "epoch": 92.98,
+      "learning_rate": 0.0001341754385964912,
+      "loss": 0.0004,
+      "step": 6230
+    },
+    {
+      "epoch": 93.0,
+      "learning_rate": 0.00013414035087719298,
+      "loss": 0.1966,
+      "step": 6231
+    },
+    {
+      "epoch": 93.01,
+      "learning_rate": 0.00013410526315789473,
+      "loss": 0.0009,
+      "step": 6232
+    },
+    {
+      "epoch": 93.03,
+      "learning_rate": 0.00013407017543859648,
+      "loss": 0.0005,
+      "step": 6233
+    },
+    {
+      "epoch": 93.04,
+      "learning_rate": 0.00013403508771929823,
+      "loss": 0.0006,
+      "step": 6234
+    },
+    {
+      "epoch": 93.06,
+      "learning_rate": 0.00013399999999999998,
+      "loss": 0.1749,
+      "step": 6235
+    },
+    {
+      "epoch": 93.07,
+      "learning_rate": 0.00013396491228070175,
+      "loss": 0.0007,
+      "step": 6236
+    },
+    {
+      "epoch": 93.09,
+      "learning_rate": 0.0001339298245614035,
+      "loss": 0.0009,
+      "step": 6237
+    },
+    {
+      "epoch": 93.1,
+      "learning_rate": 0.00013389473684210525,
+      "loss": 0.0663,
+      "step": 6238
+    },
+    {
+      "epoch": 93.12,
+      "learning_rate": 0.000133859649122807,
+      "loss": 0.0005,
+      "step": 6239
+    },
+    {
+      "epoch": 93.13,
+      "learning_rate": 0.00013382456140350878,
+      "loss": 0.0692,
+      "step": 6240
+    },
+    {
+      "epoch": 93.15,
+      "learning_rate": 0.0001337894736842105,
+      "loss": 0.001,
+      "step": 6241
+    },
+    {
+      "epoch": 93.16,
+      "learning_rate": 0.00013375438596491227,
+      "loss": 0.0006,
+      "step": 6242
+    },
+    {
+      "epoch": 93.18,
+      "learning_rate": 0.00013371929824561402,
+      "loss": 0.0683,
+      "step": 6243
+    },
+    {
+      "epoch": 93.19,
+      "learning_rate": 0.00013368421052631577,
+      "loss": 0.0006,
+      "step": 6244
+    },
+    {
+      "epoch": 93.21,
+      "learning_rate": 0.00013364912280701755,
+      "loss": 0.002,
+      "step": 6245
+    },
+    {
+      "epoch": 93.22,
+      "learning_rate": 0.0001336140350877193,
+      "loss": 0.0018,
+      "step": 6246
+    },
+    {
+      "epoch": 93.24,
+      "learning_rate": 0.00013357894736842104,
+      "loss": 0.0647,
+      "step": 6247
+    },
+    {
+      "epoch": 93.25,
+      "learning_rate": 0.0001335438596491228,
+      "loss": 0.0004,
+      "step": 6248
+    },
+    {
+      "epoch": 93.27,
+      "learning_rate": 0.00013350877192982454,
+      "loss": 0.0005,
+      "step": 6249
+    },
+    {
+      "epoch": 93.28,
+      "learning_rate": 0.0001334736842105263,
+      "loss": 0.0004,
+      "step": 6250
+    },
+    {
+      "epoch": 93.3,
+      "learning_rate": 0.00013343859649122807,
+      "loss": 0.0005,
+      "step": 6251
+    },
+    {
+      "epoch": 93.31,
+      "learning_rate": 0.00013340350877192982,
+      "loss": 0.2038,
+      "step": 6252
+    },
+    {
+      "epoch": 93.33,
+      "learning_rate": 0.00013336842105263156,
+      "loss": 0.0026,
+      "step": 6253
+    },
+    {
+      "epoch": 93.34,
+      "learning_rate": 0.0001333333333333333,
+      "loss": 0.0317,
+      "step": 6254
+    },
+    {
+      "epoch": 93.36,
+      "learning_rate": 0.0001332982456140351,
+      "loss": 0.0086,
+      "step": 6255
+    },
+    {
+      "epoch": 93.37,
+      "learning_rate": 0.00013326315789473684,
+      "loss": 0.0007,
+      "step": 6256
+    },
+    {
+      "epoch": 93.39,
+      "learning_rate": 0.00013322807017543859,
+      "loss": 0.0007,
+      "step": 6257
+    },
+    {
+      "epoch": 93.4,
+      "learning_rate": 0.00013319298245614033,
+      "loss": 0.0006,
+      "step": 6258
+    },
+    {
+      "epoch": 93.42,
+      "learning_rate": 0.00013315789473684208,
+      "loss": 0.0013,
+      "step": 6259
+    },
+    {
+      "epoch": 93.43,
+      "learning_rate": 0.00013312280701754386,
+      "loss": 0.0005,
+      "step": 6260
+    },
+    {
+      "epoch": 93.45,
+      "learning_rate": 0.0001330877192982456,
+      "loss": 0.0021,
+      "step": 6261
+    },
+    {
+      "epoch": 93.46,
+      "learning_rate": 0.00013305263157894736,
+      "loss": 0.0004,
+      "step": 6262
+    },
+    {
+      "epoch": 93.48,
+      "learning_rate": 0.0001330175438596491,
+      "loss": 0.0004,
+      "step": 6263
+    },
+    {
+      "epoch": 93.49,
+      "learning_rate": 0.00013298245614035085,
+      "loss": 0.0005,
+      "step": 6264
+    },
+    {
+      "epoch": 93.51,
+      "learning_rate": 0.0001329473684210526,
+      "loss": 0.0008,
+      "step": 6265
+    },
+    {
+      "epoch": 93.52,
+      "learning_rate": 0.00013291228070175438,
+      "loss": 0.0006,
+      "step": 6266
+    },
+    {
+      "epoch": 93.54,
+      "learning_rate": 0.00013287719298245613,
+      "loss": 0.0008,
+      "step": 6267
+    },
+    {
+      "epoch": 93.55,
+      "learning_rate": 0.00013284210526315788,
+      "loss": 0.0007,
+      "step": 6268
+    },
+    {
+      "epoch": 93.57,
+      "learning_rate": 0.00013280701754385965,
+      "loss": 0.0006,
+      "step": 6269
+    },
+    {
+      "epoch": 93.58,
+      "learning_rate": 0.0001327719298245614,
+      "loss": 0.0005,
+      "step": 6270
+    },
+    {
+      "epoch": 93.59,
+      "learning_rate": 0.00013273684210526315,
+      "loss": 0.0024,
+      "step": 6271
+    },
+    {
+      "epoch": 93.61,
+      "learning_rate": 0.0001327017543859649,
+      "loss": 0.0011,
+      "step": 6272
+    },
+    {
+      "epoch": 93.62,
+      "learning_rate": 0.00013266666666666665,
+      "loss": 0.0007,
+      "step": 6273
+    },
+    {
+      "epoch": 93.64,
+      "learning_rate": 0.0001326315789473684,
+      "loss": 0.001,
+      "step": 6274
+    },
+    {
+      "epoch": 93.65,
+      "learning_rate": 0.00013259649122807017,
+      "loss": 0.0005,
+      "step": 6275
+    },
+    {
+      "epoch": 93.67,
+      "learning_rate": 0.00013256140350877192,
+      "loss": 0.0005,
+      "step": 6276
+    },
+    {
+      "epoch": 93.68,
+      "learning_rate": 0.00013252631578947367,
+      "loss": 0.0384,
+      "step": 6277
+    },
+    {
+      "epoch": 93.7,
+      "learning_rate": 0.00013249122807017542,
+      "loss": 0.0009,
+      "step": 6278
+    },
+    {
+      "epoch": 93.71,
+      "learning_rate": 0.0001324561403508772,
+      "loss": 0.0007,
+      "step": 6279
+    },
+    {
+      "epoch": 93.73,
+      "learning_rate": 0.00013242105263157892,
+      "loss": 0.0027,
+      "step": 6280
+    },
+    {
+      "epoch": 93.74,
+      "learning_rate": 0.0001323859649122807,
+      "loss": 0.0005,
+      "step": 6281
+    },
+    {
+      "epoch": 93.76,
+      "learning_rate": 0.00013235087719298244,
+      "loss": 0.0039,
+      "step": 6282
+    },
+    {
+      "epoch": 93.77,
+      "learning_rate": 0.0001323157894736842,
+      "loss": 0.0004,
+      "step": 6283
+    },
+    {
+      "epoch": 93.79,
+      "learning_rate": 0.00013228070175438597,
+      "loss": 0.001,
+      "step": 6284
+    },
+    {
+      "epoch": 93.8,
+      "learning_rate": 0.00013224561403508771,
+      "loss": 0.001,
+      "step": 6285
+    },
+    {
+      "epoch": 93.82,
+      "learning_rate": 0.00013221052631578946,
+      "loss": 0.1333,
+      "step": 6286
+    },
+    {
+      "epoch": 93.83,
+      "learning_rate": 0.0001321754385964912,
+      "loss": 0.0008,
+      "step": 6287
+    },
+    {
+      "epoch": 93.85,
+      "learning_rate": 0.00013214035087719296,
+      "loss": 0.0007,
+      "step": 6288
+    },
+    {
+      "epoch": 93.86,
+      "learning_rate": 0.0001321052631578947,
+      "loss": 0.0018,
+      "step": 6289
+    },
+    {
+      "epoch": 93.88,
+      "learning_rate": 0.00013207017543859649,
+      "loss": 0.1526,
+      "step": 6290
+    },
+    {
+      "epoch": 93.89,
+      "learning_rate": 0.00013203508771929823,
+      "loss": 0.0005,
+      "step": 6291
+    },
+    {
+      "epoch": 93.91,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 0.0031,
+      "step": 6292
+    },
+    {
+      "epoch": 93.92,
+      "learning_rate": 0.00013196491228070173,
+      "loss": 0.0005,
+      "step": 6293
+    },
+    {
+      "epoch": 93.94,
+      "learning_rate": 0.0001319298245614035,
+      "loss": 0.0926,
+      "step": 6294
+    },
+    {
+      "epoch": 93.95,
+      "learning_rate": 0.00013189473684210526,
+      "loss": 0.0006,
+      "step": 6295
+    },
+    {
+      "epoch": 93.97,
+      "learning_rate": 0.000131859649122807,
+      "loss": 0.0005,
+      "step": 6296
+    },
+    {
+      "epoch": 93.98,
+      "learning_rate": 0.00013182456140350875,
+      "loss": 0.0016,
+      "step": 6297
+    },
+    {
+      "epoch": 94.0,
+      "learning_rate": 0.0001317894736842105,
+      "loss": 0.0028,
+      "step": 6298
+    },
+    {
+      "epoch": 94.01,
+      "learning_rate": 0.00013175438596491228,
+      "loss": 0.0095,
+      "step": 6299
+    },
+    {
+      "epoch": 94.03,
+      "learning_rate": 0.00013171929824561403,
+      "loss": 0.0258,
+      "step": 6300
+    },
+    {
+      "epoch": 94.04,
+      "learning_rate": 0.00013168421052631578,
+      "loss": 0.0004,
+      "step": 6301
+    },
+    {
+      "epoch": 94.06,
+      "learning_rate": 0.00013164912280701753,
+      "loss": 0.0005,
+      "step": 6302
+    },
+    {
+      "epoch": 94.07,
+      "learning_rate": 0.0001316140350877193,
+      "loss": 0.0438,
+      "step": 6303
+    },
+    {
+      "epoch": 94.09,
+      "learning_rate": 0.00013157894736842102,
+      "loss": 0.0025,
+      "step": 6304
+    },
+    {
+      "epoch": 94.1,
+      "learning_rate": 0.0001315438596491228,
+      "loss": 0.0166,
+      "step": 6305
+    },
+    {
+      "epoch": 94.12,
+      "learning_rate": 0.00013150877192982455,
+      "loss": 0.0006,
+      "step": 6306
+    },
+    {
+      "epoch": 94.13,
+      "learning_rate": 0.0001314736842105263,
+      "loss": 0.0005,
+      "step": 6307
+    },
+    {
+      "epoch": 94.15,
+      "learning_rate": 0.00013143859649122807,
+      "loss": 0.0006,
+      "step": 6308
+    },
+    {
+      "epoch": 94.16,
+      "learning_rate": 0.00013140350877192982,
+      "loss": 0.0017,
+      "step": 6309
+    },
+    {
+      "epoch": 94.18,
+      "learning_rate": 0.00013136842105263157,
+      "loss": 0.0004,
+      "step": 6310
+    },
+    {
+      "epoch": 94.19,
+      "learning_rate": 0.00013133333333333332,
+      "loss": 0.0006,
+      "step": 6311
+    },
+    {
+      "epoch": 94.21,
+      "learning_rate": 0.00013129824561403507,
+      "loss": 0.0006,
+      "step": 6312
+    },
+    {
+      "epoch": 94.22,
+      "learning_rate": 0.00013126315789473682,
+      "loss": 0.0004,
+      "step": 6313
+    },
+    {
+      "epoch": 94.24,
+      "learning_rate": 0.0001312280701754386,
+      "loss": 0.0004,
+      "step": 6314
+    },
+    {
+      "epoch": 94.25,
+      "learning_rate": 0.00013119298245614034,
+      "loss": 0.0006,
+      "step": 6315
+    },
+    {
+      "epoch": 94.27,
+      "learning_rate": 0.0001311578947368421,
+      "loss": 0.0004,
+      "step": 6316
+    },
+    {
+      "epoch": 94.28,
+      "learning_rate": 0.00013112280701754384,
+      "loss": 0.05,
+      "step": 6317
+    },
+    {
+      "epoch": 94.3,
+      "learning_rate": 0.00013108771929824561,
+      "loss": 0.0743,
+      "step": 6318
+    },
+    {
+      "epoch": 94.31,
+      "learning_rate": 0.00013105263157894736,
+      "loss": 0.0017,
+      "step": 6319
+    },
+    {
+      "epoch": 94.33,
+      "learning_rate": 0.0001310175438596491,
+      "loss": 0.1355,
+      "step": 6320
+    },
+    {
+      "epoch": 94.34,
+      "learning_rate": 0.00013098245614035086,
+      "loss": 0.0004,
+      "step": 6321
+    },
+    {
+      "epoch": 94.36,
+      "learning_rate": 0.0001309473684210526,
+      "loss": 0.0004,
+      "step": 6322
+    },
+    {
+      "epoch": 94.37,
+      "learning_rate": 0.00013091228070175438,
+      "loss": 0.0094,
+      "step": 6323
+    },
+    {
+      "epoch": 94.39,
+      "learning_rate": 0.00013087719298245613,
+      "loss": 0.0006,
+      "step": 6324
+    },
+    {
+      "epoch": 94.4,
+      "learning_rate": 0.00013084210526315788,
+      "loss": 0.0006,
+      "step": 6325
+    },
+    {
+      "epoch": 94.42,
+      "learning_rate": 0.00013080701754385963,
+      "loss": 0.0016,
+      "step": 6326
+    },
+    {
+      "epoch": 94.43,
+      "learning_rate": 0.0001307719298245614,
+      "loss": 0.0007,
+      "step": 6327
+    },
+    {
+      "epoch": 94.45,
+      "learning_rate": 0.00013073684210526313,
+      "loss": 0.0006,
+      "step": 6328
+    },
+    {
+      "epoch": 94.46,
+      "learning_rate": 0.0001307017543859649,
+      "loss": 0.0005,
+      "step": 6329
+    },
+    {
+      "epoch": 94.48,
+      "learning_rate": 0.00013066666666666665,
+      "loss": 0.0007,
+      "step": 6330
+    },
+    {
+      "epoch": 94.49,
+      "learning_rate": 0.0001306315789473684,
+      "loss": 0.0004,
+      "step": 6331
+    },
+    {
+      "epoch": 94.51,
+      "learning_rate": 0.00013059649122807018,
+      "loss": 0.0312,
+      "step": 6332
+    },
+    {
+      "epoch": 94.52,
+      "learning_rate": 0.00013056140350877193,
+      "loss": 0.0006,
+      "step": 6333
+    },
+    {
+      "epoch": 94.54,
+      "learning_rate": 0.00013052631578947368,
+      "loss": 0.0689,
+      "step": 6334
+    },
+    {
+      "epoch": 94.55,
+      "learning_rate": 0.00013049122807017542,
+      "loss": 0.0004,
+      "step": 6335
+    },
+    {
+      "epoch": 94.57,
+      "learning_rate": 0.00013045614035087717,
+      "loss": 0.002,
+      "step": 6336
+    },
+    {
+      "epoch": 94.58,
+      "learning_rate": 0.00013042105263157892,
+      "loss": 0.0004,
+      "step": 6337
+    },
+    {
+      "epoch": 94.59,
+      "learning_rate": 0.0001303859649122807,
+      "loss": 0.0005,
+      "step": 6338
+    },
+    {
+      "epoch": 94.61,
+      "learning_rate": 0.00013035087719298245,
+      "loss": 0.0005,
+      "step": 6339
+    },
+    {
+      "epoch": 94.62,
+      "learning_rate": 0.0001303157894736842,
+      "loss": 0.0027,
+      "step": 6340
+    },
+    {
+      "epoch": 94.64,
+      "learning_rate": 0.00013028070175438594,
+      "loss": 0.0174,
+      "step": 6341
+    },
+    {
+      "epoch": 94.65,
+      "learning_rate": 0.00013024561403508772,
+      "loss": 0.0006,
+      "step": 6342
+    },
+    {
+      "epoch": 94.67,
+      "learning_rate": 0.00013021052631578944,
+      "loss": 0.0006,
+      "step": 6343
+    },
+    {
+      "epoch": 94.68,
+      "learning_rate": 0.00013017543859649122,
+      "loss": 0.0004,
+      "step": 6344
+    },
+    {
+      "epoch": 94.7,
+      "learning_rate": 0.00013014035087719297,
+      "loss": 0.0829,
+      "step": 6345
+    },
+    {
+      "epoch": 94.71,
+      "learning_rate": 0.00013010526315789472,
+      "loss": 0.0004,
+      "step": 6346
+    },
+    {
+      "epoch": 94.73,
+      "learning_rate": 0.0001300701754385965,
+      "loss": 0.0004,
+      "step": 6347
+    },
+    {
+      "epoch": 94.74,
+      "learning_rate": 0.00013003508771929824,
+      "loss": 0.0714,
+      "step": 6348
+    },
+    {
+      "epoch": 94.76,
+      "learning_rate": 0.00013,
+      "loss": 0.0005,
+      "step": 6349
+    },
+    {
+      "epoch": 94.77,
+      "learning_rate": 0.00012996491228070174,
+      "loss": 0.0005,
+      "step": 6350
+    },
+    {
+      "epoch": 94.79,
+      "learning_rate": 0.0001299298245614035,
+      "loss": 0.048,
+      "step": 6351
+    },
+    {
+      "epoch": 94.8,
+      "learning_rate": 0.00012989473684210523,
+      "loss": 0.0013,
+      "step": 6352
+    },
+    {
+      "epoch": 94.82,
+      "learning_rate": 0.000129859649122807,
+      "loss": 0.0238,
+      "step": 6353
+    },
+    {
+      "epoch": 94.83,
+      "learning_rate": 0.00012982456140350876,
+      "loss": 0.0007,
+      "step": 6354
+    },
+    {
+      "epoch": 94.85,
+      "learning_rate": 0.0001297894736842105,
+      "loss": 0.0733,
+      "step": 6355
+    },
+    {
+      "epoch": 94.86,
+      "learning_rate": 0.00012975438596491228,
+      "loss": 0.0005,
+      "step": 6356
+    },
+    {
+      "epoch": 94.88,
+      "learning_rate": 0.00012971929824561403,
+      "loss": 0.0747,
+      "step": 6357
+    },
+    {
+      "epoch": 94.89,
+      "learning_rate": 0.00012968421052631578,
+      "loss": 0.0004,
+      "step": 6358
+    },
+    {
+      "epoch": 94.91,
+      "learning_rate": 0.00012964912280701753,
+      "loss": 0.0004,
+      "step": 6359
+    },
+    {
+      "epoch": 94.92,
+      "learning_rate": 0.00012961403508771928,
+      "loss": 0.0004,
+      "step": 6360
+    },
+    {
+      "epoch": 94.94,
+      "learning_rate": 0.00012957894736842103,
+      "loss": 0.0108,
+      "step": 6361
+    },
+    {
+      "epoch": 94.95,
+      "learning_rate": 0.0001295438596491228,
+      "loss": 0.0044,
+      "step": 6362
+    },
+    {
+      "epoch": 94.97,
+      "learning_rate": 0.00012950877192982455,
+      "loss": 0.0004,
+      "step": 6363
+    },
+    {
+      "epoch": 94.98,
+      "learning_rate": 0.0001294736842105263,
+      "loss": 0.0004,
+      "step": 6364
+    },
+    {
+      "epoch": 95.0,
+      "learning_rate": 0.00012943859649122805,
+      "loss": 0.0027,
+      "step": 6365
+    },
+    {
+      "epoch": 95.01,
+      "learning_rate": 0.00012940350877192983,
+      "loss": 0.0074,
+      "step": 6366
+    },
+    {
+      "epoch": 95.03,
+      "learning_rate": 0.00012936842105263155,
+      "loss": 0.0005,
+      "step": 6367
+    },
+    {
+      "epoch": 95.04,
+      "learning_rate": 0.00012933333333333332,
+      "loss": 0.0004,
+      "step": 6368
+    },
+    {
+      "epoch": 95.06,
+      "learning_rate": 0.00012929824561403507,
+      "loss": 0.0004,
+      "step": 6369
+    },
+    {
+      "epoch": 95.07,
+      "learning_rate": 0.00012926315789473682,
+      "loss": 0.0162,
+      "step": 6370
+    },
+    {
+      "epoch": 95.09,
+      "learning_rate": 0.0001292280701754386,
+      "loss": 0.0012,
+      "step": 6371
+    },
+    {
+      "epoch": 95.1,
+      "learning_rate": 0.00012919298245614035,
+      "loss": 0.0006,
+      "step": 6372
+    },
+    {
+      "epoch": 95.12,
+      "learning_rate": 0.0001291578947368421,
+      "loss": 0.0157,
+      "step": 6373
+    },
+    {
+      "epoch": 95.13,
+      "learning_rate": 0.00012912280701754384,
+      "loss": 0.0005,
+      "step": 6374
+    },
+    {
+      "epoch": 95.15,
+      "learning_rate": 0.00012908771929824562,
+      "loss": 0.0015,
+      "step": 6375
+    },
+    {
+      "epoch": 95.16,
+      "learning_rate": 0.00012905263157894734,
+      "loss": 0.0004,
+      "step": 6376
+    },
+    {
+      "epoch": 95.18,
+      "learning_rate": 0.00012901754385964912,
+      "loss": 0.1115,
+      "step": 6377
+    },
+    {
+      "epoch": 95.19,
+      "learning_rate": 0.00012898245614035087,
+      "loss": 0.0723,
+      "step": 6378
+    },
+    {
+      "epoch": 95.21,
+      "learning_rate": 0.00012894736842105261,
+      "loss": 0.0145,
+      "step": 6379
+    },
+    {
+      "epoch": 95.22,
+      "learning_rate": 0.00012891228070175436,
+      "loss": 0.0005,
+      "step": 6380
+    },
+    {
+      "epoch": 95.24,
+      "learning_rate": 0.00012887719298245614,
+      "loss": 0.1661,
+      "step": 6381
+    },
+    {
+      "epoch": 95.25,
+      "learning_rate": 0.0001288421052631579,
+      "loss": 0.0004,
+      "step": 6382
+    },
+    {
+      "epoch": 95.27,
+      "learning_rate": 0.00012880701754385964,
+      "loss": 0.001,
+      "step": 6383
+    },
+    {
+      "epoch": 95.28,
+      "learning_rate": 0.00012877192982456139,
+      "loss": 0.0004,
+      "step": 6384
+    },
+    {
+      "epoch": 95.3,
+      "learning_rate": 0.00012873684210526313,
+      "loss": 0.2419,
+      "step": 6385
+    },
+    {
+      "epoch": 95.31,
+      "learning_rate": 0.0001287017543859649,
+      "loss": 0.0008,
+      "step": 6386
+    },
+    {
+      "epoch": 95.33,
+      "learning_rate": 0.00012866666666666666,
+      "loss": 0.2924,
+      "step": 6387
+    },
+    {
+      "epoch": 95.34,
+      "learning_rate": 0.0001286315789473684,
+      "loss": 0.0004,
+      "step": 6388
+    },
+    {
+      "epoch": 95.36,
+      "learning_rate": 0.00012859649122807016,
+      "loss": 0.1491,
+      "step": 6389
+    },
+    {
+      "epoch": 95.37,
+      "learning_rate": 0.00012856140350877193,
+      "loss": 0.0007,
+      "step": 6390
+    },
+    {
+      "epoch": 95.39,
+      "learning_rate": 0.00012852631578947365,
+      "loss": 0.0036,
+      "step": 6391
+    },
+    {
+      "epoch": 95.4,
+      "learning_rate": 0.00012849122807017543,
+      "loss": 0.0036,
+      "step": 6392
+    },
+    {
+      "epoch": 95.42,
+      "learning_rate": 0.00012845614035087718,
+      "loss": 0.0024,
+      "step": 6393
+    },
+    {
+      "epoch": 95.43,
+      "learning_rate": 0.00012842105263157893,
+      "loss": 0.0925,
+      "step": 6394
+    },
+    {
+      "epoch": 95.45,
+      "learning_rate": 0.0001283859649122807,
+      "loss": 0.0011,
+      "step": 6395
+    },
+    {
+      "epoch": 95.46,
+      "learning_rate": 0.00012835087719298245,
+      "loss": 0.0024,
+      "step": 6396
+    },
+    {
+      "epoch": 95.48,
+      "learning_rate": 0.0001283157894736842,
+      "loss": 0.0005,
+      "step": 6397
+    },
+    {
+      "epoch": 95.49,
+      "learning_rate": 0.00012828070175438595,
+      "loss": 0.0031,
+      "step": 6398
+    },
+    {
+      "epoch": 95.51,
+      "learning_rate": 0.00012824561403508773,
+      "loss": 0.0012,
+      "step": 6399
+    },
+    {
+      "epoch": 95.52,
+      "learning_rate": 0.00012821052631578945,
+      "loss": 0.0017,
+      "step": 6400
+    },
+    {
+      "epoch": 95.52,
+      "eval_accuracy": 0.8668624571708272,
+      "eval_f1": 0.8678242564825082,
+      "eval_loss": 0.7071595191955566,
+      "eval_runtime": 345.1767,
+      "eval_samples_per_second": 11.837,
+      "eval_steps_per_second": 0.742,
+      "step": 6400
+    },
+    {
+      "epoch": 95.54,
+      "learning_rate": 0.00012817543859649122,
+      "loss": 0.0011,
+      "step": 6401
+    },
+    {
+      "epoch": 95.55,
+      "learning_rate": 0.00012814035087719297,
+      "loss": 0.1979,
+      "step": 6402
+    },
+    {
+      "epoch": 95.57,
+      "learning_rate": 0.00012810526315789472,
+      "loss": 0.205,
+      "step": 6403
+    },
+    {
+      "epoch": 95.58,
+      "learning_rate": 0.00012807017543859647,
+      "loss": 0.2207,
+      "step": 6404
+    },
+    {
+      "epoch": 95.59,
+      "learning_rate": 0.00012803508771929825,
+      "loss": 0.0012,
+      "step": 6405
+    },
+    {
+      "epoch": 95.61,
+      "learning_rate": 0.000128,
+      "loss": 0.0026,
+      "step": 6406
+    },
+    {
+      "epoch": 95.62,
+      "learning_rate": 0.00012796491228070174,
+      "loss": 0.0007,
+      "step": 6407
+    },
+    {
+      "epoch": 95.64,
+      "learning_rate": 0.0001279298245614035,
+      "loss": 0.0007,
+      "step": 6408
+    },
+    {
+      "epoch": 95.65,
+      "learning_rate": 0.00012789473684210524,
+      "loss": 0.0006,
+      "step": 6409
+    },
+    {
+      "epoch": 95.67,
+      "learning_rate": 0.00012785964912280702,
+      "loss": 0.0007,
+      "step": 6410
+    },
+    {
+      "epoch": 95.68,
+      "learning_rate": 0.00012782456140350877,
+      "loss": 0.0044,
+      "step": 6411
+    },
+    {
+      "epoch": 95.7,
+      "learning_rate": 0.00012778947368421051,
+      "loss": 0.0106,
+      "step": 6412
+    },
+    {
+      "epoch": 95.71,
+      "learning_rate": 0.00012775438596491226,
+      "loss": 0.0007,
+      "step": 6413
+    },
+    {
+      "epoch": 95.73,
+      "learning_rate": 0.00012771929824561404,
+      "loss": 0.0008,
+      "step": 6414
+    },
+    {
+      "epoch": 95.74,
+      "learning_rate": 0.00012768421052631576,
+      "loss": 0.0084,
+      "step": 6415
+    },
+    {
+      "epoch": 95.76,
+      "learning_rate": 0.00012764912280701754,
+      "loss": 0.0012,
+      "step": 6416
+    },
+    {
+      "epoch": 95.77,
+      "learning_rate": 0.00012761403508771928,
+      "loss": 0.0544,
+      "step": 6417
+    },
+    {
+      "epoch": 95.79,
+      "learning_rate": 0.00012757894736842103,
+      "loss": 0.01,
+      "step": 6418
+    },
+    {
+      "epoch": 95.8,
+      "learning_rate": 0.0001275438596491228,
+      "loss": 0.0022,
+      "step": 6419
+    },
+    {
+      "epoch": 95.82,
+      "learning_rate": 0.00012750877192982456,
+      "loss": 0.1372,
+      "step": 6420
+    },
+    {
+      "epoch": 95.83,
+      "learning_rate": 0.0001274736842105263,
+      "loss": 0.0036,
+      "step": 6421
+    },
+    {
+      "epoch": 95.85,
+      "learning_rate": 0.00012743859649122806,
+      "loss": 0.0005,
+      "step": 6422
+    },
+    {
+      "epoch": 95.86,
+      "learning_rate": 0.00012740350877192983,
+      "loss": 0.0006,
+      "step": 6423
+    },
+    {
+      "epoch": 95.88,
+      "learning_rate": 0.00012736842105263155,
+      "loss": 0.0088,
+      "step": 6424
+    },
+    {
+      "epoch": 95.89,
+      "learning_rate": 0.00012733333333333333,
+      "loss": 0.0007,
+      "step": 6425
+    },
+    {
+      "epoch": 95.91,
+      "learning_rate": 0.00012729824561403508,
+      "loss": 0.1873,
+      "step": 6426
+    },
+    {
+      "epoch": 95.92,
+      "learning_rate": 0.00012726315789473683,
+      "loss": 0.0013,
+      "step": 6427
+    },
+    {
+      "epoch": 95.94,
+      "learning_rate": 0.00012722807017543858,
+      "loss": 0.0006,
+      "step": 6428
+    },
+    {
+      "epoch": 95.95,
+      "learning_rate": 0.00012719298245614035,
+      "loss": 0.0006,
+      "step": 6429
+    },
+    {
+      "epoch": 95.97,
+      "learning_rate": 0.00012715789473684207,
+      "loss": 0.0028,
+      "step": 6430
+    },
+    {
+      "epoch": 95.98,
+      "learning_rate": 0.00012712280701754385,
+      "loss": 0.0006,
+      "step": 6431
+    },
+    {
+      "epoch": 96.0,
+      "learning_rate": 0.0001270877192982456,
+      "loss": 0.359,
+      "step": 6432
+    },
+    {
+      "epoch": 96.01,
+      "learning_rate": 0.00012705263157894735,
+      "loss": 0.0027,
+      "step": 6433
+    },
+    {
+      "epoch": 96.03,
+      "learning_rate": 0.00012701754385964912,
+      "loss": 0.0006,
+      "step": 6434
+    },
+    {
+      "epoch": 96.04,
+      "learning_rate": 0.00012698245614035087,
+      "loss": 0.0006,
+      "step": 6435
+    },
+    {
+      "epoch": 96.06,
+      "learning_rate": 0.00012694736842105262,
+      "loss": 0.1297,
+      "step": 6436
+    },
+    {
+      "epoch": 96.07,
+      "learning_rate": 0.00012691228070175437,
+      "loss": 0.0008,
+      "step": 6437
+    },
+    {
+      "epoch": 96.09,
+      "learning_rate": 0.00012687719298245614,
+      "loss": 0.0014,
+      "step": 6438
+    },
+    {
+      "epoch": 96.1,
+      "learning_rate": 0.00012684210526315787,
+      "loss": 0.045,
+      "step": 6439
+    },
+    {
+      "epoch": 96.12,
+      "learning_rate": 0.00012680701754385964,
+      "loss": 0.0009,
+      "step": 6440
+    },
+    {
+      "epoch": 96.13,
+      "learning_rate": 0.0001267719298245614,
+      "loss": 0.0013,
+      "step": 6441
+    },
+    {
+      "epoch": 96.15,
+      "learning_rate": 0.00012673684210526314,
+      "loss": 0.002,
+      "step": 6442
+    },
+    {
+      "epoch": 96.16,
+      "learning_rate": 0.0001267017543859649,
+      "loss": 0.0005,
+      "step": 6443
+    },
+    {
+      "epoch": 96.18,
+      "learning_rate": 0.00012666666666666666,
+      "loss": 0.1275,
+      "step": 6444
+    },
+    {
+      "epoch": 96.19,
+      "learning_rate": 0.0001266315789473684,
+      "loss": 0.0005,
+      "step": 6445
+    },
+    {
+      "epoch": 96.21,
+      "learning_rate": 0.00012659649122807016,
+      "loss": 0.0006,
+      "step": 6446
+    },
+    {
+      "epoch": 96.22,
+      "learning_rate": 0.00012656140350877194,
+      "loss": 0.0136,
+      "step": 6447
+    },
+    {
+      "epoch": 96.24,
+      "learning_rate": 0.00012652631578947366,
+      "loss": 0.038,
+      "step": 6448
+    },
+    {
+      "epoch": 96.25,
+      "learning_rate": 0.00012649122807017544,
+      "loss": 0.0005,
+      "step": 6449
+    },
+    {
+      "epoch": 96.27,
+      "learning_rate": 0.00012645614035087718,
+      "loss": 0.003,
+      "step": 6450
+    },
+    {
+      "epoch": 96.28,
+      "learning_rate": 0.00012642105263157893,
+      "loss": 0.0006,
+      "step": 6451
+    },
+    {
+      "epoch": 96.3,
+      "learning_rate": 0.00012638596491228068,
+      "loss": 0.0449,
+      "step": 6452
+    },
+    {
+      "epoch": 96.31,
+      "learning_rate": 0.00012635087719298246,
+      "loss": 0.1171,
+      "step": 6453
+    },
+    {
+      "epoch": 96.33,
+      "learning_rate": 0.00012631578947368418,
+      "loss": 0.0005,
+      "step": 6454
+    },
+    {
+      "epoch": 96.34,
+      "learning_rate": 0.00012628070175438596,
+      "loss": 0.0008,
+      "step": 6455
+    },
+    {
+      "epoch": 96.36,
+      "learning_rate": 0.0001262456140350877,
+      "loss": 0.0006,
+      "step": 6456
+    },
+    {
+      "epoch": 96.37,
+      "learning_rate": 0.00012621052631578945,
+      "loss": 0.0016,
+      "step": 6457
+    },
+    {
+      "epoch": 96.39,
+      "learning_rate": 0.00012617543859649123,
+      "loss": 0.0027,
+      "step": 6458
+    },
+    {
+      "epoch": 96.4,
+      "learning_rate": 0.00012614035087719298,
+      "loss": 0.1313,
+      "step": 6459
+    },
+    {
+      "epoch": 96.42,
+      "learning_rate": 0.00012610526315789473,
+      "loss": 0.0006,
+      "step": 6460
+    },
+    {
+      "epoch": 96.43,
+      "learning_rate": 0.00012607017543859648,
+      "loss": 0.0006,
+      "step": 6461
+    },
+    {
+      "epoch": 96.45,
+      "learning_rate": 0.00012603508771929825,
+      "loss": 0.0048,
+      "step": 6462
+    },
+    {
+      "epoch": 96.46,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 0.0032,
+      "step": 6463
+    },
+    {
+      "epoch": 96.48,
+      "learning_rate": 0.00012596491228070175,
+      "loss": 0.0005,
+      "step": 6464
+    },
+    {
+      "epoch": 96.49,
+      "learning_rate": 0.0001259298245614035,
+      "loss": 0.0802,
+      "step": 6465
+    },
+    {
+      "epoch": 96.51,
+      "learning_rate": 0.00012589473684210525,
+      "loss": 0.0009,
+      "step": 6466
+    },
+    {
+      "epoch": 96.52,
+      "learning_rate": 0.000125859649122807,
+      "loss": 0.0037,
+      "step": 6467
+    },
+    {
+      "epoch": 96.54,
+      "learning_rate": 0.00012582456140350877,
+      "loss": 0.0009,
+      "step": 6468
+    },
+    {
+      "epoch": 96.55,
+      "learning_rate": 0.00012578947368421052,
+      "loss": 0.0008,
+      "step": 6469
+    },
+    {
+      "epoch": 96.57,
+      "learning_rate": 0.00012575438596491227,
+      "loss": 0.0007,
+      "step": 6470
+    },
+    {
+      "epoch": 96.58,
+      "learning_rate": 0.00012571929824561404,
+      "loss": 0.0011,
+      "step": 6471
+    },
+    {
+      "epoch": 96.59,
+      "learning_rate": 0.00012568421052631577,
+      "loss": 0.0257,
+      "step": 6472
+    },
+    {
+      "epoch": 96.61,
+      "learning_rate": 0.00012564912280701754,
+      "loss": 0.014,
+      "step": 6473
+    },
+    {
+      "epoch": 96.62,
+      "learning_rate": 0.0001256140350877193,
+      "loss": 0.0006,
+      "step": 6474
+    },
+    {
+      "epoch": 96.64,
+      "learning_rate": 0.00012557894736842104,
+      "loss": 0.0185,
+      "step": 6475
+    },
+    {
+      "epoch": 96.65,
+      "learning_rate": 0.0001255438596491228,
+      "loss": 0.0005,
+      "step": 6476
+    },
+    {
+      "epoch": 96.67,
+      "learning_rate": 0.00012550877192982456,
+      "loss": 0.0009,
+      "step": 6477
+    },
+    {
+      "epoch": 96.68,
+      "learning_rate": 0.00012547368421052629,
+      "loss": 0.0007,
+      "step": 6478
+    },
+    {
+      "epoch": 96.7,
+      "learning_rate": 0.00012543859649122806,
+      "loss": 0.0401,
+      "step": 6479
+    },
+    {
+      "epoch": 96.71,
+      "learning_rate": 0.0001254035087719298,
+      "loss": 0.0007,
+      "step": 6480
+    },
+    {
+      "epoch": 96.73,
+      "learning_rate": 0.00012536842105263156,
+      "loss": 0.0008,
+      "step": 6481
+    },
+    {
+      "epoch": 96.74,
+      "learning_rate": 0.00012533333333333334,
+      "loss": 0.0511,
+      "step": 6482
+    },
+    {
+      "epoch": 96.76,
+      "learning_rate": 0.00012529824561403508,
+      "loss": 0.0219,
+      "step": 6483
+    },
+    {
+      "epoch": 96.77,
+      "learning_rate": 0.00012526315789473683,
+      "loss": 0.0034,
+      "step": 6484
+    },
+    {
+      "epoch": 96.79,
+      "learning_rate": 0.00012522807017543858,
+      "loss": 0.0456,
+      "step": 6485
+    },
+    {
+      "epoch": 96.8,
+      "learning_rate": 0.00012519298245614036,
+      "loss": 0.0006,
+      "step": 6486
+    },
+    {
+      "epoch": 96.82,
+      "learning_rate": 0.00012515789473684208,
+      "loss": 0.0041,
+      "step": 6487
+    },
+    {
+      "epoch": 96.83,
+      "learning_rate": 0.00012512280701754385,
+      "loss": 0.0008,
+      "step": 6488
+    },
+    {
+      "epoch": 96.85,
+      "learning_rate": 0.0001250877192982456,
+      "loss": 0.0006,
+      "step": 6489
+    },
+    {
+      "epoch": 96.86,
+      "learning_rate": 0.00012505263157894735,
+      "loss": 0.0044,
+      "step": 6490
+    },
+    {
+      "epoch": 96.88,
+      "learning_rate": 0.0001250175438596491,
+      "loss": 0.0006,
+      "step": 6491
+    },
+    {
+      "epoch": 96.89,
+      "learning_rate": 0.00012498245614035088,
+      "loss": 0.0007,
+      "step": 6492
+    },
+    {
+      "epoch": 96.91,
+      "learning_rate": 0.00012494736842105263,
+      "loss": 0.0006,
+      "step": 6493
+    },
+    {
+      "epoch": 96.92,
+      "learning_rate": 0.00012491228070175437,
+      "loss": 0.0076,
+      "step": 6494
+    },
+    {
+      "epoch": 96.94,
+      "learning_rate": 0.00012487719298245612,
+      "loss": 0.3258,
+      "step": 6495
+    },
+    {
+      "epoch": 96.95,
+      "learning_rate": 0.00012484210526315787,
+      "loss": 0.0007,
+      "step": 6496
+    },
+    {
+      "epoch": 96.97,
+      "learning_rate": 0.00012480701754385965,
+      "loss": 0.0005,
+      "step": 6497
+    },
+    {
+      "epoch": 96.98,
+      "learning_rate": 0.0001247719298245614,
+      "loss": 0.0013,
+      "step": 6498
+    },
+    {
+      "epoch": 97.0,
+      "learning_rate": 0.00012473684210526315,
+      "loss": 0.1592,
+      "step": 6499
+    },
+    {
+      "epoch": 97.01,
+      "learning_rate": 0.0001247017543859649,
+      "loss": 0.0008,
+      "step": 6500
+    },
+    {
+      "epoch": 97.03,
+      "learning_rate": 0.00012466666666666667,
+      "loss": 0.0006,
+      "step": 6501
+    },
+    {
+      "epoch": 97.04,
+      "learning_rate": 0.0001246315789473684,
+      "loss": 0.1534,
+      "step": 6502
+    },
+    {
+      "epoch": 97.06,
+      "learning_rate": 0.00012459649122807017,
+      "loss": 0.0008,
+      "step": 6503
+    },
+    {
+      "epoch": 97.07,
+      "learning_rate": 0.00012456140350877192,
+      "loss": 0.001,
+      "step": 6504
+    },
+    {
+      "epoch": 97.09,
+      "learning_rate": 0.00012452631578947367,
+      "loss": 0.0008,
+      "step": 6505
+    },
+    {
+      "epoch": 97.1,
+      "learning_rate": 0.00012449122807017541,
+      "loss": 0.0243,
+      "step": 6506
+    },
+    {
+      "epoch": 97.12,
+      "learning_rate": 0.0001244561403508772,
+      "loss": 0.0334,
+      "step": 6507
+    },
+    {
+      "epoch": 97.13,
+      "learning_rate": 0.00012442105263157894,
+      "loss": 0.0012,
+      "step": 6508
+    },
+    {
+      "epoch": 97.15,
+      "learning_rate": 0.0001243859649122807,
+      "loss": 0.0018,
+      "step": 6509
+    },
+    {
+      "epoch": 97.16,
+      "learning_rate": 0.00012435087719298246,
+      "loss": 0.001,
+      "step": 6510
+    },
+    {
+      "epoch": 97.18,
+      "learning_rate": 0.00012431578947368419,
+      "loss": 0.0021,
+      "step": 6511
+    },
+    {
+      "epoch": 97.19,
+      "learning_rate": 0.00012428070175438596,
+      "loss": 0.0018,
+      "step": 6512
+    },
+    {
+      "epoch": 97.21,
+      "learning_rate": 0.0001242456140350877,
+      "loss": 0.0435,
+      "step": 6513
+    },
+    {
+      "epoch": 97.22,
+      "learning_rate": 0.00012421052631578946,
+      "loss": 0.1776,
+      "step": 6514
+    },
+    {
+      "epoch": 97.24,
+      "learning_rate": 0.0001241754385964912,
+      "loss": 0.0017,
+      "step": 6515
+    },
+    {
+      "epoch": 97.25,
+      "learning_rate": 0.00012414035087719298,
+      "loss": 0.0035,
+      "step": 6516
+    },
+    {
+      "epoch": 97.27,
+      "learning_rate": 0.00012410526315789473,
+      "loss": 0.1954,
+      "step": 6517
+    },
+    {
+      "epoch": 97.28,
+      "learning_rate": 0.00012407017543859648,
+      "loss": 0.1197,
+      "step": 6518
+    },
+    {
+      "epoch": 97.3,
+      "learning_rate": 0.00012403508771929823,
+      "loss": 0.0009,
+      "step": 6519
+    },
+    {
+      "epoch": 97.31,
+      "learning_rate": 0.00012399999999999998,
+      "loss": 0.0014,
+      "step": 6520
+    },
+    {
+      "epoch": 97.33,
+      "learning_rate": 0.00012396491228070175,
+      "loss": 0.0012,
+      "step": 6521
+    },
+    {
+      "epoch": 97.34,
+      "learning_rate": 0.0001239298245614035,
+      "loss": 0.1418,
+      "step": 6522
+    },
+    {
+      "epoch": 97.36,
+      "learning_rate": 0.00012389473684210525,
+      "loss": 0.0008,
+      "step": 6523
+    },
+    {
+      "epoch": 97.37,
+      "learning_rate": 0.000123859649122807,
+      "loss": 0.0011,
+      "step": 6524
+    },
+    {
+      "epoch": 97.39,
+      "learning_rate": 0.00012382456140350878,
+      "loss": 0.1914,
+      "step": 6525
+    },
+    {
+      "epoch": 97.4,
+      "learning_rate": 0.0001237894736842105,
+      "loss": 0.0006,
+      "step": 6526
+    },
+    {
+      "epoch": 97.42,
+      "learning_rate": 0.00012375438596491227,
+      "loss": 0.0008,
+      "step": 6527
+    },
+    {
+      "epoch": 97.43,
+      "learning_rate": 0.00012371929824561402,
+      "loss": 0.0015,
+      "step": 6528
+    },
+    {
+      "epoch": 97.45,
+      "learning_rate": 0.00012368421052631577,
+      "loss": 0.0129,
+      "step": 6529
+    },
+    {
+      "epoch": 97.46,
+      "learning_rate": 0.00012364912280701752,
+      "loss": 0.0008,
+      "step": 6530
+    },
+    {
+      "epoch": 97.48,
+      "learning_rate": 0.0001236140350877193,
+      "loss": 0.1816,
+      "step": 6531
+    },
+    {
+      "epoch": 97.49,
+      "learning_rate": 0.00012357894736842104,
+      "loss": 0.2985,
+      "step": 6532
+    },
+    {
+      "epoch": 97.51,
+      "learning_rate": 0.0001235438596491228,
+      "loss": 0.1943,
+      "step": 6533
+    },
+    {
+      "epoch": 97.52,
+      "learning_rate": 0.00012350877192982457,
+      "loss": 0.0019,
+      "step": 6534
+    },
+    {
+      "epoch": 97.54,
+      "learning_rate": 0.0001234736842105263,
+      "loss": 0.0129,
+      "step": 6535
+    },
+    {
+      "epoch": 97.55,
+      "learning_rate": 0.00012343859649122807,
+      "loss": 0.3213,
+      "step": 6536
+    },
+    {
+      "epoch": 97.57,
+      "learning_rate": 0.00012340350877192982,
+      "loss": 0.0094,
+      "step": 6537
+    },
+    {
+      "epoch": 97.58,
+      "learning_rate": 0.00012336842105263156,
+      "loss": 0.0682,
+      "step": 6538
+    },
+    {
+      "epoch": 97.59,
+      "learning_rate": 0.0001233333333333333,
+      "loss": 0.0008,
+      "step": 6539
+    },
+    {
+      "epoch": 97.61,
+      "learning_rate": 0.0001232982456140351,
+      "loss": 0.0007,
+      "step": 6540
+    },
+    {
+      "epoch": 97.62,
+      "learning_rate": 0.00012326315789473684,
+      "loss": 0.0007,
+      "step": 6541
+    },
+    {
+      "epoch": 97.64,
+      "learning_rate": 0.0001232280701754386,
+      "loss": 0.0043,
+      "step": 6542
+    },
+    {
+      "epoch": 97.65,
+      "learning_rate": 0.00012319298245614034,
+      "loss": 0.0012,
+      "step": 6543
+    },
+    {
+      "epoch": 97.67,
+      "learning_rate": 0.00012315789473684208,
+      "loss": 0.0501,
+      "step": 6544
+    },
+    {
+      "epoch": 97.68,
+      "learning_rate": 0.00012312280701754386,
+      "loss": 0.0713,
+      "step": 6545
+    },
+    {
+      "epoch": 97.7,
+      "learning_rate": 0.0001230877192982456,
+      "loss": 0.0011,
+      "step": 6546
+    },
+    {
+      "epoch": 97.71,
+      "learning_rate": 0.00012305263157894736,
+      "loss": 0.0039,
+      "step": 6547
+    },
+    {
+      "epoch": 97.73,
+      "learning_rate": 0.0001230175438596491,
+      "loss": 0.0008,
+      "step": 6548
+    },
+    {
+      "epoch": 97.74,
+      "learning_rate": 0.00012298245614035088,
+      "loss": 0.0011,
+      "step": 6549
+    },
+    {
+      "epoch": 97.76,
+      "learning_rate": 0.0001229473684210526,
+      "loss": 0.0009,
+      "step": 6550
+    },
+    {
+      "epoch": 97.77,
+      "learning_rate": 0.00012291228070175438,
+      "loss": 0.001,
+      "step": 6551
+    },
+    {
+      "epoch": 97.79,
+      "learning_rate": 0.00012287719298245613,
+      "loss": 0.0026,
+      "step": 6552
+    },
+    {
+      "epoch": 97.8,
+      "learning_rate": 0.00012284210526315788,
+      "loss": 0.2741,
+      "step": 6553
+    },
+    {
+      "epoch": 97.82,
+      "learning_rate": 0.00012280701754385963,
+      "loss": 0.0012,
+      "step": 6554
+    },
+    {
+      "epoch": 97.83,
+      "learning_rate": 0.0001227719298245614,
+      "loss": 0.0008,
+      "step": 6555
+    },
+    {
+      "epoch": 97.85,
+      "learning_rate": 0.00012273684210526315,
+      "loss": 0.0007,
+      "step": 6556
+    },
+    {
+      "epoch": 97.86,
+      "learning_rate": 0.0001227017543859649,
+      "loss": 0.0024,
+      "step": 6557
+    },
+    {
+      "epoch": 97.88,
+      "learning_rate": 0.00012266666666666668,
+      "loss": 0.0008,
+      "step": 6558
+    },
+    {
+      "epoch": 97.89,
+      "learning_rate": 0.0001226315789473684,
+      "loss": 0.0038,
+      "step": 6559
+    },
+    {
+      "epoch": 97.91,
+      "learning_rate": 0.00012259649122807017,
+      "loss": 0.0148,
+      "step": 6560
+    },
+    {
+      "epoch": 97.92,
+      "learning_rate": 0.00012256140350877192,
+      "loss": 0.0027,
+      "step": 6561
+    },
+    {
+      "epoch": 97.94,
+      "learning_rate": 0.00012252631578947367,
+      "loss": 0.2488,
+      "step": 6562
+    },
+    {
+      "epoch": 97.95,
+      "learning_rate": 0.00012249122807017542,
+      "loss": 0.002,
+      "step": 6563
+    },
+    {
+      "epoch": 97.97,
+      "learning_rate": 0.0001224561403508772,
+      "loss": 0.001,
+      "step": 6564
+    },
+    {
+      "epoch": 97.98,
+      "learning_rate": 0.00012242105263157894,
+      "loss": 0.0031,
+      "step": 6565
+    },
+    {
+      "epoch": 98.0,
+      "learning_rate": 0.0001223859649122807,
+      "loss": 0.0342,
+      "step": 6566
+    },
+    {
+      "epoch": 98.01,
+      "learning_rate": 0.00012235087719298244,
+      "loss": 0.0302,
+      "step": 6567
+    },
+    {
+      "epoch": 98.03,
+      "learning_rate": 0.0001223157894736842,
+      "loss": 0.0006,
+      "step": 6568
+    },
+    {
+      "epoch": 98.04,
+      "learning_rate": 0.00012228070175438597,
+      "loss": 0.0008,
+      "step": 6569
+    },
+    {
+      "epoch": 98.06,
+      "learning_rate": 0.00012224561403508772,
+      "loss": 0.0014,
+      "step": 6570
+    },
+    {
+      "epoch": 98.07,
+      "learning_rate": 0.00012221052631578946,
+      "loss": 0.0008,
+      "step": 6571
+    },
+    {
+      "epoch": 98.09,
+      "learning_rate": 0.0001221754385964912,
+      "loss": 0.0008,
+      "step": 6572
+    },
+    {
+      "epoch": 98.1,
+      "learning_rate": 0.000122140350877193,
+      "loss": 0.0027,
+      "step": 6573
+    },
+    {
+      "epoch": 98.12,
+      "learning_rate": 0.0001221052631578947,
+      "loss": 0.0007,
+      "step": 6574
+    },
+    {
+      "epoch": 98.13,
+      "learning_rate": 0.00012207017543859649,
+      "loss": 0.001,
+      "step": 6575
+    },
+    {
+      "epoch": 98.15,
+      "learning_rate": 0.00012203508771929824,
+      "loss": 0.001,
+      "step": 6576
+    },
+    {
+      "epoch": 98.16,
+      "learning_rate": 0.000122,
+      "loss": 0.0014,
+      "step": 6577
+    },
+    {
+      "epoch": 98.18,
+      "learning_rate": 0.00012196491228070173,
+      "loss": 0.1129,
+      "step": 6578
+    },
+    {
+      "epoch": 98.19,
+      "learning_rate": 0.0001219298245614035,
+      "loss": 0.0015,
+      "step": 6579
+    },
+    {
+      "epoch": 98.21,
+      "learning_rate": 0.00012189473684210524,
+      "loss": 0.0031,
+      "step": 6580
+    },
+    {
+      "epoch": 98.22,
+      "learning_rate": 0.000121859649122807,
+      "loss": 0.0012,
+      "step": 6581
+    },
+    {
+      "epoch": 98.24,
+      "learning_rate": 0.00012182456140350877,
+      "loss": 0.0009,
+      "step": 6582
+    },
+    {
+      "epoch": 98.25,
+      "learning_rate": 0.00012178947368421052,
+      "loss": 0.0543,
+      "step": 6583
+    },
+    {
+      "epoch": 98.27,
+      "learning_rate": 0.00012175438596491228,
+      "loss": 0.0008,
+      "step": 6584
+    },
+    {
+      "epoch": 98.28,
+      "learning_rate": 0.00012171929824561403,
+      "loss": 0.0016,
+      "step": 6585
+    },
+    {
+      "epoch": 98.3,
+      "learning_rate": 0.00012168421052631579,
+      "loss": 0.0021,
+      "step": 6586
+    },
+    {
+      "epoch": 98.31,
+      "learning_rate": 0.00012164912280701753,
+      "loss": 0.2339,
+      "step": 6587
+    },
+    {
+      "epoch": 98.33,
+      "learning_rate": 0.00012161403508771929,
+      "loss": 0.0009,
+      "step": 6588
+    },
+    {
+      "epoch": 98.34,
+      "learning_rate": 0.00012157894736842104,
+      "loss": 0.0053,
+      "step": 6589
+    },
+    {
+      "epoch": 98.36,
+      "learning_rate": 0.0001215438596491228,
+      "loss": 0.001,
+      "step": 6590
+    },
+    {
+      "epoch": 98.37,
+      "learning_rate": 0.00012150877192982455,
+      "loss": 0.1344,
+      "step": 6591
+    },
+    {
+      "epoch": 98.39,
+      "learning_rate": 0.00012147368421052631,
+      "loss": 0.2065,
+      "step": 6592
+    },
+    {
+      "epoch": 98.4,
+      "learning_rate": 0.00012143859649122805,
+      "loss": 0.0011,
+      "step": 6593
+    },
+    {
+      "epoch": 98.42,
+      "learning_rate": 0.00012140350877192981,
+      "loss": 0.0007,
+      "step": 6594
+    },
+    {
+      "epoch": 98.43,
+      "learning_rate": 0.00012136842105263157,
+      "loss": 0.1541,
+      "step": 6595
+    },
+    {
+      "epoch": 98.45,
+      "learning_rate": 0.00012133333333333332,
+      "loss": 0.0015,
+      "step": 6596
+    },
+    {
+      "epoch": 98.46,
+      "learning_rate": 0.00012129824561403508,
+      "loss": 0.0012,
+      "step": 6597
+    },
+    {
+      "epoch": 98.48,
+      "learning_rate": 0.00012126315789473683,
+      "loss": 0.0543,
+      "step": 6598
+    },
+    {
+      "epoch": 98.49,
+      "learning_rate": 0.00012122807017543859,
+      "loss": 0.001,
+      "step": 6599
+    },
+    {
+      "epoch": 98.51,
+      "learning_rate": 0.00012119298245614034,
+      "loss": 0.0008,
+      "step": 6600
+    },
+    {
+      "epoch": 98.51,
+      "eval_accuracy": 0.8541360744003916,
+      "eval_f1": 0.8572461579693889,
+      "eval_loss": 0.7322733402252197,
+      "eval_runtime": 344.5259,
+      "eval_samples_per_second": 11.86,
+      "eval_steps_per_second": 0.743,
+      "step": 6600
+    },
+    {
+      "epoch": 98.52,
+      "learning_rate": 0.0001211578947368421,
+      "loss": 0.0381,
+      "step": 6601
+    },
+    {
+      "epoch": 98.54,
+      "learning_rate": 0.00012112280701754384,
+      "loss": 0.0028,
+      "step": 6602
+    },
+    {
+      "epoch": 98.55,
+      "learning_rate": 0.0001210877192982456,
+      "loss": 0.2096,
+      "step": 6603
+    },
+    {
+      "epoch": 98.57,
+      "learning_rate": 0.00012105263157894735,
+      "loss": 0.1052,
+      "step": 6604
+    },
+    {
+      "epoch": 98.58,
+      "learning_rate": 0.00012101754385964911,
+      "loss": 0.277,
+      "step": 6605
+    },
+    {
+      "epoch": 98.59,
+      "learning_rate": 0.00012098245614035086,
+      "loss": 0.0104,
+      "step": 6606
+    },
+    {
+      "epoch": 98.61,
+      "learning_rate": 0.00012094736842105262,
+      "loss": 0.0007,
+      "step": 6607
+    },
+    {
+      "epoch": 98.62,
+      "learning_rate": 0.00012091228070175439,
+      "loss": 0.004,
+      "step": 6608
+    },
+    {
+      "epoch": 98.64,
+      "learning_rate": 0.00012087719298245613,
+      "loss": 0.2461,
+      "step": 6609
+    },
+    {
+      "epoch": 98.65,
+      "learning_rate": 0.0001208421052631579,
+      "loss": 0.0009,
+      "step": 6610
+    },
+    {
+      "epoch": 98.67,
+      "learning_rate": 0.00012080701754385963,
+      "loss": 0.0634,
+      "step": 6611
+    },
+    {
+      "epoch": 98.68,
+      "learning_rate": 0.0001207719298245614,
+      "loss": 0.0097,
+      "step": 6612
+    },
+    {
+      "epoch": 98.7,
+      "learning_rate": 0.00012073684210526314,
+      "loss": 0.0934,
+      "step": 6613
+    },
+    {
+      "epoch": 98.71,
+      "learning_rate": 0.0001207017543859649,
+      "loss": 0.001,
+      "step": 6614
+    },
+    {
+      "epoch": 98.73,
+      "learning_rate": 0.00012066666666666665,
+      "loss": 0.0013,
+      "step": 6615
+    },
+    {
+      "epoch": 98.74,
+      "learning_rate": 0.00012063157894736842,
+      "loss": 0.0033,
+      "step": 6616
+    },
+    {
+      "epoch": 98.76,
+      "learning_rate": 0.00012059649122807015,
+      "loss": 0.0016,
+      "step": 6617
+    },
+    {
+      "epoch": 98.77,
+      "learning_rate": 0.00012056140350877191,
+      "loss": 0.023,
+      "step": 6618
+    },
+    {
+      "epoch": 98.79,
+      "learning_rate": 0.00012052631578947368,
+      "loss": 0.0018,
+      "step": 6619
+    },
+    {
+      "epoch": 98.8,
+      "learning_rate": 0.00012049122807017543,
+      "loss": 0.1118,
+      "step": 6620
+    },
+    {
+      "epoch": 98.82,
+      "learning_rate": 0.00012045614035087719,
+      "loss": 0.0055,
+      "step": 6621
+    },
+    {
+      "epoch": 98.83,
+      "learning_rate": 0.00012042105263157894,
+      "loss": 0.011,
+      "step": 6622
+    },
+    {
+      "epoch": 98.85,
+      "learning_rate": 0.0001203859649122807,
+      "loss": 0.0015,
+      "step": 6623
+    },
+    {
+      "epoch": 98.86,
+      "learning_rate": 0.00012035087719298245,
+      "loss": 0.2383,
+      "step": 6624
+    },
+    {
+      "epoch": 98.88,
+      "learning_rate": 0.00012031578947368421,
+      "loss": 0.0013,
+      "step": 6625
+    },
+    {
+      "epoch": 98.89,
+      "learning_rate": 0.00012028070175438595,
+      "loss": 0.0012,
+      "step": 6626
+    },
+    {
+      "epoch": 98.91,
+      "learning_rate": 0.00012024561403508771,
+      "loss": 0.0179,
+      "step": 6627
+    },
+    {
+      "epoch": 98.92,
+      "learning_rate": 0.00012021052631578946,
+      "loss": 0.0349,
+      "step": 6628
+    },
+    {
+      "epoch": 98.94,
+      "learning_rate": 0.00012017543859649122,
+      "loss": 0.0045,
+      "step": 6629
+    },
+    {
+      "epoch": 98.95,
+      "learning_rate": 0.00012014035087719297,
+      "loss": 0.0787,
+      "step": 6630
+    },
+    {
+      "epoch": 98.97,
+      "learning_rate": 0.00012010526315789473,
+      "loss": 0.0008,
+      "step": 6631
+    },
+    {
+      "epoch": 98.98,
+      "learning_rate": 0.00012007017543859649,
+      "loss": 0.1505,
+      "step": 6632
+    },
+    {
+      "epoch": 99.0,
+      "learning_rate": 0.00012003508771929824,
+      "loss": 0.0029,
+      "step": 6633
+    },
+    {
+      "epoch": 99.01,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 0.0435,
+      "step": 6634
+    },
+    {
+      "epoch": 99.03,
+      "learning_rate": 0.00011996491228070174,
+      "loss": 0.0363,
+      "step": 6635
+    },
+    {
+      "epoch": 99.04,
+      "learning_rate": 0.0001199298245614035,
+      "loss": 0.0013,
+      "step": 6636
+    },
+    {
+      "epoch": 99.06,
+      "learning_rate": 0.00011989473684210525,
+      "loss": 0.177,
+      "step": 6637
+    },
+    {
+      "epoch": 99.07,
+      "learning_rate": 0.00011985964912280701,
+      "loss": 0.001,
+      "step": 6638
+    },
+    {
+      "epoch": 99.09,
+      "learning_rate": 0.00011982456140350876,
+      "loss": 0.0011,
+      "step": 6639
+    },
+    {
+      "epoch": 99.1,
+      "learning_rate": 0.00011978947368421052,
+      "loss": 0.0017,
+      "step": 6640
+    },
+    {
+      "epoch": 99.12,
+      "learning_rate": 0.00011975438596491226,
+      "loss": 0.0028,
+      "step": 6641
+    },
+    {
+      "epoch": 99.13,
+      "learning_rate": 0.00011971929824561402,
+      "loss": 0.0211,
+      "step": 6642
+    },
+    {
+      "epoch": 99.15,
+      "learning_rate": 0.00011968421052631577,
+      "loss": 0.0025,
+      "step": 6643
+    },
+    {
+      "epoch": 99.16,
+      "learning_rate": 0.00011964912280701753,
+      "loss": 0.001,
+      "step": 6644
+    },
+    {
+      "epoch": 99.18,
+      "learning_rate": 0.0001196140350877193,
+      "loss": 0.0008,
+      "step": 6645
+    },
+    {
+      "epoch": 99.19,
+      "learning_rate": 0.00011957894736842104,
+      "loss": 0.0013,
+      "step": 6646
+    },
+    {
+      "epoch": 99.21,
+      "learning_rate": 0.0001195438596491228,
+      "loss": 0.0011,
+      "step": 6647
+    },
+    {
+      "epoch": 99.22,
+      "learning_rate": 0.00011950877192982455,
+      "loss": 0.3039,
+      "step": 6648
+    },
+    {
+      "epoch": 99.24,
+      "learning_rate": 0.00011947368421052632,
+      "loss": 0.0019,
+      "step": 6649
+    },
+    {
+      "epoch": 99.25,
+      "learning_rate": 0.00011943859649122805,
+      "loss": 0.1022,
+      "step": 6650
+    },
+    {
+      "epoch": 99.27,
+      "learning_rate": 0.00011940350877192981,
+      "loss": 0.0957,
+      "step": 6651
+    },
+    {
+      "epoch": 99.28,
+      "learning_rate": 0.00011936842105263156,
+      "loss": 0.1441,
+      "step": 6652
+    },
+    {
+      "epoch": 99.3,
+      "learning_rate": 0.00011933333333333332,
+      "loss": 0.2018,
+      "step": 6653
+    },
+    {
+      "epoch": 99.31,
+      "learning_rate": 0.00011929824561403507,
+      "loss": 0.0094,
+      "step": 6654
+    },
+    {
+      "epoch": 99.33,
+      "learning_rate": 0.00011926315789473684,
+      "loss": 0.0008,
+      "step": 6655
+    },
+    {
+      "epoch": 99.34,
+      "learning_rate": 0.00011922807017543858,
+      "loss": 0.0024,
+      "step": 6656
+    },
+    {
+      "epoch": 99.36,
+      "learning_rate": 0.00011919298245614035,
+      "loss": 0.0005,
+      "step": 6657
+    },
+    {
+      "epoch": 99.37,
+      "learning_rate": 0.0001191578947368421,
+      "loss": 0.0171,
+      "step": 6658
+    },
+    {
+      "epoch": 99.39,
+      "learning_rate": 0.00011912280701754384,
+      "loss": 0.0042,
+      "step": 6659
+    },
+    {
+      "epoch": 99.4,
+      "learning_rate": 0.00011908771929824561,
+      "loss": 0.1014,
+      "step": 6660
+    },
+    {
+      "epoch": 99.42,
+      "learning_rate": 0.00011905263157894736,
+      "loss": 0.0596,
+      "step": 6661
+    },
+    {
+      "epoch": 99.43,
+      "learning_rate": 0.00011901754385964912,
+      "loss": 0.0012,
+      "step": 6662
+    },
+    {
+      "epoch": 99.45,
+      "learning_rate": 0.00011898245614035087,
+      "loss": 0.001,
+      "step": 6663
+    },
+    {
+      "epoch": 99.46,
+      "learning_rate": 0.00011894736842105263,
+      "loss": 0.0005,
+      "step": 6664
+    },
+    {
+      "epoch": 99.48,
+      "learning_rate": 0.00011891228070175436,
+      "loss": 0.0008,
+      "step": 6665
+    },
+    {
+      "epoch": 99.49,
+      "learning_rate": 0.00011887719298245613,
+      "loss": 0.0008,
+      "step": 6666
+    },
+    {
+      "epoch": 99.51,
+      "learning_rate": 0.00011884210526315788,
+      "loss": 0.0011,
+      "step": 6667
+    },
+    {
+      "epoch": 99.52,
+      "learning_rate": 0.00011880701754385964,
+      "loss": 0.0267,
+      "step": 6668
+    },
+    {
+      "epoch": 99.54,
+      "learning_rate": 0.00011877192982456139,
+      "loss": 0.2269,
+      "step": 6669
+    },
+    {
+      "epoch": 99.55,
+      "learning_rate": 0.00011873684210526315,
+      "loss": 0.0203,
+      "step": 6670
+    },
+    {
+      "epoch": 99.57,
+      "learning_rate": 0.00011870175438596491,
+      "loss": 0.0008,
+      "step": 6671
+    },
+    {
+      "epoch": 99.58,
+      "learning_rate": 0.00011866666666666666,
+      "loss": 0.0025,
+      "step": 6672
+    },
+    {
+      "epoch": 99.59,
+      "learning_rate": 0.00011863157894736842,
+      "loss": 0.001,
+      "step": 6673
+    },
+    {
+      "epoch": 99.61,
+      "learning_rate": 0.00011859649122807016,
+      "loss": 0.0006,
+      "step": 6674
+    },
+    {
+      "epoch": 99.62,
+      "learning_rate": 0.00011856140350877192,
+      "loss": 0.0052,
+      "step": 6675
+    },
+    {
+      "epoch": 99.64,
+      "learning_rate": 0.00011852631578947367,
+      "loss": 0.0008,
+      "step": 6676
+    },
+    {
+      "epoch": 99.65,
+      "learning_rate": 0.00011849122807017543,
+      "loss": 0.0006,
+      "step": 6677
+    },
+    {
+      "epoch": 99.67,
+      "learning_rate": 0.00011845614035087718,
+      "loss": 0.1664,
+      "step": 6678
+    },
+    {
+      "epoch": 99.68,
+      "learning_rate": 0.00011842105263157894,
+      "loss": 0.0015,
+      "step": 6679
+    },
+    {
+      "epoch": 99.7,
+      "learning_rate": 0.00011838596491228069,
+      "loss": 0.0011,
+      "step": 6680
+    },
+    {
+      "epoch": 99.71,
+      "learning_rate": 0.00011835087719298244,
+      "loss": 0.0046,
+      "step": 6681
+    },
+    {
+      "epoch": 99.73,
+      "learning_rate": 0.0001183157894736842,
+      "loss": 0.0009,
+      "step": 6682
+    },
+    {
+      "epoch": 99.74,
+      "learning_rate": 0.00011828070175438595,
+      "loss": 0.0322,
+      "step": 6683
+    },
+    {
+      "epoch": 99.76,
+      "learning_rate": 0.00011824561403508771,
+      "loss": 0.0009,
+      "step": 6684
+    },
+    {
+      "epoch": 99.77,
+      "learning_rate": 0.00011821052631578946,
+      "loss": 0.0012,
+      "step": 6685
+    },
+    {
+      "epoch": 99.79,
+      "learning_rate": 0.00011817543859649122,
+      "loss": 0.0545,
+      "step": 6686
+    },
+    {
+      "epoch": 99.8,
+      "learning_rate": 0.00011814035087719297,
+      "loss": 0.0012,
+      "step": 6687
+    },
+    {
+      "epoch": 99.82,
+      "learning_rate": 0.00011810526315789474,
+      "loss": 0.0016,
+      "step": 6688
+    },
+    {
+      "epoch": 99.83,
+      "learning_rate": 0.00011807017543859647,
+      "loss": 0.007,
+      "step": 6689
+    },
+    {
+      "epoch": 99.85,
+      "learning_rate": 0.00011803508771929823,
+      "loss": 0.0365,
+      "step": 6690
+    },
+    {
+      "epoch": 99.86,
+      "learning_rate": 0.00011799999999999998,
+      "loss": 0.1005,
+      "step": 6691
+    },
+    {
+      "epoch": 99.88,
+      "learning_rate": 0.00011796491228070174,
+      "loss": 0.001,
+      "step": 6692
+    },
+    {
+      "epoch": 99.89,
+      "learning_rate": 0.00011792982456140349,
+      "loss": 0.0126,
+      "step": 6693
+    },
+    {
+      "epoch": 99.91,
+      "learning_rate": 0.00011789473684210525,
+      "loss": 0.0256,
+      "step": 6694
+    },
+    {
+      "epoch": 99.92,
+      "learning_rate": 0.00011785964912280702,
+      "loss": 0.1166,
+      "step": 6695
+    },
+    {
+      "epoch": 99.94,
+      "learning_rate": 0.00011782456140350877,
+      "loss": 0.0006,
+      "step": 6696
+    },
+    {
+      "epoch": 99.95,
+      "learning_rate": 0.00011778947368421053,
+      "loss": 0.0009,
+      "step": 6697
+    },
+    {
+      "epoch": 99.97,
+      "learning_rate": 0.00011775438596491226,
+      "loss": 0.0018,
+      "step": 6698
+    },
+    {
+      "epoch": 99.98,
+      "learning_rate": 0.00011771929824561403,
+      "loss": 0.001,
+      "step": 6699
+    },
+    {
+      "epoch": 100.0,
+      "learning_rate": 0.00011768421052631577,
+      "loss": 0.0008,
+      "step": 6700
+    },
+    {
+      "epoch": 100.01,
+      "learning_rate": 0.00011764912280701754,
+      "loss": 0.0034,
+      "step": 6701
+    },
+    {
+      "epoch": 100.03,
+      "learning_rate": 0.00011761403508771929,
+      "loss": 0.0061,
+      "step": 6702
+    },
+    {
+      "epoch": 100.04,
+      "learning_rate": 0.00011757894736842105,
+      "loss": 0.0006,
+      "step": 6703
+    },
+    {
+      "epoch": 100.06,
+      "learning_rate": 0.0001175438596491228,
+      "loss": 0.0007,
+      "step": 6704
+    },
+    {
+      "epoch": 100.07,
+      "learning_rate": 0.00011750877192982455,
+      "loss": 0.0068,
+      "step": 6705
+    },
+    {
+      "epoch": 100.09,
+      "learning_rate": 0.0001174736842105263,
+      "loss": 0.1123,
+      "step": 6706
+    },
+    {
+      "epoch": 100.1,
+      "learning_rate": 0.00011743859649122806,
+      "loss": 0.0011,
+      "step": 6707
+    },
+    {
+      "epoch": 100.12,
+      "learning_rate": 0.00011740350877192982,
+      "loss": 0.0603,
+      "step": 6708
+    },
+    {
+      "epoch": 100.13,
+      "learning_rate": 0.00011736842105263157,
+      "loss": 0.0008,
+      "step": 6709
+    },
+    {
+      "epoch": 100.15,
+      "learning_rate": 0.00011733333333333333,
+      "loss": 0.0006,
+      "step": 6710
+    },
+    {
+      "epoch": 100.16,
+      "learning_rate": 0.00011729824561403508,
+      "loss": 0.0008,
+      "step": 6711
+    },
+    {
+      "epoch": 100.18,
+      "learning_rate": 0.00011726315789473684,
+      "loss": 0.2151,
+      "step": 6712
+    },
+    {
+      "epoch": 100.19,
+      "learning_rate": 0.00011722807017543858,
+      "loss": 0.0026,
+      "step": 6713
+    },
+    {
+      "epoch": 100.21,
+      "learning_rate": 0.00011719298245614034,
+      "loss": 0.0011,
+      "step": 6714
+    },
+    {
+      "epoch": 100.22,
+      "learning_rate": 0.00011715789473684209,
+      "loss": 0.0007,
+      "step": 6715
+    },
+    {
+      "epoch": 100.24,
+      "learning_rate": 0.00011712280701754385,
+      "loss": 0.0009,
+      "step": 6716
+    },
+    {
+      "epoch": 100.25,
+      "learning_rate": 0.0001170877192982456,
+      "loss": 0.0048,
+      "step": 6717
+    },
+    {
+      "epoch": 100.27,
+      "learning_rate": 0.00011705263157894736,
+      "loss": 0.1761,
+      "step": 6718
+    },
+    {
+      "epoch": 100.28,
+      "learning_rate": 0.00011701754385964911,
+      "loss": 0.0009,
+      "step": 6719
+    },
+    {
+      "epoch": 100.3,
+      "learning_rate": 0.00011698245614035087,
+      "loss": 0.0514,
+      "step": 6720
+    },
+    {
+      "epoch": 100.31,
+      "learning_rate": 0.00011694736842105263,
+      "loss": 0.063,
+      "step": 6721
+    },
+    {
+      "epoch": 100.33,
+      "learning_rate": 0.00011691228070175437,
+      "loss": 0.0164,
+      "step": 6722
+    },
+    {
+      "epoch": 100.34,
+      "learning_rate": 0.00011687719298245613,
+      "loss": 0.0022,
+      "step": 6723
+    },
+    {
+      "epoch": 100.36,
+      "learning_rate": 0.00011684210526315788,
+      "loss": 0.0007,
+      "step": 6724
+    },
+    {
+      "epoch": 100.37,
+      "learning_rate": 0.00011680701754385964,
+      "loss": 0.0019,
+      "step": 6725
+    },
+    {
+      "epoch": 100.39,
+      "learning_rate": 0.00011677192982456139,
+      "loss": 0.0429,
+      "step": 6726
+    },
+    {
+      "epoch": 100.4,
+      "learning_rate": 0.00011673684210526315,
+      "loss": 0.0109,
+      "step": 6727
+    },
+    {
+      "epoch": 100.42,
+      "learning_rate": 0.00011670175438596489,
+      "loss": 0.0123,
+      "step": 6728
+    },
+    {
+      "epoch": 100.43,
+      "learning_rate": 0.00011666666666666665,
+      "loss": 0.0008,
+      "step": 6729
+    },
+    {
+      "epoch": 100.45,
+      "learning_rate": 0.0001166315789473684,
+      "loss": 0.0429,
+      "step": 6730
+    },
+    {
+      "epoch": 100.46,
+      "learning_rate": 0.00011659649122807016,
+      "loss": 0.3201,
+      "step": 6731
+    },
+    {
+      "epoch": 100.48,
+      "learning_rate": 0.00011656140350877193,
+      "loss": 0.0026,
+      "step": 6732
+    },
+    {
+      "epoch": 100.49,
+      "learning_rate": 0.00011652631578947367,
+      "loss": 0.002,
+      "step": 6733
+    },
+    {
+      "epoch": 100.51,
+      "learning_rate": 0.00011649122807017544,
+      "loss": 0.017,
+      "step": 6734
+    },
+    {
+      "epoch": 100.52,
+      "learning_rate": 0.00011645614035087719,
+      "loss": 0.0023,
+      "step": 6735
+    },
+    {
+      "epoch": 100.54,
+      "learning_rate": 0.00011642105263157895,
+      "loss": 0.0273,
+      "step": 6736
+    },
+    {
+      "epoch": 100.55,
+      "learning_rate": 0.00011638596491228068,
+      "loss": 0.0007,
+      "step": 6737
+    },
+    {
+      "epoch": 100.57,
+      "learning_rate": 0.00011635087719298245,
+      "loss": 0.0015,
+      "step": 6738
+    },
+    {
+      "epoch": 100.58,
+      "learning_rate": 0.0001163157894736842,
+      "loss": 0.0036,
+      "step": 6739
+    },
+    {
+      "epoch": 100.59,
+      "learning_rate": 0.00011628070175438596,
+      "loss": 0.0009,
+      "step": 6740
+    },
+    {
+      "epoch": 100.61,
+      "learning_rate": 0.0001162456140350877,
+      "loss": 0.0008,
+      "step": 6741
+    },
+    {
+      "epoch": 100.62,
+      "learning_rate": 0.00011621052631578947,
+      "loss": 0.0743,
+      "step": 6742
+    },
+    {
+      "epoch": 100.64,
+      "learning_rate": 0.00011617543859649122,
+      "loss": 0.0029,
+      "step": 6743
+    },
+    {
+      "epoch": 100.65,
+      "learning_rate": 0.00011614035087719298,
+      "loss": 0.0018,
+      "step": 6744
+    },
+    {
+      "epoch": 100.67,
+      "learning_rate": 0.00011610526315789474,
+      "loss": 0.0005,
+      "step": 6745
+    },
+    {
+      "epoch": 100.68,
+      "learning_rate": 0.00011607017543859648,
+      "loss": 0.0006,
+      "step": 6746
+    },
+    {
+      "epoch": 100.7,
+      "learning_rate": 0.00011603508771929824,
+      "loss": 0.0709,
+      "step": 6747
+    },
+    {
+      "epoch": 100.71,
+      "learning_rate": 0.00011599999999999999,
+      "loss": 0.0006,
+      "step": 6748
+    },
+    {
+      "epoch": 100.73,
+      "learning_rate": 0.00011596491228070175,
+      "loss": 0.135,
+      "step": 6749
+    },
+    {
+      "epoch": 100.74,
+      "learning_rate": 0.0001159298245614035,
+      "loss": 0.0049,
+      "step": 6750
+    },
+    {
+      "epoch": 100.76,
+      "learning_rate": 0.00011589473684210526,
+      "loss": 0.0007,
+      "step": 6751
+    },
+    {
+      "epoch": 100.77,
+      "learning_rate": 0.000115859649122807,
+      "loss": 0.0038,
+      "step": 6752
+    },
+    {
+      "epoch": 100.79,
+      "learning_rate": 0.00011582456140350876,
+      "loss": 0.0005,
+      "step": 6753
+    },
+    {
+      "epoch": 100.8,
+      "learning_rate": 0.00011578947368421051,
+      "loss": 0.0006,
+      "step": 6754
+    },
+    {
+      "epoch": 100.82,
+      "learning_rate": 0.00011575438596491227,
+      "loss": 0.0157,
+      "step": 6755
+    },
+    {
+      "epoch": 100.83,
+      "learning_rate": 0.00011571929824561402,
+      "loss": 0.0022,
+      "step": 6756
+    },
+    {
+      "epoch": 100.85,
+      "learning_rate": 0.00011568421052631578,
+      "loss": 0.0006,
+      "step": 6757
+    },
+    {
+      "epoch": 100.86,
+      "learning_rate": 0.00011564912280701754,
+      "loss": 0.2283,
+      "step": 6758
+    },
+    {
+      "epoch": 100.88,
+      "learning_rate": 0.00011561403508771929,
+      "loss": 0.0035,
+      "step": 6759
+    },
+    {
+      "epoch": 100.89,
+      "learning_rate": 0.00011557894736842105,
+      "loss": 0.001,
+      "step": 6760
+    },
+    {
+      "epoch": 100.91,
+      "learning_rate": 0.00011554385964912279,
+      "loss": 0.0009,
+      "step": 6761
+    },
+    {
+      "epoch": 100.92,
+      "learning_rate": 0.00011550877192982455,
+      "loss": 0.0014,
+      "step": 6762
+    },
+    {
+      "epoch": 100.94,
+      "learning_rate": 0.0001154736842105263,
+      "loss": 0.0028,
+      "step": 6763
+    },
+    {
+      "epoch": 100.95,
+      "learning_rate": 0.00011543859649122806,
+      "loss": 0.0009,
+      "step": 6764
+    },
+    {
+      "epoch": 100.97,
+      "learning_rate": 0.00011540350877192981,
+      "loss": 0.0024,
+      "step": 6765
+    },
+    {
+      "epoch": 100.98,
+      "learning_rate": 0.00011536842105263157,
+      "loss": 0.0009,
+      "step": 6766
+    },
+    {
+      "epoch": 101.0,
+      "learning_rate": 0.00011533333333333332,
+      "loss": 0.0269,
+      "step": 6767
+    },
+    {
+      "epoch": 101.01,
+      "learning_rate": 0.00011529824561403508,
+      "loss": 0.0018,
+      "step": 6768
+    },
+    {
+      "epoch": 101.03,
+      "learning_rate": 0.00011526315789473682,
+      "loss": 0.002,
+      "step": 6769
+    },
+    {
+      "epoch": 101.04,
+      "learning_rate": 0.00011522807017543858,
+      "loss": 0.1631,
+      "step": 6770
+    },
+    {
+      "epoch": 101.06,
+      "learning_rate": 0.00011519298245614034,
+      "loss": 0.0057,
+      "step": 6771
+    },
+    {
+      "epoch": 101.07,
+      "learning_rate": 0.0001151578947368421,
+      "loss": 0.0009,
+      "step": 6772
+    },
+    {
+      "epoch": 101.09,
+      "learning_rate": 0.00011512280701754386,
+      "loss": 0.1869,
+      "step": 6773
+    },
+    {
+      "epoch": 101.1,
+      "learning_rate": 0.0001150877192982456,
+      "loss": 0.0018,
+      "step": 6774
+    },
+    {
+      "epoch": 101.12,
+      "learning_rate": 0.00011505263157894737,
+      "loss": 0.0007,
+      "step": 6775
+    },
+    {
+      "epoch": 101.13,
+      "learning_rate": 0.0001150175438596491,
+      "loss": 0.0008,
+      "step": 6776
+    },
+    {
+      "epoch": 101.15,
+      "learning_rate": 0.00011498245614035086,
+      "loss": 0.0088,
+      "step": 6777
+    },
+    {
+      "epoch": 101.16,
+      "learning_rate": 0.00011494736842105261,
+      "loss": 0.0016,
+      "step": 6778
+    },
+    {
+      "epoch": 101.18,
+      "learning_rate": 0.00011491228070175438,
+      "loss": 0.0006,
+      "step": 6779
+    },
+    {
+      "epoch": 101.19,
+      "learning_rate": 0.00011487719298245612,
+      "loss": 0.0011,
+      "step": 6780
+    },
+    {
+      "epoch": 101.21,
+      "learning_rate": 0.00011484210526315789,
+      "loss": 0.0407,
+      "step": 6781
+    },
+    {
+      "epoch": 101.22,
+      "learning_rate": 0.00011480701754385965,
+      "loss": 0.0213,
+      "step": 6782
+    },
+    {
+      "epoch": 101.24,
+      "learning_rate": 0.0001147719298245614,
+      "loss": 0.0024,
+      "step": 6783
+    },
+    {
+      "epoch": 101.25,
+      "learning_rate": 0.00011473684210526316,
+      "loss": 0.0057,
+      "step": 6784
+    },
+    {
+      "epoch": 101.27,
+      "learning_rate": 0.0001147017543859649,
+      "loss": 0.0007,
+      "step": 6785
+    },
+    {
+      "epoch": 101.28,
+      "learning_rate": 0.00011466666666666666,
+      "loss": 0.0972,
+      "step": 6786
+    },
+    {
+      "epoch": 101.3,
+      "learning_rate": 0.0001146315789473684,
+      "loss": 0.1075,
+      "step": 6787
+    },
+    {
+      "epoch": 101.31,
+      "learning_rate": 0.00011459649122807017,
+      "loss": 0.0015,
+      "step": 6788
+    },
+    {
+      "epoch": 101.33,
+      "learning_rate": 0.00011456140350877192,
+      "loss": 0.0007,
+      "step": 6789
+    },
+    {
+      "epoch": 101.34,
+      "learning_rate": 0.00011452631578947368,
+      "loss": 0.0006,
+      "step": 6790
+    },
+    {
+      "epoch": 101.36,
+      "learning_rate": 0.00011449122807017543,
+      "loss": 0.0019,
+      "step": 6791
+    },
+    {
+      "epoch": 101.37,
+      "learning_rate": 0.00011445614035087719,
+      "loss": 0.0008,
+      "step": 6792
+    },
+    {
+      "epoch": 101.39,
+      "learning_rate": 0.00011442105263157893,
+      "loss": 0.2029,
+      "step": 6793
+    },
+    {
+      "epoch": 101.4,
+      "learning_rate": 0.00011438596491228069,
+      "loss": 0.0061,
+      "step": 6794
+    },
+    {
+      "epoch": 101.42,
+      "learning_rate": 0.00011435087719298245,
+      "loss": 0.0006,
+      "step": 6795
+    },
+    {
+      "epoch": 101.43,
+      "learning_rate": 0.0001143157894736842,
+      "loss": 0.0015,
+      "step": 6796
+    },
+    {
+      "epoch": 101.45,
+      "learning_rate": 0.00011428070175438596,
+      "loss": 0.0066,
+      "step": 6797
+    },
+    {
+      "epoch": 101.46,
+      "learning_rate": 0.00011424561403508771,
+      "loss": 0.0121,
+      "step": 6798
+    },
+    {
+      "epoch": 101.48,
+      "learning_rate": 0.00011421052631578947,
+      "loss": 0.006,
+      "step": 6799
+    },
+    {
+      "epoch": 101.49,
+      "learning_rate": 0.00011417543859649121,
+      "loss": 0.1655,
+      "step": 6800
+    },
+    {
+      "epoch": 101.49,
+      "eval_accuracy": 0.8504650024473813,
+      "eval_f1": 0.8521261710057196,
+      "eval_loss": 0.6953144073486328,
+      "eval_runtime": 343.8957,
+      "eval_samples_per_second": 11.882,
+      "eval_steps_per_second": 0.744,
+      "step": 6800
+    },
+    {
+      "epoch": 101.51,
+      "learning_rate": 0.00011414035087719297,
+      "loss": 0.004,
+      "step": 6801
+    },
+    {
+      "epoch": 101.52,
+      "learning_rate": 0.00011410526315789472,
+      "loss": 0.1086,
+      "step": 6802
+    },
+    {
+      "epoch": 101.54,
+      "learning_rate": 0.00011407017543859648,
+      "loss": 0.0871,
+      "step": 6803
+    },
+    {
+      "epoch": 101.55,
+      "learning_rate": 0.00011403508771929823,
+      "loss": 0.071,
+      "step": 6804
+    },
+    {
+      "epoch": 101.57,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 0.075,
+      "step": 6805
+    },
+    {
+      "epoch": 101.58,
+      "learning_rate": 0.00011396491228070174,
+      "loss": 0.0009,
+      "step": 6806
+    },
+    {
+      "epoch": 101.59,
+      "learning_rate": 0.0001139298245614035,
+      "loss": 0.0031,
+      "step": 6807
+    },
+    {
+      "epoch": 101.61,
+      "learning_rate": 0.00011389473684210527,
+      "loss": 0.0009,
+      "step": 6808
+    },
+    {
+      "epoch": 101.62,
+      "learning_rate": 0.000113859649122807,
+      "loss": 0.0013,
+      "step": 6809
+    },
+    {
+      "epoch": 101.64,
+      "learning_rate": 0.00011382456140350876,
+      "loss": 0.0874,
+      "step": 6810
+    },
+    {
+      "epoch": 101.65,
+      "learning_rate": 0.00011378947368421051,
+      "loss": 0.0005,
+      "step": 6811
+    },
+    {
+      "epoch": 101.67,
+      "learning_rate": 0.00011375438596491227,
+      "loss": 0.02,
+      "step": 6812
+    },
+    {
+      "epoch": 101.68,
+      "learning_rate": 0.00011371929824561402,
+      "loss": 0.0116,
+      "step": 6813
+    },
+    {
+      "epoch": 101.7,
+      "learning_rate": 0.00011368421052631579,
+      "loss": 0.001,
+      "step": 6814
+    },
+    {
+      "epoch": 101.71,
+      "learning_rate": 0.00011364912280701753,
+      "loss": 0.0017,
+      "step": 6815
+    },
+    {
+      "epoch": 101.73,
+      "learning_rate": 0.0001136140350877193,
+      "loss": 0.0016,
+      "step": 6816
+    },
+    {
+      "epoch": 101.74,
+      "learning_rate": 0.00011357894736842103,
+      "loss": 0.0042,
+      "step": 6817
+    },
+    {
+      "epoch": 101.76,
+      "learning_rate": 0.0001135438596491228,
+      "loss": 0.0054,
+      "step": 6818
+    },
+    {
+      "epoch": 101.77,
+      "learning_rate": 0.00011350877192982454,
+      "loss": 0.0103,
+      "step": 6819
+    },
+    {
+      "epoch": 101.79,
+      "learning_rate": 0.0001134736842105263,
+      "loss": 0.0009,
+      "step": 6820
+    },
+    {
+      "epoch": 101.8,
+      "learning_rate": 0.00011343859649122807,
+      "loss": 0.0088,
+      "step": 6821
+    },
+    {
+      "epoch": 101.82,
+      "learning_rate": 0.00011340350877192982,
+      "loss": 0.0007,
+      "step": 6822
+    },
+    {
+      "epoch": 101.83,
+      "learning_rate": 0.00011336842105263158,
+      "loss": 0.0122,
+      "step": 6823
+    },
+    {
+      "epoch": 101.85,
+      "learning_rate": 0.00011333333333333331,
+      "loss": 0.0046,
+      "step": 6824
+    },
+    {
+      "epoch": 101.86,
+      "learning_rate": 0.00011329824561403508,
+      "loss": 0.0005,
+      "step": 6825
+    },
+    {
+      "epoch": 101.88,
+      "learning_rate": 0.00011326315789473683,
+      "loss": 0.001,
+      "step": 6826
+    },
+    {
+      "epoch": 101.89,
+      "learning_rate": 0.00011322807017543859,
+      "loss": 0.0014,
+      "step": 6827
+    },
+    {
+      "epoch": 101.91,
+      "learning_rate": 0.00011319298245614034,
+      "loss": 0.0048,
+      "step": 6828
+    },
+    {
+      "epoch": 101.92,
+      "learning_rate": 0.0001131578947368421,
+      "loss": 0.0039,
+      "step": 6829
+    },
+    {
+      "epoch": 101.94,
+      "learning_rate": 0.00011312280701754385,
+      "loss": 0.0007,
+      "step": 6830
+    },
+    {
+      "epoch": 101.95,
+      "learning_rate": 0.00011308771929824561,
+      "loss": 0.1849,
+      "step": 6831
+    },
+    {
+      "epoch": 101.97,
+      "learning_rate": 0.00011305263157894735,
+      "loss": 0.0843,
+      "step": 6832
+    },
+    {
+      "epoch": 101.98,
+      "learning_rate": 0.00011301754385964911,
+      "loss": 0.0008,
+      "step": 6833
+    },
+    {
+      "epoch": 102.0,
+      "learning_rate": 0.00011298245614035087,
+      "loss": 0.0574,
+      "step": 6834
+    },
+    {
+      "epoch": 102.01,
+      "learning_rate": 0.00011294736842105262,
+      "loss": 0.0702,
+      "step": 6835
+    },
+    {
+      "epoch": 102.03,
+      "learning_rate": 0.00011291228070175438,
+      "loss": 0.0056,
+      "step": 6836
+    },
+    {
+      "epoch": 102.04,
+      "learning_rate": 0.00011287719298245613,
+      "loss": 0.0005,
+      "step": 6837
+    },
+    {
+      "epoch": 102.06,
+      "learning_rate": 0.00011284210526315789,
+      "loss": 0.0851,
+      "step": 6838
+    },
+    {
+      "epoch": 102.07,
+      "learning_rate": 0.00011280701754385964,
+      "loss": 0.0006,
+      "step": 6839
+    },
+    {
+      "epoch": 102.09,
+      "learning_rate": 0.0001127719298245614,
+      "loss": 0.0006,
+      "step": 6840
+    },
+    {
+      "epoch": 102.1,
+      "learning_rate": 0.00011273684210526314,
+      "loss": 0.0008,
+      "step": 6841
+    },
+    {
+      "epoch": 102.12,
+      "learning_rate": 0.0001127017543859649,
+      "loss": 0.0299,
+      "step": 6842
+    },
+    {
+      "epoch": 102.13,
+      "learning_rate": 0.00011266666666666665,
+      "loss": 0.0005,
+      "step": 6843
+    },
+    {
+      "epoch": 102.15,
+      "learning_rate": 0.00011263157894736841,
+      "loss": 0.0631,
+      "step": 6844
+    },
+    {
+      "epoch": 102.16,
+      "learning_rate": 0.00011259649122807017,
+      "loss": 0.0007,
+      "step": 6845
+    },
+    {
+      "epoch": 102.18,
+      "learning_rate": 0.00011256140350877192,
+      "loss": 0.0005,
+      "step": 6846
+    },
+    {
+      "epoch": 102.19,
+      "learning_rate": 0.00011252631578947369,
+      "loss": 0.0008,
+      "step": 6847
+    },
+    {
+      "epoch": 102.21,
+      "learning_rate": 0.00011249122807017542,
+      "loss": 0.0005,
+      "step": 6848
+    },
+    {
+      "epoch": 102.22,
+      "learning_rate": 0.00011245614035087718,
+      "loss": 0.0006,
+      "step": 6849
+    },
+    {
+      "epoch": 102.24,
+      "learning_rate": 0.00011242105263157893,
+      "loss": 0.0006,
+      "step": 6850
+    },
+    {
+      "epoch": 102.25,
+      "learning_rate": 0.0001123859649122807,
+      "loss": 0.0006,
+      "step": 6851
+    },
+    {
+      "epoch": 102.27,
+      "learning_rate": 0.00011235087719298244,
+      "loss": 0.0018,
+      "step": 6852
+    },
+    {
+      "epoch": 102.28,
+      "learning_rate": 0.0001123157894736842,
+      "loss": 0.0006,
+      "step": 6853
+    },
+    {
+      "epoch": 102.3,
+      "learning_rate": 0.00011228070175438595,
+      "loss": 0.0616,
+      "step": 6854
+    },
+    {
+      "epoch": 102.31,
+      "learning_rate": 0.00011224561403508772,
+      "loss": 0.0104,
+      "step": 6855
+    },
+    {
+      "epoch": 102.33,
+      "learning_rate": 0.00011221052631578945,
+      "loss": 0.0321,
+      "step": 6856
+    },
+    {
+      "epoch": 102.34,
+      "learning_rate": 0.00011217543859649121,
+      "loss": 0.0013,
+      "step": 6857
+    },
+    {
+      "epoch": 102.36,
+      "learning_rate": 0.00011214035087719298,
+      "loss": 0.0983,
+      "step": 6858
+    },
+    {
+      "epoch": 102.37,
+      "learning_rate": 0.00011210526315789472,
+      "loss": 0.0005,
+      "step": 6859
+    },
+    {
+      "epoch": 102.39,
+      "learning_rate": 0.00011207017543859649,
+      "loss": 0.0901,
+      "step": 6860
+    },
+    {
+      "epoch": 102.4,
+      "learning_rate": 0.00011203508771929824,
+      "loss": 0.078,
+      "step": 6861
+    },
+    {
+      "epoch": 102.42,
+      "learning_rate": 0.000112,
+      "loss": 0.0034,
+      "step": 6862
+    },
+    {
+      "epoch": 102.43,
+      "learning_rate": 0.00011196491228070175,
+      "loss": 0.0006,
+      "step": 6863
+    },
+    {
+      "epoch": 102.45,
+      "learning_rate": 0.00011192982456140351,
+      "loss": 0.0079,
+      "step": 6864
+    },
+    {
+      "epoch": 102.46,
+      "learning_rate": 0.00011189473684210524,
+      "loss": 0.0381,
+      "step": 6865
+    },
+    {
+      "epoch": 102.48,
+      "learning_rate": 0.00011185964912280701,
+      "loss": 0.0005,
+      "step": 6866
+    },
+    {
+      "epoch": 102.49,
+      "learning_rate": 0.00011182456140350876,
+      "loss": 0.0005,
+      "step": 6867
+    },
+    {
+      "epoch": 102.51,
+      "learning_rate": 0.00011178947368421052,
+      "loss": 0.0006,
+      "step": 6868
+    },
+    {
+      "epoch": 102.52,
+      "learning_rate": 0.00011175438596491227,
+      "loss": 0.0259,
+      "step": 6869
+    },
+    {
+      "epoch": 102.54,
+      "learning_rate": 0.00011171929824561403,
+      "loss": 0.0227,
+      "step": 6870
+    },
+    {
+      "epoch": 102.55,
+      "learning_rate": 0.00011168421052631579,
+      "loss": 0.0005,
+      "step": 6871
+    },
+    {
+      "epoch": 102.57,
+      "learning_rate": 0.00011164912280701753,
+      "loss": 0.0165,
+      "step": 6872
+    },
+    {
+      "epoch": 102.58,
+      "learning_rate": 0.00011161403508771929,
+      "loss": 0.0009,
+      "step": 6873
+    },
+    {
+      "epoch": 102.59,
+      "learning_rate": 0.00011157894736842104,
+      "loss": 0.0006,
+      "step": 6874
+    },
+    {
+      "epoch": 102.61,
+      "learning_rate": 0.0001115438596491228,
+      "loss": 0.0014,
+      "step": 6875
+    },
+    {
+      "epoch": 102.62,
+      "learning_rate": 0.00011150877192982455,
+      "loss": 0.2293,
+      "step": 6876
+    },
+    {
+      "epoch": 102.64,
+      "learning_rate": 0.00011147368421052631,
+      "loss": 0.0006,
+      "step": 6877
+    },
+    {
+      "epoch": 102.65,
+      "learning_rate": 0.00011143859649122806,
+      "loss": 0.0031,
+      "step": 6878
+    },
+    {
+      "epoch": 102.67,
+      "learning_rate": 0.00011140350877192982,
+      "loss": 0.0006,
+      "step": 6879
+    },
+    {
+      "epoch": 102.68,
+      "learning_rate": 0.00011136842105263156,
+      "loss": 0.0005,
+      "step": 6880
+    },
+    {
+      "epoch": 102.7,
+      "learning_rate": 0.00011133333333333332,
+      "loss": 0.0006,
+      "step": 6881
+    },
+    {
+      "epoch": 102.71,
+      "learning_rate": 0.00011129824561403507,
+      "loss": 0.0009,
+      "step": 6882
+    },
+    {
+      "epoch": 102.73,
+      "learning_rate": 0.00011126315789473683,
+      "loss": 0.0055,
+      "step": 6883
+    },
+    {
+      "epoch": 102.74,
+      "learning_rate": 0.0001112280701754386,
+      "loss": 0.029,
+      "step": 6884
+    },
+    {
+      "epoch": 102.76,
+      "learning_rate": 0.00011119298245614034,
+      "loss": 0.0482,
+      "step": 6885
+    },
+    {
+      "epoch": 102.77,
+      "learning_rate": 0.0001111578947368421,
+      "loss": 0.0178,
+      "step": 6886
+    },
+    {
+      "epoch": 102.79,
+      "learning_rate": 0.00011112280701754385,
+      "loss": 0.0007,
+      "step": 6887
+    },
+    {
+      "epoch": 102.8,
+      "learning_rate": 0.0001110877192982456,
+      "loss": 0.0005,
+      "step": 6888
+    },
+    {
+      "epoch": 102.82,
+      "learning_rate": 0.00011105263157894735,
+      "loss": 0.0006,
+      "step": 6889
+    },
+    {
+      "epoch": 102.83,
+      "learning_rate": 0.00011101754385964911,
+      "loss": 0.0014,
+      "step": 6890
+    },
+    {
+      "epoch": 102.85,
+      "learning_rate": 0.00011098245614035086,
+      "loss": 0.0006,
+      "step": 6891
+    },
+    {
+      "epoch": 102.86,
+      "learning_rate": 0.00011094736842105262,
+      "loss": 0.0005,
+      "step": 6892
+    },
+    {
+      "epoch": 102.88,
+      "learning_rate": 0.00011091228070175437,
+      "loss": 0.0545,
+      "step": 6893
+    },
+    {
+      "epoch": 102.89,
+      "learning_rate": 0.00011087719298245614,
+      "loss": 0.018,
+      "step": 6894
+    },
+    {
+      "epoch": 102.91,
+      "learning_rate": 0.0001108421052631579,
+      "loss": 0.0625,
+      "step": 6895
+    },
+    {
+      "epoch": 102.92,
+      "learning_rate": 0.00011080701754385963,
+      "loss": 0.0008,
+      "step": 6896
+    },
+    {
+      "epoch": 102.94,
+      "learning_rate": 0.0001107719298245614,
+      "loss": 0.0009,
+      "step": 6897
+    },
+    {
+      "epoch": 102.95,
+      "learning_rate": 0.00011073684210526314,
+      "loss": 0.0486,
+      "step": 6898
+    },
+    {
+      "epoch": 102.97,
+      "learning_rate": 0.0001107017543859649,
+      "loss": 0.0005,
+      "step": 6899
+    },
+    {
+      "epoch": 102.98,
+      "learning_rate": 0.00011066666666666666,
+      "loss": 0.0007,
+      "step": 6900
+    },
+    {
+      "epoch": 103.0,
+      "learning_rate": 0.00011063157894736842,
+      "loss": 0.0005,
+      "step": 6901
+    },
+    {
+      "epoch": 103.01,
+      "learning_rate": 0.00011059649122807017,
+      "loss": 0.0022,
+      "step": 6902
+    },
+    {
+      "epoch": 103.03,
+      "learning_rate": 0.00011056140350877193,
+      "loss": 0.0014,
+      "step": 6903
+    },
+    {
+      "epoch": 103.04,
+      "learning_rate": 0.00011052631578947366,
+      "loss": 0.001,
+      "step": 6904
+    },
+    {
+      "epoch": 103.06,
+      "learning_rate": 0.00011049122807017543,
+      "loss": 0.0007,
+      "step": 6905
+    },
+    {
+      "epoch": 103.07,
+      "learning_rate": 0.00011045614035087717,
+      "loss": 0.0019,
+      "step": 6906
+    },
+    {
+      "epoch": 103.09,
+      "learning_rate": 0.00011042105263157894,
+      "loss": 0.0005,
+      "step": 6907
+    },
+    {
+      "epoch": 103.1,
+      "learning_rate": 0.0001103859649122807,
+      "loss": 0.0007,
+      "step": 6908
+    },
+    {
+      "epoch": 103.12,
+      "learning_rate": 0.00011035087719298245,
+      "loss": 0.0418,
+      "step": 6909
+    },
+    {
+      "epoch": 103.13,
+      "learning_rate": 0.00011031578947368421,
+      "loss": 0.0005,
+      "step": 6910
+    },
+    {
+      "epoch": 103.15,
+      "learning_rate": 0.00011028070175438596,
+      "loss": 0.0019,
+      "step": 6911
+    },
+    {
+      "epoch": 103.16,
+      "learning_rate": 0.00011024561403508771,
+      "loss": 0.0004,
+      "step": 6912
+    },
+    {
+      "epoch": 103.18,
+      "learning_rate": 0.00011021052631578946,
+      "loss": 0.0004,
+      "step": 6913
+    },
+    {
+      "epoch": 103.19,
+      "learning_rate": 0.00011017543859649122,
+      "loss": 0.0004,
+      "step": 6914
+    },
+    {
+      "epoch": 103.21,
+      "learning_rate": 0.00011014035087719297,
+      "loss": 0.0005,
+      "step": 6915
+    },
+    {
+      "epoch": 103.22,
+      "learning_rate": 0.00011010526315789473,
+      "loss": 0.0005,
+      "step": 6916
+    },
+    {
+      "epoch": 103.24,
+      "learning_rate": 0.00011007017543859648,
+      "loss": 0.0011,
+      "step": 6917
+    },
+    {
+      "epoch": 103.25,
+      "learning_rate": 0.00011003508771929824,
+      "loss": 0.0006,
+      "step": 6918
+    },
+    {
+      "epoch": 103.27,
+      "learning_rate": 0.00010999999999999998,
+      "loss": 0.0006,
+      "step": 6919
+    },
+    {
+      "epoch": 103.28,
+      "learning_rate": 0.00010996491228070174,
+      "loss": 0.0004,
+      "step": 6920
+    },
+    {
+      "epoch": 103.3,
+      "learning_rate": 0.0001099298245614035,
+      "loss": 0.0006,
+      "step": 6921
+    },
+    {
+      "epoch": 103.31,
+      "learning_rate": 0.00010989473684210525,
+      "loss": 0.0006,
+      "step": 6922
+    },
+    {
+      "epoch": 103.33,
+      "learning_rate": 0.00010985964912280701,
+      "loss": 0.0007,
+      "step": 6923
+    },
+    {
+      "epoch": 103.34,
+      "learning_rate": 0.00010982456140350876,
+      "loss": 0.008,
+      "step": 6924
+    },
+    {
+      "epoch": 103.36,
+      "learning_rate": 0.00010978947368421052,
+      "loss": 0.0014,
+      "step": 6925
+    },
+    {
+      "epoch": 103.37,
+      "learning_rate": 0.00010975438596491227,
+      "loss": 0.0016,
+      "step": 6926
+    },
+    {
+      "epoch": 103.39,
+      "learning_rate": 0.00010971929824561403,
+      "loss": 0.0005,
+      "step": 6927
+    },
+    {
+      "epoch": 103.4,
+      "learning_rate": 0.00010968421052631577,
+      "loss": 0.0005,
+      "step": 6928
+    },
+    {
+      "epoch": 103.42,
+      "learning_rate": 0.00010964912280701753,
+      "loss": 0.0724,
+      "step": 6929
+    },
+    {
+      "epoch": 103.43,
+      "learning_rate": 0.00010961403508771928,
+      "loss": 0.0007,
+      "step": 6930
+    },
+    {
+      "epoch": 103.45,
+      "learning_rate": 0.00010957894736842104,
+      "loss": 0.0006,
+      "step": 6931
+    },
+    {
+      "epoch": 103.46,
+      "learning_rate": 0.00010954385964912279,
+      "loss": 0.0149,
+      "step": 6932
+    },
+    {
+      "epoch": 103.48,
+      "learning_rate": 0.00010950877192982455,
+      "loss": 0.0024,
+      "step": 6933
+    },
+    {
+      "epoch": 103.49,
+      "learning_rate": 0.00010947368421052632,
+      "loss": 0.1264,
+      "step": 6934
+    },
+    {
+      "epoch": 103.51,
+      "learning_rate": 0.00010943859649122805,
+      "loss": 0.0011,
+      "step": 6935
+    },
+    {
+      "epoch": 103.52,
+      "learning_rate": 0.00010940350877192981,
+      "loss": 0.0013,
+      "step": 6936
+    },
+    {
+      "epoch": 103.54,
+      "learning_rate": 0.00010936842105263156,
+      "loss": 0.0004,
+      "step": 6937
+    },
+    {
+      "epoch": 103.55,
+      "learning_rate": 0.00010933333333333333,
+      "loss": 0.1629,
+      "step": 6938
+    },
+    {
+      "epoch": 103.57,
+      "learning_rate": 0.00010929824561403507,
+      "loss": 0.0004,
+      "step": 6939
+    },
+    {
+      "epoch": 103.58,
+      "learning_rate": 0.00010926315789473684,
+      "loss": 0.0007,
+      "step": 6940
+    },
+    {
+      "epoch": 103.59,
+      "learning_rate": 0.00010922807017543859,
+      "loss": 0.0009,
+      "step": 6941
+    },
+    {
+      "epoch": 103.61,
+      "learning_rate": 0.00010919298245614035,
+      "loss": 0.0004,
+      "step": 6942
+    },
+    {
+      "epoch": 103.62,
+      "learning_rate": 0.00010915789473684208,
+      "loss": 0.0004,
+      "step": 6943
+    },
+    {
+      "epoch": 103.64,
+      "learning_rate": 0.00010912280701754385,
+      "loss": 0.0004,
+      "step": 6944
+    },
+    {
+      "epoch": 103.65,
+      "learning_rate": 0.00010908771929824561,
+      "loss": 0.0004,
+      "step": 6945
+    },
+    {
+      "epoch": 103.67,
+      "learning_rate": 0.00010905263157894736,
+      "loss": 0.0215,
+      "step": 6946
+    },
+    {
+      "epoch": 103.68,
+      "learning_rate": 0.00010901754385964912,
+      "loss": 0.0005,
+      "step": 6947
+    },
+    {
+      "epoch": 103.7,
+      "learning_rate": 0.00010898245614035087,
+      "loss": 0.0006,
+      "step": 6948
+    },
+    {
+      "epoch": 103.71,
+      "learning_rate": 0.00010894736842105263,
+      "loss": 0.0005,
+      "step": 6949
+    },
+    {
+      "epoch": 103.73,
+      "learning_rate": 0.00010891228070175438,
+      "loss": 0.0006,
+      "step": 6950
+    },
+    {
+      "epoch": 103.74,
+      "learning_rate": 0.00010887719298245614,
+      "loss": 0.0006,
+      "step": 6951
+    },
+    {
+      "epoch": 103.76,
+      "learning_rate": 0.00010884210526315788,
+      "loss": 0.0238,
+      "step": 6952
+    },
+    {
+      "epoch": 103.77,
+      "learning_rate": 0.00010880701754385964,
+      "loss": 0.0102,
+      "step": 6953
+    },
+    {
+      "epoch": 103.79,
+      "learning_rate": 0.00010877192982456139,
+      "loss": 0.0005,
+      "step": 6954
+    },
+    {
+      "epoch": 103.8,
+      "learning_rate": 0.00010873684210526315,
+      "loss": 0.0006,
+      "step": 6955
+    },
+    {
+      "epoch": 103.82,
+      "learning_rate": 0.0001087017543859649,
+      "loss": 0.0011,
+      "step": 6956
+    },
+    {
+      "epoch": 103.83,
+      "learning_rate": 0.00010866666666666666,
+      "loss": 0.0007,
+      "step": 6957
+    },
+    {
+      "epoch": 103.85,
+      "learning_rate": 0.00010863157894736842,
+      "loss": 0.0612,
+      "step": 6958
+    },
+    {
+      "epoch": 103.86,
+      "learning_rate": 0.00010859649122807016,
+      "loss": 0.0005,
+      "step": 6959
+    },
+    {
+      "epoch": 103.88,
+      "learning_rate": 0.00010856140350877192,
+      "loss": 0.0006,
+      "step": 6960
+    },
+    {
+      "epoch": 103.89,
+      "learning_rate": 0.00010852631578947367,
+      "loss": 0.0005,
+      "step": 6961
+    },
+    {
+      "epoch": 103.91,
+      "learning_rate": 0.00010849122807017543,
+      "loss": 0.0165,
+      "step": 6962
+    },
+    {
+      "epoch": 103.92,
+      "learning_rate": 0.00010845614035087718,
+      "loss": 0.0005,
+      "step": 6963
+    },
+    {
+      "epoch": 103.94,
+      "learning_rate": 0.00010842105263157894,
+      "loss": 0.0006,
+      "step": 6964
+    },
+    {
+      "epoch": 103.95,
+      "learning_rate": 0.00010838596491228069,
+      "loss": 0.0004,
+      "step": 6965
+    },
+    {
+      "epoch": 103.97,
+      "learning_rate": 0.00010835087719298245,
+      "loss": 0.0009,
+      "step": 6966
+    },
+    {
+      "epoch": 103.98,
+      "learning_rate": 0.00010831578947368419,
+      "loss": 0.0008,
+      "step": 6967
+    },
+    {
+      "epoch": 104.0,
+      "learning_rate": 0.00010828070175438595,
+      "loss": 0.0005,
+      "step": 6968
+    },
+    {
+      "epoch": 104.01,
+      "learning_rate": 0.0001082456140350877,
+      "loss": 0.0032,
+      "step": 6969
+    },
+    {
+      "epoch": 104.03,
+      "learning_rate": 0.00010821052631578946,
+      "loss": 0.0005,
+      "step": 6970
+    },
+    {
+      "epoch": 104.04,
+      "learning_rate": 0.00010817543859649122,
+      "loss": 0.0006,
+      "step": 6971
+    },
+    {
+      "epoch": 104.06,
+      "learning_rate": 0.00010814035087719297,
+      "loss": 0.0052,
+      "step": 6972
+    },
+    {
+      "epoch": 104.07,
+      "learning_rate": 0.00010810526315789474,
+      "loss": 0.0004,
+      "step": 6973
+    },
+    {
+      "epoch": 104.09,
+      "learning_rate": 0.00010807017543859648,
+      "loss": 0.0004,
+      "step": 6974
+    },
+    {
+      "epoch": 104.1,
+      "learning_rate": 0.00010803508771929825,
+      "loss": 0.0005,
+      "step": 6975
+    },
+    {
+      "epoch": 104.12,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 0.0102,
+      "step": 6976
+    },
+    {
+      "epoch": 104.13,
+      "learning_rate": 0.00010796491228070174,
+      "loss": 0.0004,
+      "step": 6977
+    },
+    {
+      "epoch": 104.15,
+      "learning_rate": 0.0001079298245614035,
+      "loss": 0.0003,
+      "step": 6978
+    },
+    {
+      "epoch": 104.16,
+      "learning_rate": 0.00010789473684210526,
+      "loss": 0.0009,
+      "step": 6979
+    },
+    {
+      "epoch": 104.18,
+      "learning_rate": 0.000107859649122807,
+      "loss": 0.0116,
+      "step": 6980
+    },
+    {
+      "epoch": 104.19,
+      "learning_rate": 0.00010782456140350877,
+      "loss": 0.0003,
+      "step": 6981
+    },
+    {
+      "epoch": 104.21,
+      "learning_rate": 0.0001077894736842105,
+      "loss": 0.0005,
+      "step": 6982
+    },
+    {
+      "epoch": 104.22,
+      "learning_rate": 0.00010775438596491226,
+      "loss": 0.0074,
+      "step": 6983
+    },
+    {
+      "epoch": 104.24,
+      "learning_rate": 0.00010771929824561403,
+      "loss": 0.0005,
+      "step": 6984
+    },
+    {
+      "epoch": 104.25,
+      "learning_rate": 0.00010768421052631578,
+      "loss": 0.0004,
+      "step": 6985
+    },
+    {
+      "epoch": 104.27,
+      "learning_rate": 0.00010764912280701754,
+      "loss": 0.0471,
+      "step": 6986
+    },
+    {
+      "epoch": 104.28,
+      "learning_rate": 0.00010761403508771929,
+      "loss": 0.0005,
+      "step": 6987
+    },
+    {
+      "epoch": 104.3,
+      "learning_rate": 0.00010757894736842105,
+      "loss": 0.0229,
+      "step": 6988
+    },
+    {
+      "epoch": 104.31,
+      "learning_rate": 0.0001075438596491228,
+      "loss": 0.0003,
+      "step": 6989
+    },
+    {
+      "epoch": 104.33,
+      "learning_rate": 0.00010750877192982456,
+      "loss": 0.0005,
+      "step": 6990
+    },
+    {
+      "epoch": 104.34,
+      "learning_rate": 0.0001074736842105263,
+      "loss": 0.0004,
+      "step": 6991
+    },
+    {
+      "epoch": 104.36,
+      "learning_rate": 0.00010743859649122806,
+      "loss": 0.0005,
+      "step": 6992
+    },
+    {
+      "epoch": 104.37,
+      "learning_rate": 0.0001074035087719298,
+      "loss": 0.1992,
+      "step": 6993
+    },
+    {
+      "epoch": 104.39,
+      "learning_rate": 0.00010736842105263157,
+      "loss": 0.0003,
+      "step": 6994
+    },
+    {
+      "epoch": 104.4,
+      "learning_rate": 0.00010733333333333332,
+      "loss": 0.0005,
+      "step": 6995
+    },
+    {
+      "epoch": 104.42,
+      "learning_rate": 0.00010729824561403508,
+      "loss": 0.0018,
+      "step": 6996
+    },
+    {
+      "epoch": 104.43,
+      "learning_rate": 0.00010726315789473684,
+      "loss": 0.0399,
+      "step": 6997
+    },
+    {
+      "epoch": 104.45,
+      "learning_rate": 0.00010722807017543859,
+      "loss": 0.0004,
+      "step": 6998
+    },
+    {
+      "epoch": 104.46,
+      "learning_rate": 0.00010719298245614035,
+      "loss": 0.1261,
+      "step": 6999
+    },
+    {
+      "epoch": 104.48,
+      "learning_rate": 0.00010715789473684209,
+      "loss": 0.01,
+      "step": 7000
+    },
+    {
+      "epoch": 104.48,
+      "eval_accuracy": 0.8673519334312286,
+      "eval_f1": 0.866532551783554,
+      "eval_loss": 0.7149230241775513,
+      "eval_runtime": 346.4155,
+      "eval_samples_per_second": 11.795,
+      "eval_steps_per_second": 0.739,
+      "step": 7000
+    },
+    {
+      "epoch": 104.49,
+      "learning_rate": 0.00010712280701754385,
+      "loss": 0.0006,
+      "step": 7001
+    },
+    {
+      "epoch": 104.51,
+      "learning_rate": 0.0001070877192982456,
+      "loss": 0.0004,
+      "step": 7002
+    },
+    {
+      "epoch": 104.52,
+      "learning_rate": 0.00010705263157894736,
+      "loss": 0.0005,
+      "step": 7003
+    },
+    {
+      "epoch": 104.54,
+      "learning_rate": 0.00010701754385964911,
+      "loss": 0.0046,
+      "step": 7004
+    },
+    {
+      "epoch": 104.55,
+      "learning_rate": 0.00010698245614035087,
+      "loss": 0.0004,
+      "step": 7005
+    },
+    {
+      "epoch": 104.57,
+      "learning_rate": 0.00010694736842105261,
+      "loss": 0.0004,
+      "step": 7006
+    },
+    {
+      "epoch": 104.58,
+      "learning_rate": 0.00010691228070175437,
+      "loss": 0.0004,
+      "step": 7007
+    },
+    {
+      "epoch": 104.59,
+      "learning_rate": 0.00010687719298245613,
+      "loss": 0.0015,
+      "step": 7008
+    },
+    {
+      "epoch": 104.61,
+      "learning_rate": 0.00010684210526315788,
+      "loss": 0.0003,
+      "step": 7009
+    },
+    {
+      "epoch": 104.62,
+      "learning_rate": 0.00010680701754385964,
+      "loss": 0.0004,
+      "step": 7010
+    },
+    {
+      "epoch": 104.64,
+      "learning_rate": 0.00010677192982456139,
+      "loss": 0.0011,
+      "step": 7011
+    },
+    {
+      "epoch": 104.65,
+      "learning_rate": 0.00010673684210526316,
+      "loss": 0.0004,
+      "step": 7012
+    },
+    {
+      "epoch": 104.67,
+      "learning_rate": 0.0001067017543859649,
+      "loss": 0.0004,
+      "step": 7013
+    },
+    {
+      "epoch": 104.68,
+      "learning_rate": 0.00010666666666666667,
+      "loss": 0.0218,
+      "step": 7014
+    },
+    {
+      "epoch": 104.7,
+      "learning_rate": 0.0001066315789473684,
+      "loss": 0.0021,
+      "step": 7015
+    },
+    {
+      "epoch": 104.71,
+      "learning_rate": 0.00010659649122807016,
+      "loss": 0.0008,
+      "step": 7016
+    },
+    {
+      "epoch": 104.73,
+      "learning_rate": 0.00010656140350877191,
+      "loss": 0.0005,
+      "step": 7017
+    },
+    {
+      "epoch": 104.74,
+      "learning_rate": 0.00010652631578947368,
+      "loss": 0.0005,
+      "step": 7018
+    },
+    {
+      "epoch": 104.76,
+      "learning_rate": 0.00010649122807017542,
+      "loss": 0.0005,
+      "step": 7019
+    },
+    {
+      "epoch": 104.77,
+      "learning_rate": 0.00010645614035087719,
+      "loss": 0.0004,
+      "step": 7020
+    },
+    {
+      "epoch": 104.79,
+      "learning_rate": 0.00010642105263157895,
+      "loss": 0.0005,
+      "step": 7021
+    },
+    {
+      "epoch": 104.8,
+      "learning_rate": 0.0001063859649122807,
+      "loss": 0.0157,
+      "step": 7022
+    },
+    {
+      "epoch": 104.82,
+      "learning_rate": 0.00010635087719298246,
+      "loss": 0.0004,
+      "step": 7023
+    },
+    {
+      "epoch": 104.83,
+      "learning_rate": 0.0001063157894736842,
+      "loss": 0.0014,
+      "step": 7024
+    },
+    {
+      "epoch": 104.85,
+      "learning_rate": 0.00010628070175438596,
+      "loss": 0.0004,
+      "step": 7025
+    },
+    {
+      "epoch": 104.86,
+      "learning_rate": 0.0001062456140350877,
+      "loss": 0.0005,
+      "step": 7026
+    },
+    {
+      "epoch": 104.88,
+      "learning_rate": 0.00010621052631578947,
+      "loss": 0.0149,
+      "step": 7027
+    },
+    {
+      "epoch": 104.89,
+      "learning_rate": 0.00010617543859649122,
+      "loss": 0.062,
+      "step": 7028
+    },
+    {
+      "epoch": 104.91,
+      "learning_rate": 0.00010614035087719298,
+      "loss": 0.0003,
+      "step": 7029
+    },
+    {
+      "epoch": 104.92,
+      "learning_rate": 0.00010610526315789471,
+      "loss": 0.0006,
+      "step": 7030
+    },
+    {
+      "epoch": 104.94,
+      "learning_rate": 0.00010607017543859648,
+      "loss": 0.0006,
+      "step": 7031
+    },
+    {
+      "epoch": 104.95,
+      "learning_rate": 0.00010603508771929823,
+      "loss": 0.0004,
+      "step": 7032
+    },
+    {
+      "epoch": 104.97,
+      "learning_rate": 0.00010599999999999999,
+      "loss": 0.3033,
+      "step": 7033
+    },
+    {
+      "epoch": 104.98,
+      "learning_rate": 0.00010596491228070175,
+      "loss": 0.0003,
+      "step": 7034
+    },
+    {
+      "epoch": 105.0,
+      "learning_rate": 0.0001059298245614035,
+      "loss": 0.1008,
+      "step": 7035
+    },
+    {
+      "epoch": 105.01,
+      "learning_rate": 0.00010589473684210526,
+      "loss": 0.0013,
+      "step": 7036
+    },
+    {
+      "epoch": 105.03,
+      "learning_rate": 0.00010585964912280701,
+      "loss": 0.0004,
+      "step": 7037
+    },
+    {
+      "epoch": 105.04,
+      "learning_rate": 0.00010582456140350877,
+      "loss": 0.0014,
+      "step": 7038
+    },
+    {
+      "epoch": 105.06,
+      "learning_rate": 0.00010578947368421051,
+      "loss": 0.0006,
+      "step": 7039
+    },
+    {
+      "epoch": 105.07,
+      "learning_rate": 0.00010575438596491227,
+      "loss": 0.0003,
+      "step": 7040
+    },
+    {
+      "epoch": 105.09,
+      "learning_rate": 0.00010571929824561402,
+      "loss": 0.0016,
+      "step": 7041
+    },
+    {
+      "epoch": 105.1,
+      "learning_rate": 0.00010568421052631578,
+      "loss": 0.0015,
+      "step": 7042
+    },
+    {
+      "epoch": 105.12,
+      "learning_rate": 0.00010564912280701753,
+      "loss": 0.0004,
+      "step": 7043
+    },
+    {
+      "epoch": 105.13,
+      "learning_rate": 0.00010561403508771929,
+      "loss": 0.0005,
+      "step": 7044
+    },
+    {
+      "epoch": 105.15,
+      "learning_rate": 0.00010557894736842104,
+      "loss": 0.0004,
+      "step": 7045
+    },
+    {
+      "epoch": 105.16,
+      "learning_rate": 0.0001055438596491228,
+      "loss": 0.0015,
+      "step": 7046
+    },
+    {
+      "epoch": 105.18,
+      "learning_rate": 0.00010550877192982457,
+      "loss": 0.0004,
+      "step": 7047
+    },
+    {
+      "epoch": 105.19,
+      "learning_rate": 0.0001054736842105263,
+      "loss": 0.0322,
+      "step": 7048
+    },
+    {
+      "epoch": 105.21,
+      "learning_rate": 0.00010543859649122806,
+      "loss": 0.0088,
+      "step": 7049
+    },
+    {
+      "epoch": 105.22,
+      "learning_rate": 0.00010540350877192981,
+      "loss": 0.0003,
+      "step": 7050
+    },
+    {
+      "epoch": 105.24,
+      "learning_rate": 0.00010536842105263157,
+      "loss": 0.0005,
+      "step": 7051
+    },
+    {
+      "epoch": 105.25,
+      "learning_rate": 0.00010533333333333332,
+      "loss": 0.0082,
+      "step": 7052
+    },
+    {
+      "epoch": 105.27,
+      "learning_rate": 0.00010529824561403509,
+      "loss": 0.0004,
+      "step": 7053
+    },
+    {
+      "epoch": 105.28,
+      "learning_rate": 0.00010526315789473682,
+      "loss": 0.0005,
+      "step": 7054
+    },
+    {
+      "epoch": 105.3,
+      "learning_rate": 0.00010522807017543858,
+      "loss": 0.0004,
+      "step": 7055
+    },
+    {
+      "epoch": 105.31,
+      "learning_rate": 0.00010519298245614033,
+      "loss": 0.2876,
+      "step": 7056
+    },
+    {
+      "epoch": 105.33,
+      "learning_rate": 0.0001051578947368421,
+      "loss": 0.0004,
+      "step": 7057
+    },
+    {
+      "epoch": 105.34,
+      "learning_rate": 0.00010512280701754386,
+      "loss": 0.0005,
+      "step": 7058
+    },
+    {
+      "epoch": 105.36,
+      "learning_rate": 0.0001050877192982456,
+      "loss": 0.0005,
+      "step": 7059
+    },
+    {
+      "epoch": 105.37,
+      "learning_rate": 0.00010505263157894737,
+      "loss": 0.0408,
+      "step": 7060
+    },
+    {
+      "epoch": 105.39,
+      "learning_rate": 0.00010501754385964912,
+      "loss": 0.0003,
+      "step": 7061
+    },
+    {
+      "epoch": 105.4,
+      "learning_rate": 0.00010498245614035088,
+      "loss": 0.0004,
+      "step": 7062
+    },
+    {
+      "epoch": 105.42,
+      "learning_rate": 0.00010494736842105261,
+      "loss": 0.0004,
+      "step": 7063
+    },
+    {
+      "epoch": 105.43,
+      "learning_rate": 0.00010491228070175438,
+      "loss": 0.0342,
+      "step": 7064
+    },
+    {
+      "epoch": 105.45,
+      "learning_rate": 0.00010487719298245613,
+      "loss": 0.0008,
+      "step": 7065
+    },
+    {
+      "epoch": 105.46,
+      "learning_rate": 0.00010484210526315789,
+      "loss": 0.0004,
+      "step": 7066
+    },
+    {
+      "epoch": 105.48,
+      "learning_rate": 0.00010480701754385964,
+      "loss": 0.0004,
+      "step": 7067
+    },
+    {
+      "epoch": 105.49,
+      "learning_rate": 0.0001047719298245614,
+      "loss": 0.0004,
+      "step": 7068
+    },
+    {
+      "epoch": 105.51,
+      "learning_rate": 0.00010473684210526315,
+      "loss": 0.0005,
+      "step": 7069
+    },
+    {
+      "epoch": 105.52,
+      "learning_rate": 0.00010470175438596491,
+      "loss": 0.044,
+      "step": 7070
+    },
+    {
+      "epoch": 105.54,
+      "learning_rate": 0.00010466666666666667,
+      "loss": 0.0004,
+      "step": 7071
+    },
+    {
+      "epoch": 105.55,
+      "learning_rate": 0.00010463157894736841,
+      "loss": 0.0051,
+      "step": 7072
+    },
+    {
+      "epoch": 105.57,
+      "learning_rate": 0.00010459649122807017,
+      "loss": 0.0007,
+      "step": 7073
+    },
+    {
+      "epoch": 105.58,
+      "learning_rate": 0.00010456140350877192,
+      "loss": 0.0009,
+      "step": 7074
+    },
+    {
+      "epoch": 105.59,
+      "learning_rate": 0.00010452631578947368,
+      "loss": 0.0004,
+      "step": 7075
+    },
+    {
+      "epoch": 105.61,
+      "learning_rate": 0.00010449122807017543,
+      "loss": 0.0009,
+      "step": 7076
+    },
+    {
+      "epoch": 105.62,
+      "learning_rate": 0.00010445614035087719,
+      "loss": 0.0007,
+      "step": 7077
+    },
+    {
+      "epoch": 105.64,
+      "learning_rate": 0.00010442105263157893,
+      "loss": 0.0501,
+      "step": 7078
+    },
+    {
+      "epoch": 105.65,
+      "learning_rate": 0.00010438596491228069,
+      "loss": 0.1193,
+      "step": 7079
+    },
+    {
+      "epoch": 105.67,
+      "learning_rate": 0.00010435087719298244,
+      "loss": 0.0006,
+      "step": 7080
+    },
+    {
+      "epoch": 105.68,
+      "learning_rate": 0.0001043157894736842,
+      "loss": 0.0005,
+      "step": 7081
+    },
+    {
+      "epoch": 105.7,
+      "learning_rate": 0.00010428070175438595,
+      "loss": 0.0393,
+      "step": 7082
+    },
+    {
+      "epoch": 105.71,
+      "learning_rate": 0.00010424561403508771,
+      "loss": 0.0015,
+      "step": 7083
+    },
+    {
+      "epoch": 105.73,
+      "learning_rate": 0.00010421052631578947,
+      "loss": 0.0004,
+      "step": 7084
+    },
+    {
+      "epoch": 105.74,
+      "learning_rate": 0.00010417543859649122,
+      "loss": 0.0006,
+      "step": 7085
+    },
+    {
+      "epoch": 105.76,
+      "learning_rate": 0.00010414035087719298,
+      "loss": 0.0011,
+      "step": 7086
+    },
+    {
+      "epoch": 105.77,
+      "learning_rate": 0.00010410526315789472,
+      "loss": 0.0026,
+      "step": 7087
+    },
+    {
+      "epoch": 105.79,
+      "learning_rate": 0.00010407017543859648,
+      "loss": 0.0008,
+      "step": 7088
+    },
+    {
+      "epoch": 105.8,
+      "learning_rate": 0.00010403508771929823,
+      "loss": 0.0011,
+      "step": 7089
+    },
+    {
+      "epoch": 105.82,
+      "learning_rate": 0.000104,
+      "loss": 0.0729,
+      "step": 7090
+    },
+    {
+      "epoch": 105.83,
+      "learning_rate": 0.00010396491228070174,
+      "loss": 0.0147,
+      "step": 7091
+    },
+    {
+      "epoch": 105.85,
+      "learning_rate": 0.0001039298245614035,
+      "loss": 0.0004,
+      "step": 7092
+    },
+    {
+      "epoch": 105.86,
+      "learning_rate": 0.00010389473684210525,
+      "loss": 0.0007,
+      "step": 7093
+    },
+    {
+      "epoch": 105.88,
+      "learning_rate": 0.00010385964912280702,
+      "loss": 0.0004,
+      "step": 7094
+    },
+    {
+      "epoch": 105.89,
+      "learning_rate": 0.00010382456140350875,
+      "loss": 0.0019,
+      "step": 7095
+    },
+    {
+      "epoch": 105.91,
+      "learning_rate": 0.00010378947368421051,
+      "loss": 0.0011,
+      "step": 7096
+    },
+    {
+      "epoch": 105.92,
+      "learning_rate": 0.00010375438596491228,
+      "loss": 0.0003,
+      "step": 7097
+    },
+    {
+      "epoch": 105.94,
+      "learning_rate": 0.00010371929824561402,
+      "loss": 0.0016,
+      "step": 7098
+    },
+    {
+      "epoch": 105.95,
+      "learning_rate": 0.00010368421052631579,
+      "loss": 0.313,
+      "step": 7099
+    },
+    {
+      "epoch": 105.97,
+      "learning_rate": 0.00010364912280701754,
+      "loss": 0.0122,
+      "step": 7100
+    },
+    {
+      "epoch": 105.98,
+      "learning_rate": 0.0001036140350877193,
+      "loss": 0.0004,
+      "step": 7101
+    },
+    {
+      "epoch": 106.0,
+      "learning_rate": 0.00010357894736842103,
+      "loss": 0.0011,
+      "step": 7102
+    },
+    {
+      "epoch": 106.01,
+      "learning_rate": 0.0001035438596491228,
+      "loss": 0.0025,
+      "step": 7103
+    },
+    {
+      "epoch": 106.03,
+      "learning_rate": 0.00010350877192982454,
+      "loss": 0.0007,
+      "step": 7104
+    },
+    {
+      "epoch": 106.04,
+      "learning_rate": 0.0001034736842105263,
+      "loss": 0.0091,
+      "step": 7105
+    },
+    {
+      "epoch": 106.06,
+      "learning_rate": 0.00010343859649122806,
+      "loss": 0.0014,
+      "step": 7106
+    },
+    {
+      "epoch": 106.07,
+      "learning_rate": 0.00010340350877192982,
+      "loss": 0.0006,
+      "step": 7107
+    },
+    {
+      "epoch": 106.09,
+      "learning_rate": 0.00010336842105263158,
+      "loss": 0.0014,
+      "step": 7108
+    },
+    {
+      "epoch": 106.1,
+      "learning_rate": 0.00010333333333333333,
+      "loss": 0.0007,
+      "step": 7109
+    },
+    {
+      "epoch": 106.12,
+      "learning_rate": 0.00010329824561403509,
+      "loss": 0.0005,
+      "step": 7110
+    },
+    {
+      "epoch": 106.13,
+      "learning_rate": 0.00010326315789473683,
+      "loss": 0.0054,
+      "step": 7111
+    },
+    {
+      "epoch": 106.15,
+      "learning_rate": 0.00010322807017543859,
+      "loss": 0.0003,
+      "step": 7112
+    },
+    {
+      "epoch": 106.16,
+      "learning_rate": 0.00010319298245614034,
+      "loss": 0.0121,
+      "step": 7113
+    },
+    {
+      "epoch": 106.18,
+      "learning_rate": 0.0001031578947368421,
+      "loss": 0.0257,
+      "step": 7114
+    },
+    {
+      "epoch": 106.19,
+      "learning_rate": 0.00010312280701754385,
+      "loss": 0.0006,
+      "step": 7115
+    },
+    {
+      "epoch": 106.21,
+      "learning_rate": 0.00010308771929824561,
+      "loss": 0.0004,
+      "step": 7116
+    },
+    {
+      "epoch": 106.22,
+      "learning_rate": 0.00010305263157894736,
+      "loss": 0.0145,
+      "step": 7117
+    },
+    {
+      "epoch": 106.24,
+      "learning_rate": 0.00010301754385964912,
+      "loss": 0.0004,
+      "step": 7118
+    },
+    {
+      "epoch": 106.25,
+      "learning_rate": 0.00010298245614035086,
+      "loss": 0.0669,
+      "step": 7119
+    },
+    {
+      "epoch": 106.27,
+      "learning_rate": 0.00010294736842105262,
+      "loss": 0.0009,
+      "step": 7120
+    },
+    {
+      "epoch": 106.28,
+      "learning_rate": 0.00010291228070175438,
+      "loss": 0.0004,
+      "step": 7121
+    },
+    {
+      "epoch": 106.3,
+      "learning_rate": 0.00010287719298245613,
+      "loss": 0.0005,
+      "step": 7122
+    },
+    {
+      "epoch": 106.31,
+      "learning_rate": 0.00010284210526315789,
+      "loss": 0.0005,
+      "step": 7123
+    },
+    {
+      "epoch": 106.33,
+      "learning_rate": 0.00010280701754385964,
+      "loss": 0.0003,
+      "step": 7124
+    },
+    {
+      "epoch": 106.34,
+      "learning_rate": 0.0001027719298245614,
+      "loss": 0.0004,
+      "step": 7125
+    },
+    {
+      "epoch": 106.36,
+      "learning_rate": 0.00010273684210526314,
+      "loss": 0.0439,
+      "step": 7126
+    },
+    {
+      "epoch": 106.37,
+      "learning_rate": 0.0001027017543859649,
+      "loss": 0.0006,
+      "step": 7127
+    },
+    {
+      "epoch": 106.39,
+      "learning_rate": 0.00010266666666666665,
+      "loss": 0.1396,
+      "step": 7128
+    },
+    {
+      "epoch": 106.4,
+      "learning_rate": 0.00010263157894736841,
+      "loss": 0.0007,
+      "step": 7129
+    },
+    {
+      "epoch": 106.42,
+      "learning_rate": 0.00010259649122807016,
+      "loss": 0.0177,
+      "step": 7130
+    },
+    {
+      "epoch": 106.43,
+      "learning_rate": 0.00010256140350877192,
+      "loss": 0.0004,
+      "step": 7131
+    },
+    {
+      "epoch": 106.45,
+      "learning_rate": 0.00010252631578947367,
+      "loss": 0.0007,
+      "step": 7132
+    },
+    {
+      "epoch": 106.46,
+      "learning_rate": 0.00010249122807017543,
+      "loss": 0.0005,
+      "step": 7133
+    },
+    {
+      "epoch": 106.48,
+      "learning_rate": 0.0001024561403508772,
+      "loss": 0.0004,
+      "step": 7134
+    },
+    {
+      "epoch": 106.49,
+      "learning_rate": 0.00010242105263157893,
+      "loss": 0.1185,
+      "step": 7135
+    },
+    {
+      "epoch": 106.51,
+      "learning_rate": 0.0001023859649122807,
+      "loss": 0.0004,
+      "step": 7136
+    },
+    {
+      "epoch": 106.52,
+      "learning_rate": 0.00010235087719298244,
+      "loss": 0.0011,
+      "step": 7137
+    },
+    {
+      "epoch": 106.54,
+      "learning_rate": 0.0001023157894736842,
+      "loss": 0.0003,
+      "step": 7138
+    },
+    {
+      "epoch": 106.55,
+      "learning_rate": 0.00010228070175438595,
+      "loss": 0.0006,
+      "step": 7139
+    },
+    {
+      "epoch": 106.57,
+      "learning_rate": 0.00010224561403508772,
+      "loss": 0.0004,
+      "step": 7140
+    },
+    {
+      "epoch": 106.58,
+      "learning_rate": 0.00010221052631578947,
+      "loss": 0.0005,
+      "step": 7141
+    },
+    {
+      "epoch": 106.59,
+      "learning_rate": 0.00010217543859649123,
+      "loss": 0.0237,
+      "step": 7142
+    },
+    {
+      "epoch": 106.61,
+      "learning_rate": 0.00010214035087719296,
+      "loss": 0.0056,
+      "step": 7143
+    },
+    {
+      "epoch": 106.62,
+      "learning_rate": 0.00010210526315789473,
+      "loss": 0.0004,
+      "step": 7144
+    },
+    {
+      "epoch": 106.64,
+      "learning_rate": 0.00010207017543859647,
+      "loss": 0.0005,
+      "step": 7145
+    },
+    {
+      "epoch": 106.65,
+      "learning_rate": 0.00010203508771929824,
+      "loss": 0.0004,
+      "step": 7146
+    },
+    {
+      "epoch": 106.67,
+      "learning_rate": 0.000102,
+      "loss": 0.0004,
+      "step": 7147
+    },
+    {
+      "epoch": 106.68,
+      "learning_rate": 0.00010196491228070175,
+      "loss": 0.0003,
+      "step": 7148
+    },
+    {
+      "epoch": 106.7,
+      "learning_rate": 0.00010192982456140351,
+      "loss": 0.1856,
+      "step": 7149
+    },
+    {
+      "epoch": 106.71,
+      "learning_rate": 0.00010189473684210525,
+      "loss": 0.0004,
+      "step": 7150
+    },
+    {
+      "epoch": 106.73,
+      "learning_rate": 0.00010185964912280701,
+      "loss": 0.0004,
+      "step": 7151
+    },
+    {
+      "epoch": 106.74,
+      "learning_rate": 0.00010182456140350876,
+      "loss": 0.0004,
+      "step": 7152
+    },
+    {
+      "epoch": 106.76,
+      "learning_rate": 0.00010178947368421052,
+      "loss": 0.0005,
+      "step": 7153
+    },
+    {
+      "epoch": 106.77,
+      "learning_rate": 0.00010175438596491227,
+      "loss": 0.0337,
+      "step": 7154
+    },
+    {
+      "epoch": 106.79,
+      "learning_rate": 0.00010171929824561403,
+      "loss": 0.0004,
+      "step": 7155
+    },
+    {
+      "epoch": 106.8,
+      "learning_rate": 0.00010168421052631578,
+      "loss": 0.0031,
+      "step": 7156
+    },
+    {
+      "epoch": 106.82,
+      "learning_rate": 0.00010164912280701754,
+      "loss": 0.0008,
+      "step": 7157
+    },
+    {
+      "epoch": 106.83,
+      "learning_rate": 0.00010161403508771928,
+      "loss": 0.0005,
+      "step": 7158
+    },
+    {
+      "epoch": 106.85,
+      "learning_rate": 0.00010157894736842104,
+      "loss": 0.0005,
+      "step": 7159
+    },
+    {
+      "epoch": 106.86,
+      "learning_rate": 0.0001015438596491228,
+      "loss": 0.0004,
+      "step": 7160
+    },
+    {
+      "epoch": 106.88,
+      "learning_rate": 0.00010150877192982455,
+      "loss": 0.0003,
+      "step": 7161
+    },
+    {
+      "epoch": 106.89,
+      "learning_rate": 0.00010147368421052631,
+      "loss": 0.0009,
+      "step": 7162
+    },
+    {
+      "epoch": 106.91,
+      "learning_rate": 0.00010143859649122806,
+      "loss": 0.0054,
+      "step": 7163
+    },
+    {
+      "epoch": 106.92,
+      "learning_rate": 0.00010140350877192982,
+      "loss": 0.0003,
+      "step": 7164
+    },
+    {
+      "epoch": 106.94,
+      "learning_rate": 0.00010136842105263157,
+      "loss": 0.0005,
+      "step": 7165
+    },
+    {
+      "epoch": 106.95,
+      "learning_rate": 0.00010133333333333332,
+      "loss": 0.0004,
+      "step": 7166
+    },
+    {
+      "epoch": 106.97,
+      "learning_rate": 0.00010129824561403507,
+      "loss": 0.0015,
+      "step": 7167
+    },
+    {
+      "epoch": 106.98,
+      "learning_rate": 0.00010126315789473683,
+      "loss": 0.0294,
+      "step": 7168
+    },
+    {
+      "epoch": 107.0,
+      "learning_rate": 0.00010122807017543858,
+      "loss": 0.0004,
+      "step": 7169
+    },
+    {
+      "epoch": 107.01,
+      "learning_rate": 0.00010119298245614034,
+      "loss": 0.0877,
+      "step": 7170
+    },
+    {
+      "epoch": 107.03,
+      "learning_rate": 0.0001011578947368421,
+      "loss": 0.0004,
+      "step": 7171
+    },
+    {
+      "epoch": 107.04,
+      "learning_rate": 0.00010112280701754385,
+      "loss": 0.0005,
+      "step": 7172
+    },
+    {
+      "epoch": 107.06,
+      "learning_rate": 0.00010108771929824562,
+      "loss": 0.0004,
+      "step": 7173
+    },
+    {
+      "epoch": 107.07,
+      "learning_rate": 0.00010105263157894735,
+      "loss": 0.0003,
+      "step": 7174
+    },
+    {
+      "epoch": 107.09,
+      "learning_rate": 0.00010101754385964911,
+      "loss": 0.0008,
+      "step": 7175
+    },
+    {
+      "epoch": 107.1,
+      "learning_rate": 0.00010098245614035086,
+      "loss": 0.0104,
+      "step": 7176
+    },
+    {
+      "epoch": 107.12,
+      "learning_rate": 0.00010094736842105263,
+      "loss": 0.0695,
+      "step": 7177
+    },
+    {
+      "epoch": 107.13,
+      "learning_rate": 0.00010091228070175437,
+      "loss": 0.0003,
+      "step": 7178
+    },
+    {
+      "epoch": 107.15,
+      "learning_rate": 0.00010087719298245614,
+      "loss": 0.1677,
+      "step": 7179
+    },
+    {
+      "epoch": 107.16,
+      "learning_rate": 0.00010084210526315789,
+      "loss": 0.1828,
+      "step": 7180
+    },
+    {
+      "epoch": 107.18,
+      "learning_rate": 0.00010080701754385965,
+      "loss": 0.0051,
+      "step": 7181
+    },
+    {
+      "epoch": 107.19,
+      "learning_rate": 0.00010077192982456138,
+      "loss": 0.0006,
+      "step": 7182
+    },
+    {
+      "epoch": 107.21,
+      "learning_rate": 0.00010073684210526314,
+      "loss": 0.0007,
+      "step": 7183
+    },
+    {
+      "epoch": 107.22,
+      "learning_rate": 0.00010070175438596491,
+      "loss": 0.0192,
+      "step": 7184
+    },
+    {
+      "epoch": 107.24,
+      "learning_rate": 0.00010066666666666666,
+      "loss": 0.0003,
+      "step": 7185
+    },
+    {
+      "epoch": 107.25,
+      "learning_rate": 0.00010063157894736842,
+      "loss": 0.0004,
+      "step": 7186
+    },
+    {
+      "epoch": 107.27,
+      "learning_rate": 0.00010059649122807017,
+      "loss": 0.1569,
+      "step": 7187
+    },
+    {
+      "epoch": 107.28,
+      "learning_rate": 0.00010056140350877193,
+      "loss": 0.0006,
+      "step": 7188
+    },
+    {
+      "epoch": 107.3,
+      "learning_rate": 0.00010052631578947368,
+      "loss": 0.193,
+      "step": 7189
+    },
+    {
+      "epoch": 107.31,
+      "learning_rate": 0.00010049122807017543,
+      "loss": 0.1177,
+      "step": 7190
+    },
+    {
+      "epoch": 107.33,
+      "learning_rate": 0.00010045614035087718,
+      "loss": 0.0047,
+      "step": 7191
+    },
+    {
+      "epoch": 107.34,
+      "learning_rate": 0.00010042105263157894,
+      "loss": 0.0005,
+      "step": 7192
+    },
+    {
+      "epoch": 107.36,
+      "learning_rate": 0.00010038596491228069,
+      "loss": 0.0005,
+      "step": 7193
+    },
+    {
+      "epoch": 107.37,
+      "learning_rate": 0.00010035087719298245,
+      "loss": 0.0005,
+      "step": 7194
+    },
+    {
+      "epoch": 107.39,
+      "learning_rate": 0.0001003157894736842,
+      "loss": 0.0082,
+      "step": 7195
+    },
+    {
+      "epoch": 107.4,
+      "learning_rate": 0.00010028070175438596,
+      "loss": 0.084,
+      "step": 7196
+    },
+    {
+      "epoch": 107.42,
+      "learning_rate": 0.00010024561403508772,
+      "loss": 0.0384,
+      "step": 7197
+    },
+    {
+      "epoch": 107.43,
+      "learning_rate": 0.00010021052631578946,
+      "loss": 0.0004,
+      "step": 7198
+    },
+    {
+      "epoch": 107.45,
+      "learning_rate": 0.00010017543859649122,
+      "loss": 0.2362,
+      "step": 7199
+    },
+    {
+      "epoch": 107.46,
+      "learning_rate": 0.00010014035087719297,
+      "loss": 0.0135,
+      "step": 7200
+    },
+    {
+      "epoch": 107.46,
+      "eval_accuracy": 0.8487518355359766,
+      "eval_f1": 0.8522686889090442,
+      "eval_loss": 0.8989996314048767,
+      "eval_runtime": 344.41,
+      "eval_samples_per_second": 11.864,
+      "eval_steps_per_second": 0.743,
+      "step": 7200
+    },
+    {
+      "epoch": 107.48,
+      "learning_rate": 0.00010010526315789473,
+      "loss": 0.0007,
+      "step": 7201
+    },
+    {
+      "epoch": 107.49,
+      "learning_rate": 0.00010007017543859648,
+      "loss": 0.0007,
+      "step": 7202
+    },
+    {
+      "epoch": 107.51,
+      "learning_rate": 0.00010003508771929824,
+      "loss": 0.0039,
+      "step": 7203
+    },
+    {
+      "epoch": 107.52,
+      "learning_rate": 9.999999999999999e-05,
+      "loss": 0.0029,
+      "step": 7204
+    },
+    {
+      "epoch": 107.54,
+      "learning_rate": 9.996491228070175e-05,
+      "loss": 0.0005,
+      "step": 7205
+    },
+    {
+      "epoch": 107.55,
+      "learning_rate": 9.992982456140349e-05,
+      "loss": 0.0929,
+      "step": 7206
+    },
+    {
+      "epoch": 107.57,
+      "learning_rate": 9.989473684210525e-05,
+      "loss": 0.0011,
+      "step": 7207
+    },
+    {
+      "epoch": 107.58,
+      "learning_rate": 9.9859649122807e-05,
+      "loss": 0.0015,
+      "step": 7208
+    },
+    {
+      "epoch": 107.59,
+      "learning_rate": 9.982456140350876e-05,
+      "loss": 0.1055,
+      "step": 7209
+    },
+    {
+      "epoch": 107.61,
+      "learning_rate": 9.978947368421052e-05,
+      "loss": 0.0126,
+      "step": 7210
+    },
+    {
+      "epoch": 107.62,
+      "learning_rate": 9.975438596491227e-05,
+      "loss": 0.0004,
+      "step": 7211
+    },
+    {
+      "epoch": 107.64,
+      "learning_rate": 9.971929824561404e-05,
+      "loss": 0.0492,
+      "step": 7212
+    },
+    {
+      "epoch": 107.65,
+      "learning_rate": 9.968421052631577e-05,
+      "loss": 0.0005,
+      "step": 7213
+    },
+    {
+      "epoch": 107.67,
+      "learning_rate": 9.964912280701753e-05,
+      "loss": 0.0005,
+      "step": 7214
+    },
+    {
+      "epoch": 107.68,
+      "learning_rate": 9.961403508771928e-05,
+      "loss": 0.0004,
+      "step": 7215
+    },
+    {
+      "epoch": 107.7,
+      "learning_rate": 9.957894736842104e-05,
+      "loss": 0.0005,
+      "step": 7216
+    },
+    {
+      "epoch": 107.71,
+      "learning_rate": 9.954385964912279e-05,
+      "loss": 0.1363,
+      "step": 7217
+    },
+    {
+      "epoch": 107.73,
+      "learning_rate": 9.950877192982456e-05,
+      "loss": 0.0007,
+      "step": 7218
+    },
+    {
+      "epoch": 107.74,
+      "learning_rate": 9.94736842105263e-05,
+      "loss": 0.0005,
+      "step": 7219
+    },
+    {
+      "epoch": 107.76,
+      "learning_rate": 9.943859649122807e-05,
+      "loss": 0.0035,
+      "step": 7220
+    },
+    {
+      "epoch": 107.77,
+      "learning_rate": 9.940350877192983e-05,
+      "loss": 0.0008,
+      "step": 7221
+    },
+    {
+      "epoch": 107.79,
+      "learning_rate": 9.936842105263156e-05,
+      "loss": 0.0004,
+      "step": 7222
+    },
+    {
+      "epoch": 107.8,
+      "learning_rate": 9.933333333333333e-05,
+      "loss": 0.0004,
+      "step": 7223
+    },
+    {
+      "epoch": 107.82,
+      "learning_rate": 9.929824561403508e-05,
+      "loss": 0.2155,
+      "step": 7224
+    },
+    {
+      "epoch": 107.83,
+      "learning_rate": 9.926315789473684e-05,
+      "loss": 0.0007,
+      "step": 7225
+    },
+    {
+      "epoch": 107.85,
+      "learning_rate": 9.922807017543859e-05,
+      "loss": 0.2681,
+      "step": 7226
+    },
+    {
+      "epoch": 107.86,
+      "learning_rate": 9.919298245614035e-05,
+      "loss": 0.0009,
+      "step": 7227
+    },
+    {
+      "epoch": 107.88,
+      "learning_rate": 9.91578947368421e-05,
+      "loss": 0.2484,
+      "step": 7228
+    },
+    {
+      "epoch": 107.89,
+      "learning_rate": 9.912280701754386e-05,
+      "loss": 0.0004,
+      "step": 7229
+    },
+    {
+      "epoch": 107.91,
+      "learning_rate": 9.90877192982456e-05,
+      "loss": 0.0012,
+      "step": 7230
+    },
+    {
+      "epoch": 107.92,
+      "learning_rate": 9.905263157894736e-05,
+      "loss": 0.0004,
+      "step": 7231
+    },
+    {
+      "epoch": 107.94,
+      "learning_rate": 9.90175438596491e-05,
+      "loss": 0.0005,
+      "step": 7232
+    },
+    {
+      "epoch": 107.95,
+      "learning_rate": 9.898245614035087e-05,
+      "loss": 0.0005,
+      "step": 7233
+    },
+    {
+      "epoch": 107.97,
+      "learning_rate": 9.894736842105263e-05,
+      "loss": 0.0008,
+      "step": 7234
+    },
+    {
+      "epoch": 107.98,
+      "learning_rate": 9.891228070175438e-05,
+      "loss": 0.2469,
+      "step": 7235
+    },
+    {
+      "epoch": 108.0,
+      "learning_rate": 9.887719298245614e-05,
+      "loss": 0.0005,
+      "step": 7236
+    },
+    {
+      "epoch": 108.01,
+      "learning_rate": 9.884210526315788e-05,
+      "loss": 0.0026,
+      "step": 7237
+    },
+    {
+      "epoch": 108.03,
+      "learning_rate": 9.880701754385964e-05,
+      "loss": 0.0469,
+      "step": 7238
+    },
+    {
+      "epoch": 108.04,
+      "learning_rate": 9.877192982456139e-05,
+      "loss": 0.0003,
+      "step": 7239
+    },
+    {
+      "epoch": 108.06,
+      "learning_rate": 9.873684210526315e-05,
+      "loss": 0.0006,
+      "step": 7240
+    },
+    {
+      "epoch": 108.07,
+      "learning_rate": 9.87017543859649e-05,
+      "loss": 0.0007,
+      "step": 7241
+    },
+    {
+      "epoch": 108.09,
+      "learning_rate": 9.866666666666666e-05,
+      "loss": 0.0048,
+      "step": 7242
+    },
+    {
+      "epoch": 108.1,
+      "learning_rate": 9.863157894736841e-05,
+      "loss": 0.0006,
+      "step": 7243
+    },
+    {
+      "epoch": 108.12,
+      "learning_rate": 9.859649122807017e-05,
+      "loss": 0.0006,
+      "step": 7244
+    },
+    {
+      "epoch": 108.13,
+      "learning_rate": 9.856140350877191e-05,
+      "loss": 0.0004,
+      "step": 7245
+    },
+    {
+      "epoch": 108.15,
+      "learning_rate": 9.852631578947367e-05,
+      "loss": 0.0757,
+      "step": 7246
+    },
+    {
+      "epoch": 108.16,
+      "learning_rate": 9.849122807017543e-05,
+      "loss": 0.0004,
+      "step": 7247
+    },
+    {
+      "epoch": 108.18,
+      "learning_rate": 9.845614035087718e-05,
+      "loss": 0.0097,
+      "step": 7248
+    },
+    {
+      "epoch": 108.19,
+      "learning_rate": 9.842105263157894e-05,
+      "loss": 0.0004,
+      "step": 7249
+    },
+    {
+      "epoch": 108.21,
+      "learning_rate": 9.838596491228069e-05,
+      "loss": 0.0033,
+      "step": 7250
+    },
+    {
+      "epoch": 108.22,
+      "learning_rate": 9.835087719298245e-05,
+      "loss": 0.0004,
+      "step": 7251
+    },
+    {
+      "epoch": 108.24,
+      "learning_rate": 9.83157894736842e-05,
+      "loss": 0.0004,
+      "step": 7252
+    },
+    {
+      "epoch": 108.25,
+      "learning_rate": 9.828070175438597e-05,
+      "loss": 0.1304,
+      "step": 7253
+    },
+    {
+      "epoch": 108.27,
+      "learning_rate": 9.82456140350877e-05,
+      "loss": 0.0005,
+      "step": 7254
+    },
+    {
+      "epoch": 108.28,
+      "learning_rate": 9.821052631578946e-05,
+      "loss": 0.0037,
+      "step": 7255
+    },
+    {
+      "epoch": 108.3,
+      "learning_rate": 9.817543859649121e-05,
+      "loss": 0.0031,
+      "step": 7256
+    },
+    {
+      "epoch": 108.31,
+      "learning_rate": 9.814035087719297e-05,
+      "loss": 0.1553,
+      "step": 7257
+    },
+    {
+      "epoch": 108.33,
+      "learning_rate": 9.810526315789472e-05,
+      "loss": 0.0009,
+      "step": 7258
+    },
+    {
+      "epoch": 108.34,
+      "learning_rate": 9.807017543859649e-05,
+      "loss": 0.0005,
+      "step": 7259
+    },
+    {
+      "epoch": 108.36,
+      "learning_rate": 9.803508771929825e-05,
+      "loss": 0.0206,
+      "step": 7260
+    },
+    {
+      "epoch": 108.37,
+      "learning_rate": 9.799999999999998e-05,
+      "loss": 0.0004,
+      "step": 7261
+    },
+    {
+      "epoch": 108.39,
+      "learning_rate": 9.796491228070175e-05,
+      "loss": 0.0011,
+      "step": 7262
+    },
+    {
+      "epoch": 108.4,
+      "learning_rate": 9.79298245614035e-05,
+      "loss": 0.0066,
+      "step": 7263
+    },
+    {
+      "epoch": 108.42,
+      "learning_rate": 9.789473684210526e-05,
+      "loss": 0.0004,
+      "step": 7264
+    },
+    {
+      "epoch": 108.43,
+      "learning_rate": 9.7859649122807e-05,
+      "loss": 0.0007,
+      "step": 7265
+    },
+    {
+      "epoch": 108.45,
+      "learning_rate": 9.782456140350877e-05,
+      "loss": 0.0005,
+      "step": 7266
+    },
+    {
+      "epoch": 108.46,
+      "learning_rate": 9.778947368421052e-05,
+      "loss": 0.0004,
+      "step": 7267
+    },
+    {
+      "epoch": 108.48,
+      "learning_rate": 9.775438596491228e-05,
+      "loss": 0.0006,
+      "step": 7268
+    },
+    {
+      "epoch": 108.49,
+      "learning_rate": 9.771929824561401e-05,
+      "loss": 0.0005,
+      "step": 7269
+    },
+    {
+      "epoch": 108.51,
+      "learning_rate": 9.768421052631578e-05,
+      "loss": 0.0004,
+      "step": 7270
+    },
+    {
+      "epoch": 108.52,
+      "learning_rate": 9.764912280701754e-05,
+      "loss": 0.0018,
+      "step": 7271
+    },
+    {
+      "epoch": 108.54,
+      "learning_rate": 9.761403508771929e-05,
+      "loss": 0.2924,
+      "step": 7272
+    },
+    {
+      "epoch": 108.55,
+      "learning_rate": 9.757894736842105e-05,
+      "loss": 0.0004,
+      "step": 7273
+    },
+    {
+      "epoch": 108.57,
+      "learning_rate": 9.75438596491228e-05,
+      "loss": 0.0018,
+      "step": 7274
+    },
+    {
+      "epoch": 108.58,
+      "learning_rate": 9.750877192982456e-05,
+      "loss": 0.0012,
+      "step": 7275
+    },
+    {
+      "epoch": 108.59,
+      "learning_rate": 9.747368421052631e-05,
+      "loss": 0.0125,
+      "step": 7276
+    },
+    {
+      "epoch": 108.61,
+      "learning_rate": 9.743859649122807e-05,
+      "loss": 0.0004,
+      "step": 7277
+    },
+    {
+      "epoch": 108.62,
+      "learning_rate": 9.740350877192981e-05,
+      "loss": 0.0005,
+      "step": 7278
+    },
+    {
+      "epoch": 108.64,
+      "learning_rate": 9.736842105263157e-05,
+      "loss": 0.0007,
+      "step": 7279
+    },
+    {
+      "epoch": 108.65,
+      "learning_rate": 9.733333333333332e-05,
+      "loss": 0.0005,
+      "step": 7280
+    },
+    {
+      "epoch": 108.67,
+      "learning_rate": 9.729824561403508e-05,
+      "loss": 0.0003,
+      "step": 7281
+    },
+    {
+      "epoch": 108.68,
+      "learning_rate": 9.726315789473683e-05,
+      "loss": 0.0005,
+      "step": 7282
+    },
+    {
+      "epoch": 108.7,
+      "learning_rate": 9.722807017543859e-05,
+      "loss": 0.0014,
+      "step": 7283
+    },
+    {
+      "epoch": 108.71,
+      "learning_rate": 9.719298245614035e-05,
+      "loss": 0.0005,
+      "step": 7284
+    },
+    {
+      "epoch": 108.73,
+      "learning_rate": 9.715789473684209e-05,
+      "loss": 0.0026,
+      "step": 7285
+    },
+    {
+      "epoch": 108.74,
+      "learning_rate": 9.712280701754385e-05,
+      "loss": 0.0128,
+      "step": 7286
+    },
+    {
+      "epoch": 108.76,
+      "learning_rate": 9.70877192982456e-05,
+      "loss": 0.0005,
+      "step": 7287
+    },
+    {
+      "epoch": 108.77,
+      "learning_rate": 9.705263157894736e-05,
+      "loss": 0.0045,
+      "step": 7288
+    },
+    {
+      "epoch": 108.79,
+      "learning_rate": 9.701754385964911e-05,
+      "loss": 0.0012,
+      "step": 7289
+    },
+    {
+      "epoch": 108.8,
+      "learning_rate": 9.698245614035087e-05,
+      "loss": 0.0269,
+      "step": 7290
+    },
+    {
+      "epoch": 108.82,
+      "learning_rate": 9.694736842105262e-05,
+      "loss": 0.0005,
+      "step": 7291
+    },
+    {
+      "epoch": 108.83,
+      "learning_rate": 9.691228070175439e-05,
+      "loss": 0.0006,
+      "step": 7292
+    },
+    {
+      "epoch": 108.85,
+      "learning_rate": 9.687719298245612e-05,
+      "loss": 0.0005,
+      "step": 7293
+    },
+    {
+      "epoch": 108.86,
+      "learning_rate": 9.684210526315788e-05,
+      "loss": 0.0004,
+      "step": 7294
+    },
+    {
+      "epoch": 108.88,
+      "learning_rate": 9.680701754385963e-05,
+      "loss": 0.0004,
+      "step": 7295
+    },
+    {
+      "epoch": 108.89,
+      "learning_rate": 9.67719298245614e-05,
+      "loss": 0.0004,
+      "step": 7296
+    },
+    {
+      "epoch": 108.91,
+      "learning_rate": 9.673684210526316e-05,
+      "loss": 0.0004,
+      "step": 7297
+    },
+    {
+      "epoch": 108.92,
+      "learning_rate": 9.67017543859649e-05,
+      "loss": 0.0051,
+      "step": 7298
+    },
+    {
+      "epoch": 108.94,
+      "learning_rate": 9.666666666666667e-05,
+      "loss": 0.0005,
+      "step": 7299
+    },
+    {
+      "epoch": 108.95,
+      "learning_rate": 9.663157894736842e-05,
+      "loss": 0.0012,
+      "step": 7300
+    },
+    {
+      "epoch": 108.97,
+      "learning_rate": 9.659649122807018e-05,
+      "loss": 0.0003,
+      "step": 7301
+    },
+    {
+      "epoch": 108.98,
+      "learning_rate": 9.656140350877191e-05,
+      "loss": 0.0017,
+      "step": 7302
+    },
+    {
+      "epoch": 109.0,
+      "learning_rate": 9.652631578947368e-05,
+      "loss": 0.0251,
+      "step": 7303
+    },
+    {
+      "epoch": 109.01,
+      "learning_rate": 9.649122807017542e-05,
+      "loss": 0.0009,
+      "step": 7304
+    },
+    {
+      "epoch": 109.03,
+      "learning_rate": 9.645614035087719e-05,
+      "loss": 0.0003,
+      "step": 7305
+    },
+    {
+      "epoch": 109.04,
+      "learning_rate": 9.642105263157894e-05,
+      "loss": 0.1369,
+      "step": 7306
+    },
+    {
+      "epoch": 109.06,
+      "learning_rate": 9.63859649122807e-05,
+      "loss": 0.0092,
+      "step": 7307
+    },
+    {
+      "epoch": 109.07,
+      "learning_rate": 9.635087719298243e-05,
+      "loss": 0.0006,
+      "step": 7308
+    },
+    {
+      "epoch": 109.09,
+      "learning_rate": 9.63157894736842e-05,
+      "loss": 0.1566,
+      "step": 7309
+    },
+    {
+      "epoch": 109.1,
+      "learning_rate": 9.628070175438596e-05,
+      "loss": 0.0004,
+      "step": 7310
+    },
+    {
+      "epoch": 109.12,
+      "learning_rate": 9.624561403508771e-05,
+      "loss": 0.0004,
+      "step": 7311
+    },
+    {
+      "epoch": 109.13,
+      "learning_rate": 9.621052631578947e-05,
+      "loss": 0.0006,
+      "step": 7312
+    },
+    {
+      "epoch": 109.15,
+      "learning_rate": 9.617543859649122e-05,
+      "loss": 0.0004,
+      "step": 7313
+    },
+    {
+      "epoch": 109.16,
+      "learning_rate": 9.614035087719298e-05,
+      "loss": 0.0004,
+      "step": 7314
+    },
+    {
+      "epoch": 109.18,
+      "learning_rate": 9.610526315789473e-05,
+      "loss": 0.0004,
+      "step": 7315
+    },
+    {
+      "epoch": 109.19,
+      "learning_rate": 9.607017543859649e-05,
+      "loss": 0.0004,
+      "step": 7316
+    },
+    {
+      "epoch": 109.21,
+      "learning_rate": 9.603508771929823e-05,
+      "loss": 0.0009,
+      "step": 7317
+    },
+    {
+      "epoch": 109.22,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 0.0004,
+      "step": 7318
+    },
+    {
+      "epoch": 109.24,
+      "learning_rate": 9.596491228070174e-05,
+      "loss": 0.0004,
+      "step": 7319
+    },
+    {
+      "epoch": 109.25,
+      "learning_rate": 9.59298245614035e-05,
+      "loss": 0.0003,
+      "step": 7320
+    },
+    {
+      "epoch": 109.27,
+      "learning_rate": 9.589473684210525e-05,
+      "loss": 0.0003,
+      "step": 7321
+    },
+    {
+      "epoch": 109.28,
+      "learning_rate": 9.585964912280701e-05,
+      "loss": 0.0005,
+      "step": 7322
+    },
+    {
+      "epoch": 109.3,
+      "learning_rate": 9.582456140350877e-05,
+      "loss": 0.0004,
+      "step": 7323
+    },
+    {
+      "epoch": 109.31,
+      "learning_rate": 9.578947368421052e-05,
+      "loss": 0.0005,
+      "step": 7324
+    },
+    {
+      "epoch": 109.33,
+      "learning_rate": 9.575438596491228e-05,
+      "loss": 0.0004,
+      "step": 7325
+    },
+    {
+      "epoch": 109.34,
+      "learning_rate": 9.571929824561402e-05,
+      "loss": 0.0006,
+      "step": 7326
+    },
+    {
+      "epoch": 109.36,
+      "learning_rate": 9.568421052631578e-05,
+      "loss": 0.0004,
+      "step": 7327
+    },
+    {
+      "epoch": 109.37,
+      "learning_rate": 9.564912280701753e-05,
+      "loss": 0.0004,
+      "step": 7328
+    },
+    {
+      "epoch": 109.39,
+      "learning_rate": 9.561403508771929e-05,
+      "loss": 0.286,
+      "step": 7329
+    },
+    {
+      "epoch": 109.4,
+      "learning_rate": 9.557894736842104e-05,
+      "loss": 0.0004,
+      "step": 7330
+    },
+    {
+      "epoch": 109.42,
+      "learning_rate": 9.55438596491228e-05,
+      "loss": 0.0005,
+      "step": 7331
+    },
+    {
+      "epoch": 109.43,
+      "learning_rate": 9.550877192982454e-05,
+      "loss": 0.0003,
+      "step": 7332
+    },
+    {
+      "epoch": 109.45,
+      "learning_rate": 9.54736842105263e-05,
+      "loss": 0.0005,
+      "step": 7333
+    },
+    {
+      "epoch": 109.46,
+      "learning_rate": 9.543859649122806e-05,
+      "loss": 0.0005,
+      "step": 7334
+    },
+    {
+      "epoch": 109.48,
+      "learning_rate": 9.540350877192981e-05,
+      "loss": 0.0004,
+      "step": 7335
+    },
+    {
+      "epoch": 109.49,
+      "learning_rate": 9.536842105263158e-05,
+      "loss": 0.1862,
+      "step": 7336
+    },
+    {
+      "epoch": 109.51,
+      "learning_rate": 9.533333333333332e-05,
+      "loss": 0.0016,
+      "step": 7337
+    },
+    {
+      "epoch": 109.52,
+      "learning_rate": 9.529824561403509e-05,
+      "loss": 0.0007,
+      "step": 7338
+    },
+    {
+      "epoch": 109.54,
+      "learning_rate": 9.526315789473684e-05,
+      "loss": 0.0004,
+      "step": 7339
+    },
+    {
+      "epoch": 109.55,
+      "learning_rate": 9.52280701754386e-05,
+      "loss": 0.0475,
+      "step": 7340
+    },
+    {
+      "epoch": 109.57,
+      "learning_rate": 9.519298245614033e-05,
+      "loss": 0.0006,
+      "step": 7341
+    },
+    {
+      "epoch": 109.58,
+      "learning_rate": 9.51578947368421e-05,
+      "loss": 0.0012,
+      "step": 7342
+    },
+    {
+      "epoch": 109.59,
+      "learning_rate": 9.512280701754384e-05,
+      "loss": 0.0007,
+      "step": 7343
+    },
+    {
+      "epoch": 109.61,
+      "learning_rate": 9.50877192982456e-05,
+      "loss": 0.0071,
+      "step": 7344
+    },
+    {
+      "epoch": 109.62,
+      "learning_rate": 9.505263157894735e-05,
+      "loss": 0.0008,
+      "step": 7345
+    },
+    {
+      "epoch": 109.64,
+      "learning_rate": 9.501754385964912e-05,
+      "loss": 0.0005,
+      "step": 7346
+    },
+    {
+      "epoch": 109.65,
+      "learning_rate": 9.498245614035088e-05,
+      "loss": 0.0004,
+      "step": 7347
+    },
+    {
+      "epoch": 109.67,
+      "learning_rate": 9.494736842105263e-05,
+      "loss": 0.0324,
+      "step": 7348
+    },
+    {
+      "epoch": 109.68,
+      "learning_rate": 9.491228070175439e-05,
+      "loss": 0.0005,
+      "step": 7349
+    },
+    {
+      "epoch": 109.7,
+      "learning_rate": 9.487719298245613e-05,
+      "loss": 0.0459,
+      "step": 7350
+    },
+    {
+      "epoch": 109.71,
+      "learning_rate": 9.484210526315789e-05,
+      "loss": 0.0004,
+      "step": 7351
+    },
+    {
+      "epoch": 109.73,
+      "learning_rate": 9.480701754385964e-05,
+      "loss": 0.0005,
+      "step": 7352
+    },
+    {
+      "epoch": 109.74,
+      "learning_rate": 9.47719298245614e-05,
+      "loss": 0.0036,
+      "step": 7353
+    },
+    {
+      "epoch": 109.76,
+      "learning_rate": 9.473684210526315e-05,
+      "loss": 0.0005,
+      "step": 7354
+    },
+    {
+      "epoch": 109.77,
+      "learning_rate": 9.470175438596491e-05,
+      "loss": 0.0005,
+      "step": 7355
+    },
+    {
+      "epoch": 109.79,
+      "learning_rate": 9.466666666666665e-05,
+      "loss": 0.0013,
+      "step": 7356
+    },
+    {
+      "epoch": 109.8,
+      "learning_rate": 9.463157894736841e-05,
+      "loss": 0.0005,
+      "step": 7357
+    },
+    {
+      "epoch": 109.82,
+      "learning_rate": 9.459649122807016e-05,
+      "loss": 0.0004,
+      "step": 7358
+    },
+    {
+      "epoch": 109.83,
+      "learning_rate": 9.456140350877192e-05,
+      "loss": 0.0004,
+      "step": 7359
+    },
+    {
+      "epoch": 109.85,
+      "learning_rate": 9.452631578947368e-05,
+      "loss": 0.0005,
+      "step": 7360
+    },
+    {
+      "epoch": 109.86,
+      "learning_rate": 9.449122807017543e-05,
+      "loss": 0.0005,
+      "step": 7361
+    },
+    {
+      "epoch": 109.88,
+      "learning_rate": 9.445614035087719e-05,
+      "loss": 0.0005,
+      "step": 7362
+    },
+    {
+      "epoch": 109.89,
+      "learning_rate": 9.442105263157894e-05,
+      "loss": 0.0006,
+      "step": 7363
+    },
+    {
+      "epoch": 109.91,
+      "learning_rate": 9.43859649122807e-05,
+      "loss": 0.0004,
+      "step": 7364
+    },
+    {
+      "epoch": 109.92,
+      "learning_rate": 9.435087719298244e-05,
+      "loss": 0.0005,
+      "step": 7365
+    },
+    {
+      "epoch": 109.94,
+      "learning_rate": 9.43157894736842e-05,
+      "loss": 0.3628,
+      "step": 7366
+    },
+    {
+      "epoch": 109.95,
+      "learning_rate": 9.428070175438595e-05,
+      "loss": 0.0009,
+      "step": 7367
+    },
+    {
+      "epoch": 109.97,
+      "learning_rate": 9.424561403508771e-05,
+      "loss": 0.0006,
+      "step": 7368
+    },
+    {
+      "epoch": 109.98,
+      "learning_rate": 9.421052631578946e-05,
+      "loss": 0.0004,
+      "step": 7369
+    },
+    {
+      "epoch": 110.0,
+      "learning_rate": 9.417543859649122e-05,
+      "loss": 0.0581,
+      "step": 7370
+    },
+    {
+      "epoch": 110.01,
+      "learning_rate": 9.414035087719297e-05,
+      "loss": 0.0037,
+      "step": 7371
+    },
+    {
+      "epoch": 110.03,
+      "learning_rate": 9.410526315789473e-05,
+      "loss": 0.0004,
+      "step": 7372
+    },
+    {
+      "epoch": 110.04,
+      "learning_rate": 9.407017543859648e-05,
+      "loss": 0.0005,
+      "step": 7373
+    },
+    {
+      "epoch": 110.06,
+      "learning_rate": 9.403508771929823e-05,
+      "loss": 0.0006,
+      "step": 7374
+    },
+    {
+      "epoch": 110.07,
+      "learning_rate": 9.4e-05,
+      "loss": 0.0005,
+      "step": 7375
+    },
+    {
+      "epoch": 110.09,
+      "learning_rate": 9.396491228070174e-05,
+      "loss": 0.0007,
+      "step": 7376
+    },
+    {
+      "epoch": 110.1,
+      "learning_rate": 9.39298245614035e-05,
+      "loss": 0.0004,
+      "step": 7377
+    },
+    {
+      "epoch": 110.12,
+      "learning_rate": 9.389473684210525e-05,
+      "loss": 0.1453,
+      "step": 7378
+    },
+    {
+      "epoch": 110.13,
+      "learning_rate": 9.385964912280702e-05,
+      "loss": 0.0018,
+      "step": 7379
+    },
+    {
+      "epoch": 110.15,
+      "learning_rate": 9.382456140350875e-05,
+      "loss": 0.0004,
+      "step": 7380
+    },
+    {
+      "epoch": 110.16,
+      "learning_rate": 9.378947368421051e-05,
+      "loss": 0.0005,
+      "step": 7381
+    },
+    {
+      "epoch": 110.18,
+      "learning_rate": 9.375438596491226e-05,
+      "loss": 0.0008,
+      "step": 7382
+    },
+    {
+      "epoch": 110.19,
+      "learning_rate": 9.371929824561403e-05,
+      "loss": 0.0005,
+      "step": 7383
+    },
+    {
+      "epoch": 110.21,
+      "learning_rate": 9.368421052631579e-05,
+      "loss": 0.0007,
+      "step": 7384
+    },
+    {
+      "epoch": 110.22,
+      "learning_rate": 9.364912280701754e-05,
+      "loss": 0.0005,
+      "step": 7385
+    },
+    {
+      "epoch": 110.24,
+      "learning_rate": 9.36140350877193e-05,
+      "loss": 0.0006,
+      "step": 7386
+    },
+    {
+      "epoch": 110.25,
+      "learning_rate": 9.357894736842105e-05,
+      "loss": 0.0199,
+      "step": 7387
+    },
+    {
+      "epoch": 110.27,
+      "learning_rate": 9.354385964912281e-05,
+      "loss": 0.0981,
+      "step": 7388
+    },
+    {
+      "epoch": 110.28,
+      "learning_rate": 9.350877192982455e-05,
+      "loss": 0.0014,
+      "step": 7389
+    },
+    {
+      "epoch": 110.3,
+      "learning_rate": 9.347368421052631e-05,
+      "loss": 0.0004,
+      "step": 7390
+    },
+    {
+      "epoch": 110.31,
+      "learning_rate": 9.343859649122806e-05,
+      "loss": 0.0005,
+      "step": 7391
+    },
+    {
+      "epoch": 110.33,
+      "learning_rate": 9.340350877192982e-05,
+      "loss": 0.001,
+      "step": 7392
+    },
+    {
+      "epoch": 110.34,
+      "learning_rate": 9.336842105263157e-05,
+      "loss": 0.0007,
+      "step": 7393
+    },
+    {
+      "epoch": 110.36,
+      "learning_rate": 9.333333333333333e-05,
+      "loss": 0.0005,
+      "step": 7394
+    },
+    {
+      "epoch": 110.37,
+      "learning_rate": 9.329824561403508e-05,
+      "loss": 0.003,
+      "step": 7395
+    },
+    {
+      "epoch": 110.39,
+      "learning_rate": 9.326315789473684e-05,
+      "loss": 0.0005,
+      "step": 7396
+    },
+    {
+      "epoch": 110.4,
+      "learning_rate": 9.322807017543859e-05,
+      "loss": 0.0031,
+      "step": 7397
+    },
+    {
+      "epoch": 110.42,
+      "learning_rate": 9.319298245614034e-05,
+      "loss": 0.1359,
+      "step": 7398
+    },
+    {
+      "epoch": 110.43,
+      "learning_rate": 9.31578947368421e-05,
+      "loss": 0.0018,
+      "step": 7399
+    },
+    {
+      "epoch": 110.45,
+      "learning_rate": 9.312280701754385e-05,
+      "loss": 0.0056,
+      "step": 7400
+    },
+    {
+      "epoch": 110.45,
+      "eval_accuracy": 0.8663729809104258,
+      "eval_f1": 0.8672745803593204,
+      "eval_loss": 0.7320371270179749,
+      "eval_runtime": 345.8375,
+      "eval_samples_per_second": 11.815,
+      "eval_steps_per_second": 0.74,
+      "step": 7400
+    },
+    {
+      "epoch": 110.46,
+      "learning_rate": 9.308771929824561e-05,
+      "loss": 0.0007,
+      "step": 7401
+    },
+    {
+      "epoch": 110.48,
+      "learning_rate": 9.305263157894736e-05,
+      "loss": 0.0004,
+      "step": 7402
+    },
+    {
+      "epoch": 110.49,
+      "learning_rate": 9.301754385964912e-05,
+      "loss": 0.0004,
+      "step": 7403
+    },
+    {
+      "epoch": 110.51,
+      "learning_rate": 9.298245614035086e-05,
+      "loss": 0.0007,
+      "step": 7404
+    },
+    {
+      "epoch": 110.52,
+      "learning_rate": 9.294736842105262e-05,
+      "loss": 0.0004,
+      "step": 7405
+    },
+    {
+      "epoch": 110.54,
+      "learning_rate": 9.291228070175437e-05,
+      "loss": 0.0021,
+      "step": 7406
+    },
+    {
+      "epoch": 110.55,
+      "learning_rate": 9.287719298245613e-05,
+      "loss": 0.0136,
+      "step": 7407
+    },
+    {
+      "epoch": 110.57,
+      "learning_rate": 9.284210526315788e-05,
+      "loss": 0.072,
+      "step": 7408
+    },
+    {
+      "epoch": 110.58,
+      "learning_rate": 9.280701754385964e-05,
+      "loss": 0.0006,
+      "step": 7409
+    },
+    {
+      "epoch": 110.59,
+      "learning_rate": 9.27719298245614e-05,
+      "loss": 0.0004,
+      "step": 7410
+    },
+    {
+      "epoch": 110.61,
+      "learning_rate": 9.273684210526315e-05,
+      "loss": 0.2028,
+      "step": 7411
+    },
+    {
+      "epoch": 110.62,
+      "learning_rate": 9.270175438596492e-05,
+      "loss": 0.0009,
+      "step": 7412
+    },
+    {
+      "epoch": 110.64,
+      "learning_rate": 9.266666666666665e-05,
+      "loss": 0.0004,
+      "step": 7413
+    },
+    {
+      "epoch": 110.65,
+      "learning_rate": 9.263157894736841e-05,
+      "loss": 0.0006,
+      "step": 7414
+    },
+    {
+      "epoch": 110.67,
+      "learning_rate": 9.259649122807016e-05,
+      "loss": 0.0006,
+      "step": 7415
+    },
+    {
+      "epoch": 110.68,
+      "learning_rate": 9.256140350877192e-05,
+      "loss": 0.0008,
+      "step": 7416
+    },
+    {
+      "epoch": 110.7,
+      "learning_rate": 9.252631578947367e-05,
+      "loss": 0.0008,
+      "step": 7417
+    },
+    {
+      "epoch": 110.71,
+      "learning_rate": 9.249122807017544e-05,
+      "loss": 0.0013,
+      "step": 7418
+    },
+    {
+      "epoch": 110.73,
+      "learning_rate": 9.245614035087718e-05,
+      "loss": 0.0007,
+      "step": 7419
+    },
+    {
+      "epoch": 110.74,
+      "learning_rate": 9.242105263157893e-05,
+      "loss": 0.0007,
+      "step": 7420
+    },
+    {
+      "epoch": 110.76,
+      "learning_rate": 9.238596491228068e-05,
+      "loss": 0.0004,
+      "step": 7421
+    },
+    {
+      "epoch": 110.77,
+      "learning_rate": 9.235087719298244e-05,
+      "loss": 0.1436,
+      "step": 7422
+    },
+    {
+      "epoch": 110.79,
+      "learning_rate": 9.231578947368421e-05,
+      "loss": 0.0007,
+      "step": 7423
+    },
+    {
+      "epoch": 110.8,
+      "learning_rate": 9.228070175438596e-05,
+      "loss": 0.0032,
+      "step": 7424
+    },
+    {
+      "epoch": 110.82,
+      "learning_rate": 9.224561403508772e-05,
+      "loss": 0.0004,
+      "step": 7425
+    },
+    {
+      "epoch": 110.83,
+      "learning_rate": 9.221052631578947e-05,
+      "loss": 0.0006,
+      "step": 7426
+    },
+    {
+      "epoch": 110.85,
+      "learning_rate": 9.217543859649123e-05,
+      "loss": 0.0038,
+      "step": 7427
+    },
+    {
+      "epoch": 110.86,
+      "learning_rate": 9.214035087719296e-05,
+      "loss": 0.0006,
+      "step": 7428
+    },
+    {
+      "epoch": 110.88,
+      "learning_rate": 9.210526315789473e-05,
+      "loss": 0.0011,
+      "step": 7429
+    },
+    {
+      "epoch": 110.89,
+      "learning_rate": 9.207017543859648e-05,
+      "loss": 0.0004,
+      "step": 7430
+    },
+    {
+      "epoch": 110.91,
+      "learning_rate": 9.203508771929824e-05,
+      "loss": 0.003,
+      "step": 7431
+    },
+    {
+      "epoch": 110.92,
+      "learning_rate": 9.199999999999999e-05,
+      "loss": 0.0009,
+      "step": 7432
+    },
+    {
+      "epoch": 110.94,
+      "learning_rate": 9.196491228070175e-05,
+      "loss": 0.0006,
+      "step": 7433
+    },
+    {
+      "epoch": 110.95,
+      "learning_rate": 9.192982456140351e-05,
+      "loss": 0.0026,
+      "step": 7434
+    },
+    {
+      "epoch": 110.97,
+      "learning_rate": 9.189473684210526e-05,
+      "loss": 0.0029,
+      "step": 7435
+    },
+    {
+      "epoch": 110.98,
+      "learning_rate": 9.185964912280702e-05,
+      "loss": 0.0007,
+      "step": 7436
+    },
+    {
+      "epoch": 111.0,
+      "learning_rate": 9.182456140350876e-05,
+      "loss": 0.0004,
+      "step": 7437
+    },
+    {
+      "epoch": 111.01,
+      "learning_rate": 9.178947368421052e-05,
+      "loss": 0.0081,
+      "step": 7438
+    },
+    {
+      "epoch": 111.03,
+      "learning_rate": 9.175438596491227e-05,
+      "loss": 0.0344,
+      "step": 7439
+    },
+    {
+      "epoch": 111.04,
+      "learning_rate": 9.171929824561403e-05,
+      "loss": 0.0006,
+      "step": 7440
+    },
+    {
+      "epoch": 111.06,
+      "learning_rate": 9.168421052631578e-05,
+      "loss": 0.002,
+      "step": 7441
+    },
+    {
+      "epoch": 111.07,
+      "learning_rate": 9.164912280701754e-05,
+      "loss": 0.0005,
+      "step": 7442
+    },
+    {
+      "epoch": 111.09,
+      "learning_rate": 9.161403508771929e-05,
+      "loss": 0.1795,
+      "step": 7443
+    },
+    {
+      "epoch": 111.1,
+      "learning_rate": 9.157894736842104e-05,
+      "loss": 0.0006,
+      "step": 7444
+    },
+    {
+      "epoch": 111.12,
+      "learning_rate": 9.154385964912279e-05,
+      "loss": 0.1568,
+      "step": 7445
+    },
+    {
+      "epoch": 111.13,
+      "learning_rate": 9.150877192982455e-05,
+      "loss": 0.0005,
+      "step": 7446
+    },
+    {
+      "epoch": 111.15,
+      "learning_rate": 9.147368421052631e-05,
+      "loss": 0.0023,
+      "step": 7447
+    },
+    {
+      "epoch": 111.16,
+      "learning_rate": 9.143859649122806e-05,
+      "loss": 0.0008,
+      "step": 7448
+    },
+    {
+      "epoch": 111.18,
+      "learning_rate": 9.140350877192982e-05,
+      "loss": 0.0029,
+      "step": 7449
+    },
+    {
+      "epoch": 111.19,
+      "learning_rate": 9.136842105263157e-05,
+      "loss": 0.0005,
+      "step": 7450
+    },
+    {
+      "epoch": 111.21,
+      "learning_rate": 9.133333333333334e-05,
+      "loss": 0.0005,
+      "step": 7451
+    },
+    {
+      "epoch": 111.22,
+      "learning_rate": 9.129824561403507e-05,
+      "loss": 0.0669,
+      "step": 7452
+    },
+    {
+      "epoch": 111.24,
+      "learning_rate": 9.126315789473683e-05,
+      "loss": 0.0004,
+      "step": 7453
+    },
+    {
+      "epoch": 111.25,
+      "learning_rate": 9.122807017543858e-05,
+      "loss": 0.0005,
+      "step": 7454
+    },
+    {
+      "epoch": 111.27,
+      "learning_rate": 9.119298245614034e-05,
+      "loss": 0.0004,
+      "step": 7455
+    },
+    {
+      "epoch": 111.28,
+      "learning_rate": 9.115789473684209e-05,
+      "loss": 0.0004,
+      "step": 7456
+    },
+    {
+      "epoch": 111.3,
+      "learning_rate": 9.112280701754386e-05,
+      "loss": 0.0012,
+      "step": 7457
+    },
+    {
+      "epoch": 111.31,
+      "learning_rate": 9.10877192982456e-05,
+      "loss": 0.0005,
+      "step": 7458
+    },
+    {
+      "epoch": 111.33,
+      "learning_rate": 9.105263157894737e-05,
+      "loss": 0.0006,
+      "step": 7459
+    },
+    {
+      "epoch": 111.34,
+      "learning_rate": 9.101754385964913e-05,
+      "loss": 0.0266,
+      "step": 7460
+    },
+    {
+      "epoch": 111.36,
+      "learning_rate": 9.098245614035086e-05,
+      "loss": 0.0007,
+      "step": 7461
+    },
+    {
+      "epoch": 111.37,
+      "learning_rate": 9.094736842105263e-05,
+      "loss": 0.0006,
+      "step": 7462
+    },
+    {
+      "epoch": 111.39,
+      "learning_rate": 9.091228070175437e-05,
+      "loss": 0.0004,
+      "step": 7463
+    },
+    {
+      "epoch": 111.4,
+      "learning_rate": 9.087719298245614e-05,
+      "loss": 0.0004,
+      "step": 7464
+    },
+    {
+      "epoch": 111.42,
+      "learning_rate": 9.084210526315789e-05,
+      "loss": 0.0004,
+      "step": 7465
+    },
+    {
+      "epoch": 111.43,
+      "learning_rate": 9.080701754385965e-05,
+      "loss": 0.0008,
+      "step": 7466
+    },
+    {
+      "epoch": 111.45,
+      "learning_rate": 9.077192982456138e-05,
+      "loss": 0.0007,
+      "step": 7467
+    },
+    {
+      "epoch": 111.46,
+      "learning_rate": 9.073684210526315e-05,
+      "loss": 0.0005,
+      "step": 7468
+    },
+    {
+      "epoch": 111.48,
+      "learning_rate": 9.07017543859649e-05,
+      "loss": 0.0004,
+      "step": 7469
+    },
+    {
+      "epoch": 111.49,
+      "learning_rate": 9.066666666666666e-05,
+      "loss": 0.0004,
+      "step": 7470
+    },
+    {
+      "epoch": 111.51,
+      "learning_rate": 9.06315789473684e-05,
+      "loss": 0.0129,
+      "step": 7471
+    },
+    {
+      "epoch": 111.52,
+      "learning_rate": 9.059649122807017e-05,
+      "loss": 0.002,
+      "step": 7472
+    },
+    {
+      "epoch": 111.54,
+      "learning_rate": 9.056140350877193e-05,
+      "loss": 0.0004,
+      "step": 7473
+    },
+    {
+      "epoch": 111.55,
+      "learning_rate": 9.052631578947368e-05,
+      "loss": 0.0004,
+      "step": 7474
+    },
+    {
+      "epoch": 111.57,
+      "learning_rate": 9.049122807017544e-05,
+      "loss": 0.0006,
+      "step": 7475
+    },
+    {
+      "epoch": 111.58,
+      "learning_rate": 9.045614035087718e-05,
+      "loss": 0.0004,
+      "step": 7476
+    },
+    {
+      "epoch": 111.59,
+      "learning_rate": 9.042105263157894e-05,
+      "loss": 0.001,
+      "step": 7477
+    },
+    {
+      "epoch": 111.61,
+      "learning_rate": 9.038596491228069e-05,
+      "loss": 0.0004,
+      "step": 7478
+    },
+    {
+      "epoch": 111.62,
+      "learning_rate": 9.035087719298245e-05,
+      "loss": 0.0007,
+      "step": 7479
+    },
+    {
+      "epoch": 111.64,
+      "learning_rate": 9.03157894736842e-05,
+      "loss": 0.0004,
+      "step": 7480
+    },
+    {
+      "epoch": 111.65,
+      "learning_rate": 9.028070175438596e-05,
+      "loss": 0.0005,
+      "step": 7481
+    },
+    {
+      "epoch": 111.67,
+      "learning_rate": 9.024561403508771e-05,
+      "loss": 0.0073,
+      "step": 7482
+    },
+    {
+      "epoch": 111.68,
+      "learning_rate": 9.021052631578947e-05,
+      "loss": 0.0004,
+      "step": 7483
+    },
+    {
+      "epoch": 111.7,
+      "learning_rate": 9.017543859649123e-05,
+      "loss": 0.0004,
+      "step": 7484
+    },
+    {
+      "epoch": 111.71,
+      "learning_rate": 9.014035087719297e-05,
+      "loss": 0.0154,
+      "step": 7485
+    },
+    {
+      "epoch": 111.73,
+      "learning_rate": 9.010526315789473e-05,
+      "loss": 0.003,
+      "step": 7486
+    },
+    {
+      "epoch": 111.74,
+      "learning_rate": 9.007017543859648e-05,
+      "loss": 0.0004,
+      "step": 7487
+    },
+    {
+      "epoch": 111.76,
+      "learning_rate": 9.003508771929824e-05,
+      "loss": 0.0004,
+      "step": 7488
+    },
+    {
+      "epoch": 111.77,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 0.0466,
+      "step": 7489
+    },
+    {
+      "epoch": 111.79,
+      "learning_rate": 8.996491228070175e-05,
+      "loss": 0.0004,
+      "step": 7490
+    },
+    {
+      "epoch": 111.8,
+      "learning_rate": 8.992982456140349e-05,
+      "loss": 0.0006,
+      "step": 7491
+    },
+    {
+      "epoch": 111.82,
+      "learning_rate": 8.989473684210525e-05,
+      "loss": 0.1431,
+      "step": 7492
+    },
+    {
+      "epoch": 111.83,
+      "learning_rate": 8.9859649122807e-05,
+      "loss": 0.0005,
+      "step": 7493
+    },
+    {
+      "epoch": 111.85,
+      "learning_rate": 8.982456140350876e-05,
+      "loss": 0.0008,
+      "step": 7494
+    },
+    {
+      "epoch": 111.86,
+      "learning_rate": 8.978947368421051e-05,
+      "loss": 0.0003,
+      "step": 7495
+    },
+    {
+      "epoch": 111.88,
+      "learning_rate": 8.975438596491227e-05,
+      "loss": 0.0004,
+      "step": 7496
+    },
+    {
+      "epoch": 111.89,
+      "learning_rate": 8.971929824561404e-05,
+      "loss": 0.0014,
+      "step": 7497
+    },
+    {
+      "epoch": 111.91,
+      "learning_rate": 8.968421052631579e-05,
+      "loss": 0.0004,
+      "step": 7498
+    },
+    {
+      "epoch": 111.92,
+      "learning_rate": 8.964912280701755e-05,
+      "loss": 0.0004,
+      "step": 7499
+    },
+    {
+      "epoch": 111.94,
+      "learning_rate": 8.961403508771928e-05,
+      "loss": 0.4048,
+      "step": 7500
+    },
+    {
+      "epoch": 111.95,
+      "learning_rate": 8.957894736842105e-05,
+      "loss": 0.0005,
+      "step": 7501
+    },
+    {
+      "epoch": 111.97,
+      "learning_rate": 8.95438596491228e-05,
+      "loss": 0.0007,
+      "step": 7502
+    },
+    {
+      "epoch": 111.98,
+      "learning_rate": 8.950877192982456e-05,
+      "loss": 0.0003,
+      "step": 7503
+    },
+    {
+      "epoch": 112.0,
+      "learning_rate": 8.94736842105263e-05,
+      "loss": 0.0005,
+      "step": 7504
+    },
+    {
+      "epoch": 112.01,
+      "learning_rate": 8.943859649122807e-05,
+      "loss": 0.0006,
+      "step": 7505
+    },
+    {
+      "epoch": 112.03,
+      "learning_rate": 8.940350877192982e-05,
+      "loss": 0.0004,
+      "step": 7506
+    },
+    {
+      "epoch": 112.04,
+      "learning_rate": 8.936842105263158e-05,
+      "loss": 0.0005,
+      "step": 7507
+    },
+    {
+      "epoch": 112.06,
+      "learning_rate": 8.933333333333331e-05,
+      "loss": 0.0004,
+      "step": 7508
+    },
+    {
+      "epoch": 112.07,
+      "learning_rate": 8.929824561403508e-05,
+      "loss": 0.0005,
+      "step": 7509
+    },
+    {
+      "epoch": 112.09,
+      "learning_rate": 8.926315789473684e-05,
+      "loss": 0.001,
+      "step": 7510
+    },
+    {
+      "epoch": 112.1,
+      "learning_rate": 8.922807017543859e-05,
+      "loss": 0.0007,
+      "step": 7511
+    },
+    {
+      "epoch": 112.12,
+      "learning_rate": 8.919298245614035e-05,
+      "loss": 0.0019,
+      "step": 7512
+    },
+    {
+      "epoch": 112.13,
+      "learning_rate": 8.91578947368421e-05,
+      "loss": 0.002,
+      "step": 7513
+    },
+    {
+      "epoch": 112.15,
+      "learning_rate": 8.912280701754386e-05,
+      "loss": 0.0004,
+      "step": 7514
+    },
+    {
+      "epoch": 112.16,
+      "learning_rate": 8.90877192982456e-05,
+      "loss": 0.0005,
+      "step": 7515
+    },
+    {
+      "epoch": 112.18,
+      "learning_rate": 8.905263157894736e-05,
+      "loss": 0.0004,
+      "step": 7516
+    },
+    {
+      "epoch": 112.19,
+      "learning_rate": 8.901754385964911e-05,
+      "loss": 0.0071,
+      "step": 7517
+    },
+    {
+      "epoch": 112.21,
+      "learning_rate": 8.898245614035087e-05,
+      "loss": 0.0004,
+      "step": 7518
+    },
+    {
+      "epoch": 112.22,
+      "learning_rate": 8.894736842105262e-05,
+      "loss": 0.0008,
+      "step": 7519
+    },
+    {
+      "epoch": 112.24,
+      "learning_rate": 8.891228070175438e-05,
+      "loss": 0.0004,
+      "step": 7520
+    },
+    {
+      "epoch": 112.25,
+      "learning_rate": 8.887719298245613e-05,
+      "loss": 0.0005,
+      "step": 7521
+    },
+    {
+      "epoch": 112.27,
+      "learning_rate": 8.884210526315789e-05,
+      "loss": 0.0009,
+      "step": 7522
+    },
+    {
+      "epoch": 112.28,
+      "learning_rate": 8.880701754385965e-05,
+      "loss": 0.0004,
+      "step": 7523
+    },
+    {
+      "epoch": 112.3,
+      "learning_rate": 8.877192982456139e-05,
+      "loss": 0.0004,
+      "step": 7524
+    },
+    {
+      "epoch": 112.31,
+      "learning_rate": 8.873684210526315e-05,
+      "loss": 0.0005,
+      "step": 7525
+    },
+    {
+      "epoch": 112.33,
+      "learning_rate": 8.87017543859649e-05,
+      "loss": 0.0004,
+      "step": 7526
+    },
+    {
+      "epoch": 112.34,
+      "learning_rate": 8.866666666666666e-05,
+      "loss": 0.0005,
+      "step": 7527
+    },
+    {
+      "epoch": 112.36,
+      "learning_rate": 8.863157894736841e-05,
+      "loss": 0.0006,
+      "step": 7528
+    },
+    {
+      "epoch": 112.37,
+      "learning_rate": 8.859649122807017e-05,
+      "loss": 0.0008,
+      "step": 7529
+    },
+    {
+      "epoch": 112.39,
+      "learning_rate": 8.856140350877192e-05,
+      "loss": 0.0005,
+      "step": 7530
+    },
+    {
+      "epoch": 112.4,
+      "learning_rate": 8.852631578947368e-05,
+      "loss": 0.0005,
+      "step": 7531
+    },
+    {
+      "epoch": 112.42,
+      "learning_rate": 8.849122807017542e-05,
+      "loss": 0.0004,
+      "step": 7532
+    },
+    {
+      "epoch": 112.43,
+      "learning_rate": 8.845614035087718e-05,
+      "loss": 0.0004,
+      "step": 7533
+    },
+    {
+      "epoch": 112.45,
+      "learning_rate": 8.842105263157893e-05,
+      "loss": 0.0003,
+      "step": 7534
+    },
+    {
+      "epoch": 112.46,
+      "learning_rate": 8.83859649122807e-05,
+      "loss": 0.0011,
+      "step": 7535
+    },
+    {
+      "epoch": 112.48,
+      "learning_rate": 8.835087719298246e-05,
+      "loss": 0.0003,
+      "step": 7536
+    },
+    {
+      "epoch": 112.49,
+      "learning_rate": 8.83157894736842e-05,
+      "loss": 0.0102,
+      "step": 7537
+    },
+    {
+      "epoch": 112.51,
+      "learning_rate": 8.828070175438597e-05,
+      "loss": 0.0007,
+      "step": 7538
+    },
+    {
+      "epoch": 112.52,
+      "learning_rate": 8.82456140350877e-05,
+      "loss": 0.0003,
+      "step": 7539
+    },
+    {
+      "epoch": 112.54,
+      "learning_rate": 8.821052631578946e-05,
+      "loss": 0.0003,
+      "step": 7540
+    },
+    {
+      "epoch": 112.55,
+      "learning_rate": 8.817543859649121e-05,
+      "loss": 0.0004,
+      "step": 7541
+    },
+    {
+      "epoch": 112.57,
+      "learning_rate": 8.814035087719298e-05,
+      "loss": 0.0003,
+      "step": 7542
+    },
+    {
+      "epoch": 112.58,
+      "learning_rate": 8.810526315789472e-05,
+      "loss": 0.0011,
+      "step": 7543
+    },
+    {
+      "epoch": 112.59,
+      "learning_rate": 8.807017543859649e-05,
+      "loss": 0.0056,
+      "step": 7544
+    },
+    {
+      "epoch": 112.61,
+      "learning_rate": 8.803508771929824e-05,
+      "loss": 0.0765,
+      "step": 7545
+    },
+    {
+      "epoch": 112.62,
+      "learning_rate": 8.8e-05,
+      "loss": 0.0003,
+      "step": 7546
+    },
+    {
+      "epoch": 112.64,
+      "learning_rate": 8.796491228070176e-05,
+      "loss": 0.0005,
+      "step": 7547
+    },
+    {
+      "epoch": 112.65,
+      "learning_rate": 8.79298245614035e-05,
+      "loss": 0.0004,
+      "step": 7548
+    },
+    {
+      "epoch": 112.67,
+      "learning_rate": 8.789473684210526e-05,
+      "loss": 0.0004,
+      "step": 7549
+    },
+    {
+      "epoch": 112.68,
+      "learning_rate": 8.7859649122807e-05,
+      "loss": 0.0005,
+      "step": 7550
+    },
+    {
+      "epoch": 112.7,
+      "learning_rate": 8.782456140350877e-05,
+      "loss": 0.0004,
+      "step": 7551
+    },
+    {
+      "epoch": 112.71,
+      "learning_rate": 8.778947368421052e-05,
+      "loss": 0.0004,
+      "step": 7552
+    },
+    {
+      "epoch": 112.73,
+      "learning_rate": 8.775438596491228e-05,
+      "loss": 0.0005,
+      "step": 7553
+    },
+    {
+      "epoch": 112.74,
+      "learning_rate": 8.771929824561403e-05,
+      "loss": 0.0005,
+      "step": 7554
+    },
+    {
+      "epoch": 112.76,
+      "learning_rate": 8.768421052631579e-05,
+      "loss": 0.0005,
+      "step": 7555
+    },
+    {
+      "epoch": 112.77,
+      "learning_rate": 8.764912280701753e-05,
+      "loss": 0.0003,
+      "step": 7556
+    },
+    {
+      "epoch": 112.79,
+      "learning_rate": 8.761403508771929e-05,
+      "loss": 0.0042,
+      "step": 7557
+    },
+    {
+      "epoch": 112.8,
+      "learning_rate": 8.757894736842104e-05,
+      "loss": 0.0004,
+      "step": 7558
+    },
+    {
+      "epoch": 112.82,
+      "learning_rate": 8.75438596491228e-05,
+      "loss": 0.0007,
+      "step": 7559
+    },
+    {
+      "epoch": 112.83,
+      "learning_rate": 8.750877192982456e-05,
+      "loss": 0.0003,
+      "step": 7560
+    },
+    {
+      "epoch": 112.85,
+      "learning_rate": 8.747368421052631e-05,
+      "loss": 0.0005,
+      "step": 7561
+    },
+    {
+      "epoch": 112.86,
+      "learning_rate": 8.743859649122807e-05,
+      "loss": 0.0005,
+      "step": 7562
+    },
+    {
+      "epoch": 112.88,
+      "learning_rate": 8.740350877192981e-05,
+      "loss": 0.0003,
+      "step": 7563
+    },
+    {
+      "epoch": 112.89,
+      "learning_rate": 8.736842105263157e-05,
+      "loss": 0.0005,
+      "step": 7564
+    },
+    {
+      "epoch": 112.91,
+      "learning_rate": 8.733333333333332e-05,
+      "loss": 0.0011,
+      "step": 7565
+    },
+    {
+      "epoch": 112.92,
+      "learning_rate": 8.729824561403508e-05,
+      "loss": 0.0009,
+      "step": 7566
+    },
+    {
+      "epoch": 112.94,
+      "learning_rate": 8.726315789473683e-05,
+      "loss": 0.0004,
+      "step": 7567
+    },
+    {
+      "epoch": 112.95,
+      "learning_rate": 8.722807017543859e-05,
+      "loss": 0.097,
+      "step": 7568
+    },
+    {
+      "epoch": 112.97,
+      "learning_rate": 8.719298245614034e-05,
+      "loss": 0.0004,
+      "step": 7569
+    },
+    {
+      "epoch": 112.98,
+      "learning_rate": 8.71578947368421e-05,
+      "loss": 0.0003,
+      "step": 7570
+    },
+    {
+      "epoch": 113.0,
+      "learning_rate": 8.712280701754384e-05,
+      "loss": 0.0004,
+      "step": 7571
+    },
+    {
+      "epoch": 113.01,
+      "learning_rate": 8.70877192982456e-05,
+      "loss": 0.0076,
+      "step": 7572
+    },
+    {
+      "epoch": 113.03,
+      "learning_rate": 8.705263157894736e-05,
+      "loss": 0.0003,
+      "step": 7573
+    },
+    {
+      "epoch": 113.04,
+      "learning_rate": 8.701754385964911e-05,
+      "loss": 0.0057,
+      "step": 7574
+    },
+    {
+      "epoch": 113.06,
+      "learning_rate": 8.698245614035087e-05,
+      "loss": 0.0005,
+      "step": 7575
+    },
+    {
+      "epoch": 113.07,
+      "learning_rate": 8.694736842105262e-05,
+      "loss": 0.0037,
+      "step": 7576
+    },
+    {
+      "epoch": 113.09,
+      "learning_rate": 8.691228070175439e-05,
+      "loss": 0.2041,
+      "step": 7577
+    },
+    {
+      "epoch": 113.1,
+      "learning_rate": 8.687719298245613e-05,
+      "loss": 0.0026,
+      "step": 7578
+    },
+    {
+      "epoch": 113.12,
+      "learning_rate": 8.68421052631579e-05,
+      "loss": 0.0009,
+      "step": 7579
+    },
+    {
+      "epoch": 113.13,
+      "learning_rate": 8.680701754385963e-05,
+      "loss": 0.0004,
+      "step": 7580
+    },
+    {
+      "epoch": 113.15,
+      "learning_rate": 8.67719298245614e-05,
+      "loss": 0.0008,
+      "step": 7581
+    },
+    {
+      "epoch": 113.16,
+      "learning_rate": 8.673684210526314e-05,
+      "loss": 0.0005,
+      "step": 7582
+    },
+    {
+      "epoch": 113.18,
+      "learning_rate": 8.67017543859649e-05,
+      "loss": 0.0003,
+      "step": 7583
+    },
+    {
+      "epoch": 113.19,
+      "learning_rate": 8.666666666666665e-05,
+      "loss": 0.0112,
+      "step": 7584
+    },
+    {
+      "epoch": 113.21,
+      "learning_rate": 8.663157894736842e-05,
+      "loss": 0.0011,
+      "step": 7585
+    },
+    {
+      "epoch": 113.22,
+      "learning_rate": 8.659649122807018e-05,
+      "loss": 0.0004,
+      "step": 7586
+    },
+    {
+      "epoch": 113.24,
+      "learning_rate": 8.656140350877191e-05,
+      "loss": 0.0005,
+      "step": 7587
+    },
+    {
+      "epoch": 113.25,
+      "learning_rate": 8.652631578947368e-05,
+      "loss": 0.0005,
+      "step": 7588
+    },
+    {
+      "epoch": 113.27,
+      "learning_rate": 8.649122807017543e-05,
+      "loss": 0.0046,
+      "step": 7589
+    },
+    {
+      "epoch": 113.28,
+      "learning_rate": 8.645614035087719e-05,
+      "loss": 0.0006,
+      "step": 7590
+    },
+    {
+      "epoch": 113.3,
+      "learning_rate": 8.642105263157894e-05,
+      "loss": 0.0848,
+      "step": 7591
+    },
+    {
+      "epoch": 113.31,
+      "learning_rate": 8.63859649122807e-05,
+      "loss": 0.0004,
+      "step": 7592
+    },
+    {
+      "epoch": 113.33,
+      "learning_rate": 8.635087719298245e-05,
+      "loss": 0.0401,
+      "step": 7593
+    },
+    {
+      "epoch": 113.34,
+      "learning_rate": 8.631578947368421e-05,
+      "loss": 0.2326,
+      "step": 7594
+    },
+    {
+      "epoch": 113.36,
+      "learning_rate": 8.628070175438595e-05,
+      "loss": 0.0007,
+      "step": 7595
+    },
+    {
+      "epoch": 113.37,
+      "learning_rate": 8.624561403508771e-05,
+      "loss": 0.0007,
+      "step": 7596
+    },
+    {
+      "epoch": 113.39,
+      "learning_rate": 8.621052631578947e-05,
+      "loss": 0.0086,
+      "step": 7597
+    },
+    {
+      "epoch": 113.4,
+      "learning_rate": 8.617543859649122e-05,
+      "loss": 0.0022,
+      "step": 7598
+    },
+    {
+      "epoch": 113.42,
+      "learning_rate": 8.614035087719298e-05,
+      "loss": 0.0006,
+      "step": 7599
+    },
+    {
+      "epoch": 113.43,
+      "learning_rate": 8.610526315789473e-05,
+      "loss": 0.0023,
+      "step": 7600
+    },
+    {
+      "epoch": 113.43,
+      "eval_accuracy": 0.8705335291238375,
+      "eval_f1": 0.8699635174956661,
+      "eval_loss": 0.7108378410339355,
+      "eval_runtime": 344.4393,
+      "eval_samples_per_second": 11.863,
+      "eval_steps_per_second": 0.743,
+      "step": 7600
+    },
+    {
+      "epoch": 113.45,
+      "learning_rate": 8.607017543859649e-05,
+      "loss": 0.0003,
+      "step": 7601
+    },
+    {
+      "epoch": 113.46,
+      "learning_rate": 8.603508771929824e-05,
+      "loss": 0.1926,
+      "step": 7602
+    },
+    {
+      "epoch": 113.48,
+      "learning_rate": 8.6e-05,
+      "loss": 0.0004,
+      "step": 7603
+    },
+    {
+      "epoch": 113.49,
+      "learning_rate": 8.596491228070174e-05,
+      "loss": 0.0003,
+      "step": 7604
+    },
+    {
+      "epoch": 113.51,
+      "learning_rate": 8.59298245614035e-05,
+      "loss": 0.0006,
+      "step": 7605
+    },
+    {
+      "epoch": 113.52,
+      "learning_rate": 8.589473684210525e-05,
+      "loss": 0.0004,
+      "step": 7606
+    },
+    {
+      "epoch": 113.54,
+      "learning_rate": 8.585964912280701e-05,
+      "loss": 0.0003,
+      "step": 7607
+    },
+    {
+      "epoch": 113.55,
+      "learning_rate": 8.582456140350876e-05,
+      "loss": 0.0006,
+      "step": 7608
+    },
+    {
+      "epoch": 113.57,
+      "learning_rate": 8.578947368421052e-05,
+      "loss": 0.0024,
+      "step": 7609
+    },
+    {
+      "epoch": 113.58,
+      "learning_rate": 8.575438596491229e-05,
+      "loss": 0.0123,
+      "step": 7610
+    },
+    {
+      "epoch": 113.59,
+      "learning_rate": 8.571929824561402e-05,
+      "loss": 0.0013,
+      "step": 7611
+    },
+    {
+      "epoch": 113.61,
+      "learning_rate": 8.568421052631578e-05,
+      "loss": 0.0004,
+      "step": 7612
+    },
+    {
+      "epoch": 113.62,
+      "learning_rate": 8.564912280701753e-05,
+      "loss": 0.1429,
+      "step": 7613
+    },
+    {
+      "epoch": 113.64,
+      "learning_rate": 8.56140350877193e-05,
+      "loss": 0.0117,
+      "step": 7614
+    },
+    {
+      "epoch": 113.65,
+      "learning_rate": 8.557894736842104e-05,
+      "loss": 0.0003,
+      "step": 7615
+    },
+    {
+      "epoch": 113.67,
+      "learning_rate": 8.55438596491228e-05,
+      "loss": 0.0345,
+      "step": 7616
+    },
+    {
+      "epoch": 113.68,
+      "learning_rate": 8.550877192982455e-05,
+      "loss": 0.0004,
+      "step": 7617
+    },
+    {
+      "epoch": 113.7,
+      "learning_rate": 8.547368421052632e-05,
+      "loss": 0.0004,
+      "step": 7618
+    },
+    {
+      "epoch": 113.71,
+      "learning_rate": 8.543859649122805e-05,
+      "loss": 0.0004,
+      "step": 7619
+    },
+    {
+      "epoch": 113.73,
+      "learning_rate": 8.540350877192981e-05,
+      "loss": 0.0003,
+      "step": 7620
+    },
+    {
+      "epoch": 113.74,
+      "learning_rate": 8.536842105263156e-05,
+      "loss": 0.0003,
+      "step": 7621
+    },
+    {
+      "epoch": 113.76,
+      "learning_rate": 8.533333333333332e-05,
+      "loss": 0.0004,
+      "step": 7622
+    },
+    {
+      "epoch": 113.77,
+      "learning_rate": 8.529824561403509e-05,
+      "loss": 0.0003,
+      "step": 7623
+    },
+    {
+      "epoch": 113.79,
+      "learning_rate": 8.526315789473684e-05,
+      "loss": 0.0011,
+      "step": 7624
+    },
+    {
+      "epoch": 113.8,
+      "learning_rate": 8.52280701754386e-05,
+      "loss": 0.0003,
+      "step": 7625
+    },
+    {
+      "epoch": 113.82,
+      "learning_rate": 8.519298245614035e-05,
+      "loss": 0.0003,
+      "step": 7626
+    },
+    {
+      "epoch": 113.83,
+      "learning_rate": 8.515789473684211e-05,
+      "loss": 0.0004,
+      "step": 7627
+    },
+    {
+      "epoch": 113.85,
+      "learning_rate": 8.512280701754384e-05,
+      "loss": 0.0003,
+      "step": 7628
+    },
+    {
+      "epoch": 113.86,
+      "learning_rate": 8.508771929824561e-05,
+      "loss": 0.0007,
+      "step": 7629
+    },
+    {
+      "epoch": 113.88,
+      "learning_rate": 8.505263157894736e-05,
+      "loss": 0.0004,
+      "step": 7630
+    },
+    {
+      "epoch": 113.89,
+      "learning_rate": 8.501754385964912e-05,
+      "loss": 0.0186,
+      "step": 7631
+    },
+    {
+      "epoch": 113.91,
+      "learning_rate": 8.498245614035087e-05,
+      "loss": 0.0004,
+      "step": 7632
+    },
+    {
+      "epoch": 113.92,
+      "learning_rate": 8.494736842105263e-05,
+      "loss": 0.0003,
+      "step": 7633
+    },
+    {
+      "epoch": 113.94,
+      "learning_rate": 8.491228070175436e-05,
+      "loss": 0.0003,
+      "step": 7634
+    },
+    {
+      "epoch": 113.95,
+      "learning_rate": 8.487719298245613e-05,
+      "loss": 0.0003,
+      "step": 7635
+    },
+    {
+      "epoch": 113.97,
+      "learning_rate": 8.484210526315789e-05,
+      "loss": 0.0005,
+      "step": 7636
+    },
+    {
+      "epoch": 113.98,
+      "learning_rate": 8.480701754385964e-05,
+      "loss": 0.0005,
+      "step": 7637
+    },
+    {
+      "epoch": 114.0,
+      "learning_rate": 8.47719298245614e-05,
+      "loss": 0.0003,
+      "step": 7638
+    },
+    {
+      "epoch": 114.01,
+      "learning_rate": 8.473684210526315e-05,
+      "loss": 0.0042,
+      "step": 7639
+    },
+    {
+      "epoch": 114.03,
+      "learning_rate": 8.470175438596491e-05,
+      "loss": 0.0004,
+      "step": 7640
+    },
+    {
+      "epoch": 114.04,
+      "learning_rate": 8.466666666666666e-05,
+      "loss": 0.0004,
+      "step": 7641
+    },
+    {
+      "epoch": 114.06,
+      "learning_rate": 8.463157894736842e-05,
+      "loss": 0.0256,
+      "step": 7642
+    },
+    {
+      "epoch": 114.07,
+      "learning_rate": 8.459649122807016e-05,
+      "loss": 0.0015,
+      "step": 7643
+    },
+    {
+      "epoch": 114.09,
+      "learning_rate": 8.456140350877192e-05,
+      "loss": 0.0005,
+      "step": 7644
+    },
+    {
+      "epoch": 114.1,
+      "learning_rate": 8.452631578947367e-05,
+      "loss": 0.0006,
+      "step": 7645
+    },
+    {
+      "epoch": 114.12,
+      "learning_rate": 8.449122807017543e-05,
+      "loss": 0.0003,
+      "step": 7646
+    },
+    {
+      "epoch": 114.13,
+      "learning_rate": 8.44561403508772e-05,
+      "loss": 0.0003,
+      "step": 7647
+    },
+    {
+      "epoch": 114.15,
+      "learning_rate": 8.442105263157894e-05,
+      "loss": 0.0051,
+      "step": 7648
+    },
+    {
+      "epoch": 114.16,
+      "learning_rate": 8.43859649122807e-05,
+      "loss": 0.0266,
+      "step": 7649
+    },
+    {
+      "epoch": 114.18,
+      "learning_rate": 8.435087719298245e-05,
+      "loss": 0.0007,
+      "step": 7650
+    },
+    {
+      "epoch": 114.19,
+      "learning_rate": 8.43157894736842e-05,
+      "loss": 0.097,
+      "step": 7651
+    },
+    {
+      "epoch": 114.21,
+      "learning_rate": 8.428070175438595e-05,
+      "loss": 0.0003,
+      "step": 7652
+    },
+    {
+      "epoch": 114.22,
+      "learning_rate": 8.424561403508771e-05,
+      "loss": 0.0006,
+      "step": 7653
+    },
+    {
+      "epoch": 114.24,
+      "learning_rate": 8.421052631578946e-05,
+      "loss": 0.2054,
+      "step": 7654
+    },
+    {
+      "epoch": 114.25,
+      "learning_rate": 8.417543859649122e-05,
+      "loss": 0.0003,
+      "step": 7655
+    },
+    {
+      "epoch": 114.27,
+      "learning_rate": 8.414035087719297e-05,
+      "loss": 0.0113,
+      "step": 7656
+    },
+    {
+      "epoch": 114.28,
+      "learning_rate": 8.410526315789474e-05,
+      "loss": 0.0003,
+      "step": 7657
+    },
+    {
+      "epoch": 114.3,
+      "learning_rate": 8.407017543859647e-05,
+      "loss": 0.1907,
+      "step": 7658
+    },
+    {
+      "epoch": 114.31,
+      "learning_rate": 8.403508771929823e-05,
+      "loss": 0.0003,
+      "step": 7659
+    },
+    {
+      "epoch": 114.33,
+      "learning_rate": 8.4e-05,
+      "loss": 0.0004,
+      "step": 7660
+    },
+    {
+      "epoch": 114.34,
+      "learning_rate": 8.396491228070174e-05,
+      "loss": 0.0004,
+      "step": 7661
+    },
+    {
+      "epoch": 114.36,
+      "learning_rate": 8.39298245614035e-05,
+      "loss": 0.0003,
+      "step": 7662
+    },
+    {
+      "epoch": 114.37,
+      "learning_rate": 8.389473684210526e-05,
+      "loss": 0.0192,
+      "step": 7663
+    },
+    {
+      "epoch": 114.39,
+      "learning_rate": 8.385964912280702e-05,
+      "loss": 0.0004,
+      "step": 7664
+    },
+    {
+      "epoch": 114.4,
+      "learning_rate": 8.382456140350877e-05,
+      "loss": 0.0003,
+      "step": 7665
+    },
+    {
+      "epoch": 114.42,
+      "learning_rate": 8.378947368421053e-05,
+      "loss": 0.0018,
+      "step": 7666
+    },
+    {
+      "epoch": 114.43,
+      "learning_rate": 8.375438596491226e-05,
+      "loss": 0.0003,
+      "step": 7667
+    },
+    {
+      "epoch": 114.45,
+      "learning_rate": 8.371929824561403e-05,
+      "loss": 0.0003,
+      "step": 7668
+    },
+    {
+      "epoch": 114.46,
+      "learning_rate": 8.368421052631578e-05,
+      "loss": 0.0003,
+      "step": 7669
+    },
+    {
+      "epoch": 114.48,
+      "learning_rate": 8.364912280701754e-05,
+      "loss": 0.0003,
+      "step": 7670
+    },
+    {
+      "epoch": 114.49,
+      "learning_rate": 8.361403508771929e-05,
+      "loss": 0.0004,
+      "step": 7671
+    },
+    {
+      "epoch": 114.51,
+      "learning_rate": 8.357894736842105e-05,
+      "loss": 0.0003,
+      "step": 7672
+    },
+    {
+      "epoch": 114.52,
+      "learning_rate": 8.354385964912281e-05,
+      "loss": 0.0003,
+      "step": 7673
+    },
+    {
+      "epoch": 114.54,
+      "learning_rate": 8.350877192982456e-05,
+      "loss": 0.0003,
+      "step": 7674
+    },
+    {
+      "epoch": 114.55,
+      "learning_rate": 8.347368421052631e-05,
+      "loss": 0.0629,
+      "step": 7675
+    },
+    {
+      "epoch": 114.57,
+      "learning_rate": 8.343859649122806e-05,
+      "loss": 0.0005,
+      "step": 7676
+    },
+    {
+      "epoch": 114.58,
+      "learning_rate": 8.340350877192982e-05,
+      "loss": 0.1229,
+      "step": 7677
+    },
+    {
+      "epoch": 114.59,
+      "learning_rate": 8.336842105263157e-05,
+      "loss": 0.0004,
+      "step": 7678
+    },
+    {
+      "epoch": 114.61,
+      "learning_rate": 8.333333333333333e-05,
+      "loss": 0.0455,
+      "step": 7679
+    },
+    {
+      "epoch": 114.62,
+      "learning_rate": 8.329824561403508e-05,
+      "loss": 0.2398,
+      "step": 7680
+    },
+    {
+      "epoch": 114.64,
+      "learning_rate": 8.326315789473684e-05,
+      "loss": 0.0346,
+      "step": 7681
+    },
+    {
+      "epoch": 114.65,
+      "learning_rate": 8.322807017543858e-05,
+      "loss": 0.0005,
+      "step": 7682
+    },
+    {
+      "epoch": 114.67,
+      "learning_rate": 8.319298245614034e-05,
+      "loss": 0.0003,
+      "step": 7683
+    },
+    {
+      "epoch": 114.68,
+      "learning_rate": 8.315789473684209e-05,
+      "loss": 0.0004,
+      "step": 7684
+    },
+    {
+      "epoch": 114.7,
+      "learning_rate": 8.312280701754385e-05,
+      "loss": 0.0018,
+      "step": 7685
+    },
+    {
+      "epoch": 114.71,
+      "learning_rate": 8.308771929824561e-05,
+      "loss": 0.0004,
+      "step": 7686
+    },
+    {
+      "epoch": 114.73,
+      "learning_rate": 8.305263157894736e-05,
+      "loss": 0.0023,
+      "step": 7687
+    },
+    {
+      "epoch": 114.74,
+      "learning_rate": 8.301754385964912e-05,
+      "loss": 0.0005,
+      "step": 7688
+    },
+    {
+      "epoch": 114.76,
+      "learning_rate": 8.298245614035087e-05,
+      "loss": 0.0005,
+      "step": 7689
+    },
+    {
+      "epoch": 114.77,
+      "learning_rate": 8.294736842105263e-05,
+      "loss": 0.0004,
+      "step": 7690
+    },
+    {
+      "epoch": 114.79,
+      "learning_rate": 8.291228070175437e-05,
+      "loss": 0.0004,
+      "step": 7691
+    },
+    {
+      "epoch": 114.8,
+      "learning_rate": 8.287719298245613e-05,
+      "loss": 0.0004,
+      "step": 7692
+    },
+    {
+      "epoch": 114.82,
+      "learning_rate": 8.284210526315788e-05,
+      "loss": 0.0004,
+      "step": 7693
+    },
+    {
+      "epoch": 114.83,
+      "learning_rate": 8.280701754385964e-05,
+      "loss": 0.0004,
+      "step": 7694
+    },
+    {
+      "epoch": 114.85,
+      "learning_rate": 8.277192982456139e-05,
+      "loss": 0.0003,
+      "step": 7695
+    },
+    {
+      "epoch": 114.86,
+      "learning_rate": 8.273684210526315e-05,
+      "loss": 0.1994,
+      "step": 7696
+    },
+    {
+      "epoch": 114.88,
+      "learning_rate": 8.27017543859649e-05,
+      "loss": 0.0004,
+      "step": 7697
+    },
+    {
+      "epoch": 114.89,
+      "learning_rate": 8.266666666666665e-05,
+      "loss": 0.0005,
+      "step": 7698
+    },
+    {
+      "epoch": 114.91,
+      "learning_rate": 8.263157894736841e-05,
+      "loss": 0.0003,
+      "step": 7699
+    },
+    {
+      "epoch": 114.92,
+      "learning_rate": 8.259649122807016e-05,
+      "loss": 0.0004,
+      "step": 7700
+    },
+    {
+      "epoch": 114.94,
+      "learning_rate": 8.256140350877193e-05,
+      "loss": 0.0004,
+      "step": 7701
+    },
+    {
+      "epoch": 114.95,
+      "learning_rate": 8.252631578947367e-05,
+      "loss": 0.0044,
+      "step": 7702
+    },
+    {
+      "epoch": 114.97,
+      "learning_rate": 8.249122807017544e-05,
+      "loss": 0.003,
+      "step": 7703
+    },
+    {
+      "epoch": 114.98,
+      "learning_rate": 8.245614035087719e-05,
+      "loss": 0.0004,
+      "step": 7704
+    },
+    {
+      "epoch": 115.0,
+      "learning_rate": 8.242105263157895e-05,
+      "loss": 0.0003,
+      "step": 7705
+    },
+    {
+      "epoch": 115.01,
+      "learning_rate": 8.238596491228068e-05,
+      "loss": 0.0008,
+      "step": 7706
+    },
+    {
+      "epoch": 115.03,
+      "learning_rate": 8.235087719298245e-05,
+      "loss": 0.0015,
+      "step": 7707
+    },
+    {
+      "epoch": 115.04,
+      "learning_rate": 8.23157894736842e-05,
+      "loss": 0.0003,
+      "step": 7708
+    },
+    {
+      "epoch": 115.06,
+      "learning_rate": 8.228070175438596e-05,
+      "loss": 0.0005,
+      "step": 7709
+    },
+    {
+      "epoch": 115.07,
+      "learning_rate": 8.224561403508772e-05,
+      "loss": 0.0029,
+      "step": 7710
+    },
+    {
+      "epoch": 115.09,
+      "learning_rate": 8.221052631578947e-05,
+      "loss": 0.0004,
+      "step": 7711
+    },
+    {
+      "epoch": 115.1,
+      "learning_rate": 8.217543859649123e-05,
+      "loss": 0.0007,
+      "step": 7712
+    },
+    {
+      "epoch": 115.12,
+      "learning_rate": 8.214035087719298e-05,
+      "loss": 0.0005,
+      "step": 7713
+    },
+    {
+      "epoch": 115.13,
+      "learning_rate": 8.210526315789474e-05,
+      "loss": 0.0004,
+      "step": 7714
+    },
+    {
+      "epoch": 115.15,
+      "learning_rate": 8.207017543859648e-05,
+      "loss": 0.0004,
+      "step": 7715
+    },
+    {
+      "epoch": 115.16,
+      "learning_rate": 8.203508771929824e-05,
+      "loss": 0.0008,
+      "step": 7716
+    },
+    {
+      "epoch": 115.18,
+      "learning_rate": 8.199999999999999e-05,
+      "loss": 0.1419,
+      "step": 7717
+    },
+    {
+      "epoch": 115.19,
+      "learning_rate": 8.196491228070175e-05,
+      "loss": 0.079,
+      "step": 7718
+    },
+    {
+      "epoch": 115.21,
+      "learning_rate": 8.19298245614035e-05,
+      "loss": 0.0003,
+      "step": 7719
+    },
+    {
+      "epoch": 115.22,
+      "learning_rate": 8.189473684210526e-05,
+      "loss": 0.0013,
+      "step": 7720
+    },
+    {
+      "epoch": 115.24,
+      "learning_rate": 8.1859649122807e-05,
+      "loss": 0.0007,
+      "step": 7721
+    },
+    {
+      "epoch": 115.25,
+      "learning_rate": 8.182456140350876e-05,
+      "loss": 0.0007,
+      "step": 7722
+    },
+    {
+      "epoch": 115.27,
+      "learning_rate": 8.178947368421052e-05,
+      "loss": 0.0003,
+      "step": 7723
+    },
+    {
+      "epoch": 115.28,
+      "learning_rate": 8.175438596491227e-05,
+      "loss": 0.0003,
+      "step": 7724
+    },
+    {
+      "epoch": 115.3,
+      "learning_rate": 8.171929824561403e-05,
+      "loss": 0.0004,
+      "step": 7725
+    },
+    {
+      "epoch": 115.31,
+      "learning_rate": 8.168421052631578e-05,
+      "loss": 0.0004,
+      "step": 7726
+    },
+    {
+      "epoch": 115.33,
+      "learning_rate": 8.164912280701754e-05,
+      "loss": 0.0043,
+      "step": 7727
+    },
+    {
+      "epoch": 115.34,
+      "learning_rate": 8.161403508771929e-05,
+      "loss": 0.0025,
+      "step": 7728
+    },
+    {
+      "epoch": 115.36,
+      "learning_rate": 8.157894736842105e-05,
+      "loss": 0.0004,
+      "step": 7729
+    },
+    {
+      "epoch": 115.37,
+      "learning_rate": 8.154385964912279e-05,
+      "loss": 0.055,
+      "step": 7730
+    },
+    {
+      "epoch": 115.39,
+      "learning_rate": 8.150877192982455e-05,
+      "loss": 0.0006,
+      "step": 7731
+    },
+    {
+      "epoch": 115.4,
+      "learning_rate": 8.14736842105263e-05,
+      "loss": 0.0004,
+      "step": 7732
+    },
+    {
+      "epoch": 115.42,
+      "learning_rate": 8.143859649122806e-05,
+      "loss": 0.0003,
+      "step": 7733
+    },
+    {
+      "epoch": 115.43,
+      "learning_rate": 8.140350877192981e-05,
+      "loss": 0.0011,
+      "step": 7734
+    },
+    {
+      "epoch": 115.45,
+      "learning_rate": 8.136842105263157e-05,
+      "loss": 0.006,
+      "step": 7735
+    },
+    {
+      "epoch": 115.46,
+      "learning_rate": 8.133333333333334e-05,
+      "loss": 0.0004,
+      "step": 7736
+    },
+    {
+      "epoch": 115.48,
+      "learning_rate": 8.129824561403508e-05,
+      "loss": 0.0711,
+      "step": 7737
+    },
+    {
+      "epoch": 115.49,
+      "learning_rate": 8.126315789473685e-05,
+      "loss": 0.0136,
+      "step": 7738
+    },
+    {
+      "epoch": 115.51,
+      "learning_rate": 8.122807017543858e-05,
+      "loss": 0.0005,
+      "step": 7739
+    },
+    {
+      "epoch": 115.52,
+      "learning_rate": 8.119298245614034e-05,
+      "loss": 0.0004,
+      "step": 7740
+    },
+    {
+      "epoch": 115.54,
+      "learning_rate": 8.11578947368421e-05,
+      "loss": 0.0004,
+      "step": 7741
+    },
+    {
+      "epoch": 115.55,
+      "learning_rate": 8.112280701754386e-05,
+      "loss": 0.0003,
+      "step": 7742
+    },
+    {
+      "epoch": 115.57,
+      "learning_rate": 8.10877192982456e-05,
+      "loss": 0.0004,
+      "step": 7743
+    },
+    {
+      "epoch": 115.58,
+      "learning_rate": 8.105263157894737e-05,
+      "loss": 0.0028,
+      "step": 7744
+    },
+    {
+      "epoch": 115.59,
+      "learning_rate": 8.10175438596491e-05,
+      "loss": 0.0022,
+      "step": 7745
+    },
+    {
+      "epoch": 115.61,
+      "learning_rate": 8.098245614035086e-05,
+      "loss": 0.0003,
+      "step": 7746
+    },
+    {
+      "epoch": 115.62,
+      "learning_rate": 8.094736842105261e-05,
+      "loss": 0.0004,
+      "step": 7747
+    },
+    {
+      "epoch": 115.64,
+      "learning_rate": 8.091228070175438e-05,
+      "loss": 0.0004,
+      "step": 7748
+    },
+    {
+      "epoch": 115.65,
+      "learning_rate": 8.087719298245614e-05,
+      "loss": 0.0003,
+      "step": 7749
+    },
+    {
+      "epoch": 115.67,
+      "learning_rate": 8.084210526315789e-05,
+      "loss": 0.0003,
+      "step": 7750
+    },
+    {
+      "epoch": 115.68,
+      "learning_rate": 8.080701754385965e-05,
+      "loss": 0.0003,
+      "step": 7751
+    },
+    {
+      "epoch": 115.7,
+      "learning_rate": 8.07719298245614e-05,
+      "loss": 0.0003,
+      "step": 7752
+    },
+    {
+      "epoch": 115.71,
+      "learning_rate": 8.073684210526316e-05,
+      "loss": 0.0003,
+      "step": 7753
+    },
+    {
+      "epoch": 115.73,
+      "learning_rate": 8.07017543859649e-05,
+      "loss": 0.0003,
+      "step": 7754
+    },
+    {
+      "epoch": 115.74,
+      "learning_rate": 8.066666666666666e-05,
+      "loss": 0.0003,
+      "step": 7755
+    },
+    {
+      "epoch": 115.76,
+      "learning_rate": 8.06315789473684e-05,
+      "loss": 0.0003,
+      "step": 7756
+    },
+    {
+      "epoch": 115.77,
+      "learning_rate": 8.059649122807017e-05,
+      "loss": 0.0017,
+      "step": 7757
+    },
+    {
+      "epoch": 115.79,
+      "learning_rate": 8.056140350877192e-05,
+      "loss": 0.001,
+      "step": 7758
+    },
+    {
+      "epoch": 115.8,
+      "learning_rate": 8.052631578947368e-05,
+      "loss": 0.0004,
+      "step": 7759
+    },
+    {
+      "epoch": 115.82,
+      "learning_rate": 8.049122807017544e-05,
+      "loss": 0.0004,
+      "step": 7760
+    },
+    {
+      "epoch": 115.83,
+      "learning_rate": 8.045614035087719e-05,
+      "loss": 0.0263,
+      "step": 7761
+    },
+    {
+      "epoch": 115.85,
+      "learning_rate": 8.042105263157895e-05,
+      "loss": 0.0004,
+      "step": 7762
+    },
+    {
+      "epoch": 115.86,
+      "learning_rate": 8.038596491228069e-05,
+      "loss": 0.0003,
+      "step": 7763
+    },
+    {
+      "epoch": 115.88,
+      "learning_rate": 8.035087719298245e-05,
+      "loss": 0.0004,
+      "step": 7764
+    },
+    {
+      "epoch": 115.89,
+      "learning_rate": 8.03157894736842e-05,
+      "loss": 0.0003,
+      "step": 7765
+    },
+    {
+      "epoch": 115.91,
+      "learning_rate": 8.028070175438596e-05,
+      "loss": 0.0015,
+      "step": 7766
+    },
+    {
+      "epoch": 115.92,
+      "learning_rate": 8.024561403508771e-05,
+      "loss": 0.0004,
+      "step": 7767
+    },
+    {
+      "epoch": 115.94,
+      "learning_rate": 8.021052631578947e-05,
+      "loss": 0.0003,
+      "step": 7768
+    },
+    {
+      "epoch": 115.95,
+      "learning_rate": 8.017543859649121e-05,
+      "loss": 0.0005,
+      "step": 7769
+    },
+    {
+      "epoch": 115.97,
+      "learning_rate": 8.014035087719297e-05,
+      "loss": 0.0004,
+      "step": 7770
+    },
+    {
+      "epoch": 115.98,
+      "learning_rate": 8.010526315789472e-05,
+      "loss": 0.0026,
+      "step": 7771
+    },
+    {
+      "epoch": 116.0,
+      "learning_rate": 8.007017543859648e-05,
+      "loss": 0.0012,
+      "step": 7772
+    },
+    {
+      "epoch": 116.01,
+      "learning_rate": 8.003508771929824e-05,
+      "loss": 0.0028,
+      "step": 7773
+    },
+    {
+      "epoch": 116.03,
+      "learning_rate": 7.999999999999999e-05,
+      "loss": 0.0005,
+      "step": 7774
+    },
+    {
+      "epoch": 116.04,
+      "learning_rate": 7.996491228070176e-05,
+      "loss": 0.0403,
+      "step": 7775
+    },
+    {
+      "epoch": 116.06,
+      "learning_rate": 7.99298245614035e-05,
+      "loss": 0.0003,
+      "step": 7776
+    },
+    {
+      "epoch": 116.07,
+      "learning_rate": 7.989473684210527e-05,
+      "loss": 0.0007,
+      "step": 7777
+    },
+    {
+      "epoch": 116.09,
+      "learning_rate": 7.9859649122807e-05,
+      "loss": 0.0003,
+      "step": 7778
+    },
+    {
+      "epoch": 116.1,
+      "learning_rate": 7.982456140350876e-05,
+      "loss": 0.0004,
+      "step": 7779
+    },
+    {
+      "epoch": 116.12,
+      "learning_rate": 7.978947368421051e-05,
+      "loss": 0.0011,
+      "step": 7780
+    },
+    {
+      "epoch": 116.13,
+      "learning_rate": 7.975438596491228e-05,
+      "loss": 0.0004,
+      "step": 7781
+    },
+    {
+      "epoch": 116.15,
+      "learning_rate": 7.971929824561402e-05,
+      "loss": 0.0003,
+      "step": 7782
+    },
+    {
+      "epoch": 116.16,
+      "learning_rate": 7.968421052631579e-05,
+      "loss": 0.0004,
+      "step": 7783
+    },
+    {
+      "epoch": 116.18,
+      "learning_rate": 7.964912280701753e-05,
+      "loss": 0.0037,
+      "step": 7784
+    },
+    {
+      "epoch": 116.19,
+      "learning_rate": 7.96140350877193e-05,
+      "loss": 0.0003,
+      "step": 7785
+    },
+    {
+      "epoch": 116.21,
+      "learning_rate": 7.957894736842106e-05,
+      "loss": 0.0003,
+      "step": 7786
+    },
+    {
+      "epoch": 116.22,
+      "learning_rate": 7.95438596491228e-05,
+      "loss": 0.0406,
+      "step": 7787
+    },
+    {
+      "epoch": 116.24,
+      "learning_rate": 7.950877192982456e-05,
+      "loss": 0.0003,
+      "step": 7788
+    },
+    {
+      "epoch": 116.25,
+      "learning_rate": 7.94736842105263e-05,
+      "loss": 0.0003,
+      "step": 7789
+    },
+    {
+      "epoch": 116.27,
+      "learning_rate": 7.943859649122807e-05,
+      "loss": 0.0103,
+      "step": 7790
+    },
+    {
+      "epoch": 116.28,
+      "learning_rate": 7.940350877192982e-05,
+      "loss": 0.0003,
+      "step": 7791
+    },
+    {
+      "epoch": 116.3,
+      "learning_rate": 7.936842105263158e-05,
+      "loss": 0.0003,
+      "step": 7792
+    },
+    {
+      "epoch": 116.31,
+      "learning_rate": 7.933333333333331e-05,
+      "loss": 0.0003,
+      "step": 7793
+    },
+    {
+      "epoch": 116.33,
+      "learning_rate": 7.929824561403508e-05,
+      "loss": 0.0004,
+      "step": 7794
+    },
+    {
+      "epoch": 116.34,
+      "learning_rate": 7.926315789473683e-05,
+      "loss": 0.0003,
+      "step": 7795
+    },
+    {
+      "epoch": 116.36,
+      "learning_rate": 7.922807017543859e-05,
+      "loss": 0.0003,
+      "step": 7796
+    },
+    {
+      "epoch": 116.37,
+      "learning_rate": 7.919298245614034e-05,
+      "loss": 0.0003,
+      "step": 7797
+    },
+    {
+      "epoch": 116.39,
+      "learning_rate": 7.91578947368421e-05,
+      "loss": 0.0027,
+      "step": 7798
+    },
+    {
+      "epoch": 116.4,
+      "learning_rate": 7.912280701754386e-05,
+      "loss": 0.0003,
+      "step": 7799
+    },
+    {
+      "epoch": 116.42,
+      "learning_rate": 7.908771929824561e-05,
+      "loss": 0.0025,
+      "step": 7800
+    },
+    {
+      "epoch": 116.42,
+      "eval_accuracy": 0.8820362212432697,
+      "eval_f1": 0.8817875307257204,
+      "eval_loss": 0.6464086174964905,
+      "eval_runtime": 346.6255,
+      "eval_samples_per_second": 11.788,
+      "eval_steps_per_second": 0.739,
+      "step": 7800
+    },
+    {
+      "epoch": 116.43,
+      "learning_rate": 7.905263157894737e-05,
+      "loss": 0.0002,
+      "step": 7801
+    },
+    {
+      "epoch": 116.45,
+      "learning_rate": 7.901754385964911e-05,
+      "loss": 0.0003,
+      "step": 7802
+    },
+    {
+      "epoch": 116.46,
+      "learning_rate": 7.898245614035087e-05,
+      "loss": 0.0003,
+      "step": 7803
+    },
+    {
+      "epoch": 116.48,
+      "learning_rate": 7.894736842105262e-05,
+      "loss": 0.0003,
+      "step": 7804
+    },
+    {
+      "epoch": 116.49,
+      "learning_rate": 7.891228070175438e-05,
+      "loss": 0.0005,
+      "step": 7805
+    },
+    {
+      "epoch": 116.51,
+      "learning_rate": 7.887719298245613e-05,
+      "loss": 0.0004,
+      "step": 7806
+    },
+    {
+      "epoch": 116.52,
+      "learning_rate": 7.884210526315789e-05,
+      "loss": 0.0003,
+      "step": 7807
+    },
+    {
+      "epoch": 116.54,
+      "learning_rate": 7.880701754385964e-05,
+      "loss": 0.0076,
+      "step": 7808
+    },
+    {
+      "epoch": 116.55,
+      "learning_rate": 7.87719298245614e-05,
+      "loss": 0.0132,
+      "step": 7809
+    },
+    {
+      "epoch": 116.57,
+      "learning_rate": 7.873684210526317e-05,
+      "loss": 0.0007,
+      "step": 7810
+    },
+    {
+      "epoch": 116.58,
+      "learning_rate": 7.87017543859649e-05,
+      "loss": 0.0003,
+      "step": 7811
+    },
+    {
+      "epoch": 116.59,
+      "learning_rate": 7.866666666666666e-05,
+      "loss": 0.0003,
+      "step": 7812
+    },
+    {
+      "epoch": 116.61,
+      "learning_rate": 7.863157894736841e-05,
+      "loss": 0.0003,
+      "step": 7813
+    },
+    {
+      "epoch": 116.62,
+      "learning_rate": 7.859649122807017e-05,
+      "loss": 0.0002,
+      "step": 7814
+    },
+    {
+      "epoch": 116.64,
+      "learning_rate": 7.856140350877192e-05,
+      "loss": 0.0003,
+      "step": 7815
+    },
+    {
+      "epoch": 116.65,
+      "learning_rate": 7.852631578947369e-05,
+      "loss": 0.0263,
+      "step": 7816
+    },
+    {
+      "epoch": 116.67,
+      "learning_rate": 7.849122807017542e-05,
+      "loss": 0.0082,
+      "step": 7817
+    },
+    {
+      "epoch": 116.68,
+      "learning_rate": 7.845614035087718e-05,
+      "loss": 0.0004,
+      "step": 7818
+    },
+    {
+      "epoch": 116.7,
+      "learning_rate": 7.842105263157893e-05,
+      "loss": 0.0004,
+      "step": 7819
+    },
+    {
+      "epoch": 116.71,
+      "learning_rate": 7.83859649122807e-05,
+      "loss": 0.0003,
+      "step": 7820
+    },
+    {
+      "epoch": 116.73,
+      "learning_rate": 7.835087719298244e-05,
+      "loss": 0.0003,
+      "step": 7821
+    },
+    {
+      "epoch": 116.74,
+      "learning_rate": 7.83157894736842e-05,
+      "loss": 0.0003,
+      "step": 7822
+    },
+    {
+      "epoch": 116.76,
+      "learning_rate": 7.828070175438597e-05,
+      "loss": 0.0003,
+      "step": 7823
+    },
+    {
+      "epoch": 116.77,
+      "learning_rate": 7.824561403508772e-05,
+      "loss": 0.0003,
+      "step": 7824
+    },
+    {
+      "epoch": 116.79,
+      "learning_rate": 7.821052631578948e-05,
+      "loss": 0.0002,
+      "step": 7825
+    },
+    {
+      "epoch": 116.8,
+      "learning_rate": 7.817543859649121e-05,
+      "loss": 0.0003,
+      "step": 7826
+    },
+    {
+      "epoch": 116.82,
+      "learning_rate": 7.814035087719298e-05,
+      "loss": 0.0002,
+      "step": 7827
+    },
+    {
+      "epoch": 116.83,
+      "learning_rate": 7.810526315789473e-05,
+      "loss": 0.0003,
+      "step": 7828
+    },
+    {
+      "epoch": 116.85,
+      "learning_rate": 7.807017543859649e-05,
+      "loss": 0.001,
+      "step": 7829
+    },
+    {
+      "epoch": 116.86,
+      "learning_rate": 7.803508771929824e-05,
+      "loss": 0.0007,
+      "step": 7830
+    },
+    {
+      "epoch": 116.88,
+      "learning_rate": 7.8e-05,
+      "loss": 0.0003,
+      "step": 7831
+    },
+    {
+      "epoch": 116.89,
+      "learning_rate": 7.796491228070175e-05,
+      "loss": 0.0004,
+      "step": 7832
+    },
+    {
+      "epoch": 116.91,
+      "learning_rate": 7.792982456140351e-05,
+      "loss": 0.0002,
+      "step": 7833
+    },
+    {
+      "epoch": 116.92,
+      "learning_rate": 7.789473684210524e-05,
+      "loss": 0.0145,
+      "step": 7834
+    },
+    {
+      "epoch": 116.94,
+      "learning_rate": 7.785964912280701e-05,
+      "loss": 0.0039,
+      "step": 7835
+    },
+    {
+      "epoch": 116.95,
+      "learning_rate": 7.782456140350877e-05,
+      "loss": 0.0012,
+      "step": 7836
+    },
+    {
+      "epoch": 116.97,
+      "learning_rate": 7.778947368421052e-05,
+      "loss": 0.0005,
+      "step": 7837
+    },
+    {
+      "epoch": 116.98,
+      "learning_rate": 7.775438596491228e-05,
+      "loss": 0.0003,
+      "step": 7838
+    },
+    {
+      "epoch": 117.0,
+      "learning_rate": 7.771929824561403e-05,
+      "loss": 0.0109,
+      "step": 7839
+    },
+    {
+      "epoch": 117.01,
+      "learning_rate": 7.768421052631579e-05,
+      "loss": 0.0004,
+      "step": 7840
+    },
+    {
+      "epoch": 117.03,
+      "learning_rate": 7.764912280701753e-05,
+      "loss": 0.0003,
+      "step": 7841
+    },
+    {
+      "epoch": 117.04,
+      "learning_rate": 7.761403508771929e-05,
+      "loss": 0.0003,
+      "step": 7842
+    },
+    {
+      "epoch": 117.06,
+      "learning_rate": 7.757894736842104e-05,
+      "loss": 0.2597,
+      "step": 7843
+    },
+    {
+      "epoch": 117.07,
+      "learning_rate": 7.75438596491228e-05,
+      "loss": 0.0002,
+      "step": 7844
+    },
+    {
+      "epoch": 117.09,
+      "learning_rate": 7.750877192982455e-05,
+      "loss": 0.0002,
+      "step": 7845
+    },
+    {
+      "epoch": 117.1,
+      "learning_rate": 7.747368421052631e-05,
+      "loss": 0.0003,
+      "step": 7846
+    },
+    {
+      "epoch": 117.12,
+      "learning_rate": 7.743859649122806e-05,
+      "loss": 0.0002,
+      "step": 7847
+    },
+    {
+      "epoch": 117.13,
+      "learning_rate": 7.740350877192982e-05,
+      "loss": 0.0002,
+      "step": 7848
+    },
+    {
+      "epoch": 117.15,
+      "learning_rate": 7.736842105263159e-05,
+      "loss": 0.0002,
+      "step": 7849
+    },
+    {
+      "epoch": 117.16,
+      "learning_rate": 7.733333333333332e-05,
+      "loss": 0.0172,
+      "step": 7850
+    },
+    {
+      "epoch": 117.18,
+      "learning_rate": 7.729824561403508e-05,
+      "loss": 0.0003,
+      "step": 7851
+    },
+    {
+      "epoch": 117.19,
+      "learning_rate": 7.726315789473683e-05,
+      "loss": 0.0009,
+      "step": 7852
+    },
+    {
+      "epoch": 117.21,
+      "learning_rate": 7.72280701754386e-05,
+      "loss": 0.0002,
+      "step": 7853
+    },
+    {
+      "epoch": 117.22,
+      "learning_rate": 7.719298245614034e-05,
+      "loss": 0.0002,
+      "step": 7854
+    },
+    {
+      "epoch": 117.24,
+      "learning_rate": 7.71578947368421e-05,
+      "loss": 0.0003,
+      "step": 7855
+    },
+    {
+      "epoch": 117.25,
+      "learning_rate": 7.712280701754385e-05,
+      "loss": 0.0021,
+      "step": 7856
+    },
+    {
+      "epoch": 117.27,
+      "learning_rate": 7.708771929824562e-05,
+      "loss": 0.0101,
+      "step": 7857
+    },
+    {
+      "epoch": 117.28,
+      "learning_rate": 7.705263157894735e-05,
+      "loss": 0.0003,
+      "step": 7858
+    },
+    {
+      "epoch": 117.3,
+      "learning_rate": 7.701754385964911e-05,
+      "loss": 0.1783,
+      "step": 7859
+    },
+    {
+      "epoch": 117.31,
+      "learning_rate": 7.698245614035086e-05,
+      "loss": 0.0002,
+      "step": 7860
+    },
+    {
+      "epoch": 117.33,
+      "learning_rate": 7.694736842105262e-05,
+      "loss": 0.1317,
+      "step": 7861
+    },
+    {
+      "epoch": 117.34,
+      "learning_rate": 7.691228070175439e-05,
+      "loss": 0.0003,
+      "step": 7862
+    },
+    {
+      "epoch": 117.36,
+      "learning_rate": 7.687719298245614e-05,
+      "loss": 0.0003,
+      "step": 7863
+    },
+    {
+      "epoch": 117.37,
+      "learning_rate": 7.68421052631579e-05,
+      "loss": 0.0002,
+      "step": 7864
+    },
+    {
+      "epoch": 117.39,
+      "learning_rate": 7.680701754385963e-05,
+      "loss": 0.0005,
+      "step": 7865
+    },
+    {
+      "epoch": 117.4,
+      "learning_rate": 7.67719298245614e-05,
+      "loss": 0.0003,
+      "step": 7866
+    },
+    {
+      "epoch": 117.42,
+      "learning_rate": 7.673684210526314e-05,
+      "loss": 0.0003,
+      "step": 7867
+    },
+    {
+      "epoch": 117.43,
+      "learning_rate": 7.67017543859649e-05,
+      "loss": 0.0005,
+      "step": 7868
+    },
+    {
+      "epoch": 117.45,
+      "learning_rate": 7.666666666666666e-05,
+      "loss": 0.0005,
+      "step": 7869
+    },
+    {
+      "epoch": 117.46,
+      "learning_rate": 7.663157894736842e-05,
+      "loss": 0.0002,
+      "step": 7870
+    },
+    {
+      "epoch": 117.48,
+      "learning_rate": 7.659649122807017e-05,
+      "loss": 0.0003,
+      "step": 7871
+    },
+    {
+      "epoch": 117.49,
+      "learning_rate": 7.656140350877193e-05,
+      "loss": 0.0002,
+      "step": 7872
+    },
+    {
+      "epoch": 117.51,
+      "learning_rate": 7.652631578947369e-05,
+      "loss": 0.0003,
+      "step": 7873
+    },
+    {
+      "epoch": 117.52,
+      "learning_rate": 7.649122807017543e-05,
+      "loss": 0.012,
+      "step": 7874
+    },
+    {
+      "epoch": 117.54,
+      "learning_rate": 7.645614035087719e-05,
+      "loss": 0.0005,
+      "step": 7875
+    },
+    {
+      "epoch": 117.55,
+      "learning_rate": 7.642105263157894e-05,
+      "loss": 0.0002,
+      "step": 7876
+    },
+    {
+      "epoch": 117.57,
+      "learning_rate": 7.63859649122807e-05,
+      "loss": 0.1101,
+      "step": 7877
+    },
+    {
+      "epoch": 117.58,
+      "learning_rate": 7.635087719298245e-05,
+      "loss": 0.048,
+      "step": 7878
+    },
+    {
+      "epoch": 117.59,
+      "learning_rate": 7.631578947368421e-05,
+      "loss": 0.0002,
+      "step": 7879
+    },
+    {
+      "epoch": 117.61,
+      "learning_rate": 7.628070175438596e-05,
+      "loss": 0.0004,
+      "step": 7880
+    },
+    {
+      "epoch": 117.62,
+      "learning_rate": 7.624561403508772e-05,
+      "loss": 0.0004,
+      "step": 7881
+    },
+    {
+      "epoch": 117.64,
+      "learning_rate": 7.621052631578946e-05,
+      "loss": 0.0018,
+      "step": 7882
+    },
+    {
+      "epoch": 117.65,
+      "learning_rate": 7.617543859649122e-05,
+      "loss": 0.0002,
+      "step": 7883
+    },
+    {
+      "epoch": 117.67,
+      "learning_rate": 7.614035087719297e-05,
+      "loss": 0.001,
+      "step": 7884
+    },
+    {
+      "epoch": 117.68,
+      "learning_rate": 7.610526315789473e-05,
+      "loss": 0.0003,
+      "step": 7885
+    },
+    {
+      "epoch": 117.7,
+      "learning_rate": 7.607017543859649e-05,
+      "loss": 0.0004,
+      "step": 7886
+    },
+    {
+      "epoch": 117.71,
+      "learning_rate": 7.603508771929824e-05,
+      "loss": 0.0002,
+      "step": 7887
+    },
+    {
+      "epoch": 117.73,
+      "learning_rate": 7.6e-05,
+      "loss": 0.0065,
+      "step": 7888
+    },
+    {
+      "epoch": 117.74,
+      "learning_rate": 7.596491228070174e-05,
+      "loss": 0.0003,
+      "step": 7889
+    },
+    {
+      "epoch": 117.76,
+      "learning_rate": 7.59298245614035e-05,
+      "loss": 0.0003,
+      "step": 7890
+    },
+    {
+      "epoch": 117.77,
+      "learning_rate": 7.589473684210525e-05,
+      "loss": 0.0003,
+      "step": 7891
+    },
+    {
+      "epoch": 117.79,
+      "learning_rate": 7.585964912280701e-05,
+      "loss": 0.0004,
+      "step": 7892
+    },
+    {
+      "epoch": 117.8,
+      "learning_rate": 7.582456140350876e-05,
+      "loss": 0.0922,
+      "step": 7893
+    },
+    {
+      "epoch": 117.82,
+      "learning_rate": 7.578947368421052e-05,
+      "loss": 0.0002,
+      "step": 7894
+    },
+    {
+      "epoch": 117.83,
+      "learning_rate": 7.575438596491227e-05,
+      "loss": 0.0011,
+      "step": 7895
+    },
+    {
+      "epoch": 117.85,
+      "learning_rate": 7.571929824561404e-05,
+      "loss": 0.0003,
+      "step": 7896
+    },
+    {
+      "epoch": 117.86,
+      "learning_rate": 7.568421052631577e-05,
+      "loss": 0.0011,
+      "step": 7897
+    },
+    {
+      "epoch": 117.88,
+      "learning_rate": 7.564912280701753e-05,
+      "loss": 0.0003,
+      "step": 7898
+    },
+    {
+      "epoch": 117.89,
+      "learning_rate": 7.56140350877193e-05,
+      "loss": 0.0002,
+      "step": 7899
+    },
+    {
+      "epoch": 117.91,
+      "learning_rate": 7.557894736842104e-05,
+      "loss": 0.1365,
+      "step": 7900
+    },
+    {
+      "epoch": 117.92,
+      "learning_rate": 7.55438596491228e-05,
+      "loss": 0.0007,
+      "step": 7901
+    },
+    {
+      "epoch": 117.94,
+      "learning_rate": 7.550877192982455e-05,
+      "loss": 0.0003,
+      "step": 7902
+    },
+    {
+      "epoch": 117.95,
+      "learning_rate": 7.547368421052632e-05,
+      "loss": 0.0003,
+      "step": 7903
+    },
+    {
+      "epoch": 117.97,
+      "learning_rate": 7.543859649122807e-05,
+      "loss": 0.0002,
+      "step": 7904
+    },
+    {
+      "epoch": 117.98,
+      "learning_rate": 7.540350877192981e-05,
+      "loss": 0.0002,
+      "step": 7905
+    },
+    {
+      "epoch": 118.0,
+      "learning_rate": 7.536842105263156e-05,
+      "loss": 0.0023,
+      "step": 7906
+    },
+    {
+      "epoch": 118.01,
+      "learning_rate": 7.533333333333333e-05,
+      "loss": 0.0019,
+      "step": 7907
+    },
+    {
+      "epoch": 118.03,
+      "learning_rate": 7.529824561403507e-05,
+      "loss": 0.0003,
+      "step": 7908
+    },
+    {
+      "epoch": 118.04,
+      "learning_rate": 7.526315789473684e-05,
+      "loss": 0.0003,
+      "step": 7909
+    },
+    {
+      "epoch": 118.06,
+      "learning_rate": 7.522807017543859e-05,
+      "loss": 0.0929,
+      "step": 7910
+    },
+    {
+      "epoch": 118.07,
+      "learning_rate": 7.519298245614035e-05,
+      "loss": 0.0003,
+      "step": 7911
+    },
+    {
+      "epoch": 118.09,
+      "learning_rate": 7.515789473684211e-05,
+      "loss": 0.0003,
+      "step": 7912
+    },
+    {
+      "epoch": 118.1,
+      "learning_rate": 7.512280701754385e-05,
+      "loss": 0.0003,
+      "step": 7913
+    },
+    {
+      "epoch": 118.12,
+      "learning_rate": 7.508771929824561e-05,
+      "loss": 0.0003,
+      "step": 7914
+    },
+    {
+      "epoch": 118.13,
+      "learning_rate": 7.505263157894736e-05,
+      "loss": 0.0005,
+      "step": 7915
+    },
+    {
+      "epoch": 118.15,
+      "learning_rate": 7.501754385964912e-05,
+      "loss": 0.0003,
+      "step": 7916
+    },
+    {
+      "epoch": 118.16,
+      "learning_rate": 7.498245614035087e-05,
+      "loss": 0.0002,
+      "step": 7917
+    },
+    {
+      "epoch": 118.18,
+      "learning_rate": 7.494736842105263e-05,
+      "loss": 0.0004,
+      "step": 7918
+    },
+    {
+      "epoch": 118.19,
+      "learning_rate": 7.491228070175438e-05,
+      "loss": 0.0237,
+      "step": 7919
+    },
+    {
+      "epoch": 118.21,
+      "learning_rate": 7.487719298245614e-05,
+      "loss": 0.0017,
+      "step": 7920
+    },
+    {
+      "epoch": 118.22,
+      "learning_rate": 7.484210526315789e-05,
+      "loss": 0.0004,
+      "step": 7921
+    },
+    {
+      "epoch": 118.24,
+      "learning_rate": 7.480701754385964e-05,
+      "loss": 0.0038,
+      "step": 7922
+    },
+    {
+      "epoch": 118.25,
+      "learning_rate": 7.47719298245614e-05,
+      "loss": 0.0003,
+      "step": 7923
+    },
+    {
+      "epoch": 118.27,
+      "learning_rate": 7.473684210526315e-05,
+      "loss": 0.2341,
+      "step": 7924
+    },
+    {
+      "epoch": 118.28,
+      "learning_rate": 7.47017543859649e-05,
+      "loss": 0.0003,
+      "step": 7925
+    },
+    {
+      "epoch": 118.3,
+      "learning_rate": 7.466666666666666e-05,
+      "loss": 0.0003,
+      "step": 7926
+    },
+    {
+      "epoch": 118.31,
+      "learning_rate": 7.463157894736841e-05,
+      "loss": 0.0008,
+      "step": 7927
+    },
+    {
+      "epoch": 118.33,
+      "learning_rate": 7.459649122807017e-05,
+      "loss": 0.0004,
+      "step": 7928
+    },
+    {
+      "epoch": 118.34,
+      "learning_rate": 7.456140350877192e-05,
+      "loss": 0.0003,
+      "step": 7929
+    },
+    {
+      "epoch": 118.36,
+      "learning_rate": 7.452631578947368e-05,
+      "loss": 0.0003,
+      "step": 7930
+    },
+    {
+      "epoch": 118.37,
+      "learning_rate": 7.449122807017543e-05,
+      "loss": 0.0003,
+      "step": 7931
+    },
+    {
+      "epoch": 118.39,
+      "learning_rate": 7.44561403508772e-05,
+      "loss": 0.0002,
+      "step": 7932
+    },
+    {
+      "epoch": 118.4,
+      "learning_rate": 7.442105263157894e-05,
+      "loss": 0.0003,
+      "step": 7933
+    },
+    {
+      "epoch": 118.42,
+      "learning_rate": 7.438596491228069e-05,
+      "loss": 0.0002,
+      "step": 7934
+    },
+    {
+      "epoch": 118.43,
+      "learning_rate": 7.435087719298245e-05,
+      "loss": 0.0367,
+      "step": 7935
+    },
+    {
+      "epoch": 118.45,
+      "learning_rate": 7.43157894736842e-05,
+      "loss": 0.0005,
+      "step": 7936
+    },
+    {
+      "epoch": 118.46,
+      "learning_rate": 7.428070175438595e-05,
+      "loss": 0.0003,
+      "step": 7937
+    },
+    {
+      "epoch": 118.48,
+      "learning_rate": 7.424561403508771e-05,
+      "loss": 0.0002,
+      "step": 7938
+    },
+    {
+      "epoch": 118.49,
+      "learning_rate": 7.421052631578946e-05,
+      "loss": 0.0002,
+      "step": 7939
+    },
+    {
+      "epoch": 118.51,
+      "learning_rate": 7.417543859649121e-05,
+      "loss": 0.0004,
+      "step": 7940
+    },
+    {
+      "epoch": 118.52,
+      "learning_rate": 7.414035087719297e-05,
+      "loss": 0.001,
+      "step": 7941
+    },
+    {
+      "epoch": 118.54,
+      "learning_rate": 7.410526315789474e-05,
+      "loss": 0.0004,
+      "step": 7942
+    },
+    {
+      "epoch": 118.55,
+      "learning_rate": 7.407017543859649e-05,
+      "loss": 0.0642,
+      "step": 7943
+    },
+    {
+      "epoch": 118.57,
+      "learning_rate": 7.403508771929825e-05,
+      "loss": 0.0002,
+      "step": 7944
+    },
+    {
+      "epoch": 118.58,
+      "learning_rate": 7.4e-05,
+      "loss": 0.0374,
+      "step": 7945
+    },
+    {
+      "epoch": 118.59,
+      "learning_rate": 7.396491228070175e-05,
+      "loss": 0.0003,
+      "step": 7946
+    },
+    {
+      "epoch": 118.61,
+      "learning_rate": 7.392982456140351e-05,
+      "loss": 0.0002,
+      "step": 7947
+    },
+    {
+      "epoch": 118.62,
+      "learning_rate": 7.389473684210526e-05,
+      "loss": 0.0003,
+      "step": 7948
+    },
+    {
+      "epoch": 118.64,
+      "learning_rate": 7.3859649122807e-05,
+      "loss": 0.0004,
+      "step": 7949
+    },
+    {
+      "epoch": 118.65,
+      "learning_rate": 7.382456140350877e-05,
+      "loss": 0.0002,
+      "step": 7950
+    },
+    {
+      "epoch": 118.67,
+      "learning_rate": 7.378947368421052e-05,
+      "loss": 0.2289,
+      "step": 7951
+    },
+    {
+      "epoch": 118.68,
+      "learning_rate": 7.375438596491226e-05,
+      "loss": 0.0002,
+      "step": 7952
+    },
+    {
+      "epoch": 118.7,
+      "learning_rate": 7.371929824561403e-05,
+      "loss": 0.0003,
+      "step": 7953
+    },
+    {
+      "epoch": 118.71,
+      "learning_rate": 7.368421052631578e-05,
+      "loss": 0.0003,
+      "step": 7954
+    },
+    {
+      "epoch": 118.73,
+      "learning_rate": 7.364912280701754e-05,
+      "loss": 0.0003,
+      "step": 7955
+    },
+    {
+      "epoch": 118.74,
+      "learning_rate": 7.36140350877193e-05,
+      "loss": 0.0003,
+      "step": 7956
+    },
+    {
+      "epoch": 118.76,
+      "learning_rate": 7.357894736842105e-05,
+      "loss": 0.0021,
+      "step": 7957
+    },
+    {
+      "epoch": 118.77,
+      "learning_rate": 7.35438596491228e-05,
+      "loss": 0.0127,
+      "step": 7958
+    },
+    {
+      "epoch": 118.79,
+      "learning_rate": 7.350877192982456e-05,
+      "loss": 0.0003,
+      "step": 7959
+    },
+    {
+      "epoch": 118.8,
+      "learning_rate": 7.347368421052631e-05,
+      "loss": 0.1613,
+      "step": 7960
+    },
+    {
+      "epoch": 118.82,
+      "learning_rate": 7.343859649122806e-05,
+      "loss": 0.0002,
+      "step": 7961
+    },
+    {
+      "epoch": 118.83,
+      "learning_rate": 7.340350877192982e-05,
+      "loss": 0.0003,
+      "step": 7962
+    },
+    {
+      "epoch": 118.85,
+      "learning_rate": 7.336842105263157e-05,
+      "loss": 0.0003,
+      "step": 7963
+    },
+    {
+      "epoch": 118.86,
+      "learning_rate": 7.333333333333332e-05,
+      "loss": 0.1024,
+      "step": 7964
+    },
+    {
+      "epoch": 118.88,
+      "learning_rate": 7.329824561403508e-05,
+      "loss": 0.0002,
+      "step": 7965
+    },
+    {
+      "epoch": 118.89,
+      "learning_rate": 7.326315789473683e-05,
+      "loss": 0.0003,
+      "step": 7966
+    },
+    {
+      "epoch": 118.91,
+      "learning_rate": 7.322807017543859e-05,
+      "loss": 0.0904,
+      "step": 7967
+    },
+    {
+      "epoch": 118.92,
+      "learning_rate": 7.319298245614035e-05,
+      "loss": 0.0003,
+      "step": 7968
+    },
+    {
+      "epoch": 118.94,
+      "learning_rate": 7.31578947368421e-05,
+      "loss": 0.0006,
+      "step": 7969
+    },
+    {
+      "epoch": 118.95,
+      "learning_rate": 7.312280701754385e-05,
+      "loss": 0.0004,
+      "step": 7970
+    },
+    {
+      "epoch": 118.97,
+      "learning_rate": 7.308771929824561e-05,
+      "loss": 0.0002,
+      "step": 7971
+    },
+    {
+      "epoch": 118.98,
+      "learning_rate": 7.305263157894736e-05,
+      "loss": 0.0002,
+      "step": 7972
+    },
+    {
+      "epoch": 119.0,
+      "learning_rate": 7.301754385964911e-05,
+      "loss": 0.057,
+      "step": 7973
+    },
+    {
+      "epoch": 119.01,
+      "learning_rate": 7.298245614035087e-05,
+      "loss": 0.0006,
+      "step": 7974
+    },
+    {
+      "epoch": 119.03,
+      "learning_rate": 7.294736842105262e-05,
+      "loss": 0.0041,
+      "step": 7975
+    },
+    {
+      "epoch": 119.04,
+      "learning_rate": 7.291228070175437e-05,
+      "loss": 0.0006,
+      "step": 7976
+    },
+    {
+      "epoch": 119.06,
+      "learning_rate": 7.287719298245613e-05,
+      "loss": 0.0002,
+      "step": 7977
+    },
+    {
+      "epoch": 119.07,
+      "learning_rate": 7.284210526315788e-05,
+      "loss": 0.2424,
+      "step": 7978
+    },
+    {
+      "epoch": 119.09,
+      "learning_rate": 7.280701754385964e-05,
+      "loss": 0.0006,
+      "step": 7979
+    },
+    {
+      "epoch": 119.1,
+      "learning_rate": 7.27719298245614e-05,
+      "loss": 0.1771,
+      "step": 7980
+    },
+    {
+      "epoch": 119.12,
+      "learning_rate": 7.273684210526316e-05,
+      "loss": 0.0004,
+      "step": 7981
+    },
+    {
+      "epoch": 119.13,
+      "learning_rate": 7.27017543859649e-05,
+      "loss": 0.0022,
+      "step": 7982
+    },
+    {
+      "epoch": 119.15,
+      "learning_rate": 7.266666666666667e-05,
+      "loss": 0.0002,
+      "step": 7983
+    },
+    {
+      "epoch": 119.16,
+      "learning_rate": 7.263157894736842e-05,
+      "loss": 0.0002,
+      "step": 7984
+    },
+    {
+      "epoch": 119.18,
+      "learning_rate": 7.259649122807016e-05,
+      "loss": 0.0003,
+      "step": 7985
+    },
+    {
+      "epoch": 119.19,
+      "learning_rate": 7.256140350877193e-05,
+      "loss": 0.0003,
+      "step": 7986
+    },
+    {
+      "epoch": 119.21,
+      "learning_rate": 7.252631578947368e-05,
+      "loss": 0.0191,
+      "step": 7987
+    },
+    {
+      "epoch": 119.22,
+      "learning_rate": 7.249122807017542e-05,
+      "loss": 0.0004,
+      "step": 7988
+    },
+    {
+      "epoch": 119.24,
+      "learning_rate": 7.245614035087719e-05,
+      "loss": 0.0003,
+      "step": 7989
+    },
+    {
+      "epoch": 119.25,
+      "learning_rate": 7.242105263157894e-05,
+      "loss": 0.0003,
+      "step": 7990
+    },
+    {
+      "epoch": 119.27,
+      "learning_rate": 7.23859649122807e-05,
+      "loss": 0.0003,
+      "step": 7991
+    },
+    {
+      "epoch": 119.28,
+      "learning_rate": 7.235087719298246e-05,
+      "loss": 0.0003,
+      "step": 7992
+    },
+    {
+      "epoch": 119.3,
+      "learning_rate": 7.231578947368421e-05,
+      "loss": 0.0003,
+      "step": 7993
+    },
+    {
+      "epoch": 119.31,
+      "learning_rate": 7.228070175438596e-05,
+      "loss": 0.0003,
+      "step": 7994
+    },
+    {
+      "epoch": 119.33,
+      "learning_rate": 7.224561403508772e-05,
+      "loss": 0.0002,
+      "step": 7995
+    },
+    {
+      "epoch": 119.34,
+      "learning_rate": 7.221052631578947e-05,
+      "loss": 0.0003,
+      "step": 7996
+    },
+    {
+      "epoch": 119.36,
+      "learning_rate": 7.217543859649122e-05,
+      "loss": 0.0003,
+      "step": 7997
+    },
+    {
+      "epoch": 119.37,
+      "learning_rate": 7.214035087719298e-05,
+      "loss": 0.0007,
+      "step": 7998
+    },
+    {
+      "epoch": 119.39,
+      "learning_rate": 7.210526315789473e-05,
+      "loss": 0.0004,
+      "step": 7999
+    },
+    {
+      "epoch": 119.4,
+      "learning_rate": 7.207017543859648e-05,
+      "loss": 0.0003,
+      "step": 8000
+    },
+    {
+      "epoch": 119.4,
+      "eval_accuracy": 0.8712677435144396,
+      "eval_f1": 0.8705504000781962,
+      "eval_loss": 0.698488175868988,
+      "eval_runtime": 343.8989,
+      "eval_samples_per_second": 11.881,
+      "eval_steps_per_second": 0.744,
+      "step": 8000
+    },
+    {
+      "epoch": 119.42,
+      "learning_rate": 7.203508771929824e-05,
+      "loss": 0.0003,
+      "step": 8001
+    },
+    {
+      "epoch": 119.43,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 0.0019,
+      "step": 8002
+    },
+    {
+      "epoch": 119.45,
+      "learning_rate": 7.196491228070175e-05,
+      "loss": 0.0002,
+      "step": 8003
+    },
+    {
+      "epoch": 119.46,
+      "learning_rate": 7.19298245614035e-05,
+      "loss": 0.0014,
+      "step": 8004
+    },
+    {
+      "epoch": 119.48,
+      "learning_rate": 7.189473684210526e-05,
+      "loss": 0.0002,
+      "step": 8005
+    },
+    {
+      "epoch": 119.49,
+      "learning_rate": 7.185964912280701e-05,
+      "loss": 0.0025,
+      "step": 8006
+    },
+    {
+      "epoch": 119.51,
+      "learning_rate": 7.182456140350877e-05,
+      "loss": 0.0003,
+      "step": 8007
+    },
+    {
+      "epoch": 119.52,
+      "learning_rate": 7.178947368421052e-05,
+      "loss": 0.1452,
+      "step": 8008
+    },
+    {
+      "epoch": 119.54,
+      "learning_rate": 7.175438596491227e-05,
+      "loss": 0.0008,
+      "step": 8009
+    },
+    {
+      "epoch": 119.55,
+      "learning_rate": 7.171929824561403e-05,
+      "loss": 0.0003,
+      "step": 8010
+    },
+    {
+      "epoch": 119.57,
+      "learning_rate": 7.168421052631578e-05,
+      "loss": 0.0041,
+      "step": 8011
+    },
+    {
+      "epoch": 119.58,
+      "learning_rate": 7.164912280701753e-05,
+      "loss": 0.0018,
+      "step": 8012
+    },
+    {
+      "epoch": 119.59,
+      "learning_rate": 7.161403508771929e-05,
+      "loss": 0.0003,
+      "step": 8013
+    },
+    {
+      "epoch": 119.61,
+      "learning_rate": 7.157894736842104e-05,
+      "loss": 0.0003,
+      "step": 8014
+    },
+    {
+      "epoch": 119.62,
+      "learning_rate": 7.15438596491228e-05,
+      "loss": 0.107,
+      "step": 8015
+    },
+    {
+      "epoch": 119.64,
+      "learning_rate": 7.150877192982455e-05,
+      "loss": 0.0005,
+      "step": 8016
+    },
+    {
+      "epoch": 119.65,
+      "learning_rate": 7.147368421052631e-05,
+      "loss": 0.0003,
+      "step": 8017
+    },
+    {
+      "epoch": 119.67,
+      "learning_rate": 7.143859649122806e-05,
+      "loss": 0.0003,
+      "step": 8018
+    },
+    {
+      "epoch": 119.68,
+      "learning_rate": 7.140350877192983e-05,
+      "loss": 0.0003,
+      "step": 8019
+    },
+    {
+      "epoch": 119.7,
+      "learning_rate": 7.136842105263157e-05,
+      "loss": 0.0003,
+      "step": 8020
+    },
+    {
+      "epoch": 119.71,
+      "learning_rate": 7.133333333333332e-05,
+      "loss": 0.0006,
+      "step": 8021
+    },
+    {
+      "epoch": 119.73,
+      "learning_rate": 7.129824561403509e-05,
+      "loss": 0.0002,
+      "step": 8022
+    },
+    {
+      "epoch": 119.74,
+      "learning_rate": 7.126315789473683e-05,
+      "loss": 0.0003,
+      "step": 8023
+    },
+    {
+      "epoch": 119.76,
+      "learning_rate": 7.122807017543858e-05,
+      "loss": 0.0003,
+      "step": 8024
+    },
+    {
+      "epoch": 119.77,
+      "learning_rate": 7.119298245614035e-05,
+      "loss": 0.0004,
+      "step": 8025
+    },
+    {
+      "epoch": 119.79,
+      "learning_rate": 7.11578947368421e-05,
+      "loss": 0.0003,
+      "step": 8026
+    },
+    {
+      "epoch": 119.8,
+      "learning_rate": 7.112280701754386e-05,
+      "loss": 0.0003,
+      "step": 8027
+    },
+    {
+      "epoch": 119.82,
+      "learning_rate": 7.10877192982456e-05,
+      "loss": 0.0029,
+      "step": 8028
+    },
+    {
+      "epoch": 119.83,
+      "learning_rate": 7.105263157894735e-05,
+      "loss": 0.0003,
+      "step": 8029
+    },
+    {
+      "epoch": 119.85,
+      "learning_rate": 7.101754385964912e-05,
+      "loss": 0.0003,
+      "step": 8030
+    },
+    {
+      "epoch": 119.86,
+      "learning_rate": 7.098245614035088e-05,
+      "loss": 0.0008,
+      "step": 8031
+    },
+    {
+      "epoch": 119.88,
+      "learning_rate": 7.094736842105263e-05,
+      "loss": 0.0009,
+      "step": 8032
+    },
+    {
+      "epoch": 119.89,
+      "learning_rate": 7.091228070175438e-05,
+      "loss": 0.0009,
+      "step": 8033
+    },
+    {
+      "epoch": 119.91,
+      "learning_rate": 7.087719298245614e-05,
+      "loss": 0.0003,
+      "step": 8034
+    },
+    {
+      "epoch": 119.92,
+      "learning_rate": 7.084210526315789e-05,
+      "loss": 0.0002,
+      "step": 8035
+    },
+    {
+      "epoch": 119.94,
+      "learning_rate": 7.080701754385964e-05,
+      "loss": 0.0003,
+      "step": 8036
+    },
+    {
+      "epoch": 119.95,
+      "learning_rate": 7.07719298245614e-05,
+      "loss": 0.0011,
+      "step": 8037
+    },
+    {
+      "epoch": 119.97,
+      "learning_rate": 7.073684210526315e-05,
+      "loss": 0.0002,
+      "step": 8038
+    },
+    {
+      "epoch": 119.98,
+      "learning_rate": 7.070175438596491e-05,
+      "loss": 0.0066,
+      "step": 8039
+    },
+    {
+      "epoch": 120.0,
+      "learning_rate": 7.066666666666666e-05,
+      "loss": 0.0071,
+      "step": 8040
+    },
+    {
+      "epoch": 120.01,
+      "learning_rate": 7.063157894736841e-05,
+      "loss": 0.0009,
+      "step": 8041
+    },
+    {
+      "epoch": 120.03,
+      "learning_rate": 7.059649122807017e-05,
+      "loss": 0.0003,
+      "step": 8042
+    },
+    {
+      "epoch": 120.04,
+      "learning_rate": 7.056140350877193e-05,
+      "loss": 0.0008,
+      "step": 8043
+    },
+    {
+      "epoch": 120.06,
+      "learning_rate": 7.052631578947368e-05,
+      "loss": 0.0004,
+      "step": 8044
+    },
+    {
+      "epoch": 120.07,
+      "learning_rate": 7.049122807017543e-05,
+      "loss": 0.0003,
+      "step": 8045
+    },
+    {
+      "epoch": 120.09,
+      "learning_rate": 7.045614035087719e-05,
+      "loss": 0.001,
+      "step": 8046
+    },
+    {
+      "epoch": 120.1,
+      "learning_rate": 7.042105263157894e-05,
+      "loss": 0.0188,
+      "step": 8047
+    },
+    {
+      "epoch": 120.12,
+      "learning_rate": 7.038596491228069e-05,
+      "loss": 0.0004,
+      "step": 8048
+    },
+    {
+      "epoch": 120.13,
+      "learning_rate": 7.035087719298245e-05,
+      "loss": 0.0003,
+      "step": 8049
+    },
+    {
+      "epoch": 120.15,
+      "learning_rate": 7.03157894736842e-05,
+      "loss": 0.0002,
+      "step": 8050
+    },
+    {
+      "epoch": 120.16,
+      "learning_rate": 7.028070175438596e-05,
+      "loss": 0.0002,
+      "step": 8051
+    },
+    {
+      "epoch": 120.18,
+      "learning_rate": 7.024561403508771e-05,
+      "loss": 0.0003,
+      "step": 8052
+    },
+    {
+      "epoch": 120.19,
+      "learning_rate": 7.021052631578946e-05,
+      "loss": 0.0003,
+      "step": 8053
+    },
+    {
+      "epoch": 120.21,
+      "learning_rate": 7.017543859649122e-05,
+      "loss": 0.0003,
+      "step": 8054
+    },
+    {
+      "epoch": 120.22,
+      "learning_rate": 7.014035087719299e-05,
+      "loss": 0.0003,
+      "step": 8055
+    },
+    {
+      "epoch": 120.24,
+      "learning_rate": 7.010526315789473e-05,
+      "loss": 0.2835,
+      "step": 8056
+    },
+    {
+      "epoch": 120.25,
+      "learning_rate": 7.007017543859648e-05,
+      "loss": 0.0002,
+      "step": 8057
+    },
+    {
+      "epoch": 120.27,
+      "learning_rate": 7.003508771929825e-05,
+      "loss": 0.0003,
+      "step": 8058
+    },
+    {
+      "epoch": 120.28,
+      "learning_rate": 7e-05,
+      "loss": 0.0004,
+      "step": 8059
+    },
+    {
+      "epoch": 120.3,
+      "learning_rate": 6.996491228070174e-05,
+      "loss": 0.0002,
+      "step": 8060
+    },
+    {
+      "epoch": 120.31,
+      "learning_rate": 6.99298245614035e-05,
+      "loss": 0.0003,
+      "step": 8061
+    },
+    {
+      "epoch": 120.33,
+      "learning_rate": 6.989473684210525e-05,
+      "loss": 0.0003,
+      "step": 8062
+    },
+    {
+      "epoch": 120.34,
+      "learning_rate": 6.985964912280702e-05,
+      "loss": 0.0009,
+      "step": 8063
+    },
+    {
+      "epoch": 120.36,
+      "learning_rate": 6.982456140350876e-05,
+      "loss": 0.0044,
+      "step": 8064
+    },
+    {
+      "epoch": 120.37,
+      "learning_rate": 6.978947368421051e-05,
+      "loss": 0.0004,
+      "step": 8065
+    },
+    {
+      "epoch": 120.39,
+      "learning_rate": 6.975438596491228e-05,
+      "loss": 0.0003,
+      "step": 8066
+    },
+    {
+      "epoch": 120.4,
+      "learning_rate": 6.971929824561404e-05,
+      "loss": 0.0003,
+      "step": 8067
+    },
+    {
+      "epoch": 120.42,
+      "learning_rate": 6.968421052631579e-05,
+      "loss": 0.0003,
+      "step": 8068
+    },
+    {
+      "epoch": 120.43,
+      "learning_rate": 6.964912280701754e-05,
+      "loss": 0.0015,
+      "step": 8069
+    },
+    {
+      "epoch": 120.45,
+      "learning_rate": 6.96140350877193e-05,
+      "loss": 0.0004,
+      "step": 8070
+    },
+    {
+      "epoch": 120.46,
+      "learning_rate": 6.957894736842105e-05,
+      "loss": 0.0004,
+      "step": 8071
+    },
+    {
+      "epoch": 120.48,
+      "learning_rate": 6.95438596491228e-05,
+      "loss": 0.0004,
+      "step": 8072
+    },
+    {
+      "epoch": 120.49,
+      "learning_rate": 6.950877192982456e-05,
+      "loss": 0.0011,
+      "step": 8073
+    },
+    {
+      "epoch": 120.51,
+      "learning_rate": 6.947368421052631e-05,
+      "loss": 0.0021,
+      "step": 8074
+    },
+    {
+      "epoch": 120.52,
+      "learning_rate": 6.943859649122807e-05,
+      "loss": 0.0526,
+      "step": 8075
+    },
+    {
+      "epoch": 120.54,
+      "learning_rate": 6.940350877192982e-05,
+      "loss": 0.006,
+      "step": 8076
+    },
+    {
+      "epoch": 120.55,
+      "learning_rate": 6.936842105263157e-05,
+      "loss": 0.0005,
+      "step": 8077
+    },
+    {
+      "epoch": 120.57,
+      "learning_rate": 6.933333333333333e-05,
+      "loss": 0.0004,
+      "step": 8078
+    },
+    {
+      "epoch": 120.58,
+      "learning_rate": 6.929824561403508e-05,
+      "loss": 0.0048,
+      "step": 8079
+    },
+    {
+      "epoch": 120.59,
+      "learning_rate": 6.926315789473684e-05,
+      "loss": 0.0005,
+      "step": 8080
+    },
+    {
+      "epoch": 120.61,
+      "learning_rate": 6.922807017543859e-05,
+      "loss": 0.0004,
+      "step": 8081
+    },
+    {
+      "epoch": 120.62,
+      "learning_rate": 6.919298245614035e-05,
+      "loss": 0.0125,
+      "step": 8082
+    },
+    {
+      "epoch": 120.64,
+      "learning_rate": 6.91578947368421e-05,
+      "loss": 0.0353,
+      "step": 8083
+    },
+    {
+      "epoch": 120.65,
+      "learning_rate": 6.912280701754385e-05,
+      "loss": 0.015,
+      "step": 8084
+    },
+    {
+      "epoch": 120.67,
+      "learning_rate": 6.908771929824561e-05,
+      "loss": 0.0004,
+      "step": 8085
+    },
+    {
+      "epoch": 120.68,
+      "learning_rate": 6.905263157894736e-05,
+      "loss": 0.0691,
+      "step": 8086
+    },
+    {
+      "epoch": 120.7,
+      "learning_rate": 6.901754385964912e-05,
+      "loss": 0.0005,
+      "step": 8087
+    },
+    {
+      "epoch": 120.71,
+      "learning_rate": 6.898245614035087e-05,
+      "loss": 0.0014,
+      "step": 8088
+    },
+    {
+      "epoch": 120.73,
+      "learning_rate": 6.894736842105262e-05,
+      "loss": 0.3338,
+      "step": 8089
+    },
+    {
+      "epoch": 120.74,
+      "learning_rate": 6.891228070175438e-05,
+      "loss": 0.0003,
+      "step": 8090
+    },
+    {
+      "epoch": 120.76,
+      "learning_rate": 6.887719298245613e-05,
+      "loss": 0.0003,
+      "step": 8091
+    },
+    {
+      "epoch": 120.77,
+      "learning_rate": 6.884210526315788e-05,
+      "loss": 0.0003,
+      "step": 8092
+    },
+    {
+      "epoch": 120.79,
+      "learning_rate": 6.880701754385964e-05,
+      "loss": 0.0298,
+      "step": 8093
+    },
+    {
+      "epoch": 120.8,
+      "learning_rate": 6.87719298245614e-05,
+      "loss": 0.001,
+      "step": 8094
+    },
+    {
+      "epoch": 120.82,
+      "learning_rate": 6.873684210526315e-05,
+      "loss": 0.0004,
+      "step": 8095
+    },
+    {
+      "epoch": 120.83,
+      "learning_rate": 6.87017543859649e-05,
+      "loss": 0.0004,
+      "step": 8096
+    },
+    {
+      "epoch": 120.85,
+      "learning_rate": 6.866666666666666e-05,
+      "loss": 0.0509,
+      "step": 8097
+    },
+    {
+      "epoch": 120.86,
+      "learning_rate": 6.863157894736841e-05,
+      "loss": 0.1401,
+      "step": 8098
+    },
+    {
+      "epoch": 120.88,
+      "learning_rate": 6.859649122807018e-05,
+      "loss": 0.0007,
+      "step": 8099
+    },
+    {
+      "epoch": 120.89,
+      "learning_rate": 6.856140350877192e-05,
+      "loss": 0.0005,
+      "step": 8100
+    },
+    {
+      "epoch": 120.91,
+      "learning_rate": 6.852631578947367e-05,
+      "loss": 0.0005,
+      "step": 8101
+    },
+    {
+      "epoch": 120.92,
+      "learning_rate": 6.849122807017544e-05,
+      "loss": 0.001,
+      "step": 8102
+    },
+    {
+      "epoch": 120.94,
+      "learning_rate": 6.845614035087718e-05,
+      "loss": 0.0003,
+      "step": 8103
+    },
+    {
+      "epoch": 120.95,
+      "learning_rate": 6.842105263157893e-05,
+      "loss": 0.174,
+      "step": 8104
+    },
+    {
+      "epoch": 120.97,
+      "learning_rate": 6.83859649122807e-05,
+      "loss": 0.001,
+      "step": 8105
+    },
+    {
+      "epoch": 120.98,
+      "learning_rate": 6.835087719298246e-05,
+      "loss": 0.0006,
+      "step": 8106
+    },
+    {
+      "epoch": 121.0,
+      "learning_rate": 6.83157894736842e-05,
+      "loss": 0.0003,
+      "step": 8107
+    },
+    {
+      "epoch": 121.01,
+      "learning_rate": 6.828070175438596e-05,
+      "loss": 0.0004,
+      "step": 8108
+    },
+    {
+      "epoch": 121.03,
+      "learning_rate": 6.824561403508772e-05,
+      "loss": 0.0005,
+      "step": 8109
+    },
+    {
+      "epoch": 121.04,
+      "learning_rate": 6.821052631578947e-05,
+      "loss": 0.0006,
+      "step": 8110
+    },
+    {
+      "epoch": 121.06,
+      "learning_rate": 6.817543859649123e-05,
+      "loss": 0.0008,
+      "step": 8111
+    },
+    {
+      "epoch": 121.07,
+      "learning_rate": 6.814035087719298e-05,
+      "loss": 0.0035,
+      "step": 8112
+    },
+    {
+      "epoch": 121.09,
+      "learning_rate": 6.810526315789473e-05,
+      "loss": 0.0006,
+      "step": 8113
+    },
+    {
+      "epoch": 121.1,
+      "learning_rate": 6.807017543859649e-05,
+      "loss": 0.0008,
+      "step": 8114
+    },
+    {
+      "epoch": 121.12,
+      "learning_rate": 6.803508771929824e-05,
+      "loss": 0.001,
+      "step": 8115
+    },
+    {
+      "epoch": 121.13,
+      "learning_rate": 6.799999999999999e-05,
+      "loss": 0.0007,
+      "step": 8116
+    },
+    {
+      "epoch": 121.15,
+      "learning_rate": 6.796491228070175e-05,
+      "loss": 0.0006,
+      "step": 8117
+    },
+    {
+      "epoch": 121.16,
+      "learning_rate": 6.792982456140351e-05,
+      "loss": 0.0007,
+      "step": 8118
+    },
+    {
+      "epoch": 121.18,
+      "learning_rate": 6.789473684210526e-05,
+      "loss": 0.0272,
+      "step": 8119
+    },
+    {
+      "epoch": 121.19,
+      "learning_rate": 6.785964912280701e-05,
+      "loss": 0.0264,
+      "step": 8120
+    },
+    {
+      "epoch": 121.21,
+      "learning_rate": 6.782456140350877e-05,
+      "loss": 0.0007,
+      "step": 8121
+    },
+    {
+      "epoch": 121.22,
+      "learning_rate": 6.778947368421052e-05,
+      "loss": 0.0007,
+      "step": 8122
+    },
+    {
+      "epoch": 121.24,
+      "learning_rate": 6.775438596491228e-05,
+      "loss": 0.0009,
+      "step": 8123
+    },
+    {
+      "epoch": 121.25,
+      "learning_rate": 6.771929824561403e-05,
+      "loss": 0.0006,
+      "step": 8124
+    },
+    {
+      "epoch": 121.27,
+      "learning_rate": 6.768421052631578e-05,
+      "loss": 0.0005,
+      "step": 8125
+    },
+    {
+      "epoch": 121.28,
+      "learning_rate": 6.764912280701754e-05,
+      "loss": 0.0193,
+      "step": 8126
+    },
+    {
+      "epoch": 121.3,
+      "learning_rate": 6.761403508771929e-05,
+      "loss": 0.0092,
+      "step": 8127
+    },
+    {
+      "epoch": 121.31,
+      "learning_rate": 6.757894736842104e-05,
+      "loss": 0.0099,
+      "step": 8128
+    },
+    {
+      "epoch": 121.33,
+      "learning_rate": 6.75438596491228e-05,
+      "loss": 0.0004,
+      "step": 8129
+    },
+    {
+      "epoch": 121.34,
+      "learning_rate": 6.750877192982456e-05,
+      "loss": 0.0005,
+      "step": 8130
+    },
+    {
+      "epoch": 121.36,
+      "learning_rate": 6.747368421052631e-05,
+      "loss": 0.0005,
+      "step": 8131
+    },
+    {
+      "epoch": 121.37,
+      "learning_rate": 6.743859649122806e-05,
+      "loss": 0.0004,
+      "step": 8132
+    },
+    {
+      "epoch": 121.39,
+      "learning_rate": 6.740350877192982e-05,
+      "loss": 0.001,
+      "step": 8133
+    },
+    {
+      "epoch": 121.4,
+      "learning_rate": 6.736842105263157e-05,
+      "loss": 0.0005,
+      "step": 8134
+    },
+    {
+      "epoch": 121.42,
+      "learning_rate": 6.733333333333333e-05,
+      "loss": 0.0004,
+      "step": 8135
+    },
+    {
+      "epoch": 121.43,
+      "learning_rate": 6.729824561403508e-05,
+      "loss": 0.0005,
+      "step": 8136
+    },
+    {
+      "epoch": 121.45,
+      "learning_rate": 6.726315789473683e-05,
+      "loss": 0.4882,
+      "step": 8137
+    },
+    {
+      "epoch": 121.46,
+      "learning_rate": 6.72280701754386e-05,
+      "loss": 0.0033,
+      "step": 8138
+    },
+    {
+      "epoch": 121.48,
+      "learning_rate": 6.719298245614034e-05,
+      "loss": 0.0003,
+      "step": 8139
+    },
+    {
+      "epoch": 121.49,
+      "learning_rate": 6.715789473684209e-05,
+      "loss": 0.0005,
+      "step": 8140
+    },
+    {
+      "epoch": 121.51,
+      "learning_rate": 6.712280701754385e-05,
+      "loss": 0.0008,
+      "step": 8141
+    },
+    {
+      "epoch": 121.52,
+      "learning_rate": 6.70877192982456e-05,
+      "loss": 0.0311,
+      "step": 8142
+    },
+    {
+      "epoch": 121.54,
+      "learning_rate": 6.705263157894737e-05,
+      "loss": 0.0003,
+      "step": 8143
+    },
+    {
+      "epoch": 121.55,
+      "learning_rate": 6.701754385964911e-05,
+      "loss": 0.0004,
+      "step": 8144
+    },
+    {
+      "epoch": 121.57,
+      "learning_rate": 6.698245614035088e-05,
+      "loss": 0.0012,
+      "step": 8145
+    },
+    {
+      "epoch": 121.58,
+      "learning_rate": 6.694736842105263e-05,
+      "loss": 0.0004,
+      "step": 8146
+    },
+    {
+      "epoch": 121.59,
+      "learning_rate": 6.691228070175439e-05,
+      "loss": 0.0003,
+      "step": 8147
+    },
+    {
+      "epoch": 121.61,
+      "learning_rate": 6.687719298245614e-05,
+      "loss": 0.0031,
+      "step": 8148
+    },
+    {
+      "epoch": 121.62,
+      "learning_rate": 6.684210526315789e-05,
+      "loss": 0.0031,
+      "step": 8149
+    },
+    {
+      "epoch": 121.64,
+      "learning_rate": 6.680701754385965e-05,
+      "loss": 0.0003,
+      "step": 8150
+    },
+    {
+      "epoch": 121.65,
+      "learning_rate": 6.67719298245614e-05,
+      "loss": 0.0003,
+      "step": 8151
+    },
+    {
+      "epoch": 121.67,
+      "learning_rate": 6.673684210526315e-05,
+      "loss": 0.0006,
+      "step": 8152
+    },
+    {
+      "epoch": 121.68,
+      "learning_rate": 6.670175438596491e-05,
+      "loss": 0.0005,
+      "step": 8153
+    },
+    {
+      "epoch": 121.7,
+      "learning_rate": 6.666666666666666e-05,
+      "loss": 0.0005,
+      "step": 8154
+    },
+    {
+      "epoch": 121.71,
+      "learning_rate": 6.663157894736842e-05,
+      "loss": 0.0006,
+      "step": 8155
+    },
+    {
+      "epoch": 121.73,
+      "learning_rate": 6.659649122807017e-05,
+      "loss": 0.0005,
+      "step": 8156
+    },
+    {
+      "epoch": 121.74,
+      "learning_rate": 6.656140350877193e-05,
+      "loss": 0.0004,
+      "step": 8157
+    },
+    {
+      "epoch": 121.76,
+      "learning_rate": 6.652631578947368e-05,
+      "loss": 0.0384,
+      "step": 8158
+    },
+    {
+      "epoch": 121.77,
+      "learning_rate": 6.649122807017543e-05,
+      "loss": 0.0011,
+      "step": 8159
+    },
+    {
+      "epoch": 121.79,
+      "learning_rate": 6.645614035087719e-05,
+      "loss": 0.0004,
+      "step": 8160
+    },
+    {
+      "epoch": 121.8,
+      "learning_rate": 6.642105263157894e-05,
+      "loss": 0.0032,
+      "step": 8161
+    },
+    {
+      "epoch": 121.82,
+      "learning_rate": 6.63859649122807e-05,
+      "loss": 0.0018,
+      "step": 8162
+    },
+    {
+      "epoch": 121.83,
+      "learning_rate": 6.635087719298245e-05,
+      "loss": 0.0003,
+      "step": 8163
+    },
+    {
+      "epoch": 121.85,
+      "learning_rate": 6.63157894736842e-05,
+      "loss": 0.0392,
+      "step": 8164
+    },
+    {
+      "epoch": 121.86,
+      "learning_rate": 6.628070175438596e-05,
+      "loss": 0.0005,
+      "step": 8165
+    },
+    {
+      "epoch": 121.88,
+      "learning_rate": 6.624561403508771e-05,
+      "loss": 0.0014,
+      "step": 8166
+    },
+    {
+      "epoch": 121.89,
+      "learning_rate": 6.621052631578946e-05,
+      "loss": 0.0004,
+      "step": 8167
+    },
+    {
+      "epoch": 121.91,
+      "learning_rate": 6.617543859649122e-05,
+      "loss": 0.0008,
+      "step": 8168
+    },
+    {
+      "epoch": 121.92,
+      "learning_rate": 6.614035087719298e-05,
+      "loss": 0.0003,
+      "step": 8169
+    },
+    {
+      "epoch": 121.94,
+      "learning_rate": 6.610526315789473e-05,
+      "loss": 0.001,
+      "step": 8170
+    },
+    {
+      "epoch": 121.95,
+      "learning_rate": 6.607017543859648e-05,
+      "loss": 0.0003,
+      "step": 8171
+    },
+    {
+      "epoch": 121.97,
+      "learning_rate": 6.603508771929824e-05,
+      "loss": 0.0003,
+      "step": 8172
+    },
+    {
+      "epoch": 121.98,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 0.0005,
+      "step": 8173
+    },
+    {
+      "epoch": 122.0,
+      "learning_rate": 6.596491228070175e-05,
+      "loss": 0.0004,
+      "step": 8174
+    },
+    {
+      "epoch": 122.01,
+      "learning_rate": 6.59298245614035e-05,
+      "loss": 0.0211,
+      "step": 8175
+    },
+    {
+      "epoch": 122.03,
+      "learning_rate": 6.589473684210525e-05,
+      "loss": 0.0003,
+      "step": 8176
+    },
+    {
+      "epoch": 122.04,
+      "learning_rate": 6.585964912280701e-05,
+      "loss": 0.002,
+      "step": 8177
+    },
+    {
+      "epoch": 122.06,
+      "learning_rate": 6.582456140350876e-05,
+      "loss": 0.0005,
+      "step": 8178
+    },
+    {
+      "epoch": 122.07,
+      "learning_rate": 6.578947368421051e-05,
+      "loss": 0.0003,
+      "step": 8179
+    },
+    {
+      "epoch": 122.09,
+      "learning_rate": 6.575438596491227e-05,
+      "loss": 0.0008,
+      "step": 8180
+    },
+    {
+      "epoch": 122.1,
+      "learning_rate": 6.571929824561404e-05,
+      "loss": 0.0044,
+      "step": 8181
+    },
+    {
+      "epoch": 122.12,
+      "learning_rate": 6.568421052631578e-05,
+      "loss": 0.0005,
+      "step": 8182
+    },
+    {
+      "epoch": 122.13,
+      "learning_rate": 6.564912280701753e-05,
+      "loss": 0.0003,
+      "step": 8183
+    },
+    {
+      "epoch": 122.15,
+      "learning_rate": 6.56140350877193e-05,
+      "loss": 0.0019,
+      "step": 8184
+    },
+    {
+      "epoch": 122.16,
+      "learning_rate": 6.557894736842104e-05,
+      "loss": 0.0017,
+      "step": 8185
+    },
+    {
+      "epoch": 122.18,
+      "learning_rate": 6.554385964912281e-05,
+      "loss": 0.0003,
+      "step": 8186
+    },
+    {
+      "epoch": 122.19,
+      "learning_rate": 6.550877192982456e-05,
+      "loss": 0.0005,
+      "step": 8187
+    },
+    {
+      "epoch": 122.21,
+      "learning_rate": 6.54736842105263e-05,
+      "loss": 0.0005,
+      "step": 8188
+    },
+    {
+      "epoch": 122.22,
+      "learning_rate": 6.543859649122807e-05,
+      "loss": 0.0096,
+      "step": 8189
+    },
+    {
+      "epoch": 122.24,
+      "learning_rate": 6.540350877192982e-05,
+      "loss": 0.0004,
+      "step": 8190
+    },
+    {
+      "epoch": 122.25,
+      "learning_rate": 6.536842105263156e-05,
+      "loss": 0.0006,
+      "step": 8191
+    },
+    {
+      "epoch": 122.27,
+      "learning_rate": 6.533333333333333e-05,
+      "loss": 0.0184,
+      "step": 8192
+    },
+    {
+      "epoch": 122.28,
+      "learning_rate": 6.529824561403509e-05,
+      "loss": 0.0003,
+      "step": 8193
+    },
+    {
+      "epoch": 122.3,
+      "learning_rate": 6.526315789473684e-05,
+      "loss": 0.0005,
+      "step": 8194
+    },
+    {
+      "epoch": 122.31,
+      "learning_rate": 6.522807017543859e-05,
+      "loss": 0.0003,
+      "step": 8195
+    },
+    {
+      "epoch": 122.33,
+      "learning_rate": 6.519298245614035e-05,
+      "loss": 0.0007,
+      "step": 8196
+    },
+    {
+      "epoch": 122.34,
+      "learning_rate": 6.51578947368421e-05,
+      "loss": 0.004,
+      "step": 8197
+    },
+    {
+      "epoch": 122.36,
+      "learning_rate": 6.512280701754386e-05,
+      "loss": 0.0003,
+      "step": 8198
+    },
+    {
+      "epoch": 122.37,
+      "learning_rate": 6.508771929824561e-05,
+      "loss": 0.0003,
+      "step": 8199
+    },
+    {
+      "epoch": 122.39,
+      "learning_rate": 6.505263157894736e-05,
+      "loss": 0.0048,
+      "step": 8200
+    },
+    {
+      "epoch": 122.39,
+      "eval_accuracy": 0.8739598629466471,
+      "eval_f1": 0.8764769996728009,
+      "eval_loss": 0.6620244979858398,
+      "eval_runtime": 344.6899,
+      "eval_samples_per_second": 11.854,
+      "eval_steps_per_second": 0.743,
+      "step": 8200
+    },
+    {
+      "epoch": 122.4,
+      "learning_rate": 6.501754385964912e-05,
+      "loss": 0.001,
+      "step": 8201
+    },
+    {
+      "epoch": 122.42,
+      "learning_rate": 6.498245614035087e-05,
+      "loss": 0.0004,
+      "step": 8202
+    },
+    {
+      "epoch": 122.43,
+      "learning_rate": 6.494736842105262e-05,
+      "loss": 0.0022,
+      "step": 8203
+    },
+    {
+      "epoch": 122.45,
+      "learning_rate": 6.491228070175438e-05,
+      "loss": 0.0004,
+      "step": 8204
+    },
+    {
+      "epoch": 122.46,
+      "learning_rate": 6.487719298245614e-05,
+      "loss": 0.0034,
+      "step": 8205
+    },
+    {
+      "epoch": 122.48,
+      "learning_rate": 6.484210526315789e-05,
+      "loss": 0.0013,
+      "step": 8206
+    },
+    {
+      "epoch": 122.49,
+      "learning_rate": 6.480701754385964e-05,
+      "loss": 0.0004,
+      "step": 8207
+    },
+    {
+      "epoch": 122.51,
+      "learning_rate": 6.47719298245614e-05,
+      "loss": 0.0004,
+      "step": 8208
+    },
+    {
+      "epoch": 122.52,
+      "learning_rate": 6.473684210526315e-05,
+      "loss": 0.0003,
+      "step": 8209
+    },
+    {
+      "epoch": 122.54,
+      "learning_rate": 6.470175438596491e-05,
+      "loss": 0.0003,
+      "step": 8210
+    },
+    {
+      "epoch": 122.55,
+      "learning_rate": 6.466666666666666e-05,
+      "loss": 0.0003,
+      "step": 8211
+    },
+    {
+      "epoch": 122.57,
+      "learning_rate": 6.463157894736841e-05,
+      "loss": 0.0003,
+      "step": 8212
+    },
+    {
+      "epoch": 122.58,
+      "learning_rate": 6.459649122807017e-05,
+      "loss": 0.1371,
+      "step": 8213
+    },
+    {
+      "epoch": 122.59,
+      "learning_rate": 6.456140350877192e-05,
+      "loss": 0.0026,
+      "step": 8214
+    },
+    {
+      "epoch": 122.61,
+      "learning_rate": 6.452631578947367e-05,
+      "loss": 0.0022,
+      "step": 8215
+    },
+    {
+      "epoch": 122.62,
+      "learning_rate": 6.449122807017543e-05,
+      "loss": 0.031,
+      "step": 8216
+    },
+    {
+      "epoch": 122.64,
+      "learning_rate": 6.445614035087718e-05,
+      "loss": 0.0003,
+      "step": 8217
+    },
+    {
+      "epoch": 122.65,
+      "learning_rate": 6.442105263157894e-05,
+      "loss": 0.0004,
+      "step": 8218
+    },
+    {
+      "epoch": 122.67,
+      "learning_rate": 6.438596491228069e-05,
+      "loss": 0.177,
+      "step": 8219
+    },
+    {
+      "epoch": 122.68,
+      "learning_rate": 6.435087719298246e-05,
+      "loss": 0.0004,
+      "step": 8220
+    },
+    {
+      "epoch": 122.7,
+      "learning_rate": 6.43157894736842e-05,
+      "loss": 0.0004,
+      "step": 8221
+    },
+    {
+      "epoch": 122.71,
+      "learning_rate": 6.428070175438597e-05,
+      "loss": 0.0004,
+      "step": 8222
+    },
+    {
+      "epoch": 122.73,
+      "learning_rate": 6.424561403508772e-05,
+      "loss": 0.0004,
+      "step": 8223
+    },
+    {
+      "epoch": 122.74,
+      "learning_rate": 6.421052631578946e-05,
+      "loss": 0.0002,
+      "step": 8224
+    },
+    {
+      "epoch": 122.76,
+      "learning_rate": 6.417543859649123e-05,
+      "loss": 0.0923,
+      "step": 8225
+    },
+    {
+      "epoch": 122.77,
+      "learning_rate": 6.414035087719297e-05,
+      "loss": 0.0009,
+      "step": 8226
+    },
+    {
+      "epoch": 122.79,
+      "learning_rate": 6.410526315789472e-05,
+      "loss": 0.0002,
+      "step": 8227
+    },
+    {
+      "epoch": 122.8,
+      "learning_rate": 6.407017543859649e-05,
+      "loss": 0.0006,
+      "step": 8228
+    },
+    {
+      "epoch": 122.82,
+      "learning_rate": 6.403508771929823e-05,
+      "loss": 0.0004,
+      "step": 8229
+    },
+    {
+      "epoch": 122.83,
+      "learning_rate": 6.4e-05,
+      "loss": 0.0003,
+      "step": 8230
+    },
+    {
+      "epoch": 122.85,
+      "learning_rate": 6.396491228070175e-05,
+      "loss": 0.0004,
+      "step": 8231
+    },
+    {
+      "epoch": 122.86,
+      "learning_rate": 6.392982456140351e-05,
+      "loss": 0.0664,
+      "step": 8232
+    },
+    {
+      "epoch": 122.88,
+      "learning_rate": 6.389473684210526e-05,
+      "loss": 0.0004,
+      "step": 8233
+    },
+    {
+      "epoch": 122.89,
+      "learning_rate": 6.385964912280702e-05,
+      "loss": 0.0002,
+      "step": 8234
+    },
+    {
+      "epoch": 122.91,
+      "learning_rate": 6.382456140350877e-05,
+      "loss": 0.0003,
+      "step": 8235
+    },
+    {
+      "epoch": 122.92,
+      "learning_rate": 6.378947368421052e-05,
+      "loss": 0.0003,
+      "step": 8236
+    },
+    {
+      "epoch": 122.94,
+      "learning_rate": 6.375438596491228e-05,
+      "loss": 0.0005,
+      "step": 8237
+    },
+    {
+      "epoch": 122.95,
+      "learning_rate": 6.371929824561403e-05,
+      "loss": 0.0081,
+      "step": 8238
+    },
+    {
+      "epoch": 122.97,
+      "learning_rate": 6.368421052631578e-05,
+      "loss": 0.0003,
+      "step": 8239
+    },
+    {
+      "epoch": 122.98,
+      "learning_rate": 6.364912280701754e-05,
+      "loss": 0.0004,
+      "step": 8240
+    },
+    {
+      "epoch": 123.0,
+      "learning_rate": 6.361403508771929e-05,
+      "loss": 0.0003,
+      "step": 8241
+    },
+    {
+      "epoch": 123.01,
+      "learning_rate": 6.357894736842104e-05,
+      "loss": 0.0003,
+      "step": 8242
+    },
+    {
+      "epoch": 123.03,
+      "learning_rate": 6.35438596491228e-05,
+      "loss": 0.0003,
+      "step": 8243
+    },
+    {
+      "epoch": 123.04,
+      "learning_rate": 6.350877192982456e-05,
+      "loss": 0.0072,
+      "step": 8244
+    },
+    {
+      "epoch": 123.06,
+      "learning_rate": 6.347368421052631e-05,
+      "loss": 0.0005,
+      "step": 8245
+    },
+    {
+      "epoch": 123.07,
+      "learning_rate": 6.343859649122807e-05,
+      "loss": 0.0003,
+      "step": 8246
+    },
+    {
+      "epoch": 123.09,
+      "learning_rate": 6.340350877192982e-05,
+      "loss": 0.0003,
+      "step": 8247
+    },
+    {
+      "epoch": 123.1,
+      "learning_rate": 6.336842105263157e-05,
+      "loss": 0.0015,
+      "step": 8248
+    },
+    {
+      "epoch": 123.12,
+      "learning_rate": 6.333333333333333e-05,
+      "loss": 0.0003,
+      "step": 8249
+    },
+    {
+      "epoch": 123.13,
+      "learning_rate": 6.329824561403508e-05,
+      "loss": 0.0004,
+      "step": 8250
+    },
+    {
+      "epoch": 123.15,
+      "learning_rate": 6.326315789473683e-05,
+      "loss": 0.0003,
+      "step": 8251
+    },
+    {
+      "epoch": 123.16,
+      "learning_rate": 6.322807017543859e-05,
+      "loss": 0.1065,
+      "step": 8252
+    },
+    {
+      "epoch": 123.18,
+      "learning_rate": 6.319298245614034e-05,
+      "loss": 0.0002,
+      "step": 8253
+    },
+    {
+      "epoch": 123.19,
+      "learning_rate": 6.315789473684209e-05,
+      "loss": 0.0003,
+      "step": 8254
+    },
+    {
+      "epoch": 123.21,
+      "learning_rate": 6.312280701754385e-05,
+      "loss": 0.0321,
+      "step": 8255
+    },
+    {
+      "epoch": 123.22,
+      "learning_rate": 6.308771929824561e-05,
+      "loss": 0.0003,
+      "step": 8256
+    },
+    {
+      "epoch": 123.24,
+      "learning_rate": 6.305263157894736e-05,
+      "loss": 0.0003,
+      "step": 8257
+    },
+    {
+      "epoch": 123.25,
+      "learning_rate": 6.301754385964913e-05,
+      "loss": 0.0003,
+      "step": 8258
+    },
+    {
+      "epoch": 123.27,
+      "learning_rate": 6.298245614035087e-05,
+      "loss": 0.0003,
+      "step": 8259
+    },
+    {
+      "epoch": 123.28,
+      "learning_rate": 6.294736842105262e-05,
+      "loss": 0.0002,
+      "step": 8260
+    },
+    {
+      "epoch": 123.3,
+      "learning_rate": 6.291228070175439e-05,
+      "loss": 0.0002,
+      "step": 8261
+    },
+    {
+      "epoch": 123.31,
+      "learning_rate": 6.287719298245613e-05,
+      "loss": 0.0003,
+      "step": 8262
+    },
+    {
+      "epoch": 123.33,
+      "learning_rate": 6.284210526315788e-05,
+      "loss": 0.0004,
+      "step": 8263
+    },
+    {
+      "epoch": 123.34,
+      "learning_rate": 6.280701754385965e-05,
+      "loss": 0.0018,
+      "step": 8264
+    },
+    {
+      "epoch": 123.36,
+      "learning_rate": 6.27719298245614e-05,
+      "loss": 0.0003,
+      "step": 8265
+    },
+    {
+      "epoch": 123.37,
+      "learning_rate": 6.273684210526314e-05,
+      "loss": 0.0003,
+      "step": 8266
+    },
+    {
+      "epoch": 123.39,
+      "learning_rate": 6.27017543859649e-05,
+      "loss": 0.0002,
+      "step": 8267
+    },
+    {
+      "epoch": 123.4,
+      "learning_rate": 6.266666666666667e-05,
+      "loss": 0.0226,
+      "step": 8268
+    },
+    {
+      "epoch": 123.42,
+      "learning_rate": 6.263157894736842e-05,
+      "loss": 0.0003,
+      "step": 8269
+    },
+    {
+      "epoch": 123.43,
+      "learning_rate": 6.259649122807018e-05,
+      "loss": 0.0127,
+      "step": 8270
+    },
+    {
+      "epoch": 123.45,
+      "learning_rate": 6.256140350877193e-05,
+      "loss": 0.0004,
+      "step": 8271
+    },
+    {
+      "epoch": 123.46,
+      "learning_rate": 6.252631578947368e-05,
+      "loss": 0.0003,
+      "step": 8272
+    },
+    {
+      "epoch": 123.48,
+      "learning_rate": 6.249122807017544e-05,
+      "loss": 0.0003,
+      "step": 8273
+    },
+    {
+      "epoch": 123.49,
+      "learning_rate": 6.245614035087719e-05,
+      "loss": 0.0002,
+      "step": 8274
+    },
+    {
+      "epoch": 123.51,
+      "learning_rate": 6.242105263157894e-05,
+      "loss": 0.0004,
+      "step": 8275
+    },
+    {
+      "epoch": 123.52,
+      "learning_rate": 6.23859649122807e-05,
+      "loss": 0.0002,
+      "step": 8276
+    },
+    {
+      "epoch": 123.54,
+      "learning_rate": 6.235087719298245e-05,
+      "loss": 0.0003,
+      "step": 8277
+    },
+    {
+      "epoch": 123.55,
+      "learning_rate": 6.23157894736842e-05,
+      "loss": 0.0004,
+      "step": 8278
+    },
+    {
+      "epoch": 123.57,
+      "learning_rate": 6.228070175438596e-05,
+      "loss": 0.0002,
+      "step": 8279
+    },
+    {
+      "epoch": 123.58,
+      "learning_rate": 6.224561403508771e-05,
+      "loss": 0.0003,
+      "step": 8280
+    },
+    {
+      "epoch": 123.59,
+      "learning_rate": 6.221052631578947e-05,
+      "loss": 0.0018,
+      "step": 8281
+    },
+    {
+      "epoch": 123.61,
+      "learning_rate": 6.217543859649123e-05,
+      "loss": 0.0004,
+      "step": 8282
+    },
+    {
+      "epoch": 123.62,
+      "learning_rate": 6.214035087719298e-05,
+      "loss": 0.0003,
+      "step": 8283
+    },
+    {
+      "epoch": 123.64,
+      "learning_rate": 6.210526315789473e-05,
+      "loss": 0.0025,
+      "step": 8284
+    },
+    {
+      "epoch": 123.65,
+      "learning_rate": 6.207017543859649e-05,
+      "loss": 0.0004,
+      "step": 8285
+    },
+    {
+      "epoch": 123.67,
+      "learning_rate": 6.203508771929824e-05,
+      "loss": 0.0005,
+      "step": 8286
+    },
+    {
+      "epoch": 123.68,
+      "learning_rate": 6.199999999999999e-05,
+      "loss": 0.0003,
+      "step": 8287
+    },
+    {
+      "epoch": 123.7,
+      "learning_rate": 6.196491228070175e-05,
+      "loss": 0.0003,
+      "step": 8288
+    },
+    {
+      "epoch": 123.71,
+      "learning_rate": 6.19298245614035e-05,
+      "loss": 0.0253,
+      "step": 8289
+    },
+    {
+      "epoch": 123.73,
+      "learning_rate": 6.189473684210525e-05,
+      "loss": 0.0003,
+      "step": 8290
+    },
+    {
+      "epoch": 123.74,
+      "learning_rate": 6.185964912280701e-05,
+      "loss": 0.0003,
+      "step": 8291
+    },
+    {
+      "epoch": 123.76,
+      "learning_rate": 6.182456140350876e-05,
+      "loss": 0.0003,
+      "step": 8292
+    },
+    {
+      "epoch": 123.77,
+      "learning_rate": 6.178947368421052e-05,
+      "loss": 0.0007,
+      "step": 8293
+    },
+    {
+      "epoch": 123.79,
+      "learning_rate": 6.175438596491228e-05,
+      "loss": 0.0003,
+      "step": 8294
+    },
+    {
+      "epoch": 123.8,
+      "learning_rate": 6.171929824561403e-05,
+      "loss": 0.0003,
+      "step": 8295
+    },
+    {
+      "epoch": 123.82,
+      "learning_rate": 6.168421052631578e-05,
+      "loss": 0.0003,
+      "step": 8296
+    },
+    {
+      "epoch": 123.83,
+      "learning_rate": 6.164912280701754e-05,
+      "loss": 0.0003,
+      "step": 8297
+    },
+    {
+      "epoch": 123.85,
+      "learning_rate": 6.16140350877193e-05,
+      "loss": 0.0003,
+      "step": 8298
+    },
+    {
+      "epoch": 123.86,
+      "learning_rate": 6.157894736842104e-05,
+      "loss": 0.0003,
+      "step": 8299
+    },
+    {
+      "epoch": 123.88,
+      "learning_rate": 6.15438596491228e-05,
+      "loss": 0.0003,
+      "step": 8300
+    },
+    {
+      "epoch": 123.89,
+      "learning_rate": 6.150877192982455e-05,
+      "loss": 0.0002,
+      "step": 8301
+    },
+    {
+      "epoch": 123.91,
+      "learning_rate": 6.14736842105263e-05,
+      "loss": 0.0002,
+      "step": 8302
+    },
+    {
+      "epoch": 123.92,
+      "learning_rate": 6.143859649122806e-05,
+      "loss": 0.0003,
+      "step": 8303
+    },
+    {
+      "epoch": 123.94,
+      "learning_rate": 6.140350877192981e-05,
+      "loss": 0.0003,
+      "step": 8304
+    },
+    {
+      "epoch": 123.95,
+      "learning_rate": 6.136842105263158e-05,
+      "loss": 0.0003,
+      "step": 8305
+    },
+    {
+      "epoch": 123.97,
+      "learning_rate": 6.133333333333334e-05,
+      "loss": 0.0003,
+      "step": 8306
+    },
+    {
+      "epoch": 123.98,
+      "learning_rate": 6.129824561403509e-05,
+      "loss": 0.0003,
+      "step": 8307
+    },
+    {
+      "epoch": 124.0,
+      "learning_rate": 6.126315789473684e-05,
+      "loss": 0.0002,
+      "step": 8308
+    },
+    {
+      "epoch": 124.01,
+      "learning_rate": 6.12280701754386e-05,
+      "loss": 0.0016,
+      "step": 8309
+    },
+    {
+      "epoch": 124.03,
+      "learning_rate": 6.119298245614035e-05,
+      "loss": 0.0002,
+      "step": 8310
+    },
+    {
+      "epoch": 124.04,
+      "learning_rate": 6.11578947368421e-05,
+      "loss": 0.0002,
+      "step": 8311
+    },
+    {
+      "epoch": 124.06,
+      "learning_rate": 6.112280701754386e-05,
+      "loss": 0.0002,
+      "step": 8312
+    },
+    {
+      "epoch": 124.07,
+      "learning_rate": 6.10877192982456e-05,
+      "loss": 0.0003,
+      "step": 8313
+    },
+    {
+      "epoch": 124.09,
+      "learning_rate": 6.105263157894736e-05,
+      "loss": 0.0004,
+      "step": 8314
+    },
+    {
+      "epoch": 124.1,
+      "learning_rate": 6.101754385964912e-05,
+      "loss": 0.0002,
+      "step": 8315
+    },
+    {
+      "epoch": 124.12,
+      "learning_rate": 6.0982456140350866e-05,
+      "loss": 0.0003,
+      "step": 8316
+    },
+    {
+      "epoch": 124.13,
+      "learning_rate": 6.094736842105262e-05,
+      "loss": 0.0003,
+      "step": 8317
+    },
+    {
+      "epoch": 124.15,
+      "learning_rate": 6.0912280701754384e-05,
+      "loss": 0.0002,
+      "step": 8318
+    },
+    {
+      "epoch": 124.16,
+      "learning_rate": 6.087719298245614e-05,
+      "loss": 0.0003,
+      "step": 8319
+    },
+    {
+      "epoch": 124.18,
+      "learning_rate": 6.0842105263157895e-05,
+      "loss": 0.0003,
+      "step": 8320
+    },
+    {
+      "epoch": 124.19,
+      "learning_rate": 6.0807017543859644e-05,
+      "loss": 0.0035,
+      "step": 8321
+    },
+    {
+      "epoch": 124.21,
+      "learning_rate": 6.07719298245614e-05,
+      "loss": 0.0002,
+      "step": 8322
+    },
+    {
+      "epoch": 124.22,
+      "learning_rate": 6.0736842105263155e-05,
+      "loss": 0.0031,
+      "step": 8323
+    },
+    {
+      "epoch": 124.24,
+      "learning_rate": 6.0701754385964904e-05,
+      "loss": 0.1618,
+      "step": 8324
+    },
+    {
+      "epoch": 124.25,
+      "learning_rate": 6.066666666666666e-05,
+      "loss": 0.0002,
+      "step": 8325
+    },
+    {
+      "epoch": 124.27,
+      "learning_rate": 6.0631578947368415e-05,
+      "loss": 0.0002,
+      "step": 8326
+    },
+    {
+      "epoch": 124.28,
+      "learning_rate": 6.059649122807017e-05,
+      "loss": 0.0002,
+      "step": 8327
+    },
+    {
+      "epoch": 124.3,
+      "learning_rate": 6.056140350877192e-05,
+      "loss": 0.0004,
+      "step": 8328
+    },
+    {
+      "epoch": 124.31,
+      "learning_rate": 6.0526315789473675e-05,
+      "loss": 0.0002,
+      "step": 8329
+    },
+    {
+      "epoch": 124.33,
+      "learning_rate": 6.049122807017543e-05,
+      "loss": 0.0002,
+      "step": 8330
+    },
+    {
+      "epoch": 124.34,
+      "learning_rate": 6.045614035087719e-05,
+      "loss": 0.0003,
+      "step": 8331
+    },
+    {
+      "epoch": 124.36,
+      "learning_rate": 6.042105263157895e-05,
+      "loss": 0.0002,
+      "step": 8332
+    },
+    {
+      "epoch": 124.37,
+      "learning_rate": 6.03859649122807e-05,
+      "loss": 0.0002,
+      "step": 8333
+    },
+    {
+      "epoch": 124.39,
+      "learning_rate": 6.035087719298245e-05,
+      "loss": 0.0003,
+      "step": 8334
+    },
+    {
+      "epoch": 124.4,
+      "learning_rate": 6.031578947368421e-05,
+      "loss": 0.0009,
+      "step": 8335
+    },
+    {
+      "epoch": 124.42,
+      "learning_rate": 6.028070175438596e-05,
+      "loss": 0.0004,
+      "step": 8336
+    },
+    {
+      "epoch": 124.43,
+      "learning_rate": 6.024561403508771e-05,
+      "loss": 0.0003,
+      "step": 8337
+    },
+    {
+      "epoch": 124.45,
+      "learning_rate": 6.021052631578947e-05,
+      "loss": 0.0004,
+      "step": 8338
+    },
+    {
+      "epoch": 124.46,
+      "learning_rate": 6.0175438596491224e-05,
+      "loss": 0.0003,
+      "step": 8339
+    },
+    {
+      "epoch": 124.48,
+      "learning_rate": 6.014035087719297e-05,
+      "loss": 0.0002,
+      "step": 8340
+    },
+    {
+      "epoch": 124.49,
+      "learning_rate": 6.010526315789473e-05,
+      "loss": 0.008,
+      "step": 8341
+    },
+    {
+      "epoch": 124.51,
+      "learning_rate": 6.0070175438596484e-05,
+      "loss": 0.0002,
+      "step": 8342
+    },
+    {
+      "epoch": 124.52,
+      "learning_rate": 6.0035087719298246e-05,
+      "loss": 0.0002,
+      "step": 8343
+    },
+    {
+      "epoch": 124.54,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 0.0002,
+      "step": 8344
+    },
+    {
+      "epoch": 124.55,
+      "learning_rate": 5.996491228070175e-05,
+      "loss": 0.0002,
+      "step": 8345
+    },
+    {
+      "epoch": 124.57,
+      "learning_rate": 5.9929824561403506e-05,
+      "loss": 0.0002,
+      "step": 8346
+    },
+    {
+      "epoch": 124.58,
+      "learning_rate": 5.989473684210526e-05,
+      "loss": 0.0005,
+      "step": 8347
+    },
+    {
+      "epoch": 124.59,
+      "learning_rate": 5.985964912280701e-05,
+      "loss": 0.0002,
+      "step": 8348
+    },
+    {
+      "epoch": 124.61,
+      "learning_rate": 5.9824561403508766e-05,
+      "loss": 0.0002,
+      "step": 8349
+    },
+    {
+      "epoch": 124.62,
+      "learning_rate": 5.978947368421052e-05,
+      "loss": 0.0003,
+      "step": 8350
+    },
+    {
+      "epoch": 124.64,
+      "learning_rate": 5.975438596491228e-05,
+      "loss": 0.0004,
+      "step": 8351
+    },
+    {
+      "epoch": 124.65,
+      "learning_rate": 5.9719298245614026e-05,
+      "loss": 0.0002,
+      "step": 8352
+    },
+    {
+      "epoch": 124.67,
+      "learning_rate": 5.968421052631578e-05,
+      "loss": 0.0002,
+      "step": 8353
+    },
+    {
+      "epoch": 124.68,
+      "learning_rate": 5.964912280701754e-05,
+      "loss": 0.0061,
+      "step": 8354
+    },
+    {
+      "epoch": 124.7,
+      "learning_rate": 5.961403508771929e-05,
+      "loss": 0.0002,
+      "step": 8355
+    },
+    {
+      "epoch": 124.71,
+      "learning_rate": 5.957894736842105e-05,
+      "loss": 0.0002,
+      "step": 8356
+    },
+    {
+      "epoch": 124.73,
+      "learning_rate": 5.9543859649122803e-05,
+      "loss": 0.0002,
+      "step": 8357
+    },
+    {
+      "epoch": 124.74,
+      "learning_rate": 5.950877192982456e-05,
+      "loss": 0.0003,
+      "step": 8358
+    },
+    {
+      "epoch": 124.76,
+      "learning_rate": 5.9473684210526315e-05,
+      "loss": 0.0002,
+      "step": 8359
+    },
+    {
+      "epoch": 124.77,
+      "learning_rate": 5.943859649122806e-05,
+      "loss": 0.0002,
+      "step": 8360
+    },
+    {
+      "epoch": 124.79,
+      "learning_rate": 5.940350877192982e-05,
+      "loss": 0.0098,
+      "step": 8361
+    },
+    {
+      "epoch": 124.8,
+      "learning_rate": 5.9368421052631574e-05,
+      "loss": 0.0003,
+      "step": 8362
+    },
+    {
+      "epoch": 124.82,
+      "learning_rate": 5.933333333333333e-05,
+      "loss": 0.0002,
+      "step": 8363
+    },
+    {
+      "epoch": 124.83,
+      "learning_rate": 5.929824561403508e-05,
+      "loss": 0.0034,
+      "step": 8364
+    },
+    {
+      "epoch": 124.85,
+      "learning_rate": 5.9263157894736834e-05,
+      "loss": 0.0003,
+      "step": 8365
+    },
+    {
+      "epoch": 124.86,
+      "learning_rate": 5.922807017543859e-05,
+      "loss": 0.0002,
+      "step": 8366
+    },
+    {
+      "epoch": 124.88,
+      "learning_rate": 5.9192982456140345e-05,
+      "loss": 0.0004,
+      "step": 8367
+    },
+    {
+      "epoch": 124.89,
+      "learning_rate": 5.91578947368421e-05,
+      "loss": 0.0002,
+      "step": 8368
+    },
+    {
+      "epoch": 124.91,
+      "learning_rate": 5.9122807017543856e-05,
+      "loss": 0.0003,
+      "step": 8369
+    },
+    {
+      "epoch": 124.92,
+      "learning_rate": 5.908771929824561e-05,
+      "loss": 0.0002,
+      "step": 8370
+    },
+    {
+      "epoch": 124.94,
+      "learning_rate": 5.905263157894737e-05,
+      "loss": 0.0003,
+      "step": 8371
+    },
+    {
+      "epoch": 124.95,
+      "learning_rate": 5.9017543859649116e-05,
+      "loss": 0.0002,
+      "step": 8372
+    },
+    {
+      "epoch": 124.97,
+      "learning_rate": 5.898245614035087e-05,
+      "loss": 0.0002,
+      "step": 8373
+    },
+    {
+      "epoch": 124.98,
+      "learning_rate": 5.894736842105263e-05,
+      "loss": 0.0002,
+      "step": 8374
+    },
+    {
+      "epoch": 125.0,
+      "learning_rate": 5.891228070175438e-05,
+      "loss": 0.0002,
+      "step": 8375
+    },
+    {
+      "epoch": 125.01,
+      "learning_rate": 5.887719298245613e-05,
+      "loss": 0.0003,
+      "step": 8376
+    },
+    {
+      "epoch": 125.03,
+      "learning_rate": 5.884210526315789e-05,
+      "loss": 0.0006,
+      "step": 8377
+    },
+    {
+      "epoch": 125.04,
+      "learning_rate": 5.880701754385964e-05,
+      "loss": 0.0002,
+      "step": 8378
+    },
+    {
+      "epoch": 125.06,
+      "learning_rate": 5.87719298245614e-05,
+      "loss": 0.0002,
+      "step": 8379
+    },
+    {
+      "epoch": 125.07,
+      "learning_rate": 5.873684210526315e-05,
+      "loss": 0.0002,
+      "step": 8380
+    },
+    {
+      "epoch": 125.09,
+      "learning_rate": 5.870175438596491e-05,
+      "loss": 0.1178,
+      "step": 8381
+    },
+    {
+      "epoch": 125.1,
+      "learning_rate": 5.8666666666666665e-05,
+      "loss": 0.0002,
+      "step": 8382
+    },
+    {
+      "epoch": 125.12,
+      "learning_rate": 5.863157894736842e-05,
+      "loss": 0.0002,
+      "step": 8383
+    },
+    {
+      "epoch": 125.13,
+      "learning_rate": 5.859649122807017e-05,
+      "loss": 0.0003,
+      "step": 8384
+    },
+    {
+      "epoch": 125.15,
+      "learning_rate": 5.8561403508771925e-05,
+      "loss": 0.0006,
+      "step": 8385
+    },
+    {
+      "epoch": 125.16,
+      "learning_rate": 5.852631578947368e-05,
+      "loss": 0.0002,
+      "step": 8386
+    },
+    {
+      "epoch": 125.18,
+      "learning_rate": 5.8491228070175436e-05,
+      "loss": 0.0027,
+      "step": 8387
+    },
+    {
+      "epoch": 125.19,
+      "learning_rate": 5.8456140350877185e-05,
+      "loss": 0.0004,
+      "step": 8388
+    },
+    {
+      "epoch": 125.21,
+      "learning_rate": 5.842105263157894e-05,
+      "loss": 0.0002,
+      "step": 8389
+    },
+    {
+      "epoch": 125.22,
+      "learning_rate": 5.8385964912280696e-05,
+      "loss": 0.0004,
+      "step": 8390
+    },
+    {
+      "epoch": 125.24,
+      "learning_rate": 5.8350877192982445e-05,
+      "loss": 0.0002,
+      "step": 8391
+    },
+    {
+      "epoch": 125.25,
+      "learning_rate": 5.83157894736842e-05,
+      "loss": 0.0002,
+      "step": 8392
+    },
+    {
+      "epoch": 125.27,
+      "learning_rate": 5.828070175438596e-05,
+      "loss": 0.0002,
+      "step": 8393
+    },
+    {
+      "epoch": 125.28,
+      "learning_rate": 5.824561403508772e-05,
+      "loss": 0.0007,
+      "step": 8394
+    },
+    {
+      "epoch": 125.3,
+      "learning_rate": 5.8210526315789474e-05,
+      "loss": 0.0002,
+      "step": 8395
+    },
+    {
+      "epoch": 125.31,
+      "learning_rate": 5.817543859649122e-05,
+      "loss": 0.0003,
+      "step": 8396
+    },
+    {
+      "epoch": 125.33,
+      "learning_rate": 5.814035087719298e-05,
+      "loss": 0.0009,
+      "step": 8397
+    },
+    {
+      "epoch": 125.34,
+      "learning_rate": 5.8105263157894734e-05,
+      "loss": 0.0002,
+      "step": 8398
+    },
+    {
+      "epoch": 125.36,
+      "learning_rate": 5.807017543859649e-05,
+      "loss": 0.0003,
+      "step": 8399
+    },
+    {
+      "epoch": 125.37,
+      "learning_rate": 5.803508771929824e-05,
+      "loss": 0.2335,
+      "step": 8400
+    },
+    {
+      "epoch": 125.37,
+      "eval_accuracy": 0.8827704356338718,
+      "eval_f1": 0.8831895482310441,
+      "eval_loss": 0.6515348553657532,
+      "eval_runtime": 348.7658,
+      "eval_samples_per_second": 11.716,
+      "eval_steps_per_second": 0.734,
+      "step": 8400
+    },
+    {
+      "epoch": 125.39,
+      "learning_rate": 5.7999999999999994e-05,
+      "loss": 0.0005,
+      "step": 8401
+    },
+    {
+      "epoch": 125.4,
+      "learning_rate": 5.796491228070175e-05,
+      "loss": 0.0002,
+      "step": 8402
+    },
+    {
+      "epoch": 125.42,
+      "learning_rate": 5.79298245614035e-05,
+      "loss": 0.0002,
+      "step": 8403
+    },
+    {
+      "epoch": 125.43,
+      "learning_rate": 5.7894736842105253e-05,
+      "loss": 0.0002,
+      "step": 8404
+    },
+    {
+      "epoch": 125.45,
+      "learning_rate": 5.785964912280701e-05,
+      "loss": 0.0003,
+      "step": 8405
+    },
+    {
+      "epoch": 125.46,
+      "learning_rate": 5.782456140350877e-05,
+      "loss": 0.0002,
+      "step": 8406
+    },
+    {
+      "epoch": 125.48,
+      "learning_rate": 5.778947368421053e-05,
+      "loss": 0.0008,
+      "step": 8407
+    },
+    {
+      "epoch": 125.49,
+      "learning_rate": 5.7754385964912276e-05,
+      "loss": 0.0002,
+      "step": 8408
+    },
+    {
+      "epoch": 125.51,
+      "learning_rate": 5.771929824561403e-05,
+      "loss": 0.0002,
+      "step": 8409
+    },
+    {
+      "epoch": 125.52,
+      "learning_rate": 5.768421052631579e-05,
+      "loss": 0.0029,
+      "step": 8410
+    },
+    {
+      "epoch": 125.54,
+      "learning_rate": 5.764912280701754e-05,
+      "loss": 0.0002,
+      "step": 8411
+    },
+    {
+      "epoch": 125.55,
+      "learning_rate": 5.761403508771929e-05,
+      "loss": 0.0044,
+      "step": 8412
+    },
+    {
+      "epoch": 125.57,
+      "learning_rate": 5.757894736842105e-05,
+      "loss": 0.0002,
+      "step": 8413
+    },
+    {
+      "epoch": 125.58,
+      "learning_rate": 5.75438596491228e-05,
+      "loss": 0.0003,
+      "step": 8414
+    },
+    {
+      "epoch": 125.59,
+      "learning_rate": 5.750877192982455e-05,
+      "loss": 0.0002,
+      "step": 8415
+    },
+    {
+      "epoch": 125.61,
+      "learning_rate": 5.7473684210526307e-05,
+      "loss": 0.0002,
+      "step": 8416
+    },
+    {
+      "epoch": 125.62,
+      "learning_rate": 5.743859649122806e-05,
+      "loss": 0.0199,
+      "step": 8417
+    },
+    {
+      "epoch": 125.64,
+      "learning_rate": 5.7403508771929824e-05,
+      "loss": 0.0004,
+      "step": 8418
+    },
+    {
+      "epoch": 125.65,
+      "learning_rate": 5.736842105263158e-05,
+      "loss": 0.0621,
+      "step": 8419
+    },
+    {
+      "epoch": 125.67,
+      "learning_rate": 5.733333333333333e-05,
+      "loss": 0.0003,
+      "step": 8420
+    },
+    {
+      "epoch": 125.68,
+      "learning_rate": 5.7298245614035084e-05,
+      "loss": 0.0002,
+      "step": 8421
+    },
+    {
+      "epoch": 125.7,
+      "learning_rate": 5.726315789473684e-05,
+      "loss": 0.0002,
+      "step": 8422
+    },
+    {
+      "epoch": 125.71,
+      "learning_rate": 5.7228070175438595e-05,
+      "loss": 0.0008,
+      "step": 8423
+    },
+    {
+      "epoch": 125.73,
+      "learning_rate": 5.7192982456140344e-05,
+      "loss": 0.0002,
+      "step": 8424
+    },
+    {
+      "epoch": 125.74,
+      "learning_rate": 5.71578947368421e-05,
+      "loss": 0.0014,
+      "step": 8425
+    },
+    {
+      "epoch": 125.76,
+      "learning_rate": 5.7122807017543855e-05,
+      "loss": 0.0848,
+      "step": 8426
+    },
+    {
+      "epoch": 125.77,
+      "learning_rate": 5.7087719298245604e-05,
+      "loss": 0.0004,
+      "step": 8427
+    },
+    {
+      "epoch": 125.79,
+      "learning_rate": 5.705263157894736e-05,
+      "loss": 0.0003,
+      "step": 8428
+    },
+    {
+      "epoch": 125.8,
+      "learning_rate": 5.7017543859649115e-05,
+      "loss": 0.0092,
+      "step": 8429
+    },
+    {
+      "epoch": 125.82,
+      "learning_rate": 5.698245614035087e-05,
+      "loss": 0.0004,
+      "step": 8430
+    },
+    {
+      "epoch": 125.83,
+      "learning_rate": 5.694736842105263e-05,
+      "loss": 0.0072,
+      "step": 8431
+    },
+    {
+      "epoch": 125.85,
+      "learning_rate": 5.691228070175438e-05,
+      "loss": 0.0002,
+      "step": 8432
+    },
+    {
+      "epoch": 125.86,
+      "learning_rate": 5.687719298245614e-05,
+      "loss": 0.0003,
+      "step": 8433
+    },
+    {
+      "epoch": 125.88,
+      "learning_rate": 5.684210526315789e-05,
+      "loss": 0.0002,
+      "step": 8434
+    },
+    {
+      "epoch": 125.89,
+      "learning_rate": 5.680701754385965e-05,
+      "loss": 0.0003,
+      "step": 8435
+    },
+    {
+      "epoch": 125.91,
+      "learning_rate": 5.67719298245614e-05,
+      "loss": 0.0002,
+      "step": 8436
+    },
+    {
+      "epoch": 125.92,
+      "learning_rate": 5.673684210526315e-05,
+      "loss": 0.0002,
+      "step": 8437
+    },
+    {
+      "epoch": 125.94,
+      "learning_rate": 5.670175438596491e-05,
+      "loss": 0.0012,
+      "step": 8438
+    },
+    {
+      "epoch": 125.95,
+      "learning_rate": 5.666666666666666e-05,
+      "loss": 0.0005,
+      "step": 8439
+    },
+    {
+      "epoch": 125.97,
+      "learning_rate": 5.663157894736841e-05,
+      "loss": 0.0003,
+      "step": 8440
+    },
+    {
+      "epoch": 125.98,
+      "learning_rate": 5.659649122807017e-05,
+      "loss": 0.0003,
+      "step": 8441
+    },
+    {
+      "epoch": 126.0,
+      "learning_rate": 5.6561403508771924e-05,
+      "loss": 0.0002,
+      "step": 8442
+    },
+    {
+      "epoch": 126.01,
+      "learning_rate": 5.652631578947367e-05,
+      "loss": 0.0003,
+      "step": 8443
+    },
+    {
+      "epoch": 126.03,
+      "learning_rate": 5.6491228070175435e-05,
+      "loss": 0.0004,
+      "step": 8444
+    },
+    {
+      "epoch": 126.04,
+      "learning_rate": 5.645614035087719e-05,
+      "loss": 0.0143,
+      "step": 8445
+    },
+    {
+      "epoch": 126.06,
+      "learning_rate": 5.6421052631578946e-05,
+      "loss": 0.0083,
+      "step": 8446
+    },
+    {
+      "epoch": 126.07,
+      "learning_rate": 5.63859649122807e-05,
+      "loss": 0.0032,
+      "step": 8447
+    },
+    {
+      "epoch": 126.09,
+      "learning_rate": 5.635087719298245e-05,
+      "loss": 0.0002,
+      "step": 8448
+    },
+    {
+      "epoch": 126.1,
+      "learning_rate": 5.6315789473684206e-05,
+      "loss": 0.0004,
+      "step": 8449
+    },
+    {
+      "epoch": 126.12,
+      "learning_rate": 5.628070175438596e-05,
+      "loss": 0.0003,
+      "step": 8450
+    },
+    {
+      "epoch": 126.13,
+      "learning_rate": 5.624561403508771e-05,
+      "loss": 0.0002,
+      "step": 8451
+    },
+    {
+      "epoch": 126.15,
+      "learning_rate": 5.6210526315789466e-05,
+      "loss": 0.0003,
+      "step": 8452
+    },
+    {
+      "epoch": 126.16,
+      "learning_rate": 5.617543859649122e-05,
+      "loss": 0.0002,
+      "step": 8453
+    },
+    {
+      "epoch": 126.18,
+      "learning_rate": 5.614035087719298e-05,
+      "loss": 0.0013,
+      "step": 8454
+    },
+    {
+      "epoch": 126.19,
+      "learning_rate": 5.6105263157894726e-05,
+      "loss": 0.0013,
+      "step": 8455
+    },
+    {
+      "epoch": 126.21,
+      "learning_rate": 5.607017543859649e-05,
+      "loss": 0.0003,
+      "step": 8456
+    },
+    {
+      "epoch": 126.22,
+      "learning_rate": 5.6035087719298244e-05,
+      "loss": 0.0002,
+      "step": 8457
+    },
+    {
+      "epoch": 126.24,
+      "learning_rate": 5.6e-05,
+      "loss": 0.0002,
+      "step": 8458
+    },
+    {
+      "epoch": 126.25,
+      "learning_rate": 5.5964912280701755e-05,
+      "loss": 0.0002,
+      "step": 8459
+    },
+    {
+      "epoch": 126.27,
+      "learning_rate": 5.5929824561403503e-05,
+      "loss": 0.0002,
+      "step": 8460
+    },
+    {
+      "epoch": 126.28,
+      "learning_rate": 5.589473684210526e-05,
+      "loss": 0.001,
+      "step": 8461
+    },
+    {
+      "epoch": 126.3,
+      "learning_rate": 5.5859649122807015e-05,
+      "loss": 0.001,
+      "step": 8462
+    },
+    {
+      "epoch": 126.31,
+      "learning_rate": 5.582456140350876e-05,
+      "loss": 0.0002,
+      "step": 8463
+    },
+    {
+      "epoch": 126.33,
+      "learning_rate": 5.578947368421052e-05,
+      "loss": 0.0002,
+      "step": 8464
+    },
+    {
+      "epoch": 126.34,
+      "learning_rate": 5.5754385964912274e-05,
+      "loss": 0.0002,
+      "step": 8465
+    },
+    {
+      "epoch": 126.36,
+      "learning_rate": 5.571929824561403e-05,
+      "loss": 0.0002,
+      "step": 8466
+    },
+    {
+      "epoch": 126.37,
+      "learning_rate": 5.568421052631578e-05,
+      "loss": 0.0003,
+      "step": 8467
+    },
+    {
+      "epoch": 126.39,
+      "learning_rate": 5.5649122807017534e-05,
+      "loss": 0.0002,
+      "step": 8468
+    },
+    {
+      "epoch": 126.4,
+      "learning_rate": 5.56140350877193e-05,
+      "loss": 0.0003,
+      "step": 8469
+    },
+    {
+      "epoch": 126.42,
+      "learning_rate": 5.557894736842105e-05,
+      "loss": 0.0002,
+      "step": 8470
+    },
+    {
+      "epoch": 126.43,
+      "learning_rate": 5.55438596491228e-05,
+      "loss": 0.0366,
+      "step": 8471
+    },
+    {
+      "epoch": 126.45,
+      "learning_rate": 5.5508771929824557e-05,
+      "loss": 0.0002,
+      "step": 8472
+    },
+    {
+      "epoch": 126.46,
+      "learning_rate": 5.547368421052631e-05,
+      "loss": 0.0002,
+      "step": 8473
+    },
+    {
+      "epoch": 126.48,
+      "learning_rate": 5.543859649122807e-05,
+      "loss": 0.0004,
+      "step": 8474
+    },
+    {
+      "epoch": 126.49,
+      "learning_rate": 5.5403508771929816e-05,
+      "loss": 0.0002,
+      "step": 8475
+    },
+    {
+      "epoch": 126.51,
+      "learning_rate": 5.536842105263157e-05,
+      "loss": 0.0003,
+      "step": 8476
+    },
+    {
+      "epoch": 126.52,
+      "learning_rate": 5.533333333333333e-05,
+      "loss": 0.0003,
+      "step": 8477
+    },
+    {
+      "epoch": 126.54,
+      "learning_rate": 5.529824561403508e-05,
+      "loss": 0.0002,
+      "step": 8478
+    },
+    {
+      "epoch": 126.55,
+      "learning_rate": 5.526315789473683e-05,
+      "loss": 0.0002,
+      "step": 8479
+    },
+    {
+      "epoch": 126.57,
+      "learning_rate": 5.522807017543859e-05,
+      "loss": 0.0002,
+      "step": 8480
+    },
+    {
+      "epoch": 126.58,
+      "learning_rate": 5.519298245614035e-05,
+      "loss": 0.0002,
+      "step": 8481
+    },
+    {
+      "epoch": 126.59,
+      "learning_rate": 5.5157894736842105e-05,
+      "loss": 0.0005,
+      "step": 8482
+    },
+    {
+      "epoch": 126.61,
+      "learning_rate": 5.5122807017543854e-05,
+      "loss": 0.0002,
+      "step": 8483
+    },
+    {
+      "epoch": 126.62,
+      "learning_rate": 5.508771929824561e-05,
+      "loss": 0.0003,
+      "step": 8484
+    },
+    {
+      "epoch": 126.64,
+      "learning_rate": 5.5052631578947365e-05,
+      "loss": 0.0002,
+      "step": 8485
+    },
+    {
+      "epoch": 126.65,
+      "learning_rate": 5.501754385964912e-05,
+      "loss": 0.0002,
+      "step": 8486
+    },
+    {
+      "epoch": 126.67,
+      "learning_rate": 5.498245614035087e-05,
+      "loss": 0.0002,
+      "step": 8487
+    },
+    {
+      "epoch": 126.68,
+      "learning_rate": 5.4947368421052625e-05,
+      "loss": 0.0002,
+      "step": 8488
+    },
+    {
+      "epoch": 126.7,
+      "learning_rate": 5.491228070175438e-05,
+      "loss": 0.0003,
+      "step": 8489
+    },
+    {
+      "epoch": 126.71,
+      "learning_rate": 5.4877192982456136e-05,
+      "loss": 0.0002,
+      "step": 8490
+    },
+    {
+      "epoch": 126.73,
+      "learning_rate": 5.4842105263157885e-05,
+      "loss": 0.0002,
+      "step": 8491
+    },
+    {
+      "epoch": 126.74,
+      "learning_rate": 5.480701754385964e-05,
+      "loss": 0.0003,
+      "step": 8492
+    },
+    {
+      "epoch": 126.76,
+      "learning_rate": 5.4771929824561396e-05,
+      "loss": 0.2602,
+      "step": 8493
+    },
+    {
+      "epoch": 126.77,
+      "learning_rate": 5.473684210526316e-05,
+      "loss": 0.0003,
+      "step": 8494
+    },
+    {
+      "epoch": 126.79,
+      "learning_rate": 5.470175438596491e-05,
+      "loss": 0.0002,
+      "step": 8495
+    },
+    {
+      "epoch": 126.8,
+      "learning_rate": 5.466666666666666e-05,
+      "loss": 0.0003,
+      "step": 8496
+    },
+    {
+      "epoch": 126.82,
+      "learning_rate": 5.463157894736842e-05,
+      "loss": 0.0137,
+      "step": 8497
+    },
+    {
+      "epoch": 126.83,
+      "learning_rate": 5.4596491228070174e-05,
+      "loss": 0.2638,
+      "step": 8498
+    },
+    {
+      "epoch": 126.85,
+      "learning_rate": 5.456140350877192e-05,
+      "loss": 0.0002,
+      "step": 8499
+    },
+    {
+      "epoch": 126.86,
+      "learning_rate": 5.452631578947368e-05,
+      "loss": 0.0003,
+      "step": 8500
+    },
+    {
+      "epoch": 126.88,
+      "learning_rate": 5.4491228070175434e-05,
+      "loss": 0.0002,
+      "step": 8501
+    },
+    {
+      "epoch": 126.89,
+      "learning_rate": 5.445614035087719e-05,
+      "loss": 0.0003,
+      "step": 8502
+    },
+    {
+      "epoch": 126.91,
+      "learning_rate": 5.442105263157894e-05,
+      "loss": 0.0003,
+      "step": 8503
+    },
+    {
+      "epoch": 126.92,
+      "learning_rate": 5.4385964912280694e-05,
+      "loss": 0.0051,
+      "step": 8504
+    },
+    {
+      "epoch": 126.94,
+      "learning_rate": 5.435087719298245e-05,
+      "loss": 0.0003,
+      "step": 8505
+    },
+    {
+      "epoch": 126.95,
+      "learning_rate": 5.431578947368421e-05,
+      "loss": 0.0003,
+      "step": 8506
+    },
+    {
+      "epoch": 126.97,
+      "learning_rate": 5.428070175438596e-05,
+      "loss": 0.0007,
+      "step": 8507
+    },
+    {
+      "epoch": 126.98,
+      "learning_rate": 5.4245614035087716e-05,
+      "loss": 0.0003,
+      "step": 8508
+    },
+    {
+      "epoch": 127.0,
+      "learning_rate": 5.421052631578947e-05,
+      "loss": 0.0002,
+      "step": 8509
+    },
+    {
+      "epoch": 127.01,
+      "learning_rate": 5.417543859649123e-05,
+      "loss": 0.0003,
+      "step": 8510
+    },
+    {
+      "epoch": 127.03,
+      "learning_rate": 5.4140350877192976e-05,
+      "loss": 0.0003,
+      "step": 8511
+    },
+    {
+      "epoch": 127.04,
+      "learning_rate": 5.410526315789473e-05,
+      "loss": 0.0003,
+      "step": 8512
+    },
+    {
+      "epoch": 127.06,
+      "learning_rate": 5.407017543859649e-05,
+      "loss": 0.0002,
+      "step": 8513
+    },
+    {
+      "epoch": 127.07,
+      "learning_rate": 5.403508771929824e-05,
+      "loss": 0.0004,
+      "step": 8514
+    },
+    {
+      "epoch": 127.09,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 0.0002,
+      "step": 8515
+    },
+    {
+      "epoch": 127.1,
+      "learning_rate": 5.396491228070175e-05,
+      "loss": 0.0003,
+      "step": 8516
+    },
+    {
+      "epoch": 127.12,
+      "learning_rate": 5.39298245614035e-05,
+      "loss": 0.0004,
+      "step": 8517
+    },
+    {
+      "epoch": 127.13,
+      "learning_rate": 5.389473684210525e-05,
+      "loss": 0.0003,
+      "step": 8518
+    },
+    {
+      "epoch": 127.15,
+      "learning_rate": 5.3859649122807013e-05,
+      "loss": 0.0004,
+      "step": 8519
+    },
+    {
+      "epoch": 127.16,
+      "learning_rate": 5.382456140350877e-05,
+      "loss": 0.0003,
+      "step": 8520
+    },
+    {
+      "epoch": 127.18,
+      "learning_rate": 5.3789473684210525e-05,
+      "loss": 0.0003,
+      "step": 8521
+    },
+    {
+      "epoch": 127.19,
+      "learning_rate": 5.375438596491228e-05,
+      "loss": 0.0003,
+      "step": 8522
+    },
+    {
+      "epoch": 127.21,
+      "learning_rate": 5.371929824561403e-05,
+      "loss": 0.0004,
+      "step": 8523
+    },
+    {
+      "epoch": 127.22,
+      "learning_rate": 5.3684210526315784e-05,
+      "loss": 0.1857,
+      "step": 8524
+    },
+    {
+      "epoch": 127.24,
+      "learning_rate": 5.364912280701754e-05,
+      "loss": 0.0003,
+      "step": 8525
+    },
+    {
+      "epoch": 127.25,
+      "learning_rate": 5.3614035087719296e-05,
+      "loss": 0.0003,
+      "step": 8526
+    },
+    {
+      "epoch": 127.27,
+      "learning_rate": 5.3578947368421044e-05,
+      "loss": 0.0005,
+      "step": 8527
+    },
+    {
+      "epoch": 127.28,
+      "learning_rate": 5.35438596491228e-05,
+      "loss": 0.0003,
+      "step": 8528
+    },
+    {
+      "epoch": 127.3,
+      "learning_rate": 5.3508771929824555e-05,
+      "loss": 0.0003,
+      "step": 8529
+    },
+    {
+      "epoch": 127.31,
+      "learning_rate": 5.3473684210526304e-05,
+      "loss": 0.0004,
+      "step": 8530
+    },
+    {
+      "epoch": 127.33,
+      "learning_rate": 5.3438596491228067e-05,
+      "loss": 0.0003,
+      "step": 8531
+    },
+    {
+      "epoch": 127.34,
+      "learning_rate": 5.340350877192982e-05,
+      "loss": 0.1111,
+      "step": 8532
+    },
+    {
+      "epoch": 127.36,
+      "learning_rate": 5.336842105263158e-05,
+      "loss": 0.0005,
+      "step": 8533
+    },
+    {
+      "epoch": 127.37,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 0.0003,
+      "step": 8534
+    },
+    {
+      "epoch": 127.39,
+      "learning_rate": 5.329824561403508e-05,
+      "loss": 0.0002,
+      "step": 8535
+    },
+    {
+      "epoch": 127.4,
+      "learning_rate": 5.326315789473684e-05,
+      "loss": 0.0004,
+      "step": 8536
+    },
+    {
+      "epoch": 127.42,
+      "learning_rate": 5.322807017543859e-05,
+      "loss": 0.0017,
+      "step": 8537
+    },
+    {
+      "epoch": 127.43,
+      "learning_rate": 5.319298245614035e-05,
+      "loss": 0.0004,
+      "step": 8538
+    },
+    {
+      "epoch": 127.45,
+      "learning_rate": 5.31578947368421e-05,
+      "loss": 0.0003,
+      "step": 8539
+    },
+    {
+      "epoch": 127.46,
+      "learning_rate": 5.312280701754385e-05,
+      "loss": 0.0003,
+      "step": 8540
+    },
+    {
+      "epoch": 127.48,
+      "learning_rate": 5.308771929824561e-05,
+      "loss": 0.0003,
+      "step": 8541
+    },
+    {
+      "epoch": 127.49,
+      "learning_rate": 5.305263157894736e-05,
+      "loss": 0.0003,
+      "step": 8542
+    },
+    {
+      "epoch": 127.51,
+      "learning_rate": 5.301754385964911e-05,
+      "loss": 0.0003,
+      "step": 8543
+    },
+    {
+      "epoch": 127.52,
+      "learning_rate": 5.2982456140350875e-05,
+      "loss": 0.0016,
+      "step": 8544
+    },
+    {
+      "epoch": 127.54,
+      "learning_rate": 5.294736842105263e-05,
+      "loss": 0.0003,
+      "step": 8545
+    },
+    {
+      "epoch": 127.55,
+      "learning_rate": 5.2912280701754386e-05,
+      "loss": 0.0004,
+      "step": 8546
+    },
+    {
+      "epoch": 127.57,
+      "learning_rate": 5.2877192982456135e-05,
+      "loss": 0.0003,
+      "step": 8547
+    },
+    {
+      "epoch": 127.58,
+      "learning_rate": 5.284210526315789e-05,
+      "loss": 0.0004,
+      "step": 8548
+    },
+    {
+      "epoch": 127.59,
+      "learning_rate": 5.2807017543859646e-05,
+      "loss": 0.0014,
+      "step": 8549
+    },
+    {
+      "epoch": 127.61,
+      "learning_rate": 5.27719298245614e-05,
+      "loss": 0.0004,
+      "step": 8550
+    },
+    {
+      "epoch": 127.62,
+      "learning_rate": 5.273684210526315e-05,
+      "loss": 0.0132,
+      "step": 8551
+    },
+    {
+      "epoch": 127.64,
+      "learning_rate": 5.2701754385964906e-05,
+      "loss": 0.0012,
+      "step": 8552
+    },
+    {
+      "epoch": 127.65,
+      "learning_rate": 5.266666666666666e-05,
+      "loss": 0.1389,
+      "step": 8553
+    },
+    {
+      "epoch": 127.67,
+      "learning_rate": 5.263157894736841e-05,
+      "loss": 0.0003,
+      "step": 8554
+    },
+    {
+      "epoch": 127.68,
+      "learning_rate": 5.2596491228070166e-05,
+      "loss": 0.0417,
+      "step": 8555
+    },
+    {
+      "epoch": 127.7,
+      "learning_rate": 5.256140350877193e-05,
+      "loss": 0.0004,
+      "step": 8556
+    },
+    {
+      "epoch": 127.71,
+      "learning_rate": 5.2526315789473684e-05,
+      "loss": 0.0003,
+      "step": 8557
+    },
+    {
+      "epoch": 127.73,
+      "learning_rate": 5.249122807017544e-05,
+      "loss": 0.0003,
+      "step": 8558
+    },
+    {
+      "epoch": 127.74,
+      "learning_rate": 5.245614035087719e-05,
+      "loss": 0.0005,
+      "step": 8559
+    },
+    {
+      "epoch": 127.76,
+      "learning_rate": 5.2421052631578944e-05,
+      "loss": 0.0002,
+      "step": 8560
+    },
+    {
+      "epoch": 127.77,
+      "learning_rate": 5.23859649122807e-05,
+      "loss": 0.0005,
+      "step": 8561
+    },
+    {
+      "epoch": 127.79,
+      "learning_rate": 5.2350877192982455e-05,
+      "loss": 0.0003,
+      "step": 8562
+    },
+    {
+      "epoch": 127.8,
+      "learning_rate": 5.2315789473684204e-05,
+      "loss": 0.0007,
+      "step": 8563
+    },
+    {
+      "epoch": 127.82,
+      "learning_rate": 5.228070175438596e-05,
+      "loss": 0.0002,
+      "step": 8564
+    },
+    {
+      "epoch": 127.83,
+      "learning_rate": 5.2245614035087715e-05,
+      "loss": 0.0004,
+      "step": 8565
+    },
+    {
+      "epoch": 127.85,
+      "learning_rate": 5.2210526315789463e-05,
+      "loss": 0.0015,
+      "step": 8566
+    },
+    {
+      "epoch": 127.86,
+      "learning_rate": 5.217543859649122e-05,
+      "loss": 0.0009,
+      "step": 8567
+    },
+    {
+      "epoch": 127.88,
+      "learning_rate": 5.2140350877192975e-05,
+      "loss": 0.0003,
+      "step": 8568
+    },
+    {
+      "epoch": 127.89,
+      "learning_rate": 5.210526315789474e-05,
+      "loss": 0.3038,
+      "step": 8569
+    },
+    {
+      "epoch": 127.91,
+      "learning_rate": 5.207017543859649e-05,
+      "loss": 0.0003,
+      "step": 8570
+    },
+    {
+      "epoch": 127.92,
+      "learning_rate": 5.203508771929824e-05,
+      "loss": 0.0012,
+      "step": 8571
+    },
+    {
+      "epoch": 127.94,
+      "learning_rate": 5.2e-05,
+      "loss": 0.0003,
+      "step": 8572
+    },
+    {
+      "epoch": 127.95,
+      "learning_rate": 5.196491228070175e-05,
+      "loss": 0.0003,
+      "step": 8573
+    },
+    {
+      "epoch": 127.97,
+      "learning_rate": 5.192982456140351e-05,
+      "loss": 0.0003,
+      "step": 8574
+    },
+    {
+      "epoch": 127.98,
+      "learning_rate": 5.189473684210526e-05,
+      "loss": 0.1019,
+      "step": 8575
+    },
+    {
+      "epoch": 128.0,
+      "learning_rate": 5.185964912280701e-05,
+      "loss": 0.0004,
+      "step": 8576
+    },
+    {
+      "epoch": 128.01,
+      "learning_rate": 5.182456140350877e-05,
+      "loss": 0.0004,
+      "step": 8577
+    },
+    {
+      "epoch": 128.03,
+      "learning_rate": 5.1789473684210517e-05,
+      "loss": 0.0119,
+      "step": 8578
+    },
+    {
+      "epoch": 128.04,
+      "learning_rate": 5.175438596491227e-05,
+      "loss": 0.0004,
+      "step": 8579
+    },
+    {
+      "epoch": 128.06,
+      "learning_rate": 5.171929824561403e-05,
+      "loss": 0.0018,
+      "step": 8580
+    },
+    {
+      "epoch": 128.07,
+      "learning_rate": 5.168421052631579e-05,
+      "loss": 0.078,
+      "step": 8581
+    },
+    {
+      "epoch": 128.09,
+      "learning_rate": 5.1649122807017546e-05,
+      "loss": 0.0004,
+      "step": 8582
+    },
+    {
+      "epoch": 128.1,
+      "learning_rate": 5.1614035087719294e-05,
+      "loss": 0.0005,
+      "step": 8583
+    },
+    {
+      "epoch": 128.12,
+      "learning_rate": 5.157894736842105e-05,
+      "loss": 0.2706,
+      "step": 8584
+    },
+    {
+      "epoch": 128.13,
+      "learning_rate": 5.1543859649122805e-05,
+      "loss": 0.0003,
+      "step": 8585
+    },
+    {
+      "epoch": 128.15,
+      "learning_rate": 5.150877192982456e-05,
+      "loss": 0.0099,
+      "step": 8586
+    },
+    {
+      "epoch": 128.16,
+      "learning_rate": 5.147368421052631e-05,
+      "loss": 0.0003,
+      "step": 8587
+    },
+    {
+      "epoch": 128.18,
+      "learning_rate": 5.1438596491228065e-05,
+      "loss": 0.0438,
+      "step": 8588
+    },
+    {
+      "epoch": 128.19,
+      "learning_rate": 5.140350877192982e-05,
+      "loss": 0.0003,
+      "step": 8589
+    },
+    {
+      "epoch": 128.21,
+      "learning_rate": 5.136842105263157e-05,
+      "loss": 0.0004,
+      "step": 8590
+    },
+    {
+      "epoch": 128.22,
+      "learning_rate": 5.1333333333333325e-05,
+      "loss": 0.0004,
+      "step": 8591
+    },
+    {
+      "epoch": 128.24,
+      "learning_rate": 5.129824561403508e-05,
+      "loss": 0.0004,
+      "step": 8592
+    },
+    {
+      "epoch": 128.25,
+      "learning_rate": 5.1263157894736836e-05,
+      "loss": 0.0004,
+      "step": 8593
+    },
+    {
+      "epoch": 128.27,
+      "learning_rate": 5.12280701754386e-05,
+      "loss": 0.0684,
+      "step": 8594
+    },
+    {
+      "epoch": 128.28,
+      "learning_rate": 5.119298245614035e-05,
+      "loss": 0.001,
+      "step": 8595
+    },
+    {
+      "epoch": 128.3,
+      "learning_rate": 5.11578947368421e-05,
+      "loss": 0.0004,
+      "step": 8596
+    },
+    {
+      "epoch": 128.31,
+      "learning_rate": 5.112280701754386e-05,
+      "loss": 0.0004,
+      "step": 8597
+    },
+    {
+      "epoch": 128.33,
+      "learning_rate": 5.1087719298245614e-05,
+      "loss": 0.0007,
+      "step": 8598
+    },
+    {
+      "epoch": 128.34,
+      "learning_rate": 5.105263157894736e-05,
+      "loss": 0.0036,
+      "step": 8599
+    },
+    {
+      "epoch": 128.36,
+      "learning_rate": 5.101754385964912e-05,
+      "loss": 0.0005,
+      "step": 8600
+    },
+    {
+      "epoch": 128.36,
+      "eval_accuracy": 0.8761625061184533,
+      "eval_f1": 0.8776023752481238,
+      "eval_loss": 0.6961445808410645,
+      "eval_runtime": 345.4659,
+      "eval_samples_per_second": 11.828,
+      "eval_steps_per_second": 0.741,
+      "step": 8600
+    },
+    {
+      "epoch": 128.37,
+      "learning_rate": 5.0982456140350874e-05,
+      "loss": 0.0066,
+      "step": 8601
+    },
+    {
+      "epoch": 128.39,
+      "learning_rate": 5.094736842105262e-05,
+      "loss": 0.0003,
+      "step": 8602
+    },
+    {
+      "epoch": 128.4,
+      "learning_rate": 5.091228070175438e-05,
+      "loss": 0.0004,
+      "step": 8603
+    },
+    {
+      "epoch": 128.42,
+      "learning_rate": 5.0877192982456134e-05,
+      "loss": 0.0009,
+      "step": 8604
+    },
+    {
+      "epoch": 128.43,
+      "learning_rate": 5.084210526315789e-05,
+      "loss": 0.0017,
+      "step": 8605
+    },
+    {
+      "epoch": 128.45,
+      "learning_rate": 5.080701754385964e-05,
+      "loss": 0.0003,
+      "step": 8606
+    },
+    {
+      "epoch": 128.46,
+      "learning_rate": 5.07719298245614e-05,
+      "loss": 0.0008,
+      "step": 8607
+    },
+    {
+      "epoch": 128.48,
+      "learning_rate": 5.0736842105263156e-05,
+      "loss": 0.0005,
+      "step": 8608
+    },
+    {
+      "epoch": 128.49,
+      "learning_rate": 5.070175438596491e-05,
+      "loss": 0.0004,
+      "step": 8609
+    },
+    {
+      "epoch": 128.51,
+      "learning_rate": 5.066666666666666e-05,
+      "loss": 0.0004,
+      "step": 8610
+    },
+    {
+      "epoch": 128.52,
+      "learning_rate": 5.0631578947368416e-05,
+      "loss": 0.0007,
+      "step": 8611
+    },
+    {
+      "epoch": 128.54,
+      "learning_rate": 5.059649122807017e-05,
+      "loss": 0.0006,
+      "step": 8612
+    },
+    {
+      "epoch": 128.55,
+      "learning_rate": 5.056140350877193e-05,
+      "loss": 0.0003,
+      "step": 8613
+    },
+    {
+      "epoch": 128.57,
+      "learning_rate": 5.0526315789473676e-05,
+      "loss": 0.0003,
+      "step": 8614
+    },
+    {
+      "epoch": 128.58,
+      "learning_rate": 5.049122807017543e-05,
+      "loss": 0.0002,
+      "step": 8615
+    },
+    {
+      "epoch": 128.59,
+      "learning_rate": 5.045614035087719e-05,
+      "loss": 0.0004,
+      "step": 8616
+    },
+    {
+      "epoch": 128.61,
+      "learning_rate": 5.042105263157894e-05,
+      "loss": 0.0194,
+      "step": 8617
+    },
+    {
+      "epoch": 128.62,
+      "learning_rate": 5.038596491228069e-05,
+      "loss": 0.0003,
+      "step": 8618
+    },
+    {
+      "epoch": 128.64,
+      "learning_rate": 5.0350877192982454e-05,
+      "loss": 0.001,
+      "step": 8619
+    },
+    {
+      "epoch": 128.65,
+      "learning_rate": 5.031578947368421e-05,
+      "loss": 0.0004,
+      "step": 8620
+    },
+    {
+      "epoch": 128.67,
+      "learning_rate": 5.0280701754385965e-05,
+      "loss": 0.0152,
+      "step": 8621
+    },
+    {
+      "epoch": 128.68,
+      "learning_rate": 5.0245614035087714e-05,
+      "loss": 0.0003,
+      "step": 8622
+    },
+    {
+      "epoch": 128.7,
+      "learning_rate": 5.021052631578947e-05,
+      "loss": 0.0003,
+      "step": 8623
+    },
+    {
+      "epoch": 128.71,
+      "learning_rate": 5.0175438596491225e-05,
+      "loss": 0.0004,
+      "step": 8624
+    },
+    {
+      "epoch": 128.73,
+      "learning_rate": 5.014035087719298e-05,
+      "loss": 0.0003,
+      "step": 8625
+    },
+    {
+      "epoch": 128.74,
+      "learning_rate": 5.010526315789473e-05,
+      "loss": 0.0013,
+      "step": 8626
+    },
+    {
+      "epoch": 128.76,
+      "learning_rate": 5.0070175438596485e-05,
+      "loss": 0.0003,
+      "step": 8627
+    },
+    {
+      "epoch": 128.77,
+      "learning_rate": 5.003508771929824e-05,
+      "loss": 0.0002,
+      "step": 8628
+    },
+    {
+      "epoch": 128.79,
+      "learning_rate": 4.9999999999999996e-05,
+      "loss": 0.0003,
+      "step": 8629
+    },
+    {
+      "epoch": 128.8,
+      "learning_rate": 4.9964912280701744e-05,
+      "loss": 0.0003,
+      "step": 8630
+    },
+    {
+      "epoch": 128.82,
+      "learning_rate": 4.99298245614035e-05,
+      "loss": 0.0007,
+      "step": 8631
+    },
+    {
+      "epoch": 128.83,
+      "learning_rate": 4.989473684210526e-05,
+      "loss": 0.0075,
+      "step": 8632
+    },
+    {
+      "epoch": 128.85,
+      "learning_rate": 4.985964912280702e-05,
+      "loss": 0.0009,
+      "step": 8633
+    },
+    {
+      "epoch": 128.86,
+      "learning_rate": 4.982456140350877e-05,
+      "loss": 0.0003,
+      "step": 8634
+    },
+    {
+      "epoch": 128.88,
+      "learning_rate": 4.978947368421052e-05,
+      "loss": 0.0003,
+      "step": 8635
+    },
+    {
+      "epoch": 128.89,
+      "learning_rate": 4.975438596491228e-05,
+      "loss": 0.0004,
+      "step": 8636
+    },
+    {
+      "epoch": 128.91,
+      "learning_rate": 4.971929824561403e-05,
+      "loss": 0.0004,
+      "step": 8637
+    },
+    {
+      "epoch": 128.92,
+      "learning_rate": 4.968421052631578e-05,
+      "loss": 0.0004,
+      "step": 8638
+    },
+    {
+      "epoch": 128.94,
+      "learning_rate": 4.964912280701754e-05,
+      "loss": 0.0003,
+      "step": 8639
+    },
+    {
+      "epoch": 128.95,
+      "learning_rate": 4.961403508771929e-05,
+      "loss": 0.0004,
+      "step": 8640
+    },
+    {
+      "epoch": 128.97,
+      "learning_rate": 4.957894736842105e-05,
+      "loss": 0.0004,
+      "step": 8641
+    },
+    {
+      "epoch": 128.98,
+      "learning_rate": 4.95438596491228e-05,
+      "loss": 0.0004,
+      "step": 8642
+    },
+    {
+      "epoch": 129.0,
+      "learning_rate": 4.950877192982455e-05,
+      "loss": 0.0004,
+      "step": 8643
+    },
+    {
+      "epoch": 129.01,
+      "learning_rate": 4.9473684210526315e-05,
+      "loss": 0.0003,
+      "step": 8644
+    },
+    {
+      "epoch": 129.03,
+      "learning_rate": 4.943859649122807e-05,
+      "loss": 0.0003,
+      "step": 8645
+    },
+    {
+      "epoch": 129.04,
+      "learning_rate": 4.940350877192982e-05,
+      "loss": 0.0002,
+      "step": 8646
+    },
+    {
+      "epoch": 129.06,
+      "learning_rate": 4.9368421052631575e-05,
+      "loss": 0.0003,
+      "step": 8647
+    },
+    {
+      "epoch": 129.07,
+      "learning_rate": 4.933333333333333e-05,
+      "loss": 0.0957,
+      "step": 8648
+    },
+    {
+      "epoch": 129.09,
+      "learning_rate": 4.9298245614035086e-05,
+      "loss": 0.0003,
+      "step": 8649
+    },
+    {
+      "epoch": 129.1,
+      "learning_rate": 4.9263157894736835e-05,
+      "loss": 0.0003,
+      "step": 8650
+    },
+    {
+      "epoch": 129.12,
+      "learning_rate": 4.922807017543859e-05,
+      "loss": 0.0009,
+      "step": 8651
+    },
+    {
+      "epoch": 129.13,
+      "learning_rate": 4.9192982456140346e-05,
+      "loss": 0.0003,
+      "step": 8652
+    },
+    {
+      "epoch": 129.15,
+      "learning_rate": 4.91578947368421e-05,
+      "loss": 0.0002,
+      "step": 8653
+    },
+    {
+      "epoch": 129.16,
+      "learning_rate": 4.912280701754385e-05,
+      "loss": 0.0002,
+      "step": 8654
+    },
+    {
+      "epoch": 129.18,
+      "learning_rate": 4.9087719298245606e-05,
+      "loss": 0.0002,
+      "step": 8655
+    },
+    {
+      "epoch": 129.19,
+      "learning_rate": 4.905263157894736e-05,
+      "loss": 0.0004,
+      "step": 8656
+    },
+    {
+      "epoch": 129.21,
+      "learning_rate": 4.9017543859649124e-05,
+      "loss": 0.0003,
+      "step": 8657
+    },
+    {
+      "epoch": 129.22,
+      "learning_rate": 4.898245614035087e-05,
+      "loss": 0.0003,
+      "step": 8658
+    },
+    {
+      "epoch": 129.24,
+      "learning_rate": 4.894736842105263e-05,
+      "loss": 0.0002,
+      "step": 8659
+    },
+    {
+      "epoch": 129.25,
+      "learning_rate": 4.8912280701754384e-05,
+      "loss": 0.0003,
+      "step": 8660
+    },
+    {
+      "epoch": 129.27,
+      "learning_rate": 4.887719298245614e-05,
+      "loss": 0.0005,
+      "step": 8661
+    },
+    {
+      "epoch": 129.28,
+      "learning_rate": 4.884210526315789e-05,
+      "loss": 0.1397,
+      "step": 8662
+    },
+    {
+      "epoch": 129.3,
+      "learning_rate": 4.8807017543859644e-05,
+      "loss": 0.0002,
+      "step": 8663
+    },
+    {
+      "epoch": 129.31,
+      "learning_rate": 4.87719298245614e-05,
+      "loss": 0.001,
+      "step": 8664
+    },
+    {
+      "epoch": 129.33,
+      "learning_rate": 4.8736842105263155e-05,
+      "loss": 0.0003,
+      "step": 8665
+    },
+    {
+      "epoch": 129.34,
+      "learning_rate": 4.8701754385964904e-05,
+      "loss": 0.0003,
+      "step": 8666
+    },
+    {
+      "epoch": 129.36,
+      "learning_rate": 4.866666666666666e-05,
+      "loss": 0.0002,
+      "step": 8667
+    },
+    {
+      "epoch": 129.37,
+      "learning_rate": 4.8631578947368415e-05,
+      "loss": 0.0002,
+      "step": 8668
+    },
+    {
+      "epoch": 129.39,
+      "learning_rate": 4.859649122807018e-05,
+      "loss": 0.0004,
+      "step": 8669
+    },
+    {
+      "epoch": 129.4,
+      "learning_rate": 4.8561403508771926e-05,
+      "loss": 0.0003,
+      "step": 8670
+    },
+    {
+      "epoch": 129.42,
+      "learning_rate": 4.852631578947368e-05,
+      "loss": 0.0003,
+      "step": 8671
+    },
+    {
+      "epoch": 129.43,
+      "learning_rate": 4.849122807017544e-05,
+      "loss": 0.0003,
+      "step": 8672
+    },
+    {
+      "epoch": 129.45,
+      "learning_rate": 4.845614035087719e-05,
+      "loss": 0.0002,
+      "step": 8673
+    },
+    {
+      "epoch": 129.46,
+      "learning_rate": 4.842105263157894e-05,
+      "loss": 0.0004,
+      "step": 8674
+    },
+    {
+      "epoch": 129.48,
+      "learning_rate": 4.83859649122807e-05,
+      "loss": 0.0006,
+      "step": 8675
+    },
+    {
+      "epoch": 129.49,
+      "learning_rate": 4.835087719298245e-05,
+      "loss": 0.0077,
+      "step": 8676
+    },
+    {
+      "epoch": 129.51,
+      "learning_rate": 4.831578947368421e-05,
+      "loss": 0.0002,
+      "step": 8677
+    },
+    {
+      "epoch": 129.52,
+      "learning_rate": 4.828070175438596e-05,
+      "loss": 0.0004,
+      "step": 8678
+    },
+    {
+      "epoch": 129.54,
+      "learning_rate": 4.824561403508771e-05,
+      "loss": 0.0003,
+      "step": 8679
+    },
+    {
+      "epoch": 129.55,
+      "learning_rate": 4.821052631578947e-05,
+      "loss": 0.0004,
+      "step": 8680
+    },
+    {
+      "epoch": 129.57,
+      "learning_rate": 4.817543859649122e-05,
+      "loss": 0.0004,
+      "step": 8681
+    },
+    {
+      "epoch": 129.58,
+      "learning_rate": 4.814035087719298e-05,
+      "loss": 0.0004,
+      "step": 8682
+    },
+    {
+      "epoch": 129.59,
+      "learning_rate": 4.8105263157894735e-05,
+      "loss": 0.1145,
+      "step": 8683
+    },
+    {
+      "epoch": 129.61,
+      "learning_rate": 4.807017543859649e-05,
+      "loss": 0.0003,
+      "step": 8684
+    },
+    {
+      "epoch": 129.62,
+      "learning_rate": 4.8035087719298246e-05,
+      "loss": 0.1198,
+      "step": 8685
+    },
+    {
+      "epoch": 129.64,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 0.0003,
+      "step": 8686
+    },
+    {
+      "epoch": 129.65,
+      "learning_rate": 4.796491228070175e-05,
+      "loss": 0.0002,
+      "step": 8687
+    },
+    {
+      "epoch": 129.67,
+      "learning_rate": 4.7929824561403506e-05,
+      "loss": 0.0003,
+      "step": 8688
+    },
+    {
+      "epoch": 129.68,
+      "learning_rate": 4.789473684210526e-05,
+      "loss": 0.0002,
+      "step": 8689
+    },
+    {
+      "epoch": 129.7,
+      "learning_rate": 4.785964912280701e-05,
+      "loss": 0.0005,
+      "step": 8690
+    },
+    {
+      "epoch": 129.71,
+      "learning_rate": 4.7824561403508765e-05,
+      "loss": 0.0006,
+      "step": 8691
+    },
+    {
+      "epoch": 129.73,
+      "learning_rate": 4.778947368421052e-05,
+      "loss": 0.0003,
+      "step": 8692
+    },
+    {
+      "epoch": 129.74,
+      "learning_rate": 4.775438596491227e-05,
+      "loss": 0.0003,
+      "step": 8693
+    },
+    {
+      "epoch": 129.76,
+      "learning_rate": 4.771929824561403e-05,
+      "loss": 0.0003,
+      "step": 8694
+    },
+    {
+      "epoch": 129.77,
+      "learning_rate": 4.768421052631579e-05,
+      "loss": 0.0003,
+      "step": 8695
+    },
+    {
+      "epoch": 129.79,
+      "learning_rate": 4.764912280701754e-05,
+      "loss": 0.0003,
+      "step": 8696
+    },
+    {
+      "epoch": 129.8,
+      "learning_rate": 4.76140350877193e-05,
+      "loss": 0.03,
+      "step": 8697
+    },
+    {
+      "epoch": 129.82,
+      "learning_rate": 4.757894736842105e-05,
+      "loss": 0.0003,
+      "step": 8698
+    },
+    {
+      "epoch": 129.83,
+      "learning_rate": 4.75438596491228e-05,
+      "loss": 0.0003,
+      "step": 8699
+    },
+    {
+      "epoch": 129.85,
+      "learning_rate": 4.750877192982456e-05,
+      "loss": 0.0003,
+      "step": 8700
+    },
+    {
+      "epoch": 129.86,
+      "learning_rate": 4.7473684210526314e-05,
+      "loss": 0.0004,
+      "step": 8701
+    },
+    {
+      "epoch": 129.88,
+      "learning_rate": 4.743859649122806e-05,
+      "loss": 0.0003,
+      "step": 8702
+    },
+    {
+      "epoch": 129.89,
+      "learning_rate": 4.740350877192982e-05,
+      "loss": 0.0002,
+      "step": 8703
+    },
+    {
+      "epoch": 129.91,
+      "learning_rate": 4.7368421052631574e-05,
+      "loss": 0.0003,
+      "step": 8704
+    },
+    {
+      "epoch": 129.92,
+      "learning_rate": 4.733333333333332e-05,
+      "loss": 0.0004,
+      "step": 8705
+    },
+    {
+      "epoch": 129.94,
+      "learning_rate": 4.729824561403508e-05,
+      "loss": 0.0012,
+      "step": 8706
+    },
+    {
+      "epoch": 129.95,
+      "learning_rate": 4.726315789473684e-05,
+      "loss": 0.0003,
+      "step": 8707
+    },
+    {
+      "epoch": 129.97,
+      "learning_rate": 4.7228070175438596e-05,
+      "loss": 0.0024,
+      "step": 8708
+    },
+    {
+      "epoch": 129.98,
+      "learning_rate": 4.719298245614035e-05,
+      "loss": 0.0003,
+      "step": 8709
+    },
+    {
+      "epoch": 130.0,
+      "learning_rate": 4.71578947368421e-05,
+      "loss": 0.0003,
+      "step": 8710
+    },
+    {
+      "epoch": 130.01,
+      "learning_rate": 4.7122807017543856e-05,
+      "loss": 0.0006,
+      "step": 8711
+    },
+    {
+      "epoch": 130.03,
+      "learning_rate": 4.708771929824561e-05,
+      "loss": 0.0012,
+      "step": 8712
+    },
+    {
+      "epoch": 130.04,
+      "learning_rate": 4.705263157894737e-05,
+      "loss": 0.0002,
+      "step": 8713
+    },
+    {
+      "epoch": 130.06,
+      "learning_rate": 4.7017543859649116e-05,
+      "loss": 0.0005,
+      "step": 8714
+    },
+    {
+      "epoch": 130.07,
+      "learning_rate": 4.698245614035087e-05,
+      "loss": 0.0003,
+      "step": 8715
+    },
+    {
+      "epoch": 130.09,
+      "learning_rate": 4.694736842105263e-05,
+      "loss": 0.0003,
+      "step": 8716
+    },
+    {
+      "epoch": 130.1,
+      "learning_rate": 4.6912280701754376e-05,
+      "loss": 0.0004,
+      "step": 8717
+    },
+    {
+      "epoch": 130.12,
+      "learning_rate": 4.687719298245613e-05,
+      "loss": 0.0058,
+      "step": 8718
+    },
+    {
+      "epoch": 130.13,
+      "learning_rate": 4.6842105263157894e-05,
+      "loss": 0.0002,
+      "step": 8719
+    },
+    {
+      "epoch": 130.15,
+      "learning_rate": 4.680701754385965e-05,
+      "loss": 0.0009,
+      "step": 8720
+    },
+    {
+      "epoch": 130.16,
+      "learning_rate": 4.6771929824561405e-05,
+      "loss": 0.0003,
+      "step": 8721
+    },
+    {
+      "epoch": 130.18,
+      "learning_rate": 4.6736842105263154e-05,
+      "loss": 0.0002,
+      "step": 8722
+    },
+    {
+      "epoch": 130.19,
+      "learning_rate": 4.670175438596491e-05,
+      "loss": 0.1649,
+      "step": 8723
+    },
+    {
+      "epoch": 130.21,
+      "learning_rate": 4.6666666666666665e-05,
+      "loss": 0.0003,
+      "step": 8724
+    },
+    {
+      "epoch": 130.22,
+      "learning_rate": 4.663157894736842e-05,
+      "loss": 0.0006,
+      "step": 8725
+    },
+    {
+      "epoch": 130.24,
+      "learning_rate": 4.659649122807017e-05,
+      "loss": 0.0003,
+      "step": 8726
+    },
+    {
+      "epoch": 130.25,
+      "learning_rate": 4.6561403508771925e-05,
+      "loss": 0.0003,
+      "step": 8727
+    },
+    {
+      "epoch": 130.27,
+      "learning_rate": 4.652631578947368e-05,
+      "loss": 0.0002,
+      "step": 8728
+    },
+    {
+      "epoch": 130.28,
+      "learning_rate": 4.649122807017543e-05,
+      "loss": 0.0008,
+      "step": 8729
+    },
+    {
+      "epoch": 130.3,
+      "learning_rate": 4.6456140350877185e-05,
+      "loss": 0.0003,
+      "step": 8730
+    },
+    {
+      "epoch": 130.31,
+      "learning_rate": 4.642105263157894e-05,
+      "loss": 0.0005,
+      "step": 8731
+    },
+    {
+      "epoch": 130.33,
+      "learning_rate": 4.63859649122807e-05,
+      "loss": 0.0003,
+      "step": 8732
+    },
+    {
+      "epoch": 130.34,
+      "learning_rate": 4.635087719298246e-05,
+      "loss": 0.0004,
+      "step": 8733
+    },
+    {
+      "epoch": 130.36,
+      "learning_rate": 4.631578947368421e-05,
+      "loss": 0.2316,
+      "step": 8734
+    },
+    {
+      "epoch": 130.37,
+      "learning_rate": 4.628070175438596e-05,
+      "loss": 0.0004,
+      "step": 8735
+    },
+    {
+      "epoch": 130.39,
+      "learning_rate": 4.624561403508772e-05,
+      "loss": 0.0003,
+      "step": 8736
+    },
+    {
+      "epoch": 130.4,
+      "learning_rate": 4.621052631578947e-05,
+      "loss": 0.0002,
+      "step": 8737
+    },
+    {
+      "epoch": 130.42,
+      "learning_rate": 4.617543859649122e-05,
+      "loss": 0.0002,
+      "step": 8738
+    },
+    {
+      "epoch": 130.43,
+      "learning_rate": 4.614035087719298e-05,
+      "loss": 0.0881,
+      "step": 8739
+    },
+    {
+      "epoch": 130.45,
+      "learning_rate": 4.610526315789473e-05,
+      "loss": 0.0137,
+      "step": 8740
+    },
+    {
+      "epoch": 130.46,
+      "learning_rate": 4.607017543859648e-05,
+      "loss": 0.0003,
+      "step": 8741
+    },
+    {
+      "epoch": 130.48,
+      "learning_rate": 4.603508771929824e-05,
+      "loss": 0.0003,
+      "step": 8742
+    },
+    {
+      "epoch": 130.49,
+      "learning_rate": 4.599999999999999e-05,
+      "loss": 0.0003,
+      "step": 8743
+    },
+    {
+      "epoch": 130.51,
+      "learning_rate": 4.5964912280701756e-05,
+      "loss": 0.0003,
+      "step": 8744
+    },
+    {
+      "epoch": 130.52,
+      "learning_rate": 4.592982456140351e-05,
+      "loss": 0.0003,
+      "step": 8745
+    },
+    {
+      "epoch": 130.54,
+      "learning_rate": 4.589473684210526e-05,
+      "loss": 0.0013,
+      "step": 8746
+    },
+    {
+      "epoch": 130.55,
+      "learning_rate": 4.5859649122807015e-05,
+      "loss": 0.0004,
+      "step": 8747
+    },
+    {
+      "epoch": 130.57,
+      "learning_rate": 4.582456140350877e-05,
+      "loss": 0.0003,
+      "step": 8748
+    },
+    {
+      "epoch": 130.58,
+      "learning_rate": 4.578947368421052e-05,
+      "loss": 0.0011,
+      "step": 8749
+    },
+    {
+      "epoch": 130.59,
+      "learning_rate": 4.5754385964912275e-05,
+      "loss": 0.0002,
+      "step": 8750
+    },
+    {
+      "epoch": 130.61,
+      "learning_rate": 4.571929824561403e-05,
+      "loss": 0.0004,
+      "step": 8751
+    },
+    {
+      "epoch": 130.62,
+      "learning_rate": 4.5684210526315786e-05,
+      "loss": 0.0002,
+      "step": 8752
+    },
+    {
+      "epoch": 130.64,
+      "learning_rate": 4.5649122807017535e-05,
+      "loss": 0.0005,
+      "step": 8753
+    },
+    {
+      "epoch": 130.65,
+      "learning_rate": 4.561403508771929e-05,
+      "loss": 0.0003,
+      "step": 8754
+    },
+    {
+      "epoch": 130.67,
+      "learning_rate": 4.5578947368421046e-05,
+      "loss": 0.0003,
+      "step": 8755
+    },
+    {
+      "epoch": 130.68,
+      "learning_rate": 4.55438596491228e-05,
+      "loss": 0.0003,
+      "step": 8756
+    },
+    {
+      "epoch": 130.7,
+      "learning_rate": 4.5508771929824564e-05,
+      "loss": 0.0004,
+      "step": 8757
+    },
+    {
+      "epoch": 130.71,
+      "learning_rate": 4.547368421052631e-05,
+      "loss": 0.0004,
+      "step": 8758
+    },
+    {
+      "epoch": 130.73,
+      "learning_rate": 4.543859649122807e-05,
+      "loss": 0.0003,
+      "step": 8759
+    },
+    {
+      "epoch": 130.74,
+      "learning_rate": 4.5403508771929824e-05,
+      "loss": 0.0366,
+      "step": 8760
+    },
+    {
+      "epoch": 130.76,
+      "learning_rate": 4.536842105263157e-05,
+      "loss": 0.0002,
+      "step": 8761
+    },
+    {
+      "epoch": 130.77,
+      "learning_rate": 4.533333333333333e-05,
+      "loss": 0.0003,
+      "step": 8762
+    },
+    {
+      "epoch": 130.79,
+      "learning_rate": 4.5298245614035084e-05,
+      "loss": 0.0003,
+      "step": 8763
+    },
+    {
+      "epoch": 130.8,
+      "learning_rate": 4.526315789473684e-05,
+      "loss": 0.0003,
+      "step": 8764
+    },
+    {
+      "epoch": 130.82,
+      "learning_rate": 4.522807017543859e-05,
+      "loss": 0.0002,
+      "step": 8765
+    },
+    {
+      "epoch": 130.83,
+      "learning_rate": 4.5192982456140344e-05,
+      "loss": 0.0002,
+      "step": 8766
+    },
+    {
+      "epoch": 130.85,
+      "learning_rate": 4.51578947368421e-05,
+      "loss": 0.0065,
+      "step": 8767
+    },
+    {
+      "epoch": 130.86,
+      "learning_rate": 4.5122807017543855e-05,
+      "loss": 0.0002,
+      "step": 8768
+    },
+    {
+      "epoch": 130.88,
+      "learning_rate": 4.508771929824562e-05,
+      "loss": 0.0002,
+      "step": 8769
+    },
+    {
+      "epoch": 130.89,
+      "learning_rate": 4.5052631578947366e-05,
+      "loss": 0.0002,
+      "step": 8770
+    },
+    {
+      "epoch": 130.91,
+      "learning_rate": 4.501754385964912e-05,
+      "loss": 0.0002,
+      "step": 8771
+    },
+    {
+      "epoch": 130.92,
+      "learning_rate": 4.498245614035088e-05,
+      "loss": 0.0002,
+      "step": 8772
+    },
+    {
+      "epoch": 130.94,
+      "learning_rate": 4.4947368421052626e-05,
+      "loss": 0.0002,
+      "step": 8773
+    },
+    {
+      "epoch": 130.95,
+      "learning_rate": 4.491228070175438e-05,
+      "loss": 0.0003,
+      "step": 8774
+    },
+    {
+      "epoch": 130.97,
+      "learning_rate": 4.487719298245614e-05,
+      "loss": 0.0003,
+      "step": 8775
+    },
+    {
+      "epoch": 130.98,
+      "learning_rate": 4.484210526315789e-05,
+      "loss": 0.0007,
+      "step": 8776
+    },
+    {
+      "epoch": 131.0,
+      "learning_rate": 4.480701754385964e-05,
+      "loss": 0.0003,
+      "step": 8777
+    },
+    {
+      "epoch": 131.01,
+      "learning_rate": 4.47719298245614e-05,
+      "loss": 0.0003,
+      "step": 8778
+    },
+    {
+      "epoch": 131.03,
+      "learning_rate": 4.473684210526315e-05,
+      "loss": 0.0042,
+      "step": 8779
+    },
+    {
+      "epoch": 131.04,
+      "learning_rate": 4.470175438596491e-05,
+      "loss": 0.0002,
+      "step": 8780
+    },
+    {
+      "epoch": 131.06,
+      "learning_rate": 4.466666666666666e-05,
+      "loss": 0.0003,
+      "step": 8781
+    },
+    {
+      "epoch": 131.07,
+      "learning_rate": 4.463157894736842e-05,
+      "loss": 0.0002,
+      "step": 8782
+    },
+    {
+      "epoch": 131.09,
+      "learning_rate": 4.4596491228070175e-05,
+      "loss": 0.0002,
+      "step": 8783
+    },
+    {
+      "epoch": 131.1,
+      "learning_rate": 4.456140350877193e-05,
+      "loss": 0.0003,
+      "step": 8784
+    },
+    {
+      "epoch": 131.12,
+      "learning_rate": 4.452631578947368e-05,
+      "loss": 0.0004,
+      "step": 8785
+    },
+    {
+      "epoch": 131.13,
+      "learning_rate": 4.4491228070175435e-05,
+      "loss": 0.0002,
+      "step": 8786
+    },
+    {
+      "epoch": 131.15,
+      "learning_rate": 4.445614035087719e-05,
+      "loss": 0.0003,
+      "step": 8787
+    },
+    {
+      "epoch": 131.16,
+      "learning_rate": 4.4421052631578946e-05,
+      "loss": 0.0003,
+      "step": 8788
+    },
+    {
+      "epoch": 131.18,
+      "learning_rate": 4.4385964912280695e-05,
+      "loss": 0.0003,
+      "step": 8789
+    },
+    {
+      "epoch": 131.19,
+      "learning_rate": 4.435087719298245e-05,
+      "loss": 0.0013,
+      "step": 8790
+    },
+    {
+      "epoch": 131.21,
+      "learning_rate": 4.4315789473684206e-05,
+      "loss": 0.0467,
+      "step": 8791
+    },
+    {
+      "epoch": 131.22,
+      "learning_rate": 4.428070175438596e-05,
+      "loss": 0.0002,
+      "step": 8792
+    },
+    {
+      "epoch": 131.24,
+      "learning_rate": 4.424561403508771e-05,
+      "loss": 0.0002,
+      "step": 8793
+    },
+    {
+      "epoch": 131.25,
+      "learning_rate": 4.4210526315789466e-05,
+      "loss": 0.0003,
+      "step": 8794
+    },
+    {
+      "epoch": 131.27,
+      "learning_rate": 4.417543859649123e-05,
+      "loss": 0.0003,
+      "step": 8795
+    },
+    {
+      "epoch": 131.28,
+      "learning_rate": 4.4140350877192983e-05,
+      "loss": 0.0009,
+      "step": 8796
+    },
+    {
+      "epoch": 131.3,
+      "learning_rate": 4.410526315789473e-05,
+      "loss": 0.0005,
+      "step": 8797
+    },
+    {
+      "epoch": 131.31,
+      "learning_rate": 4.407017543859649e-05,
+      "loss": 0.0003,
+      "step": 8798
+    },
+    {
+      "epoch": 131.33,
+      "learning_rate": 4.403508771929824e-05,
+      "loss": 0.0019,
+      "step": 8799
+    },
+    {
+      "epoch": 131.34,
+      "learning_rate": 4.4e-05,
+      "loss": 0.0003,
+      "step": 8800
+    },
+    {
+      "epoch": 131.34,
+      "eval_accuracy": 0.8881546744982868,
+      "eval_f1": 0.887829841036129,
+      "eval_loss": 0.5989698171615601,
+      "eval_runtime": 343.81,
+      "eval_samples_per_second": 11.884,
+      "eval_steps_per_second": 0.745,
+      "step": 8800
+    },
+    {
+      "epoch": 131.36,
+      "learning_rate": 4.396491228070175e-05,
+      "loss": 0.0003,
+      "step": 8801
+    },
+    {
+      "epoch": 131.37,
+      "learning_rate": 4.39298245614035e-05,
+      "loss": 0.0002,
+      "step": 8802
+    },
+    {
+      "epoch": 131.39,
+      "learning_rate": 4.389473684210526e-05,
+      "loss": 0.0002,
+      "step": 8803
+    },
+    {
+      "epoch": 131.4,
+      "learning_rate": 4.3859649122807014e-05,
+      "loss": 0.079,
+      "step": 8804
+    },
+    {
+      "epoch": 131.42,
+      "learning_rate": 4.382456140350876e-05,
+      "loss": 0.0002,
+      "step": 8805
+    },
+    {
+      "epoch": 131.43,
+      "learning_rate": 4.378947368421052e-05,
+      "loss": 0.0009,
+      "step": 8806
+    },
+    {
+      "epoch": 131.45,
+      "learning_rate": 4.375438596491228e-05,
+      "loss": 0.0003,
+      "step": 8807
+    },
+    {
+      "epoch": 131.46,
+      "learning_rate": 4.3719298245614037e-05,
+      "loss": 0.0005,
+      "step": 8808
+    },
+    {
+      "epoch": 131.48,
+      "learning_rate": 4.3684210526315785e-05,
+      "loss": 0.0002,
+      "step": 8809
+    },
+    {
+      "epoch": 131.49,
+      "learning_rate": 4.364912280701754e-05,
+      "loss": 0.0003,
+      "step": 8810
+    },
+    {
+      "epoch": 131.51,
+      "learning_rate": 4.3614035087719296e-05,
+      "loss": 0.003,
+      "step": 8811
+    },
+    {
+      "epoch": 131.52,
+      "learning_rate": 4.357894736842105e-05,
+      "loss": 0.0004,
+      "step": 8812
+    },
+    {
+      "epoch": 131.54,
+      "learning_rate": 4.35438596491228e-05,
+      "loss": 0.0002,
+      "step": 8813
+    },
+    {
+      "epoch": 131.55,
+      "learning_rate": 4.3508771929824556e-05,
+      "loss": 0.0002,
+      "step": 8814
+    },
+    {
+      "epoch": 131.57,
+      "learning_rate": 4.347368421052631e-05,
+      "loss": 0.0003,
+      "step": 8815
+    },
+    {
+      "epoch": 131.58,
+      "learning_rate": 4.343859649122807e-05,
+      "loss": 0.0003,
+      "step": 8816
+    },
+    {
+      "epoch": 131.59,
+      "learning_rate": 4.3403508771929816e-05,
+      "loss": 0.0004,
+      "step": 8817
+    },
+    {
+      "epoch": 131.61,
+      "learning_rate": 4.336842105263157e-05,
+      "loss": 0.0003,
+      "step": 8818
+    },
+    {
+      "epoch": 131.62,
+      "learning_rate": 4.333333333333333e-05,
+      "loss": 0.0003,
+      "step": 8819
+    },
+    {
+      "epoch": 131.64,
+      "learning_rate": 4.329824561403509e-05,
+      "loss": 0.0004,
+      "step": 8820
+    },
+    {
+      "epoch": 131.65,
+      "learning_rate": 4.326315789473684e-05,
+      "loss": 0.0003,
+      "step": 8821
+    },
+    {
+      "epoch": 131.67,
+      "learning_rate": 4.3228070175438594e-05,
+      "loss": 0.0002,
+      "step": 8822
+    },
+    {
+      "epoch": 131.68,
+      "learning_rate": 4.319298245614035e-05,
+      "loss": 0.009,
+      "step": 8823
+    },
+    {
+      "epoch": 131.7,
+      "learning_rate": 4.3157894736842105e-05,
+      "loss": 0.0358,
+      "step": 8824
+    },
+    {
+      "epoch": 131.71,
+      "learning_rate": 4.3122807017543854e-05,
+      "loss": 0.0002,
+      "step": 8825
+    },
+    {
+      "epoch": 131.73,
+      "learning_rate": 4.308771929824561e-05,
+      "loss": 0.0003,
+      "step": 8826
+    },
+    {
+      "epoch": 131.74,
+      "learning_rate": 4.3052631578947365e-05,
+      "loss": 0.1321,
+      "step": 8827
+    },
+    {
+      "epoch": 131.76,
+      "learning_rate": 4.301754385964912e-05,
+      "loss": 0.0002,
+      "step": 8828
+    },
+    {
+      "epoch": 131.77,
+      "learning_rate": 4.298245614035087e-05,
+      "loss": 0.0002,
+      "step": 8829
+    },
+    {
+      "epoch": 131.79,
+      "learning_rate": 4.2947368421052625e-05,
+      "loss": 0.0002,
+      "step": 8830
+    },
+    {
+      "epoch": 131.8,
+      "learning_rate": 4.291228070175438e-05,
+      "loss": 0.0002,
+      "step": 8831
+    },
+    {
+      "epoch": 131.82,
+      "learning_rate": 4.287719298245614e-05,
+      "loss": 0.0039,
+      "step": 8832
+    },
+    {
+      "epoch": 131.83,
+      "learning_rate": 4.284210526315789e-05,
+      "loss": 0.0006,
+      "step": 8833
+    },
+    {
+      "epoch": 131.85,
+      "learning_rate": 4.280701754385965e-05,
+      "loss": 0.0003,
+      "step": 8834
+    },
+    {
+      "epoch": 131.86,
+      "learning_rate": 4.27719298245614e-05,
+      "loss": 0.0007,
+      "step": 8835
+    },
+    {
+      "epoch": 131.88,
+      "learning_rate": 4.273684210526316e-05,
+      "loss": 0.0002,
+      "step": 8836
+    },
+    {
+      "epoch": 131.89,
+      "learning_rate": 4.270175438596491e-05,
+      "loss": 0.0002,
+      "step": 8837
+    },
+    {
+      "epoch": 131.91,
+      "learning_rate": 4.266666666666666e-05,
+      "loss": 0.0003,
+      "step": 8838
+    },
+    {
+      "epoch": 131.92,
+      "learning_rate": 4.263157894736842e-05,
+      "loss": 0.0003,
+      "step": 8839
+    },
+    {
+      "epoch": 131.94,
+      "learning_rate": 4.2596491228070174e-05,
+      "loss": 0.0007,
+      "step": 8840
+    },
+    {
+      "epoch": 131.95,
+      "learning_rate": 4.256140350877192e-05,
+      "loss": 0.0003,
+      "step": 8841
+    },
+    {
+      "epoch": 131.97,
+      "learning_rate": 4.252631578947368e-05,
+      "loss": 0.0002,
+      "step": 8842
+    },
+    {
+      "epoch": 131.98,
+      "learning_rate": 4.2491228070175433e-05,
+      "loss": 0.0002,
+      "step": 8843
+    },
+    {
+      "epoch": 132.0,
+      "learning_rate": 4.245614035087718e-05,
+      "loss": 0.0003,
+      "step": 8844
+    },
+    {
+      "epoch": 132.01,
+      "learning_rate": 4.2421052631578945e-05,
+      "loss": 0.0459,
+      "step": 8845
+    },
+    {
+      "epoch": 132.03,
+      "learning_rate": 4.23859649122807e-05,
+      "loss": 0.0036,
+      "step": 8846
+    },
+    {
+      "epoch": 132.04,
+      "learning_rate": 4.2350877192982456e-05,
+      "loss": 0.2391,
+      "step": 8847
+    },
+    {
+      "epoch": 132.06,
+      "learning_rate": 4.231578947368421e-05,
+      "loss": 0.0002,
+      "step": 8848
+    },
+    {
+      "epoch": 132.07,
+      "learning_rate": 4.228070175438596e-05,
+      "loss": 0.0002,
+      "step": 8849
+    },
+    {
+      "epoch": 132.09,
+      "learning_rate": 4.2245614035087716e-05,
+      "loss": 0.0002,
+      "step": 8850
+    },
+    {
+      "epoch": 132.1,
+      "learning_rate": 4.221052631578947e-05,
+      "loss": 0.0003,
+      "step": 8851
+    },
+    {
+      "epoch": 132.12,
+      "learning_rate": 4.217543859649123e-05,
+      "loss": 0.0002,
+      "step": 8852
+    },
+    {
+      "epoch": 132.13,
+      "learning_rate": 4.2140350877192975e-05,
+      "loss": 0.0003,
+      "step": 8853
+    },
+    {
+      "epoch": 132.15,
+      "learning_rate": 4.210526315789473e-05,
+      "loss": 0.0002,
+      "step": 8854
+    },
+    {
+      "epoch": 132.16,
+      "learning_rate": 4.2070175438596487e-05,
+      "loss": 0.0003,
+      "step": 8855
+    },
+    {
+      "epoch": 132.18,
+      "learning_rate": 4.2035087719298235e-05,
+      "loss": 0.0002,
+      "step": 8856
+    },
+    {
+      "epoch": 132.19,
+      "learning_rate": 4.2e-05,
+      "loss": 0.0002,
+      "step": 8857
+    },
+    {
+      "epoch": 132.21,
+      "learning_rate": 4.196491228070175e-05,
+      "loss": 0.0002,
+      "step": 8858
+    },
+    {
+      "epoch": 132.22,
+      "learning_rate": 4.192982456140351e-05,
+      "loss": 0.0742,
+      "step": 8859
+    },
+    {
+      "epoch": 132.24,
+      "learning_rate": 4.1894736842105264e-05,
+      "loss": 0.0009,
+      "step": 8860
+    },
+    {
+      "epoch": 132.25,
+      "learning_rate": 4.185964912280701e-05,
+      "loss": 0.0268,
+      "step": 8861
+    },
+    {
+      "epoch": 132.27,
+      "learning_rate": 4.182456140350877e-05,
+      "loss": 0.0003,
+      "step": 8862
+    },
+    {
+      "epoch": 132.28,
+      "learning_rate": 4.1789473684210524e-05,
+      "loss": 0.0003,
+      "step": 8863
+    },
+    {
+      "epoch": 132.3,
+      "learning_rate": 4.175438596491228e-05,
+      "loss": 0.0065,
+      "step": 8864
+    },
+    {
+      "epoch": 132.31,
+      "learning_rate": 4.171929824561403e-05,
+      "loss": 0.0002,
+      "step": 8865
+    },
+    {
+      "epoch": 132.33,
+      "learning_rate": 4.1684210526315784e-05,
+      "loss": 0.0004,
+      "step": 8866
+    },
+    {
+      "epoch": 132.34,
+      "learning_rate": 4.164912280701754e-05,
+      "loss": 0.0506,
+      "step": 8867
+    },
+    {
+      "epoch": 132.36,
+      "learning_rate": 4.161403508771929e-05,
+      "loss": 0.0004,
+      "step": 8868
+    },
+    {
+      "epoch": 132.37,
+      "learning_rate": 4.1578947368421044e-05,
+      "loss": 0.0002,
+      "step": 8869
+    },
+    {
+      "epoch": 132.39,
+      "learning_rate": 4.1543859649122806e-05,
+      "loss": 0.0002,
+      "step": 8870
+    },
+    {
+      "epoch": 132.4,
+      "learning_rate": 4.150877192982456e-05,
+      "loss": 0.0006,
+      "step": 8871
+    },
+    {
+      "epoch": 132.42,
+      "learning_rate": 4.147368421052632e-05,
+      "loss": 0.0002,
+      "step": 8872
+    },
+    {
+      "epoch": 132.43,
+      "learning_rate": 4.1438596491228066e-05,
+      "loss": 0.0002,
+      "step": 8873
+    },
+    {
+      "epoch": 132.45,
+      "learning_rate": 4.140350877192982e-05,
+      "loss": 0.0003,
+      "step": 8874
+    },
+    {
+      "epoch": 132.46,
+      "learning_rate": 4.136842105263158e-05,
+      "loss": 0.0003,
+      "step": 8875
+    },
+    {
+      "epoch": 132.48,
+      "learning_rate": 4.1333333333333326e-05,
+      "loss": 0.0238,
+      "step": 8876
+    },
+    {
+      "epoch": 132.49,
+      "learning_rate": 4.129824561403508e-05,
+      "loss": 0.0003,
+      "step": 8877
+    },
+    {
+      "epoch": 132.51,
+      "learning_rate": 4.126315789473684e-05,
+      "loss": 0.0002,
+      "step": 8878
+    },
+    {
+      "epoch": 132.52,
+      "learning_rate": 4.122807017543859e-05,
+      "loss": 0.0006,
+      "step": 8879
+    },
+    {
+      "epoch": 132.54,
+      "learning_rate": 4.119298245614034e-05,
+      "loss": 0.0003,
+      "step": 8880
+    },
+    {
+      "epoch": 132.55,
+      "learning_rate": 4.11578947368421e-05,
+      "loss": 0.0003,
+      "step": 8881
+    },
+    {
+      "epoch": 132.57,
+      "learning_rate": 4.112280701754386e-05,
+      "loss": 0.0003,
+      "step": 8882
+    },
+    {
+      "epoch": 132.58,
+      "learning_rate": 4.1087719298245615e-05,
+      "loss": 0.0002,
+      "step": 8883
+    },
+    {
+      "epoch": 132.59,
+      "learning_rate": 4.105263157894737e-05,
+      "loss": 0.0003,
+      "step": 8884
+    },
+    {
+      "epoch": 132.61,
+      "learning_rate": 4.101754385964912e-05,
+      "loss": 0.0002,
+      "step": 8885
+    },
+    {
+      "epoch": 132.62,
+      "learning_rate": 4.0982456140350875e-05,
+      "loss": 0.0002,
+      "step": 8886
+    },
+    {
+      "epoch": 132.64,
+      "learning_rate": 4.094736842105263e-05,
+      "loss": 0.0003,
+      "step": 8887
+    },
+    {
+      "epoch": 132.65,
+      "learning_rate": 4.091228070175438e-05,
+      "loss": 0.0002,
+      "step": 8888
+    },
+    {
+      "epoch": 132.67,
+      "learning_rate": 4.0877192982456135e-05,
+      "loss": 0.0002,
+      "step": 8889
+    },
+    {
+      "epoch": 132.68,
+      "learning_rate": 4.084210526315789e-05,
+      "loss": 0.0003,
+      "step": 8890
+    },
+    {
+      "epoch": 132.7,
+      "learning_rate": 4.0807017543859646e-05,
+      "loss": 0.0003,
+      "step": 8891
+    },
+    {
+      "epoch": 132.71,
+      "learning_rate": 4.0771929824561395e-05,
+      "loss": 0.0002,
+      "step": 8892
+    },
+    {
+      "epoch": 132.73,
+      "learning_rate": 4.073684210526315e-05,
+      "loss": 0.0002,
+      "step": 8893
+    },
+    {
+      "epoch": 132.74,
+      "learning_rate": 4.0701754385964906e-05,
+      "loss": 0.0003,
+      "step": 8894
+    },
+    {
+      "epoch": 132.76,
+      "learning_rate": 4.066666666666667e-05,
+      "loss": 0.0003,
+      "step": 8895
+    },
+    {
+      "epoch": 132.77,
+      "learning_rate": 4.0631578947368424e-05,
+      "loss": 0.0002,
+      "step": 8896
+    },
+    {
+      "epoch": 132.79,
+      "learning_rate": 4.059649122807017e-05,
+      "loss": 0.0002,
+      "step": 8897
+    },
+    {
+      "epoch": 132.8,
+      "learning_rate": 4.056140350877193e-05,
+      "loss": 0.0062,
+      "step": 8898
+    },
+    {
+      "epoch": 132.82,
+      "learning_rate": 4.0526315789473684e-05,
+      "loss": 0.0178,
+      "step": 8899
+    },
+    {
+      "epoch": 132.83,
+      "learning_rate": 4.049122807017543e-05,
+      "loss": 0.0005,
+      "step": 8900
+    },
+    {
+      "epoch": 132.85,
+      "learning_rate": 4.045614035087719e-05,
+      "loss": 0.0003,
+      "step": 8901
+    },
+    {
+      "epoch": 132.86,
+      "learning_rate": 4.0421052631578943e-05,
+      "loss": 0.0003,
+      "step": 8902
+    },
+    {
+      "epoch": 132.88,
+      "learning_rate": 4.03859649122807e-05,
+      "loss": 0.0002,
+      "step": 8903
+    },
+    {
+      "epoch": 132.89,
+      "learning_rate": 4.035087719298245e-05,
+      "loss": 0.0002,
+      "step": 8904
+    },
+    {
+      "epoch": 132.91,
+      "learning_rate": 4.03157894736842e-05,
+      "loss": 0.0003,
+      "step": 8905
+    },
+    {
+      "epoch": 132.92,
+      "learning_rate": 4.028070175438596e-05,
+      "loss": 0.0002,
+      "step": 8906
+    },
+    {
+      "epoch": 132.94,
+      "learning_rate": 4.024561403508772e-05,
+      "loss": 0.0039,
+      "step": 8907
+    },
+    {
+      "epoch": 132.95,
+      "learning_rate": 4.021052631578948e-05,
+      "loss": 0.0004,
+      "step": 8908
+    },
+    {
+      "epoch": 132.97,
+      "learning_rate": 4.0175438596491226e-05,
+      "loss": 0.0048,
+      "step": 8909
+    },
+    {
+      "epoch": 132.98,
+      "learning_rate": 4.014035087719298e-05,
+      "loss": 0.0053,
+      "step": 8910
+    },
+    {
+      "epoch": 133.0,
+      "learning_rate": 4.0105263157894737e-05,
+      "loss": 0.0002,
+      "step": 8911
+    },
+    {
+      "epoch": 133.01,
+      "learning_rate": 4.0070175438596485e-05,
+      "loss": 0.0002,
+      "step": 8912
+    },
+    {
+      "epoch": 133.03,
+      "learning_rate": 4.003508771929824e-05,
+      "loss": 0.0002,
+      "step": 8913
+    },
+    {
+      "epoch": 133.04,
+      "learning_rate": 3.9999999999999996e-05,
+      "loss": 0.0003,
+      "step": 8914
+    },
+    {
+      "epoch": 133.06,
+      "learning_rate": 3.996491228070175e-05,
+      "loss": 0.0011,
+      "step": 8915
+    },
+    {
+      "epoch": 133.07,
+      "learning_rate": 3.99298245614035e-05,
+      "loss": 0.0002,
+      "step": 8916
+    },
+    {
+      "epoch": 133.09,
+      "learning_rate": 3.9894736842105256e-05,
+      "loss": 0.0004,
+      "step": 8917
+    },
+    {
+      "epoch": 133.1,
+      "learning_rate": 3.985964912280701e-05,
+      "loss": 0.0003,
+      "step": 8918
+    },
+    {
+      "epoch": 133.12,
+      "learning_rate": 3.982456140350877e-05,
+      "loss": 0.0002,
+      "step": 8919
+    },
+    {
+      "epoch": 133.13,
+      "learning_rate": 3.978947368421053e-05,
+      "loss": 0.0002,
+      "step": 8920
+    },
+    {
+      "epoch": 133.15,
+      "learning_rate": 3.975438596491228e-05,
+      "loss": 0.0002,
+      "step": 8921
+    },
+    {
+      "epoch": 133.16,
+      "learning_rate": 3.9719298245614034e-05,
+      "loss": 0.0348,
+      "step": 8922
+    },
+    {
+      "epoch": 133.18,
+      "learning_rate": 3.968421052631579e-05,
+      "loss": 0.0002,
+      "step": 8923
+    },
+    {
+      "epoch": 133.19,
+      "learning_rate": 3.964912280701754e-05,
+      "loss": 0.0002,
+      "step": 8924
+    },
+    {
+      "epoch": 133.21,
+      "learning_rate": 3.9614035087719294e-05,
+      "loss": 0.0002,
+      "step": 8925
+    },
+    {
+      "epoch": 133.22,
+      "learning_rate": 3.957894736842105e-05,
+      "loss": 0.0017,
+      "step": 8926
+    },
+    {
+      "epoch": 133.24,
+      "learning_rate": 3.9543859649122805e-05,
+      "loss": 0.0003,
+      "step": 8927
+    },
+    {
+      "epoch": 133.25,
+      "learning_rate": 3.9508771929824554e-05,
+      "loss": 0.0003,
+      "step": 8928
+    },
+    {
+      "epoch": 133.27,
+      "learning_rate": 3.947368421052631e-05,
+      "loss": 0.0003,
+      "step": 8929
+    },
+    {
+      "epoch": 133.28,
+      "learning_rate": 3.9438596491228065e-05,
+      "loss": 0.0002,
+      "step": 8930
+    },
+    {
+      "epoch": 133.3,
+      "learning_rate": 3.940350877192982e-05,
+      "loss": 0.0002,
+      "step": 8931
+    },
+    {
+      "epoch": 133.31,
+      "learning_rate": 3.936842105263158e-05,
+      "loss": 0.0002,
+      "step": 8932
+    },
+    {
+      "epoch": 133.33,
+      "learning_rate": 3.933333333333333e-05,
+      "loss": 0.0003,
+      "step": 8933
+    },
+    {
+      "epoch": 133.34,
+      "learning_rate": 3.929824561403509e-05,
+      "loss": 0.0003,
+      "step": 8934
+    },
+    {
+      "epoch": 133.36,
+      "learning_rate": 3.926315789473684e-05,
+      "loss": 0.0003,
+      "step": 8935
+    },
+    {
+      "epoch": 133.37,
+      "learning_rate": 3.922807017543859e-05,
+      "loss": 0.0002,
+      "step": 8936
+    },
+    {
+      "epoch": 133.39,
+      "learning_rate": 3.919298245614035e-05,
+      "loss": 0.0002,
+      "step": 8937
+    },
+    {
+      "epoch": 133.4,
+      "learning_rate": 3.91578947368421e-05,
+      "loss": 0.0003,
+      "step": 8938
+    },
+    {
+      "epoch": 133.42,
+      "learning_rate": 3.912280701754386e-05,
+      "loss": 0.0002,
+      "step": 8939
+    },
+    {
+      "epoch": 133.43,
+      "learning_rate": 3.908771929824561e-05,
+      "loss": 0.0002,
+      "step": 8940
+    },
+    {
+      "epoch": 133.45,
+      "learning_rate": 3.905263157894736e-05,
+      "loss": 0.0374,
+      "step": 8941
+    },
+    {
+      "epoch": 133.46,
+      "learning_rate": 3.901754385964912e-05,
+      "loss": 0.0002,
+      "step": 8942
+    },
+    {
+      "epoch": 133.48,
+      "learning_rate": 3.8982456140350874e-05,
+      "loss": 0.0004,
+      "step": 8943
+    },
+    {
+      "epoch": 133.49,
+      "learning_rate": 3.894736842105262e-05,
+      "loss": 0.0002,
+      "step": 8944
+    },
+    {
+      "epoch": 133.51,
+      "learning_rate": 3.8912280701754385e-05,
+      "loss": 0.173,
+      "step": 8945
+    },
+    {
+      "epoch": 133.52,
+      "learning_rate": 3.887719298245614e-05,
+      "loss": 0.0002,
+      "step": 8946
+    },
+    {
+      "epoch": 133.54,
+      "learning_rate": 3.8842105263157896e-05,
+      "loss": 0.0002,
+      "step": 8947
+    },
+    {
+      "epoch": 133.55,
+      "learning_rate": 3.8807017543859645e-05,
+      "loss": 0.0006,
+      "step": 8948
+    },
+    {
+      "epoch": 133.57,
+      "learning_rate": 3.87719298245614e-05,
+      "loss": 0.0002,
+      "step": 8949
+    },
+    {
+      "epoch": 133.58,
+      "learning_rate": 3.8736842105263156e-05,
+      "loss": 0.0002,
+      "step": 8950
+    },
+    {
+      "epoch": 133.59,
+      "learning_rate": 3.870175438596491e-05,
+      "loss": 0.0002,
+      "step": 8951
+    },
+    {
+      "epoch": 133.61,
+      "learning_rate": 3.866666666666666e-05,
+      "loss": 0.0002,
+      "step": 8952
+    },
+    {
+      "epoch": 133.62,
+      "learning_rate": 3.8631578947368416e-05,
+      "loss": 0.0002,
+      "step": 8953
+    },
+    {
+      "epoch": 133.64,
+      "learning_rate": 3.859649122807017e-05,
+      "loss": 0.0003,
+      "step": 8954
+    },
+    {
+      "epoch": 133.65,
+      "learning_rate": 3.856140350877193e-05,
+      "loss": 0.0003,
+      "step": 8955
+    },
+    {
+      "epoch": 133.67,
+      "learning_rate": 3.8526315789473676e-05,
+      "loss": 0.0003,
+      "step": 8956
+    },
+    {
+      "epoch": 133.68,
+      "learning_rate": 3.849122807017543e-05,
+      "loss": 0.0003,
+      "step": 8957
+    },
+    {
+      "epoch": 133.7,
+      "learning_rate": 3.8456140350877193e-05,
+      "loss": 0.0002,
+      "step": 8958
+    },
+    {
+      "epoch": 133.71,
+      "learning_rate": 3.842105263157895e-05,
+      "loss": 0.0002,
+      "step": 8959
+    },
+    {
+      "epoch": 133.73,
+      "learning_rate": 3.83859649122807e-05,
+      "loss": 0.0002,
+      "step": 8960
+    },
+    {
+      "epoch": 133.74,
+      "learning_rate": 3.835087719298245e-05,
+      "loss": 0.0729,
+      "step": 8961
+    },
+    {
+      "epoch": 133.76,
+      "learning_rate": 3.831578947368421e-05,
+      "loss": 0.0002,
+      "step": 8962
+    },
+    {
+      "epoch": 133.77,
+      "learning_rate": 3.8280701754385964e-05,
+      "loss": 0.0002,
+      "step": 8963
+    },
+    {
+      "epoch": 133.79,
+      "learning_rate": 3.824561403508771e-05,
+      "loss": 0.0002,
+      "step": 8964
+    },
+    {
+      "epoch": 133.8,
+      "learning_rate": 3.821052631578947e-05,
+      "loss": 0.0002,
+      "step": 8965
+    },
+    {
+      "epoch": 133.82,
+      "learning_rate": 3.8175438596491224e-05,
+      "loss": 0.0399,
+      "step": 8966
+    },
+    {
+      "epoch": 133.83,
+      "learning_rate": 3.814035087719298e-05,
+      "loss": 0.0002,
+      "step": 8967
+    },
+    {
+      "epoch": 133.85,
+      "learning_rate": 3.810526315789473e-05,
+      "loss": 0.0005,
+      "step": 8968
+    },
+    {
+      "epoch": 133.86,
+      "learning_rate": 3.8070175438596484e-05,
+      "loss": 0.0002,
+      "step": 8969
+    },
+    {
+      "epoch": 133.88,
+      "learning_rate": 3.8035087719298247e-05,
+      "loss": 0.0002,
+      "step": 8970
+    },
+    {
+      "epoch": 133.89,
+      "learning_rate": 3.8e-05,
+      "loss": 0.0002,
+      "step": 8971
+    },
+    {
+      "epoch": 133.91,
+      "learning_rate": 3.796491228070175e-05,
+      "loss": 0.0002,
+      "step": 8972
+    },
+    {
+      "epoch": 133.92,
+      "learning_rate": 3.7929824561403506e-05,
+      "loss": 0.0004,
+      "step": 8973
+    },
+    {
+      "epoch": 133.94,
+      "learning_rate": 3.789473684210526e-05,
+      "loss": 0.0011,
+      "step": 8974
+    },
+    {
+      "epoch": 133.95,
+      "learning_rate": 3.785964912280702e-05,
+      "loss": 0.0014,
+      "step": 8975
+    },
+    {
+      "epoch": 133.97,
+      "learning_rate": 3.7824561403508766e-05,
+      "loss": 0.0002,
+      "step": 8976
+    },
+    {
+      "epoch": 133.98,
+      "learning_rate": 3.778947368421052e-05,
+      "loss": 0.0002,
+      "step": 8977
+    },
+    {
+      "epoch": 134.0,
+      "learning_rate": 3.775438596491228e-05,
+      "loss": 0.0002,
+      "step": 8978
+    },
+    {
+      "epoch": 134.01,
+      "learning_rate": 3.771929824561403e-05,
+      "loss": 0.0004,
+      "step": 8979
+    },
+    {
+      "epoch": 134.03,
+      "learning_rate": 3.768421052631578e-05,
+      "loss": 0.0003,
+      "step": 8980
+    },
+    {
+      "epoch": 134.04,
+      "learning_rate": 3.764912280701754e-05,
+      "loss": 0.0003,
+      "step": 8981
+    },
+    {
+      "epoch": 134.06,
+      "learning_rate": 3.761403508771929e-05,
+      "loss": 0.0002,
+      "step": 8982
+    },
+    {
+      "epoch": 134.07,
+      "learning_rate": 3.7578947368421055e-05,
+      "loss": 0.0002,
+      "step": 8983
+    },
+    {
+      "epoch": 134.09,
+      "learning_rate": 3.7543859649122804e-05,
+      "loss": 0.0002,
+      "step": 8984
+    },
+    {
+      "epoch": 134.1,
+      "learning_rate": 3.750877192982456e-05,
+      "loss": 0.0003,
+      "step": 8985
+    },
+    {
+      "epoch": 134.12,
+      "learning_rate": 3.7473684210526315e-05,
+      "loss": 0.0003,
+      "step": 8986
+    },
+    {
+      "epoch": 134.13,
+      "learning_rate": 3.743859649122807e-05,
+      "loss": 0.0002,
+      "step": 8987
+    },
+    {
+      "epoch": 134.15,
+      "learning_rate": 3.740350877192982e-05,
+      "loss": 0.0003,
+      "step": 8988
+    },
+    {
+      "epoch": 134.16,
+      "learning_rate": 3.7368421052631575e-05,
+      "loss": 0.0002,
+      "step": 8989
+    },
+    {
+      "epoch": 134.18,
+      "learning_rate": 3.733333333333333e-05,
+      "loss": 0.0002,
+      "step": 8990
+    },
+    {
+      "epoch": 134.19,
+      "learning_rate": 3.7298245614035086e-05,
+      "loss": 0.0002,
+      "step": 8991
+    },
+    {
+      "epoch": 134.21,
+      "learning_rate": 3.726315789473684e-05,
+      "loss": 0.0004,
+      "step": 8992
+    },
+    {
+      "epoch": 134.22,
+      "learning_rate": 3.72280701754386e-05,
+      "loss": 0.0002,
+      "step": 8993
+    },
+    {
+      "epoch": 134.24,
+      "learning_rate": 3.7192982456140346e-05,
+      "loss": 0.0002,
+      "step": 8994
+    },
+    {
+      "epoch": 134.25,
+      "learning_rate": 3.71578947368421e-05,
+      "loss": 0.0002,
+      "step": 8995
+    },
+    {
+      "epoch": 134.27,
+      "learning_rate": 3.712280701754386e-05,
+      "loss": 0.0002,
+      "step": 8996
+    },
+    {
+      "epoch": 134.28,
+      "learning_rate": 3.7087719298245606e-05,
+      "loss": 0.0002,
+      "step": 8997
+    },
+    {
+      "epoch": 134.3,
+      "learning_rate": 3.705263157894737e-05,
+      "loss": 0.001,
+      "step": 8998
+    },
+    {
+      "epoch": 134.31,
+      "learning_rate": 3.7017543859649124e-05,
+      "loss": 0.0002,
+      "step": 8999
+    },
+    {
+      "epoch": 134.33,
+      "learning_rate": 3.698245614035087e-05,
+      "loss": 0.0002,
+      "step": 9000
+    },
+    {
+      "epoch": 134.33,
+      "eval_accuracy": 0.8888888888888888,
+      "eval_f1": 0.8887363564634383,
+      "eval_loss": 0.6235546469688416,
+      "eval_runtime": 344.0974,
+      "eval_samples_per_second": 11.875,
+      "eval_steps_per_second": 0.744,
+      "step": 9000
+    },
+    {
+      "epoch": 134.34,
+      "learning_rate": 3.694736842105263e-05,
+      "loss": 0.0051,
+      "step": 9001
+    },
+    {
+      "epoch": 134.36,
+      "learning_rate": 3.6912280701754384e-05,
+      "loss": 0.0002,
+      "step": 9002
+    },
+    {
+      "epoch": 134.37,
+      "learning_rate": 3.687719298245613e-05,
+      "loss": 0.0003,
+      "step": 9003
+    },
+    {
+      "epoch": 134.39,
+      "learning_rate": 3.684210526315789e-05,
+      "loss": 0.0002,
+      "step": 9004
+    },
+    {
+      "epoch": 134.4,
+      "learning_rate": 3.680701754385965e-05,
+      "loss": 0.0002,
+      "step": 9005
+    },
+    {
+      "epoch": 134.42,
+      "learning_rate": 3.67719298245614e-05,
+      "loss": 0.0002,
+      "step": 9006
+    },
+    {
+      "epoch": 134.43,
+      "learning_rate": 3.6736842105263155e-05,
+      "loss": 0.0002,
+      "step": 9007
+    },
+    {
+      "epoch": 134.45,
+      "learning_rate": 3.670175438596491e-05,
+      "loss": 0.0003,
+      "step": 9008
+    },
+    {
+      "epoch": 134.46,
+      "learning_rate": 3.666666666666666e-05,
+      "loss": 0.0003,
+      "step": 9009
+    },
+    {
+      "epoch": 134.48,
+      "learning_rate": 3.6631578947368414e-05,
+      "loss": 0.139,
+      "step": 9010
+    },
+    {
+      "epoch": 134.49,
+      "learning_rate": 3.659649122807018e-05,
+      "loss": 0.0002,
+      "step": 9011
+    },
+    {
+      "epoch": 134.51,
+      "learning_rate": 3.6561403508771926e-05,
+      "loss": 0.0003,
+      "step": 9012
+    },
+    {
+      "epoch": 134.52,
+      "learning_rate": 3.652631578947368e-05,
+      "loss": 0.0002,
+      "step": 9013
+    },
+    {
+      "epoch": 134.54,
+      "learning_rate": 3.649122807017544e-05,
+      "loss": 0.0002,
+      "step": 9014
+    },
+    {
+      "epoch": 134.55,
+      "learning_rate": 3.6456140350877185e-05,
+      "loss": 0.0002,
+      "step": 9015
+    },
+    {
+      "epoch": 134.57,
+      "learning_rate": 3.642105263157894e-05,
+      "loss": 0.0002,
+      "step": 9016
+    },
+    {
+      "epoch": 134.58,
+      "learning_rate": 3.63859649122807e-05,
+      "loss": 0.0002,
+      "step": 9017
+    },
+    {
+      "epoch": 134.59,
+      "learning_rate": 3.635087719298245e-05,
+      "loss": 0.0004,
+      "step": 9018
+    },
+    {
+      "epoch": 134.61,
+      "learning_rate": 3.631578947368421e-05,
+      "loss": 0.0007,
+      "step": 9019
+    },
+    {
+      "epoch": 134.62,
+      "learning_rate": 3.628070175438596e-05,
+      "loss": 0.0002,
+      "step": 9020
+    },
+    {
+      "epoch": 134.64,
+      "learning_rate": 3.624561403508771e-05,
+      "loss": 0.0566,
+      "step": 9021
+    },
+    {
+      "epoch": 134.65,
+      "learning_rate": 3.621052631578947e-05,
+      "loss": 0.0002,
+      "step": 9022
+    },
+    {
+      "epoch": 134.67,
+      "learning_rate": 3.617543859649123e-05,
+      "loss": 0.0002,
+      "step": 9023
+    },
+    {
+      "epoch": 134.68,
+      "learning_rate": 3.614035087719298e-05,
+      "loss": 0.0002,
+      "step": 9024
+    },
+    {
+      "epoch": 134.7,
+      "learning_rate": 3.6105263157894734e-05,
+      "loss": 0.0005,
+      "step": 9025
+    },
+    {
+      "epoch": 134.71,
+      "learning_rate": 3.607017543859649e-05,
+      "loss": 0.0002,
+      "step": 9026
+    },
+    {
+      "epoch": 134.73,
+      "learning_rate": 3.603508771929824e-05,
+      "loss": 0.0002,
+      "step": 9027
+    },
+    {
+      "epoch": 134.74,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 0.0002,
+      "step": 9028
+    },
+    {
+      "epoch": 134.76,
+      "learning_rate": 3.596491228070175e-05,
+      "loss": 0.0003,
+      "step": 9029
+    },
+    {
+      "epoch": 134.77,
+      "learning_rate": 3.5929824561403505e-05,
+      "loss": 0.0003,
+      "step": 9030
+    },
+    {
+      "epoch": 134.79,
+      "learning_rate": 3.589473684210526e-05,
+      "loss": 0.0002,
+      "step": 9031
+    },
+    {
+      "epoch": 134.8,
+      "learning_rate": 3.5859649122807016e-05,
+      "loss": 0.0005,
+      "step": 9032
+    },
+    {
+      "epoch": 134.82,
+      "learning_rate": 3.5824561403508765e-05,
+      "loss": 0.0002,
+      "step": 9033
+    },
+    {
+      "epoch": 134.83,
+      "learning_rate": 3.578947368421052e-05,
+      "loss": 0.0002,
+      "step": 9034
+    },
+    {
+      "epoch": 134.85,
+      "learning_rate": 3.5754385964912276e-05,
+      "loss": 0.0002,
+      "step": 9035
+    },
+    {
+      "epoch": 134.86,
+      "learning_rate": 3.571929824561403e-05,
+      "loss": 0.0002,
+      "step": 9036
+    },
+    {
+      "epoch": 134.88,
+      "learning_rate": 3.568421052631579e-05,
+      "loss": 0.0002,
+      "step": 9037
+    },
+    {
+      "epoch": 134.89,
+      "learning_rate": 3.564912280701754e-05,
+      "loss": 0.0002,
+      "step": 9038
+    },
+    {
+      "epoch": 134.91,
+      "learning_rate": 3.561403508771929e-05,
+      "loss": 0.0002,
+      "step": 9039
+    },
+    {
+      "epoch": 134.92,
+      "learning_rate": 3.557894736842105e-05,
+      "loss": 0.0002,
+      "step": 9040
+    },
+    {
+      "epoch": 134.94,
+      "learning_rate": 3.55438596491228e-05,
+      "loss": 0.0002,
+      "step": 9041
+    },
+    {
+      "epoch": 134.95,
+      "learning_rate": 3.550877192982456e-05,
+      "loss": 0.0003,
+      "step": 9042
+    },
+    {
+      "epoch": 134.97,
+      "learning_rate": 3.5473684210526314e-05,
+      "loss": 0.0002,
+      "step": 9043
+    },
+    {
+      "epoch": 134.98,
+      "learning_rate": 3.543859649122807e-05,
+      "loss": 0.0002,
+      "step": 9044
+    },
+    {
+      "epoch": 135.0,
+      "learning_rate": 3.540350877192982e-05,
+      "loss": 0.0002,
+      "step": 9045
+    },
+    {
+      "epoch": 135.01,
+      "learning_rate": 3.5368421052631574e-05,
+      "loss": 0.0003,
+      "step": 9046
+    },
+    {
+      "epoch": 135.03,
+      "learning_rate": 3.533333333333333e-05,
+      "loss": 0.0002,
+      "step": 9047
+    },
+    {
+      "epoch": 135.04,
+      "learning_rate": 3.5298245614035085e-05,
+      "loss": 0.0644,
+      "step": 9048
+    },
+    {
+      "epoch": 135.06,
+      "learning_rate": 3.526315789473684e-05,
+      "loss": 0.0002,
+      "step": 9049
+    },
+    {
+      "epoch": 135.07,
+      "learning_rate": 3.5228070175438596e-05,
+      "loss": 0.0002,
+      "step": 9050
+    },
+    {
+      "epoch": 135.09,
+      "learning_rate": 3.5192982456140345e-05,
+      "loss": 0.0006,
+      "step": 9051
+    },
+    {
+      "epoch": 135.1,
+      "learning_rate": 3.51578947368421e-05,
+      "loss": 0.0004,
+      "step": 9052
+    },
+    {
+      "epoch": 135.12,
+      "learning_rate": 3.5122807017543856e-05,
+      "loss": 0.0002,
+      "step": 9053
+    },
+    {
+      "epoch": 135.13,
+      "learning_rate": 3.508771929824561e-05,
+      "loss": 0.0004,
+      "step": 9054
+    },
+    {
+      "epoch": 135.15,
+      "learning_rate": 3.505263157894737e-05,
+      "loss": 0.0006,
+      "step": 9055
+    },
+    {
+      "epoch": 135.16,
+      "learning_rate": 3.501754385964912e-05,
+      "loss": 0.0002,
+      "step": 9056
+    },
+    {
+      "epoch": 135.18,
+      "learning_rate": 3.498245614035087e-05,
+      "loss": 0.0002,
+      "step": 9057
+    },
+    {
+      "epoch": 135.19,
+      "learning_rate": 3.494736842105263e-05,
+      "loss": 0.0002,
+      "step": 9058
+    },
+    {
+      "epoch": 135.21,
+      "learning_rate": 3.491228070175438e-05,
+      "loss": 0.0002,
+      "step": 9059
+    },
+    {
+      "epoch": 135.22,
+      "learning_rate": 3.487719298245614e-05,
+      "loss": 0.0829,
+      "step": 9060
+    },
+    {
+      "epoch": 135.24,
+      "learning_rate": 3.4842105263157894e-05,
+      "loss": 0.0002,
+      "step": 9061
+    },
+    {
+      "epoch": 135.25,
+      "learning_rate": 3.480701754385965e-05,
+      "loss": 0.0005,
+      "step": 9062
+    },
+    {
+      "epoch": 135.27,
+      "learning_rate": 3.47719298245614e-05,
+      "loss": 0.0003,
+      "step": 9063
+    },
+    {
+      "epoch": 135.28,
+      "learning_rate": 3.4736842105263153e-05,
+      "loss": 0.0002,
+      "step": 9064
+    },
+    {
+      "epoch": 135.3,
+      "learning_rate": 3.470175438596491e-05,
+      "loss": 0.0003,
+      "step": 9065
+    },
+    {
+      "epoch": 135.31,
+      "learning_rate": 3.4666666666666665e-05,
+      "loss": 0.0002,
+      "step": 9066
+    },
+    {
+      "epoch": 135.33,
+      "learning_rate": 3.463157894736842e-05,
+      "loss": 0.0004,
+      "step": 9067
+    },
+    {
+      "epoch": 135.34,
+      "learning_rate": 3.4596491228070176e-05,
+      "loss": 0.0002,
+      "step": 9068
+    },
+    {
+      "epoch": 135.36,
+      "learning_rate": 3.4561403508771924e-05,
+      "loss": 0.0002,
+      "step": 9069
+    },
+    {
+      "epoch": 135.37,
+      "learning_rate": 3.452631578947368e-05,
+      "loss": 0.001,
+      "step": 9070
+    },
+    {
+      "epoch": 135.39,
+      "learning_rate": 3.4491228070175436e-05,
+      "loss": 0.0002,
+      "step": 9071
+    },
+    {
+      "epoch": 135.4,
+      "learning_rate": 3.445614035087719e-05,
+      "loss": 0.0002,
+      "step": 9072
+    },
+    {
+      "epoch": 135.42,
+      "learning_rate": 3.442105263157894e-05,
+      "loss": 0.0002,
+      "step": 9073
+    },
+    {
+      "epoch": 135.43,
+      "learning_rate": 3.43859649122807e-05,
+      "loss": 0.0002,
+      "step": 9074
+    },
+    {
+      "epoch": 135.45,
+      "learning_rate": 3.435087719298245e-05,
+      "loss": 0.0002,
+      "step": 9075
+    },
+    {
+      "epoch": 135.46,
+      "learning_rate": 3.4315789473684207e-05,
+      "loss": 0.0002,
+      "step": 9076
+    },
+    {
+      "epoch": 135.48,
+      "learning_rate": 3.428070175438596e-05,
+      "loss": 0.0002,
+      "step": 9077
+    },
+    {
+      "epoch": 135.49,
+      "learning_rate": 3.424561403508772e-05,
+      "loss": 0.0002,
+      "step": 9078
+    },
+    {
+      "epoch": 135.51,
+      "learning_rate": 3.4210526315789466e-05,
+      "loss": 0.0002,
+      "step": 9079
+    },
+    {
+      "epoch": 135.52,
+      "learning_rate": 3.417543859649123e-05,
+      "loss": 0.0002,
+      "step": 9080
+    },
+    {
+      "epoch": 135.54,
+      "learning_rate": 3.414035087719298e-05,
+      "loss": 0.0002,
+      "step": 9081
+    },
+    {
+      "epoch": 135.55,
+      "learning_rate": 3.410526315789473e-05,
+      "loss": 0.0002,
+      "step": 9082
+    },
+    {
+      "epoch": 135.57,
+      "learning_rate": 3.407017543859649e-05,
+      "loss": 0.0002,
+      "step": 9083
+    },
+    {
+      "epoch": 135.58,
+      "learning_rate": 3.4035087719298244e-05,
+      "loss": 0.0002,
+      "step": 9084
+    },
+    {
+      "epoch": 135.59,
+      "learning_rate": 3.399999999999999e-05,
+      "loss": 0.0007,
+      "step": 9085
+    },
+    {
+      "epoch": 135.61,
+      "learning_rate": 3.3964912280701755e-05,
+      "loss": 0.0005,
+      "step": 9086
+    },
+    {
+      "epoch": 135.62,
+      "learning_rate": 3.3929824561403504e-05,
+      "loss": 0.0002,
+      "step": 9087
+    },
+    {
+      "epoch": 135.64,
+      "learning_rate": 3.389473684210526e-05,
+      "loss": 0.0002,
+      "step": 9088
+    },
+    {
+      "epoch": 135.65,
+      "learning_rate": 3.3859649122807015e-05,
+      "loss": 0.0002,
+      "step": 9089
+    },
+    {
+      "epoch": 135.67,
+      "learning_rate": 3.382456140350877e-05,
+      "loss": 0.0002,
+      "step": 9090
+    },
+    {
+      "epoch": 135.68,
+      "learning_rate": 3.378947368421052e-05,
+      "loss": 0.0006,
+      "step": 9091
+    },
+    {
+      "epoch": 135.7,
+      "learning_rate": 3.375438596491228e-05,
+      "loss": 0.0002,
+      "step": 9092
+    },
+    {
+      "epoch": 135.71,
+      "learning_rate": 3.371929824561403e-05,
+      "loss": 0.1007,
+      "step": 9093
+    },
+    {
+      "epoch": 135.73,
+      "learning_rate": 3.3684210526315786e-05,
+      "loss": 0.0003,
+      "step": 9094
+    },
+    {
+      "epoch": 135.74,
+      "learning_rate": 3.364912280701754e-05,
+      "loss": 0.0002,
+      "step": 9095
+    },
+    {
+      "epoch": 135.76,
+      "learning_rate": 3.36140350877193e-05,
+      "loss": 0.0002,
+      "step": 9096
+    },
+    {
+      "epoch": 135.77,
+      "learning_rate": 3.3578947368421046e-05,
+      "loss": 0.0002,
+      "step": 9097
+    },
+    {
+      "epoch": 135.79,
+      "learning_rate": 3.35438596491228e-05,
+      "loss": 0.0138,
+      "step": 9098
+    },
+    {
+      "epoch": 135.8,
+      "learning_rate": 3.350877192982456e-05,
+      "loss": 0.0002,
+      "step": 9099
+    },
+    {
+      "epoch": 135.82,
+      "learning_rate": 3.347368421052631e-05,
+      "loss": 0.0002,
+      "step": 9100
+    },
+    {
+      "epoch": 135.83,
+      "learning_rate": 3.343859649122807e-05,
+      "loss": 0.0618,
+      "step": 9101
+    },
+    {
+      "epoch": 135.85,
+      "learning_rate": 3.3403508771929824e-05,
+      "loss": 0.0002,
+      "step": 9102
+    },
+    {
+      "epoch": 135.86,
+      "learning_rate": 3.336842105263157e-05,
+      "loss": 0.0002,
+      "step": 9103
+    },
+    {
+      "epoch": 135.88,
+      "learning_rate": 3.333333333333333e-05,
+      "loss": 0.0002,
+      "step": 9104
+    },
+    {
+      "epoch": 135.89,
+      "learning_rate": 3.3298245614035084e-05,
+      "loss": 0.0002,
+      "step": 9105
+    },
+    {
+      "epoch": 135.91,
+      "learning_rate": 3.326315789473684e-05,
+      "loss": 0.0003,
+      "step": 9106
+    },
+    {
+      "epoch": 135.92,
+      "learning_rate": 3.3228070175438595e-05,
+      "loss": 0.0002,
+      "step": 9107
+    },
+    {
+      "epoch": 135.94,
+      "learning_rate": 3.319298245614035e-05,
+      "loss": 0.0002,
+      "step": 9108
+    },
+    {
+      "epoch": 135.95,
+      "learning_rate": 3.31578947368421e-05,
+      "loss": 0.0054,
+      "step": 9109
+    },
+    {
+      "epoch": 135.97,
+      "learning_rate": 3.3122807017543855e-05,
+      "loss": 0.0002,
+      "step": 9110
+    },
+    {
+      "epoch": 135.98,
+      "learning_rate": 3.308771929824561e-05,
+      "loss": 0.0002,
+      "step": 9111
+    },
+    {
+      "epoch": 136.0,
+      "learning_rate": 3.3052631578947366e-05,
+      "loss": 0.0002,
+      "step": 9112
+    },
+    {
+      "epoch": 136.01,
+      "learning_rate": 3.301754385964912e-05,
+      "loss": 0.0008,
+      "step": 9113
+    },
+    {
+      "epoch": 136.03,
+      "learning_rate": 3.298245614035088e-05,
+      "loss": 0.0002,
+      "step": 9114
+    },
+    {
+      "epoch": 136.04,
+      "learning_rate": 3.2947368421052626e-05,
+      "loss": 0.0002,
+      "step": 9115
+    },
+    {
+      "epoch": 136.06,
+      "learning_rate": 3.291228070175438e-05,
+      "loss": 0.0002,
+      "step": 9116
+    },
+    {
+      "epoch": 136.07,
+      "learning_rate": 3.287719298245614e-05,
+      "loss": 0.0002,
+      "step": 9117
+    },
+    {
+      "epoch": 136.09,
+      "learning_rate": 3.284210526315789e-05,
+      "loss": 0.0845,
+      "step": 9118
+    },
+    {
+      "epoch": 136.1,
+      "learning_rate": 3.280701754385965e-05,
+      "loss": 0.0002,
+      "step": 9119
+    },
+    {
+      "epoch": 136.12,
+      "learning_rate": 3.2771929824561403e-05,
+      "loss": 0.0002,
+      "step": 9120
+    },
+    {
+      "epoch": 136.13,
+      "learning_rate": 3.273684210526315e-05,
+      "loss": 0.0002,
+      "step": 9121
+    },
+    {
+      "epoch": 136.15,
+      "learning_rate": 3.270175438596491e-05,
+      "loss": 0.0019,
+      "step": 9122
+    },
+    {
+      "epoch": 136.16,
+      "learning_rate": 3.266666666666666e-05,
+      "loss": 0.0002,
+      "step": 9123
+    },
+    {
+      "epoch": 136.18,
+      "learning_rate": 3.263157894736842e-05,
+      "loss": 0.0002,
+      "step": 9124
+    },
+    {
+      "epoch": 136.19,
+      "learning_rate": 3.2596491228070174e-05,
+      "loss": 0.0002,
+      "step": 9125
+    },
+    {
+      "epoch": 136.21,
+      "learning_rate": 3.256140350877193e-05,
+      "loss": 0.1942,
+      "step": 9126
+    },
+    {
+      "epoch": 136.22,
+      "learning_rate": 3.252631578947368e-05,
+      "loss": 0.0002,
+      "step": 9127
+    },
+    {
+      "epoch": 136.24,
+      "learning_rate": 3.2491228070175434e-05,
+      "loss": 0.0002,
+      "step": 9128
+    },
+    {
+      "epoch": 136.25,
+      "learning_rate": 3.245614035087719e-05,
+      "loss": 0.0002,
+      "step": 9129
+    },
+    {
+      "epoch": 136.27,
+      "learning_rate": 3.2421052631578945e-05,
+      "loss": 0.0002,
+      "step": 9130
+    },
+    {
+      "epoch": 136.28,
+      "learning_rate": 3.23859649122807e-05,
+      "loss": 0.0007,
+      "step": 9131
+    },
+    {
+      "epoch": 136.3,
+      "learning_rate": 3.2350877192982457e-05,
+      "loss": 0.0002,
+      "step": 9132
+    },
+    {
+      "epoch": 136.31,
+      "learning_rate": 3.2315789473684205e-05,
+      "loss": 0.0002,
+      "step": 9133
+    },
+    {
+      "epoch": 136.33,
+      "learning_rate": 3.228070175438596e-05,
+      "loss": 0.0002,
+      "step": 9134
+    },
+    {
+      "epoch": 136.34,
+      "learning_rate": 3.2245614035087716e-05,
+      "loss": 0.0002,
+      "step": 9135
+    },
+    {
+      "epoch": 136.36,
+      "learning_rate": 3.221052631578947e-05,
+      "loss": 0.0002,
+      "step": 9136
+    },
+    {
+      "epoch": 136.37,
+      "learning_rate": 3.217543859649123e-05,
+      "loss": 0.0002,
+      "step": 9137
+    },
+    {
+      "epoch": 136.39,
+      "learning_rate": 3.214035087719298e-05,
+      "loss": 0.0002,
+      "step": 9138
+    },
+    {
+      "epoch": 136.4,
+      "learning_rate": 3.210526315789473e-05,
+      "loss": 0.0002,
+      "step": 9139
+    },
+    {
+      "epoch": 136.42,
+      "learning_rate": 3.207017543859649e-05,
+      "loss": 0.0002,
+      "step": 9140
+    },
+    {
+      "epoch": 136.43,
+      "learning_rate": 3.203508771929824e-05,
+      "loss": 0.0002,
+      "step": 9141
+    },
+    {
+      "epoch": 136.45,
+      "learning_rate": 3.2e-05,
+      "loss": 0.0004,
+      "step": 9142
+    },
+    {
+      "epoch": 136.46,
+      "learning_rate": 3.1964912280701754e-05,
+      "loss": 0.0007,
+      "step": 9143
+    },
+    {
+      "epoch": 136.48,
+      "learning_rate": 3.192982456140351e-05,
+      "loss": 0.0018,
+      "step": 9144
+    },
+    {
+      "epoch": 136.49,
+      "learning_rate": 3.189473684210526e-05,
+      "loss": 0.0002,
+      "step": 9145
+    },
+    {
+      "epoch": 136.51,
+      "learning_rate": 3.1859649122807014e-05,
+      "loss": 0.0003,
+      "step": 9146
+    },
+    {
+      "epoch": 136.52,
+      "learning_rate": 3.182456140350877e-05,
+      "loss": 0.0002,
+      "step": 9147
+    },
+    {
+      "epoch": 136.54,
+      "learning_rate": 3.178947368421052e-05,
+      "loss": 0.0002,
+      "step": 9148
+    },
+    {
+      "epoch": 136.55,
+      "learning_rate": 3.175438596491228e-05,
+      "loss": 0.0002,
+      "step": 9149
+    },
+    {
+      "epoch": 136.57,
+      "learning_rate": 3.1719298245614036e-05,
+      "loss": 0.19,
+      "step": 9150
+    },
+    {
+      "epoch": 136.58,
+      "learning_rate": 3.1684210526315785e-05,
+      "loss": 0.0011,
+      "step": 9151
+    },
+    {
+      "epoch": 136.59,
+      "learning_rate": 3.164912280701754e-05,
+      "loss": 0.0005,
+      "step": 9152
+    },
+    {
+      "epoch": 136.61,
+      "learning_rate": 3.1614035087719296e-05,
+      "loss": 0.0002,
+      "step": 9153
+    },
+    {
+      "epoch": 136.62,
+      "learning_rate": 3.1578947368421045e-05,
+      "loss": 0.0002,
+      "step": 9154
+    },
+    {
+      "epoch": 136.64,
+      "learning_rate": 3.154385964912281e-05,
+      "loss": 0.0002,
+      "step": 9155
+    },
+    {
+      "epoch": 136.65,
+      "learning_rate": 3.150877192982456e-05,
+      "loss": 0.0042,
+      "step": 9156
+    },
+    {
+      "epoch": 136.67,
+      "learning_rate": 3.147368421052631e-05,
+      "loss": 0.0002,
+      "step": 9157
+    },
+    {
+      "epoch": 136.68,
+      "learning_rate": 3.143859649122807e-05,
+      "loss": 0.0016,
+      "step": 9158
+    },
+    {
+      "epoch": 136.7,
+      "learning_rate": 3.140350877192982e-05,
+      "loss": 0.0003,
+      "step": 9159
+    },
+    {
+      "epoch": 136.71,
+      "learning_rate": 3.136842105263157e-05,
+      "loss": 0.0002,
+      "step": 9160
+    },
+    {
+      "epoch": 136.73,
+      "learning_rate": 3.1333333333333334e-05,
+      "loss": 0.0006,
+      "step": 9161
+    },
+    {
+      "epoch": 136.74,
+      "learning_rate": 3.129824561403509e-05,
+      "loss": 0.0002,
+      "step": 9162
+    },
+    {
+      "epoch": 136.76,
+      "learning_rate": 3.126315789473684e-05,
+      "loss": 0.0002,
+      "step": 9163
+    },
+    {
+      "epoch": 136.77,
+      "learning_rate": 3.1228070175438594e-05,
+      "loss": 0.0015,
+      "step": 9164
+    },
+    {
+      "epoch": 136.79,
+      "learning_rate": 3.119298245614035e-05,
+      "loss": 0.0002,
+      "step": 9165
+    },
+    {
+      "epoch": 136.8,
+      "learning_rate": 3.11578947368421e-05,
+      "loss": 0.0027,
+      "step": 9166
+    },
+    {
+      "epoch": 136.82,
+      "learning_rate": 3.1122807017543854e-05,
+      "loss": 0.0004,
+      "step": 9167
+    },
+    {
+      "epoch": 136.83,
+      "learning_rate": 3.1087719298245616e-05,
+      "loss": 0.0005,
+      "step": 9168
+    },
+    {
+      "epoch": 136.85,
+      "learning_rate": 3.1052631578947365e-05,
+      "loss": 0.0003,
+      "step": 9169
+    },
+    {
+      "epoch": 136.86,
+      "learning_rate": 3.101754385964912e-05,
+      "loss": 0.0002,
+      "step": 9170
+    },
+    {
+      "epoch": 136.88,
+      "learning_rate": 3.0982456140350876e-05,
+      "loss": 0.0002,
+      "step": 9171
+    },
+    {
+      "epoch": 136.89,
+      "learning_rate": 3.0947368421052625e-05,
+      "loss": 0.0002,
+      "step": 9172
+    },
+    {
+      "epoch": 136.91,
+      "learning_rate": 3.091228070175438e-05,
+      "loss": 0.0002,
+      "step": 9173
+    },
+    {
+      "epoch": 136.92,
+      "learning_rate": 3.087719298245614e-05,
+      "loss": 0.0053,
+      "step": 9174
+    },
+    {
+      "epoch": 136.94,
+      "learning_rate": 3.084210526315789e-05,
+      "loss": 0.0007,
+      "step": 9175
+    },
+    {
+      "epoch": 136.95,
+      "learning_rate": 3.080701754385965e-05,
+      "loss": 0.0002,
+      "step": 9176
+    },
+    {
+      "epoch": 136.97,
+      "learning_rate": 3.07719298245614e-05,
+      "loss": 0.0003,
+      "step": 9177
+    },
+    {
+      "epoch": 136.98,
+      "learning_rate": 3.073684210526315e-05,
+      "loss": 0.0002,
+      "step": 9178
+    },
+    {
+      "epoch": 137.0,
+      "learning_rate": 3.070175438596491e-05,
+      "loss": 0.0002,
+      "step": 9179
+    },
+    {
+      "epoch": 137.01,
+      "learning_rate": 3.066666666666667e-05,
+      "loss": 0.0855,
+      "step": 9180
+    },
+    {
+      "epoch": 137.03,
+      "learning_rate": 3.063157894736842e-05,
+      "loss": 0.0021,
+      "step": 9181
+    },
+    {
+      "epoch": 137.04,
+      "learning_rate": 3.059649122807017e-05,
+      "loss": 0.0239,
+      "step": 9182
+    },
+    {
+      "epoch": 137.06,
+      "learning_rate": 3.056140350877193e-05,
+      "loss": 0.0002,
+      "step": 9183
+    },
+    {
+      "epoch": 137.07,
+      "learning_rate": 3.052631578947368e-05,
+      "loss": 0.0002,
+      "step": 9184
+    },
+    {
+      "epoch": 137.09,
+      "learning_rate": 3.0491228070175433e-05,
+      "loss": 0.0031,
+      "step": 9185
+    },
+    {
+      "epoch": 137.1,
+      "learning_rate": 3.0456140350877192e-05,
+      "loss": 0.0011,
+      "step": 9186
+    },
+    {
+      "epoch": 137.12,
+      "learning_rate": 3.0421052631578948e-05,
+      "loss": 0.0002,
+      "step": 9187
+    },
+    {
+      "epoch": 137.13,
+      "learning_rate": 3.03859649122807e-05,
+      "loss": 0.0182,
+      "step": 9188
+    },
+    {
+      "epoch": 137.15,
+      "learning_rate": 3.0350877192982452e-05,
+      "loss": 0.0387,
+      "step": 9189
+    },
+    {
+      "epoch": 137.16,
+      "learning_rate": 3.0315789473684208e-05,
+      "loss": 0.0002,
+      "step": 9190
+    },
+    {
+      "epoch": 137.18,
+      "learning_rate": 3.028070175438596e-05,
+      "loss": 0.0002,
+      "step": 9191
+    },
+    {
+      "epoch": 137.19,
+      "learning_rate": 3.0245614035087715e-05,
+      "loss": 0.0002,
+      "step": 9192
+    },
+    {
+      "epoch": 137.21,
+      "learning_rate": 3.0210526315789474e-05,
+      "loss": 0.0002,
+      "step": 9193
+    },
+    {
+      "epoch": 137.22,
+      "learning_rate": 3.0175438596491226e-05,
+      "loss": 0.0003,
+      "step": 9194
+    },
+    {
+      "epoch": 137.24,
+      "learning_rate": 3.014035087719298e-05,
+      "loss": 0.0002,
+      "step": 9195
+    },
+    {
+      "epoch": 137.25,
+      "learning_rate": 3.0105263157894734e-05,
+      "loss": 0.0215,
+      "step": 9196
+    },
+    {
+      "epoch": 137.27,
+      "learning_rate": 3.0070175438596486e-05,
+      "loss": 0.0002,
+      "step": 9197
+    },
+    {
+      "epoch": 137.28,
+      "learning_rate": 3.0035087719298242e-05,
+      "loss": 0.0002,
+      "step": 9198
+    },
+    {
+      "epoch": 137.3,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 0.0002,
+      "step": 9199
+    },
+    {
+      "epoch": 137.31,
+      "learning_rate": 2.9964912280701753e-05,
+      "loss": 0.002,
+      "step": 9200
+    },
+    {
+      "epoch": 137.31,
+      "eval_accuracy": 0.8844836025452766,
+      "eval_f1": 0.8846804745314188,
+      "eval_loss": 0.667094349861145,
+      "eval_runtime": 345.8776,
+      "eval_samples_per_second": 11.813,
+      "eval_steps_per_second": 0.74,
+      "step": 9200
+    },
+    {
+      "epoch": 137.33,
+      "learning_rate": 2.9929824561403505e-05,
+      "loss": 0.0002,
+      "step": 9201
+    },
+    {
+      "epoch": 137.34,
+      "learning_rate": 2.989473684210526e-05,
+      "loss": 0.0002,
+      "step": 9202
+    },
+    {
+      "epoch": 137.36,
+      "learning_rate": 2.9859649122807013e-05,
+      "loss": 0.0007,
+      "step": 9203
+    },
+    {
+      "epoch": 137.37,
+      "learning_rate": 2.982456140350877e-05,
+      "loss": 0.0002,
+      "step": 9204
+    },
+    {
+      "epoch": 137.39,
+      "learning_rate": 2.9789473684210524e-05,
+      "loss": 0.0002,
+      "step": 9205
+    },
+    {
+      "epoch": 137.4,
+      "learning_rate": 2.975438596491228e-05,
+      "loss": 0.0003,
+      "step": 9206
+    },
+    {
+      "epoch": 137.42,
+      "learning_rate": 2.971929824561403e-05,
+      "loss": 0.0002,
+      "step": 9207
+    },
+    {
+      "epoch": 137.43,
+      "learning_rate": 2.9684210526315787e-05,
+      "loss": 0.1348,
+      "step": 9208
+    },
+    {
+      "epoch": 137.45,
+      "learning_rate": 2.964912280701754e-05,
+      "loss": 0.0002,
+      "step": 9209
+    },
+    {
+      "epoch": 137.46,
+      "learning_rate": 2.9614035087719295e-05,
+      "loss": 0.0002,
+      "step": 9210
+    },
+    {
+      "epoch": 137.48,
+      "learning_rate": 2.957894736842105e-05,
+      "loss": 0.0002,
+      "step": 9211
+    },
+    {
+      "epoch": 137.49,
+      "learning_rate": 2.9543859649122806e-05,
+      "loss": 0.0002,
+      "step": 9212
+    },
+    {
+      "epoch": 137.51,
+      "learning_rate": 2.9508771929824558e-05,
+      "loss": 0.0002,
+      "step": 9213
+    },
+    {
+      "epoch": 137.52,
+      "learning_rate": 2.9473684210526314e-05,
+      "loss": 0.0002,
+      "step": 9214
+    },
+    {
+      "epoch": 137.54,
+      "learning_rate": 2.9438596491228066e-05,
+      "loss": 0.0002,
+      "step": 9215
+    },
+    {
+      "epoch": 137.55,
+      "learning_rate": 2.940350877192982e-05,
+      "loss": 0.0002,
+      "step": 9216
+    },
+    {
+      "epoch": 137.57,
+      "learning_rate": 2.9368421052631574e-05,
+      "loss": 0.067,
+      "step": 9217
+    },
+    {
+      "epoch": 137.58,
+      "learning_rate": 2.9333333333333333e-05,
+      "loss": 0.0002,
+      "step": 9218
+    },
+    {
+      "epoch": 137.59,
+      "learning_rate": 2.9298245614035085e-05,
+      "loss": 0.0002,
+      "step": 9219
+    },
+    {
+      "epoch": 137.61,
+      "learning_rate": 2.926315789473684e-05,
+      "loss": 0.0002,
+      "step": 9220
+    },
+    {
+      "epoch": 137.62,
+      "learning_rate": 2.9228070175438592e-05,
+      "loss": 0.0002,
+      "step": 9221
+    },
+    {
+      "epoch": 137.64,
+      "learning_rate": 2.9192982456140348e-05,
+      "loss": 0.0002,
+      "step": 9222
+    },
+    {
+      "epoch": 137.65,
+      "learning_rate": 2.91578947368421e-05,
+      "loss": 0.0002,
+      "step": 9223
+    },
+    {
+      "epoch": 137.67,
+      "learning_rate": 2.912280701754386e-05,
+      "loss": 0.0008,
+      "step": 9224
+    },
+    {
+      "epoch": 137.68,
+      "learning_rate": 2.908771929824561e-05,
+      "loss": 0.0002,
+      "step": 9225
+    },
+    {
+      "epoch": 137.7,
+      "learning_rate": 2.9052631578947367e-05,
+      "loss": 0.0771,
+      "step": 9226
+    },
+    {
+      "epoch": 137.71,
+      "learning_rate": 2.901754385964912e-05,
+      "loss": 0.0005,
+      "step": 9227
+    },
+    {
+      "epoch": 137.73,
+      "learning_rate": 2.8982456140350875e-05,
+      "loss": 0.0002,
+      "step": 9228
+    },
+    {
+      "epoch": 137.74,
+      "learning_rate": 2.8947368421052627e-05,
+      "loss": 0.0006,
+      "step": 9229
+    },
+    {
+      "epoch": 137.76,
+      "learning_rate": 2.8912280701754386e-05,
+      "loss": 0.0002,
+      "step": 9230
+    },
+    {
+      "epoch": 137.77,
+      "learning_rate": 2.8877192982456138e-05,
+      "loss": 0.0002,
+      "step": 9231
+    },
+    {
+      "epoch": 137.79,
+      "learning_rate": 2.8842105263157893e-05,
+      "loss": 0.0002,
+      "step": 9232
+    },
+    {
+      "epoch": 137.8,
+      "learning_rate": 2.8807017543859646e-05,
+      "loss": 0.0002,
+      "step": 9233
+    },
+    {
+      "epoch": 137.82,
+      "learning_rate": 2.87719298245614e-05,
+      "loss": 0.0002,
+      "step": 9234
+    },
+    {
+      "epoch": 137.83,
+      "learning_rate": 2.8736842105263153e-05,
+      "loss": 0.0002,
+      "step": 9235
+    },
+    {
+      "epoch": 137.85,
+      "learning_rate": 2.8701754385964912e-05,
+      "loss": 0.0208,
+      "step": 9236
+    },
+    {
+      "epoch": 137.86,
+      "learning_rate": 2.8666666666666664e-05,
+      "loss": 0.1129,
+      "step": 9237
+    },
+    {
+      "epoch": 137.88,
+      "learning_rate": 2.863157894736842e-05,
+      "loss": 0.0002,
+      "step": 9238
+    },
+    {
+      "epoch": 137.89,
+      "learning_rate": 2.8596491228070172e-05,
+      "loss": 0.0002,
+      "step": 9239
+    },
+    {
+      "epoch": 137.91,
+      "learning_rate": 2.8561403508771928e-05,
+      "loss": 0.0002,
+      "step": 9240
+    },
+    {
+      "epoch": 137.92,
+      "learning_rate": 2.852631578947368e-05,
+      "loss": 0.0003,
+      "step": 9241
+    },
+    {
+      "epoch": 137.94,
+      "learning_rate": 2.8491228070175435e-05,
+      "loss": 0.0004,
+      "step": 9242
+    },
+    {
+      "epoch": 137.95,
+      "learning_rate": 2.845614035087719e-05,
+      "loss": 0.0002,
+      "step": 9243
+    },
+    {
+      "epoch": 137.97,
+      "learning_rate": 2.8421052631578946e-05,
+      "loss": 0.0002,
+      "step": 9244
+    },
+    {
+      "epoch": 137.98,
+      "learning_rate": 2.83859649122807e-05,
+      "loss": 0.0002,
+      "step": 9245
+    },
+    {
+      "epoch": 138.0,
+      "learning_rate": 2.8350877192982454e-05,
+      "loss": 0.0002,
+      "step": 9246
+    },
+    {
+      "epoch": 138.01,
+      "learning_rate": 2.8315789473684206e-05,
+      "loss": 0.0003,
+      "step": 9247
+    },
+    {
+      "epoch": 138.03,
+      "learning_rate": 2.8280701754385962e-05,
+      "loss": 0.0003,
+      "step": 9248
+    },
+    {
+      "epoch": 138.04,
+      "learning_rate": 2.8245614035087717e-05,
+      "loss": 0.0002,
+      "step": 9249
+    },
+    {
+      "epoch": 138.06,
+      "learning_rate": 2.8210526315789473e-05,
+      "loss": 0.0004,
+      "step": 9250
+    },
+    {
+      "epoch": 138.07,
+      "learning_rate": 2.8175438596491225e-05,
+      "loss": 0.0002,
+      "step": 9251
+    },
+    {
+      "epoch": 138.09,
+      "learning_rate": 2.814035087719298e-05,
+      "loss": 0.0002,
+      "step": 9252
+    },
+    {
+      "epoch": 138.1,
+      "learning_rate": 2.8105263157894733e-05,
+      "loss": 0.0016,
+      "step": 9253
+    },
+    {
+      "epoch": 138.12,
+      "learning_rate": 2.807017543859649e-05,
+      "loss": 0.0002,
+      "step": 9254
+    },
+    {
+      "epoch": 138.13,
+      "learning_rate": 2.8035087719298244e-05,
+      "loss": 0.0002,
+      "step": 9255
+    },
+    {
+      "epoch": 138.15,
+      "learning_rate": 2.8e-05,
+      "loss": 0.0002,
+      "step": 9256
+    },
+    {
+      "epoch": 138.16,
+      "learning_rate": 2.7964912280701752e-05,
+      "loss": 0.0002,
+      "step": 9257
+    },
+    {
+      "epoch": 138.18,
+      "learning_rate": 2.7929824561403507e-05,
+      "loss": 0.0001,
+      "step": 9258
+    },
+    {
+      "epoch": 138.19,
+      "learning_rate": 2.789473684210526e-05,
+      "loss": 0.019,
+      "step": 9259
+    },
+    {
+      "epoch": 138.21,
+      "learning_rate": 2.7859649122807015e-05,
+      "loss": 0.0002,
+      "step": 9260
+    },
+    {
+      "epoch": 138.22,
+      "learning_rate": 2.7824561403508767e-05,
+      "loss": 0.0003,
+      "step": 9261
+    },
+    {
+      "epoch": 138.24,
+      "learning_rate": 2.7789473684210526e-05,
+      "loss": 0.0002,
+      "step": 9262
+    },
+    {
+      "epoch": 138.25,
+      "learning_rate": 2.7754385964912278e-05,
+      "loss": 0.0003,
+      "step": 9263
+    },
+    {
+      "epoch": 138.27,
+      "learning_rate": 2.7719298245614034e-05,
+      "loss": 0.0002,
+      "step": 9264
+    },
+    {
+      "epoch": 138.28,
+      "learning_rate": 2.7684210526315786e-05,
+      "loss": 0.1113,
+      "step": 9265
+    },
+    {
+      "epoch": 138.3,
+      "learning_rate": 2.764912280701754e-05,
+      "loss": 0.0003,
+      "step": 9266
+    },
+    {
+      "epoch": 138.31,
+      "learning_rate": 2.7614035087719294e-05,
+      "loss": 0.0002,
+      "step": 9267
+    },
+    {
+      "epoch": 138.33,
+      "learning_rate": 2.7578947368421053e-05,
+      "loss": 0.0002,
+      "step": 9268
+    },
+    {
+      "epoch": 138.34,
+      "learning_rate": 2.7543859649122805e-05,
+      "loss": 0.0002,
+      "step": 9269
+    },
+    {
+      "epoch": 138.36,
+      "learning_rate": 2.750877192982456e-05,
+      "loss": 0.0002,
+      "step": 9270
+    },
+    {
+      "epoch": 138.37,
+      "learning_rate": 2.7473684210526313e-05,
+      "loss": 0.0002,
+      "step": 9271
+    },
+    {
+      "epoch": 138.39,
+      "learning_rate": 2.7438596491228068e-05,
+      "loss": 0.0002,
+      "step": 9272
+    },
+    {
+      "epoch": 138.4,
+      "learning_rate": 2.740350877192982e-05,
+      "loss": 0.0736,
+      "step": 9273
+    },
+    {
+      "epoch": 138.42,
+      "learning_rate": 2.736842105263158e-05,
+      "loss": 0.0002,
+      "step": 9274
+    },
+    {
+      "epoch": 138.43,
+      "learning_rate": 2.733333333333333e-05,
+      "loss": 0.0002,
+      "step": 9275
+    },
+    {
+      "epoch": 138.45,
+      "learning_rate": 2.7298245614035087e-05,
+      "loss": 0.0002,
+      "step": 9276
+    },
+    {
+      "epoch": 138.46,
+      "learning_rate": 2.726315789473684e-05,
+      "loss": 0.0002,
+      "step": 9277
+    },
+    {
+      "epoch": 138.48,
+      "learning_rate": 2.7228070175438595e-05,
+      "loss": 0.0002,
+      "step": 9278
+    },
+    {
+      "epoch": 138.49,
+      "learning_rate": 2.7192982456140347e-05,
+      "loss": 0.0002,
+      "step": 9279
+    },
+    {
+      "epoch": 138.51,
+      "learning_rate": 2.7157894736842106e-05,
+      "loss": 0.0002,
+      "step": 9280
+    },
+    {
+      "epoch": 138.52,
+      "learning_rate": 2.7122807017543858e-05,
+      "loss": 0.0001,
+      "step": 9281
+    },
+    {
+      "epoch": 138.54,
+      "learning_rate": 2.7087719298245613e-05,
+      "loss": 0.0002,
+      "step": 9282
+    },
+    {
+      "epoch": 138.55,
+      "learning_rate": 2.7052631578947366e-05,
+      "loss": 0.0003,
+      "step": 9283
+    },
+    {
+      "epoch": 138.57,
+      "learning_rate": 2.701754385964912e-05,
+      "loss": 0.0003,
+      "step": 9284
+    },
+    {
+      "epoch": 138.58,
+      "learning_rate": 2.6982456140350873e-05,
+      "loss": 0.0005,
+      "step": 9285
+    },
+    {
+      "epoch": 138.59,
+      "learning_rate": 2.6947368421052626e-05,
+      "loss": 0.0002,
+      "step": 9286
+    },
+    {
+      "epoch": 138.61,
+      "learning_rate": 2.6912280701754384e-05,
+      "loss": 0.0002,
+      "step": 9287
+    },
+    {
+      "epoch": 138.62,
+      "learning_rate": 2.687719298245614e-05,
+      "loss": 0.0003,
+      "step": 9288
+    },
+    {
+      "epoch": 138.64,
+      "learning_rate": 2.6842105263157892e-05,
+      "loss": 0.0002,
+      "step": 9289
+    },
+    {
+      "epoch": 138.65,
+      "learning_rate": 2.6807017543859648e-05,
+      "loss": 0.0002,
+      "step": 9290
+    },
+    {
+      "epoch": 138.67,
+      "learning_rate": 2.67719298245614e-05,
+      "loss": 0.0002,
+      "step": 9291
+    },
+    {
+      "epoch": 138.68,
+      "learning_rate": 2.6736842105263152e-05,
+      "loss": 0.001,
+      "step": 9292
+    },
+    {
+      "epoch": 138.7,
+      "learning_rate": 2.670175438596491e-05,
+      "loss": 0.0002,
+      "step": 9293
+    },
+    {
+      "epoch": 138.71,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.0002,
+      "step": 9294
+    },
+    {
+      "epoch": 138.73,
+      "learning_rate": 2.663157894736842e-05,
+      "loss": 0.0006,
+      "step": 9295
+    },
+    {
+      "epoch": 138.74,
+      "learning_rate": 2.6596491228070174e-05,
+      "loss": 0.0002,
+      "step": 9296
+    },
+    {
+      "epoch": 138.76,
+      "learning_rate": 2.6561403508771926e-05,
+      "loss": 0.0002,
+      "step": 9297
+    },
+    {
+      "epoch": 138.77,
+      "learning_rate": 2.652631578947368e-05,
+      "loss": 0.0002,
+      "step": 9298
+    },
+    {
+      "epoch": 138.79,
+      "learning_rate": 2.6491228070175438e-05,
+      "loss": 0.0002,
+      "step": 9299
+    },
+    {
+      "epoch": 138.8,
+      "learning_rate": 2.6456140350877193e-05,
+      "loss": 0.0002,
+      "step": 9300
+    },
+    {
+      "epoch": 138.82,
+      "learning_rate": 2.6421052631578945e-05,
+      "loss": 0.0003,
+      "step": 9301
+    },
+    {
+      "epoch": 138.83,
+      "learning_rate": 2.63859649122807e-05,
+      "loss": 0.0002,
+      "step": 9302
+    },
+    {
+      "epoch": 138.85,
+      "learning_rate": 2.6350877192982453e-05,
+      "loss": 0.0097,
+      "step": 9303
+    },
+    {
+      "epoch": 138.86,
+      "learning_rate": 2.6315789473684205e-05,
+      "loss": 0.0002,
+      "step": 9304
+    },
+    {
+      "epoch": 138.88,
+      "learning_rate": 2.6280701754385964e-05,
+      "loss": 0.0002,
+      "step": 9305
+    },
+    {
+      "epoch": 138.89,
+      "learning_rate": 2.624561403508772e-05,
+      "loss": 0.0002,
+      "step": 9306
+    },
+    {
+      "epoch": 138.91,
+      "learning_rate": 2.6210526315789472e-05,
+      "loss": 0.0002,
+      "step": 9307
+    },
+    {
+      "epoch": 138.92,
+      "learning_rate": 2.6175438596491227e-05,
+      "loss": 0.0002,
+      "step": 9308
+    },
+    {
+      "epoch": 138.94,
+      "learning_rate": 2.614035087719298e-05,
+      "loss": 0.0002,
+      "step": 9309
+    },
+    {
+      "epoch": 138.95,
+      "learning_rate": 2.6105263157894732e-05,
+      "loss": 0.0002,
+      "step": 9310
+    },
+    {
+      "epoch": 138.97,
+      "learning_rate": 2.6070175438596487e-05,
+      "loss": 0.0002,
+      "step": 9311
+    },
+    {
+      "epoch": 138.98,
+      "learning_rate": 2.6035087719298246e-05,
+      "loss": 0.0002,
+      "step": 9312
+    },
+    {
+      "epoch": 139.0,
+      "learning_rate": 2.6e-05,
+      "loss": 0.0002,
+      "step": 9313
+    },
+    {
+      "epoch": 139.01,
+      "learning_rate": 2.5964912280701754e-05,
+      "loss": 0.0005,
+      "step": 9314
+    },
+    {
+      "epoch": 139.03,
+      "learning_rate": 2.5929824561403506e-05,
+      "loss": 0.0003,
+      "step": 9315
+    },
+    {
+      "epoch": 139.04,
+      "learning_rate": 2.5894736842105258e-05,
+      "loss": 0.0002,
+      "step": 9316
+    },
+    {
+      "epoch": 139.06,
+      "learning_rate": 2.5859649122807014e-05,
+      "loss": 0.0002,
+      "step": 9317
+    },
+    {
+      "epoch": 139.07,
+      "learning_rate": 2.5824561403508773e-05,
+      "loss": 0.0017,
+      "step": 9318
+    },
+    {
+      "epoch": 139.09,
+      "learning_rate": 2.5789473684210525e-05,
+      "loss": 0.0002,
+      "step": 9319
+    },
+    {
+      "epoch": 139.1,
+      "learning_rate": 2.575438596491228e-05,
+      "loss": 0.0002,
+      "step": 9320
+    },
+    {
+      "epoch": 139.12,
+      "learning_rate": 2.5719298245614033e-05,
+      "loss": 0.0002,
+      "step": 9321
+    },
+    {
+      "epoch": 139.13,
+      "learning_rate": 2.5684210526315785e-05,
+      "loss": 0.0002,
+      "step": 9322
+    },
+    {
+      "epoch": 139.15,
+      "learning_rate": 2.564912280701754e-05,
+      "loss": 0.0002,
+      "step": 9323
+    },
+    {
+      "epoch": 139.16,
+      "learning_rate": 2.56140350877193e-05,
+      "loss": 0.0001,
+      "step": 9324
+    },
+    {
+      "epoch": 139.18,
+      "learning_rate": 2.557894736842105e-05,
+      "loss": 0.0002,
+      "step": 9325
+    },
+    {
+      "epoch": 139.19,
+      "learning_rate": 2.5543859649122807e-05,
+      "loss": 0.0002,
+      "step": 9326
+    },
+    {
+      "epoch": 139.21,
+      "learning_rate": 2.550877192982456e-05,
+      "loss": 0.0002,
+      "step": 9327
+    },
+    {
+      "epoch": 139.22,
+      "learning_rate": 2.547368421052631e-05,
+      "loss": 0.0002,
+      "step": 9328
+    },
+    {
+      "epoch": 139.24,
+      "learning_rate": 2.5438596491228067e-05,
+      "loss": 0.0002,
+      "step": 9329
+    },
+    {
+      "epoch": 139.25,
+      "learning_rate": 2.540350877192982e-05,
+      "loss": 0.0002,
+      "step": 9330
+    },
+    {
+      "epoch": 139.27,
+      "learning_rate": 2.5368421052631578e-05,
+      "loss": 0.0002,
+      "step": 9331
+    },
+    {
+      "epoch": 139.28,
+      "learning_rate": 2.533333333333333e-05,
+      "loss": 0.0002,
+      "step": 9332
+    },
+    {
+      "epoch": 139.3,
+      "learning_rate": 2.5298245614035086e-05,
+      "loss": 0.0002,
+      "step": 9333
+    },
+    {
+      "epoch": 139.31,
+      "learning_rate": 2.5263157894736838e-05,
+      "loss": 0.0002,
+      "step": 9334
+    },
+    {
+      "epoch": 139.33,
+      "learning_rate": 2.5228070175438593e-05,
+      "loss": 0.0002,
+      "step": 9335
+    },
+    {
+      "epoch": 139.34,
+      "learning_rate": 2.5192982456140346e-05,
+      "loss": 0.0002,
+      "step": 9336
+    },
+    {
+      "epoch": 139.36,
+      "learning_rate": 2.5157894736842105e-05,
+      "loss": 0.0002,
+      "step": 9337
+    },
+    {
+      "epoch": 139.37,
+      "learning_rate": 2.5122807017543857e-05,
+      "loss": 0.0002,
+      "step": 9338
+    },
+    {
+      "epoch": 139.39,
+      "learning_rate": 2.5087719298245612e-05,
+      "loss": 0.0002,
+      "step": 9339
+    },
+    {
+      "epoch": 139.4,
+      "learning_rate": 2.5052631578947364e-05,
+      "loss": 0.0002,
+      "step": 9340
+    },
+    {
+      "epoch": 139.42,
+      "learning_rate": 2.501754385964912e-05,
+      "loss": 0.0002,
+      "step": 9341
+    },
+    {
+      "epoch": 139.43,
+      "learning_rate": 2.4982456140350872e-05,
+      "loss": 0.0002,
+      "step": 9342
+    },
+    {
+      "epoch": 139.45,
+      "learning_rate": 2.494736842105263e-05,
+      "loss": 0.0002,
+      "step": 9343
+    },
+    {
+      "epoch": 139.46,
+      "learning_rate": 2.4912280701754383e-05,
+      "loss": 0.0003,
+      "step": 9344
+    },
+    {
+      "epoch": 139.48,
+      "learning_rate": 2.487719298245614e-05,
+      "loss": 0.0002,
+      "step": 9345
+    },
+    {
+      "epoch": 139.49,
+      "learning_rate": 2.484210526315789e-05,
+      "loss": 0.0002,
+      "step": 9346
+    },
+    {
+      "epoch": 139.51,
+      "learning_rate": 2.4807017543859647e-05,
+      "loss": 0.0002,
+      "step": 9347
+    },
+    {
+      "epoch": 139.52,
+      "learning_rate": 2.47719298245614e-05,
+      "loss": 0.0002,
+      "step": 9348
+    },
+    {
+      "epoch": 139.54,
+      "learning_rate": 2.4736842105263158e-05,
+      "loss": 0.0002,
+      "step": 9349
+    },
+    {
+      "epoch": 139.55,
+      "learning_rate": 2.470175438596491e-05,
+      "loss": 0.0002,
+      "step": 9350
+    },
+    {
+      "epoch": 139.57,
+      "learning_rate": 2.4666666666666665e-05,
+      "loss": 0.0015,
+      "step": 9351
+    },
+    {
+      "epoch": 139.58,
+      "learning_rate": 2.4631578947368418e-05,
+      "loss": 0.0007,
+      "step": 9352
+    },
+    {
+      "epoch": 139.59,
+      "learning_rate": 2.4596491228070173e-05,
+      "loss": 0.0002,
+      "step": 9353
+    },
+    {
+      "epoch": 139.61,
+      "learning_rate": 2.4561403508771925e-05,
+      "loss": 0.0002,
+      "step": 9354
+    },
+    {
+      "epoch": 139.62,
+      "learning_rate": 2.452631578947368e-05,
+      "loss": 0.0002,
+      "step": 9355
+    },
+    {
+      "epoch": 139.64,
+      "learning_rate": 2.4491228070175436e-05,
+      "loss": 0.0002,
+      "step": 9356
+    },
+    {
+      "epoch": 139.65,
+      "learning_rate": 2.4456140350877192e-05,
+      "loss": 0.0002,
+      "step": 9357
+    },
+    {
+      "epoch": 139.67,
+      "learning_rate": 2.4421052631578944e-05,
+      "loss": 0.0002,
+      "step": 9358
+    },
+    {
+      "epoch": 139.68,
+      "learning_rate": 2.43859649122807e-05,
+      "loss": 0.0002,
+      "step": 9359
+    },
+    {
+      "epoch": 139.7,
+      "learning_rate": 2.4350877192982452e-05,
+      "loss": 0.0002,
+      "step": 9360
+    },
+    {
+      "epoch": 139.71,
+      "learning_rate": 2.4315789473684207e-05,
+      "loss": 0.0002,
+      "step": 9361
+    },
+    {
+      "epoch": 139.73,
+      "learning_rate": 2.4280701754385963e-05,
+      "loss": 0.0002,
+      "step": 9362
+    },
+    {
+      "epoch": 139.74,
+      "learning_rate": 2.424561403508772e-05,
+      "loss": 0.0002,
+      "step": 9363
+    },
+    {
+      "epoch": 139.76,
+      "learning_rate": 2.421052631578947e-05,
+      "loss": 0.0002,
+      "step": 9364
+    },
+    {
+      "epoch": 139.77,
+      "learning_rate": 2.4175438596491226e-05,
+      "loss": 0.0002,
+      "step": 9365
+    },
+    {
+      "epoch": 139.79,
+      "learning_rate": 2.414035087719298e-05,
+      "loss": 0.0251,
+      "step": 9366
+    },
+    {
+      "epoch": 139.8,
+      "learning_rate": 2.4105263157894734e-05,
+      "loss": 0.0002,
+      "step": 9367
+    },
+    {
+      "epoch": 139.82,
+      "learning_rate": 2.407017543859649e-05,
+      "loss": 0.0118,
+      "step": 9368
+    },
+    {
+      "epoch": 139.83,
+      "learning_rate": 2.4035087719298245e-05,
+      "loss": 0.2565,
+      "step": 9369
+    },
+    {
+      "epoch": 139.85,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 0.0004,
+      "step": 9370
+    },
+    {
+      "epoch": 139.86,
+      "learning_rate": 2.3964912280701753e-05,
+      "loss": 0.0002,
+      "step": 9371
+    },
+    {
+      "epoch": 139.88,
+      "learning_rate": 2.3929824561403505e-05,
+      "loss": 0.0003,
+      "step": 9372
+    },
+    {
+      "epoch": 139.89,
+      "learning_rate": 2.389473684210526e-05,
+      "loss": 0.0002,
+      "step": 9373
+    },
+    {
+      "epoch": 139.91,
+      "learning_rate": 2.3859649122807016e-05,
+      "loss": 0.0033,
+      "step": 9374
+    },
+    {
+      "epoch": 139.92,
+      "learning_rate": 2.382456140350877e-05,
+      "loss": 0.0002,
+      "step": 9375
+    },
+    {
+      "epoch": 139.94,
+      "learning_rate": 2.3789473684210524e-05,
+      "loss": 0.0002,
+      "step": 9376
+    },
+    {
+      "epoch": 139.95,
+      "learning_rate": 2.375438596491228e-05,
+      "loss": 0.0002,
+      "step": 9377
+    },
+    {
+      "epoch": 139.97,
+      "learning_rate": 2.371929824561403e-05,
+      "loss": 0.0001,
+      "step": 9378
+    },
+    {
+      "epoch": 139.98,
+      "learning_rate": 2.3684210526315787e-05,
+      "loss": 0.0002,
+      "step": 9379
+    },
+    {
+      "epoch": 140.0,
+      "learning_rate": 2.364912280701754e-05,
+      "loss": 0.0002,
+      "step": 9380
+    },
+    {
+      "epoch": 140.01,
+      "learning_rate": 2.3614035087719298e-05,
+      "loss": 0.0002,
+      "step": 9381
+    },
+    {
+      "epoch": 140.03,
+      "learning_rate": 2.357894736842105e-05,
+      "loss": 0.0002,
+      "step": 9382
+    },
+    {
+      "epoch": 140.04,
+      "learning_rate": 2.3543859649122806e-05,
+      "loss": 0.0002,
+      "step": 9383
+    },
+    {
+      "epoch": 140.06,
+      "learning_rate": 2.3508771929824558e-05,
+      "loss": 0.0002,
+      "step": 9384
+    },
+    {
+      "epoch": 140.07,
+      "learning_rate": 2.3473684210526314e-05,
+      "loss": 0.0032,
+      "step": 9385
+    },
+    {
+      "epoch": 140.09,
+      "learning_rate": 2.3438596491228066e-05,
+      "loss": 0.0002,
+      "step": 9386
+    },
+    {
+      "epoch": 140.1,
+      "learning_rate": 2.3403508771929825e-05,
+      "loss": 0.0002,
+      "step": 9387
+    },
+    {
+      "epoch": 140.12,
+      "learning_rate": 2.3368421052631577e-05,
+      "loss": 0.0002,
+      "step": 9388
+    },
+    {
+      "epoch": 140.13,
+      "learning_rate": 2.3333333333333332e-05,
+      "loss": 0.0005,
+      "step": 9389
+    },
+    {
+      "epoch": 140.15,
+      "learning_rate": 2.3298245614035085e-05,
+      "loss": 0.0003,
+      "step": 9390
+    },
+    {
+      "epoch": 140.16,
+      "learning_rate": 2.326315789473684e-05,
+      "loss": 0.0002,
+      "step": 9391
+    },
+    {
+      "epoch": 140.18,
+      "learning_rate": 2.3228070175438592e-05,
+      "loss": 0.0007,
+      "step": 9392
+    },
+    {
+      "epoch": 140.19,
+      "learning_rate": 2.319298245614035e-05,
+      "loss": 0.0002,
+      "step": 9393
+    },
+    {
+      "epoch": 140.21,
+      "learning_rate": 2.3157894736842103e-05,
+      "loss": 0.0003,
+      "step": 9394
+    },
+    {
+      "epoch": 140.22,
+      "learning_rate": 2.312280701754386e-05,
+      "loss": 0.0002,
+      "step": 9395
+    },
+    {
+      "epoch": 140.24,
+      "learning_rate": 2.308771929824561e-05,
+      "loss": 0.0014,
+      "step": 9396
+    },
+    {
+      "epoch": 140.25,
+      "learning_rate": 2.3052631578947367e-05,
+      "loss": 0.0002,
+      "step": 9397
+    },
+    {
+      "epoch": 140.27,
+      "learning_rate": 2.301754385964912e-05,
+      "loss": 0.0002,
+      "step": 9398
+    },
+    {
+      "epoch": 140.28,
+      "learning_rate": 2.2982456140350878e-05,
+      "loss": 0.0002,
+      "step": 9399
+    },
+    {
+      "epoch": 140.3,
+      "learning_rate": 2.294736842105263e-05,
+      "loss": 0.0002,
+      "step": 9400
+    },
+    {
+      "epoch": 140.3,
+      "eval_accuracy": 0.8935389133627019,
+      "eval_f1": 0.8931026363559365,
+      "eval_loss": 0.5969720482826233,
+      "eval_runtime": 345.1416,
+      "eval_samples_per_second": 11.839,
+      "eval_steps_per_second": 0.742,
+      "step": 9400
+    },
+    {
+      "epoch": 140.31,
+      "learning_rate": 2.2912280701754386e-05,
+      "loss": 0.0002,
+      "step": 9401
+    },
+    {
+      "epoch": 140.33,
+      "learning_rate": 2.2877192982456138e-05,
+      "loss": 0.0002,
+      "step": 9402
+    },
+    {
+      "epoch": 140.34,
+      "learning_rate": 2.2842105263157893e-05,
+      "loss": 0.0182,
+      "step": 9403
+    },
+    {
+      "epoch": 140.36,
+      "learning_rate": 2.2807017543859645e-05,
+      "loss": 0.1147,
+      "step": 9404
+    },
+    {
+      "epoch": 140.37,
+      "learning_rate": 2.27719298245614e-05,
+      "loss": 0.0002,
+      "step": 9405
+    },
+    {
+      "epoch": 140.39,
+      "learning_rate": 2.2736842105263157e-05,
+      "loss": 0.0019,
+      "step": 9406
+    },
+    {
+      "epoch": 140.4,
+      "learning_rate": 2.2701754385964912e-05,
+      "loss": 0.0002,
+      "step": 9407
+    },
+    {
+      "epoch": 140.42,
+      "learning_rate": 2.2666666666666664e-05,
+      "loss": 0.0543,
+      "step": 9408
+    },
+    {
+      "epoch": 140.43,
+      "learning_rate": 2.263157894736842e-05,
+      "loss": 0.0002,
+      "step": 9409
+    },
+    {
+      "epoch": 140.45,
+      "learning_rate": 2.2596491228070172e-05,
+      "loss": 0.0002,
+      "step": 9410
+    },
+    {
+      "epoch": 140.46,
+      "learning_rate": 2.2561403508771928e-05,
+      "loss": 0.0002,
+      "step": 9411
+    },
+    {
+      "epoch": 140.48,
+      "learning_rate": 2.2526315789473683e-05,
+      "loss": 0.0002,
+      "step": 9412
+    },
+    {
+      "epoch": 140.49,
+      "learning_rate": 2.249122807017544e-05,
+      "loss": 0.0002,
+      "step": 9413
+    },
+    {
+      "epoch": 140.51,
+      "learning_rate": 2.245614035087719e-05,
+      "loss": 0.0002,
+      "step": 9414
+    },
+    {
+      "epoch": 140.52,
+      "learning_rate": 2.2421052631578946e-05,
+      "loss": 0.0002,
+      "step": 9415
+    },
+    {
+      "epoch": 140.54,
+      "learning_rate": 2.23859649122807e-05,
+      "loss": 0.0001,
+      "step": 9416
+    },
+    {
+      "epoch": 140.55,
+      "learning_rate": 2.2350877192982454e-05,
+      "loss": 0.0002,
+      "step": 9417
+    },
+    {
+      "epoch": 140.57,
+      "learning_rate": 2.231578947368421e-05,
+      "loss": 0.0002,
+      "step": 9418
+    },
+    {
+      "epoch": 140.58,
+      "learning_rate": 2.2280701754385965e-05,
+      "loss": 0.0002,
+      "step": 9419
+    },
+    {
+      "epoch": 140.59,
+      "learning_rate": 2.2245614035087717e-05,
+      "loss": 0.0002,
+      "step": 9420
+    },
+    {
+      "epoch": 140.61,
+      "learning_rate": 2.2210526315789473e-05,
+      "loss": 0.0002,
+      "step": 9421
+    },
+    {
+      "epoch": 140.62,
+      "learning_rate": 2.2175438596491225e-05,
+      "loss": 0.0002,
+      "step": 9422
+    },
+    {
+      "epoch": 140.64,
+      "learning_rate": 2.214035087719298e-05,
+      "loss": 0.0002,
+      "step": 9423
+    },
+    {
+      "epoch": 140.65,
+      "learning_rate": 2.2105263157894733e-05,
+      "loss": 0.0002,
+      "step": 9424
+    },
+    {
+      "epoch": 140.67,
+      "learning_rate": 2.2070175438596492e-05,
+      "loss": 0.0002,
+      "step": 9425
+    },
+    {
+      "epoch": 140.68,
+      "learning_rate": 2.2035087719298244e-05,
+      "loss": 0.0002,
+      "step": 9426
+    },
+    {
+      "epoch": 140.7,
+      "learning_rate": 2.2e-05,
+      "loss": 0.0003,
+      "step": 9427
+    },
+    {
+      "epoch": 140.71,
+      "learning_rate": 2.196491228070175e-05,
+      "loss": 0.0002,
+      "step": 9428
+    },
+    {
+      "epoch": 140.73,
+      "learning_rate": 2.1929824561403507e-05,
+      "loss": 0.0002,
+      "step": 9429
+    },
+    {
+      "epoch": 140.74,
+      "learning_rate": 2.189473684210526e-05,
+      "loss": 0.0002,
+      "step": 9430
+    },
+    {
+      "epoch": 140.76,
+      "learning_rate": 2.1859649122807018e-05,
+      "loss": 0.0003,
+      "step": 9431
+    },
+    {
+      "epoch": 140.77,
+      "learning_rate": 2.182456140350877e-05,
+      "loss": 0.0002,
+      "step": 9432
+    },
+    {
+      "epoch": 140.79,
+      "learning_rate": 2.1789473684210526e-05,
+      "loss": 0.0002,
+      "step": 9433
+    },
+    {
+      "epoch": 140.8,
+      "learning_rate": 2.1754385964912278e-05,
+      "loss": 0.0064,
+      "step": 9434
+    },
+    {
+      "epoch": 140.82,
+      "learning_rate": 2.1719298245614034e-05,
+      "loss": 0.0004,
+      "step": 9435
+    },
+    {
+      "epoch": 140.83,
+      "learning_rate": 2.1684210526315786e-05,
+      "loss": 0.0002,
+      "step": 9436
+    },
+    {
+      "epoch": 140.85,
+      "learning_rate": 2.1649122807017545e-05,
+      "loss": 0.0002,
+      "step": 9437
+    },
+    {
+      "epoch": 140.86,
+      "learning_rate": 2.1614035087719297e-05,
+      "loss": 0.0002,
+      "step": 9438
+    },
+    {
+      "epoch": 140.88,
+      "learning_rate": 2.1578947368421053e-05,
+      "loss": 0.0001,
+      "step": 9439
+    },
+    {
+      "epoch": 140.89,
+      "learning_rate": 2.1543859649122805e-05,
+      "loss": 0.0014,
+      "step": 9440
+    },
+    {
+      "epoch": 140.91,
+      "learning_rate": 2.150877192982456e-05,
+      "loss": 0.0654,
+      "step": 9441
+    },
+    {
+      "epoch": 140.92,
+      "learning_rate": 2.1473684210526312e-05,
+      "loss": 0.0002,
+      "step": 9442
+    },
+    {
+      "epoch": 140.94,
+      "learning_rate": 2.143859649122807e-05,
+      "loss": 0.0004,
+      "step": 9443
+    },
+    {
+      "epoch": 140.95,
+      "learning_rate": 2.1403508771929824e-05,
+      "loss": 0.0003,
+      "step": 9444
+    },
+    {
+      "epoch": 140.97,
+      "learning_rate": 2.136842105263158e-05,
+      "loss": 0.0002,
+      "step": 9445
+    },
+    {
+      "epoch": 140.98,
+      "learning_rate": 2.133333333333333e-05,
+      "loss": 0.0002,
+      "step": 9446
+    },
+    {
+      "epoch": 141.0,
+      "learning_rate": 2.1298245614035087e-05,
+      "loss": 0.0002,
+      "step": 9447
+    },
+    {
+      "epoch": 141.01,
+      "learning_rate": 2.126315789473684e-05,
+      "loss": 0.0003,
+      "step": 9448
+    },
+    {
+      "epoch": 141.03,
+      "learning_rate": 2.122807017543859e-05,
+      "loss": 0.0002,
+      "step": 9449
+    },
+    {
+      "epoch": 141.04,
+      "learning_rate": 2.119298245614035e-05,
+      "loss": 0.0002,
+      "step": 9450
+    },
+    {
+      "epoch": 141.06,
+      "learning_rate": 2.1157894736842106e-05,
+      "loss": 0.0002,
+      "step": 9451
+    },
+    {
+      "epoch": 141.07,
+      "learning_rate": 2.1122807017543858e-05,
+      "loss": 0.0002,
+      "step": 9452
+    },
+    {
+      "epoch": 141.09,
+      "learning_rate": 2.1087719298245613e-05,
+      "loss": 0.1253,
+      "step": 9453
+    },
+    {
+      "epoch": 141.1,
+      "learning_rate": 2.1052631578947366e-05,
+      "loss": 0.0002,
+      "step": 9454
+    },
+    {
+      "epoch": 141.12,
+      "learning_rate": 2.1017543859649118e-05,
+      "loss": 0.0002,
+      "step": 9455
+    },
+    {
+      "epoch": 141.13,
+      "learning_rate": 2.0982456140350877e-05,
+      "loss": 0.0002,
+      "step": 9456
+    },
+    {
+      "epoch": 141.15,
+      "learning_rate": 2.0947368421052632e-05,
+      "loss": 0.1106,
+      "step": 9457
+    },
+    {
+      "epoch": 141.16,
+      "learning_rate": 2.0912280701754384e-05,
+      "loss": 0.0002,
+      "step": 9458
+    },
+    {
+      "epoch": 141.18,
+      "learning_rate": 2.087719298245614e-05,
+      "loss": 0.0002,
+      "step": 9459
+    },
+    {
+      "epoch": 141.19,
+      "learning_rate": 2.0842105263157892e-05,
+      "loss": 0.0002,
+      "step": 9460
+    },
+    {
+      "epoch": 141.21,
+      "learning_rate": 2.0807017543859644e-05,
+      "loss": 0.0002,
+      "step": 9461
+    },
+    {
+      "epoch": 141.22,
+      "learning_rate": 2.0771929824561403e-05,
+      "loss": 0.0002,
+      "step": 9462
+    },
+    {
+      "epoch": 141.24,
+      "learning_rate": 2.073684210526316e-05,
+      "loss": 0.0007,
+      "step": 9463
+    },
+    {
+      "epoch": 141.25,
+      "learning_rate": 2.070175438596491e-05,
+      "loss": 0.0002,
+      "step": 9464
+    },
+    {
+      "epoch": 141.27,
+      "learning_rate": 2.0666666666666663e-05,
+      "loss": 0.0003,
+      "step": 9465
+    },
+    {
+      "epoch": 141.28,
+      "learning_rate": 2.063157894736842e-05,
+      "loss": 0.0561,
+      "step": 9466
+    },
+    {
+      "epoch": 141.3,
+      "learning_rate": 2.059649122807017e-05,
+      "loss": 0.0002,
+      "step": 9467
+    },
+    {
+      "epoch": 141.31,
+      "learning_rate": 2.056140350877193e-05,
+      "loss": 0.0003,
+      "step": 9468
+    },
+    {
+      "epoch": 141.33,
+      "learning_rate": 2.0526315789473685e-05,
+      "loss": 0.0002,
+      "step": 9469
+    },
+    {
+      "epoch": 141.34,
+      "learning_rate": 2.0491228070175437e-05,
+      "loss": 0.0002,
+      "step": 9470
+    },
+    {
+      "epoch": 141.36,
+      "learning_rate": 2.045614035087719e-05,
+      "loss": 0.0002,
+      "step": 9471
+    },
+    {
+      "epoch": 141.37,
+      "learning_rate": 2.0421052631578945e-05,
+      "loss": 0.0002,
+      "step": 9472
+    },
+    {
+      "epoch": 141.39,
+      "learning_rate": 2.0385964912280697e-05,
+      "loss": 0.0002,
+      "step": 9473
+    },
+    {
+      "epoch": 141.4,
+      "learning_rate": 2.0350877192982453e-05,
+      "loss": 0.0002,
+      "step": 9474
+    },
+    {
+      "epoch": 141.42,
+      "learning_rate": 2.0315789473684212e-05,
+      "loss": 0.0002,
+      "step": 9475
+    },
+    {
+      "epoch": 141.43,
+      "learning_rate": 2.0280701754385964e-05,
+      "loss": 0.0002,
+      "step": 9476
+    },
+    {
+      "epoch": 141.45,
+      "learning_rate": 2.0245614035087716e-05,
+      "loss": 0.0002,
+      "step": 9477
+    },
+    {
+      "epoch": 141.46,
+      "learning_rate": 2.0210526315789472e-05,
+      "loss": 0.0018,
+      "step": 9478
+    },
+    {
+      "epoch": 141.48,
+      "learning_rate": 2.0175438596491224e-05,
+      "loss": 0.0002,
+      "step": 9479
+    },
+    {
+      "epoch": 141.49,
+      "learning_rate": 2.014035087719298e-05,
+      "loss": 0.0002,
+      "step": 9480
+    },
+    {
+      "epoch": 141.51,
+      "learning_rate": 2.010526315789474e-05,
+      "loss": 0.0002,
+      "step": 9481
+    },
+    {
+      "epoch": 141.52,
+      "learning_rate": 2.007017543859649e-05,
+      "loss": 0.0002,
+      "step": 9482
+    },
+    {
+      "epoch": 141.54,
+      "learning_rate": 2.0035087719298243e-05,
+      "loss": 0.0002,
+      "step": 9483
+    },
+    {
+      "epoch": 141.55,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 0.0002,
+      "step": 9484
+    },
+    {
+      "epoch": 141.57,
+      "learning_rate": 1.996491228070175e-05,
+      "loss": 0.0002,
+      "step": 9485
+    },
+    {
+      "epoch": 141.58,
+      "learning_rate": 1.9929824561403506e-05,
+      "loss": 0.0002,
+      "step": 9486
+    },
+    {
+      "epoch": 141.59,
+      "learning_rate": 1.9894736842105265e-05,
+      "loss": 0.0002,
+      "step": 9487
+    },
+    {
+      "epoch": 141.61,
+      "learning_rate": 1.9859649122807017e-05,
+      "loss": 0.0002,
+      "step": 9488
+    },
+    {
+      "epoch": 141.62,
+      "learning_rate": 1.982456140350877e-05,
+      "loss": 0.0002,
+      "step": 9489
+    },
+    {
+      "epoch": 141.64,
+      "learning_rate": 1.9789473684210525e-05,
+      "loss": 0.0002,
+      "step": 9490
+    },
+    {
+      "epoch": 141.65,
+      "learning_rate": 1.9754385964912277e-05,
+      "loss": 0.0002,
+      "step": 9491
+    },
+    {
+      "epoch": 141.67,
+      "learning_rate": 1.9719298245614033e-05,
+      "loss": 0.0002,
+      "step": 9492
+    },
+    {
+      "epoch": 141.68,
+      "learning_rate": 1.968421052631579e-05,
+      "loss": 0.0002,
+      "step": 9493
+    },
+    {
+      "epoch": 141.7,
+      "learning_rate": 1.9649122807017544e-05,
+      "loss": 0.0002,
+      "step": 9494
+    },
+    {
+      "epoch": 141.71,
+      "learning_rate": 1.9614035087719296e-05,
+      "loss": 0.0002,
+      "step": 9495
+    },
+    {
+      "epoch": 141.73,
+      "learning_rate": 1.957894736842105e-05,
+      "loss": 0.0002,
+      "step": 9496
+    },
+    {
+      "epoch": 141.74,
+      "learning_rate": 1.9543859649122804e-05,
+      "loss": 0.0002,
+      "step": 9497
+    },
+    {
+      "epoch": 141.76,
+      "learning_rate": 1.950877192982456e-05,
+      "loss": 0.0005,
+      "step": 9498
+    },
+    {
+      "epoch": 141.77,
+      "learning_rate": 1.947368421052631e-05,
+      "loss": 0.0002,
+      "step": 9499
+    },
+    {
+      "epoch": 141.79,
+      "learning_rate": 1.943859649122807e-05,
+      "loss": 0.0002,
+      "step": 9500
+    },
+    {
+      "epoch": 141.8,
+      "learning_rate": 1.9403508771929822e-05,
+      "loss": 0.0002,
+      "step": 9501
+    },
+    {
+      "epoch": 141.82,
+      "learning_rate": 1.9368421052631578e-05,
+      "loss": 0.001,
+      "step": 9502
+    },
+    {
+      "epoch": 141.83,
+      "learning_rate": 1.933333333333333e-05,
+      "loss": 0.0002,
+      "step": 9503
+    },
+    {
+      "epoch": 141.85,
+      "learning_rate": 1.9298245614035086e-05,
+      "loss": 0.0014,
+      "step": 9504
+    },
+    {
+      "epoch": 141.86,
+      "learning_rate": 1.9263157894736838e-05,
+      "loss": 0.0002,
+      "step": 9505
+    },
+    {
+      "epoch": 141.88,
+      "learning_rate": 1.9228070175438597e-05,
+      "loss": 0.0002,
+      "step": 9506
+    },
+    {
+      "epoch": 141.89,
+      "learning_rate": 1.919298245614035e-05,
+      "loss": 0.0002,
+      "step": 9507
+    },
+    {
+      "epoch": 141.91,
+      "learning_rate": 1.9157894736842104e-05,
+      "loss": 0.0002,
+      "step": 9508
+    },
+    {
+      "epoch": 141.92,
+      "learning_rate": 1.9122807017543857e-05,
+      "loss": 0.0001,
+      "step": 9509
+    },
+    {
+      "epoch": 141.94,
+      "learning_rate": 1.9087719298245612e-05,
+      "loss": 0.0002,
+      "step": 9510
+    },
+    {
+      "epoch": 141.95,
+      "learning_rate": 1.9052631578947364e-05,
+      "loss": 0.0002,
+      "step": 9511
+    },
+    {
+      "epoch": 141.97,
+      "learning_rate": 1.9017543859649123e-05,
+      "loss": 0.0002,
+      "step": 9512
+    },
+    {
+      "epoch": 141.98,
+      "learning_rate": 1.8982456140350875e-05,
+      "loss": 0.0002,
+      "step": 9513
+    },
+    {
+      "epoch": 142.0,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 0.0014,
+      "step": 9514
+    },
+    {
+      "epoch": 142.01,
+      "learning_rate": 1.8912280701754383e-05,
+      "loss": 0.0002,
+      "step": 9515
+    },
+    {
+      "epoch": 142.03,
+      "learning_rate": 1.887719298245614e-05,
+      "loss": 0.0002,
+      "step": 9516
+    },
+    {
+      "epoch": 142.04,
+      "learning_rate": 1.884210526315789e-05,
+      "loss": 0.0002,
+      "step": 9517
+    },
+    {
+      "epoch": 142.06,
+      "learning_rate": 1.8807017543859646e-05,
+      "loss": 0.0002,
+      "step": 9518
+    },
+    {
+      "epoch": 142.07,
+      "learning_rate": 1.8771929824561402e-05,
+      "loss": 0.0002,
+      "step": 9519
+    },
+    {
+      "epoch": 142.09,
+      "learning_rate": 1.8736842105263158e-05,
+      "loss": 0.0003,
+      "step": 9520
+    },
+    {
+      "epoch": 142.1,
+      "learning_rate": 1.870175438596491e-05,
+      "loss": 0.0041,
+      "step": 9521
+    },
+    {
+      "epoch": 142.12,
+      "learning_rate": 1.8666666666666665e-05,
+      "loss": 0.0002,
+      "step": 9522
+    },
+    {
+      "epoch": 142.13,
+      "learning_rate": 1.863157894736842e-05,
+      "loss": 0.0011,
+      "step": 9523
+    },
+    {
+      "epoch": 142.15,
+      "learning_rate": 1.8596491228070173e-05,
+      "loss": 0.0002,
+      "step": 9524
+    },
+    {
+      "epoch": 142.16,
+      "learning_rate": 1.856140350877193e-05,
+      "loss": 0.0002,
+      "step": 9525
+    },
+    {
+      "epoch": 142.18,
+      "learning_rate": 1.8526315789473684e-05,
+      "loss": 0.0002,
+      "step": 9526
+    },
+    {
+      "epoch": 142.19,
+      "learning_rate": 1.8491228070175436e-05,
+      "loss": 0.0002,
+      "step": 9527
+    },
+    {
+      "epoch": 142.21,
+      "learning_rate": 1.8456140350877192e-05,
+      "loss": 0.0002,
+      "step": 9528
+    },
+    {
+      "epoch": 142.22,
+      "learning_rate": 1.8421052631578944e-05,
+      "loss": 0.0002,
+      "step": 9529
+    },
+    {
+      "epoch": 142.24,
+      "learning_rate": 1.83859649122807e-05,
+      "loss": 0.0003,
+      "step": 9530
+    },
+    {
+      "epoch": 142.25,
+      "learning_rate": 1.8350877192982455e-05,
+      "loss": 0.008,
+      "step": 9531
+    },
+    {
+      "epoch": 142.27,
+      "learning_rate": 1.8315789473684207e-05,
+      "loss": 0.0002,
+      "step": 9532
+    },
+    {
+      "epoch": 142.28,
+      "learning_rate": 1.8280701754385963e-05,
+      "loss": 0.0002,
+      "step": 9533
+    },
+    {
+      "epoch": 142.3,
+      "learning_rate": 1.824561403508772e-05,
+      "loss": 0.0002,
+      "step": 9534
+    },
+    {
+      "epoch": 142.31,
+      "learning_rate": 1.821052631578947e-05,
+      "loss": 0.0002,
+      "step": 9535
+    },
+    {
+      "epoch": 142.33,
+      "learning_rate": 1.8175438596491226e-05,
+      "loss": 0.0004,
+      "step": 9536
+    },
+    {
+      "epoch": 142.34,
+      "learning_rate": 1.814035087719298e-05,
+      "loss": 0.0002,
+      "step": 9537
+    },
+    {
+      "epoch": 142.36,
+      "learning_rate": 1.8105263157894734e-05,
+      "loss": 0.0004,
+      "step": 9538
+    },
+    {
+      "epoch": 142.37,
+      "learning_rate": 1.807017543859649e-05,
+      "loss": 0.0002,
+      "step": 9539
+    },
+    {
+      "epoch": 142.39,
+      "learning_rate": 1.8035087719298245e-05,
+      "loss": 0.0004,
+      "step": 9540
+    },
+    {
+      "epoch": 142.4,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 0.0002,
+      "step": 9541
+    },
+    {
+      "epoch": 142.42,
+      "learning_rate": 1.7964912280701753e-05,
+      "loss": 0.0002,
+      "step": 9542
+    },
+    {
+      "epoch": 142.43,
+      "learning_rate": 1.7929824561403508e-05,
+      "loss": 0.0002,
+      "step": 9543
+    },
+    {
+      "epoch": 142.45,
+      "learning_rate": 1.789473684210526e-05,
+      "loss": 0.0058,
+      "step": 9544
+    },
+    {
+      "epoch": 142.46,
+      "learning_rate": 1.7859649122807016e-05,
+      "loss": 0.0002,
+      "step": 9545
+    },
+    {
+      "epoch": 142.48,
+      "learning_rate": 1.782456140350877e-05,
+      "loss": 0.0003,
+      "step": 9546
+    },
+    {
+      "epoch": 142.49,
+      "learning_rate": 1.7789473684210524e-05,
+      "loss": 0.0002,
+      "step": 9547
+    },
+    {
+      "epoch": 142.51,
+      "learning_rate": 1.775438596491228e-05,
+      "loss": 0.0005,
+      "step": 9548
+    },
+    {
+      "epoch": 142.52,
+      "learning_rate": 1.7719298245614035e-05,
+      "loss": 0.0001,
+      "step": 9549
+    },
+    {
+      "epoch": 142.54,
+      "learning_rate": 1.7684210526315787e-05,
+      "loss": 0.0003,
+      "step": 9550
+    },
+    {
+      "epoch": 142.55,
+      "learning_rate": 1.7649122807017542e-05,
+      "loss": 0.0002,
+      "step": 9551
+    },
+    {
+      "epoch": 142.57,
+      "learning_rate": 1.7614035087719298e-05,
+      "loss": 0.0003,
+      "step": 9552
+    },
+    {
+      "epoch": 142.58,
+      "learning_rate": 1.757894736842105e-05,
+      "loss": 0.0002,
+      "step": 9553
+    },
+    {
+      "epoch": 142.59,
+      "learning_rate": 1.7543859649122806e-05,
+      "loss": 0.0002,
+      "step": 9554
+    },
+    {
+      "epoch": 142.61,
+      "learning_rate": 1.750877192982456e-05,
+      "loss": 0.0002,
+      "step": 9555
+    },
+    {
+      "epoch": 142.62,
+      "learning_rate": 1.7473684210526313e-05,
+      "loss": 0.0585,
+      "step": 9556
+    },
+    {
+      "epoch": 142.64,
+      "learning_rate": 1.743859649122807e-05,
+      "loss": 0.0004,
+      "step": 9557
+    },
+    {
+      "epoch": 142.65,
+      "learning_rate": 1.7403508771929825e-05,
+      "loss": 0.0002,
+      "step": 9558
+    },
+    {
+      "epoch": 142.67,
+      "learning_rate": 1.7368421052631577e-05,
+      "loss": 0.0002,
+      "step": 9559
+    },
+    {
+      "epoch": 142.68,
+      "learning_rate": 1.7333333333333332e-05,
+      "loss": 0.0002,
+      "step": 9560
+    },
+    {
+      "epoch": 142.7,
+      "learning_rate": 1.7298245614035088e-05,
+      "loss": 0.0002,
+      "step": 9561
+    },
+    {
+      "epoch": 142.71,
+      "learning_rate": 1.726315789473684e-05,
+      "loss": 0.0002,
+      "step": 9562
+    },
+    {
+      "epoch": 142.73,
+      "learning_rate": 1.7228070175438596e-05,
+      "loss": 0.0002,
+      "step": 9563
+    },
+    {
+      "epoch": 142.74,
+      "learning_rate": 1.719298245614035e-05,
+      "loss": 0.0002,
+      "step": 9564
+    },
+    {
+      "epoch": 142.76,
+      "learning_rate": 1.7157894736842103e-05,
+      "loss": 0.0002,
+      "step": 9565
+    },
+    {
+      "epoch": 142.77,
+      "learning_rate": 1.712280701754386e-05,
+      "loss": 0.0328,
+      "step": 9566
+    },
+    {
+      "epoch": 142.79,
+      "learning_rate": 1.7087719298245614e-05,
+      "loss": 0.0002,
+      "step": 9567
+    },
+    {
+      "epoch": 142.8,
+      "learning_rate": 1.7052631578947367e-05,
+      "loss": 0.0002,
+      "step": 9568
+    },
+    {
+      "epoch": 142.82,
+      "learning_rate": 1.7017543859649122e-05,
+      "loss": 0.0002,
+      "step": 9569
+    },
+    {
+      "epoch": 142.83,
+      "learning_rate": 1.6982456140350878e-05,
+      "loss": 0.0002,
+      "step": 9570
+    },
+    {
+      "epoch": 142.85,
+      "learning_rate": 1.694736842105263e-05,
+      "loss": 0.0002,
+      "step": 9571
+    },
+    {
+      "epoch": 142.86,
+      "learning_rate": 1.6912280701754385e-05,
+      "loss": 0.0002,
+      "step": 9572
+    },
+    {
+      "epoch": 142.88,
+      "learning_rate": 1.687719298245614e-05,
+      "loss": 0.0002,
+      "step": 9573
+    },
+    {
+      "epoch": 142.89,
+      "learning_rate": 1.6842105263157893e-05,
+      "loss": 0.0001,
+      "step": 9574
+    },
+    {
+      "epoch": 142.91,
+      "learning_rate": 1.680701754385965e-05,
+      "loss": 0.0001,
+      "step": 9575
+    },
+    {
+      "epoch": 142.92,
+      "learning_rate": 1.67719298245614e-05,
+      "loss": 0.0002,
+      "step": 9576
+    },
+    {
+      "epoch": 142.94,
+      "learning_rate": 1.6736842105263156e-05,
+      "loss": 0.0002,
+      "step": 9577
+    },
+    {
+      "epoch": 142.95,
+      "learning_rate": 1.6701754385964912e-05,
+      "loss": 0.0002,
+      "step": 9578
+    },
+    {
+      "epoch": 142.97,
+      "learning_rate": 1.6666666666666664e-05,
+      "loss": 0.0002,
+      "step": 9579
+    },
+    {
+      "epoch": 142.98,
+      "learning_rate": 1.663157894736842e-05,
+      "loss": 0.0002,
+      "step": 9580
+    },
+    {
+      "epoch": 143.0,
+      "learning_rate": 1.6596491228070175e-05,
+      "loss": 0.0002,
+      "step": 9581
+    },
+    {
+      "epoch": 143.01,
+      "learning_rate": 1.6561403508771927e-05,
+      "loss": 0.0002,
+      "step": 9582
+    },
+    {
+      "epoch": 143.03,
+      "learning_rate": 1.6526315789473683e-05,
+      "loss": 0.0002,
+      "step": 9583
+    },
+    {
+      "epoch": 143.04,
+      "learning_rate": 1.649122807017544e-05,
+      "loss": 0.0018,
+      "step": 9584
+    },
+    {
+      "epoch": 143.06,
+      "learning_rate": 1.645614035087719e-05,
+      "loss": 0.0002,
+      "step": 9585
+    },
+    {
+      "epoch": 143.07,
+      "learning_rate": 1.6421052631578946e-05,
+      "loss": 0.0169,
+      "step": 9586
+    },
+    {
+      "epoch": 143.09,
+      "learning_rate": 1.6385964912280702e-05,
+      "loss": 0.0004,
+      "step": 9587
+    },
+    {
+      "epoch": 143.1,
+      "learning_rate": 1.6350877192982454e-05,
+      "loss": 0.0002,
+      "step": 9588
+    },
+    {
+      "epoch": 143.12,
+      "learning_rate": 1.631578947368421e-05,
+      "loss": 0.0002,
+      "step": 9589
+    },
+    {
+      "epoch": 143.13,
+      "learning_rate": 1.6280701754385965e-05,
+      "loss": 0.0002,
+      "step": 9590
+    },
+    {
+      "epoch": 143.15,
+      "learning_rate": 1.6245614035087717e-05,
+      "loss": 0.0002,
+      "step": 9591
+    },
+    {
+      "epoch": 143.16,
+      "learning_rate": 1.6210526315789473e-05,
+      "loss": 0.0002,
+      "step": 9592
+    },
+    {
+      "epoch": 143.18,
+      "learning_rate": 1.6175438596491228e-05,
+      "loss": 0.0002,
+      "step": 9593
+    },
+    {
+      "epoch": 143.19,
+      "learning_rate": 1.614035087719298e-05,
+      "loss": 0.0002,
+      "step": 9594
+    },
+    {
+      "epoch": 143.21,
+      "learning_rate": 1.6105263157894736e-05,
+      "loss": 0.0002,
+      "step": 9595
+    },
+    {
+      "epoch": 143.22,
+      "learning_rate": 1.607017543859649e-05,
+      "loss": 0.0002,
+      "step": 9596
+    },
+    {
+      "epoch": 143.24,
+      "learning_rate": 1.6035087719298244e-05,
+      "loss": 0.0865,
+      "step": 9597
+    },
+    {
+      "epoch": 143.25,
+      "learning_rate": 1.6e-05,
+      "loss": 0.0002,
+      "step": 9598
+    },
+    {
+      "epoch": 143.27,
+      "learning_rate": 1.5964912280701755e-05,
+      "loss": 0.0003,
+      "step": 9599
+    },
+    {
+      "epoch": 143.28,
+      "learning_rate": 1.5929824561403507e-05,
+      "loss": 0.0002,
+      "step": 9600
+    },
+    {
+      "epoch": 143.28,
+      "eval_accuracy": 0.8913362701908958,
+      "eval_f1": 0.8906088969378417,
+      "eval_loss": 0.6095054745674133,
+      "eval_runtime": 344.9289,
+      "eval_samples_per_second": 11.846,
+      "eval_steps_per_second": 0.742,
+      "step": 9600
+    },
+    {
+      "epoch": 143.3,
+      "learning_rate": 1.589473684210526e-05,
+      "loss": 0.0002,
+      "step": 9601
+    },
+    {
+      "epoch": 143.31,
+      "learning_rate": 1.5859649122807018e-05,
+      "loss": 0.0002,
+      "step": 9602
+    },
+    {
+      "epoch": 143.33,
+      "learning_rate": 1.582456140350877e-05,
+      "loss": 0.0002,
+      "step": 9603
+    },
+    {
+      "epoch": 143.34,
+      "learning_rate": 1.5789473684210522e-05,
+      "loss": 0.0002,
+      "step": 9604
+    },
+    {
+      "epoch": 143.36,
+      "learning_rate": 1.575438596491228e-05,
+      "loss": 0.0002,
+      "step": 9605
+    },
+    {
+      "epoch": 143.37,
+      "learning_rate": 1.5719298245614034e-05,
+      "loss": 0.0002,
+      "step": 9606
+    },
+    {
+      "epoch": 143.39,
+      "learning_rate": 1.5684210526315786e-05,
+      "loss": 0.0002,
+      "step": 9607
+    },
+    {
+      "epoch": 143.4,
+      "learning_rate": 1.5649122807017545e-05,
+      "loss": 0.0006,
+      "step": 9608
+    },
+    {
+      "epoch": 143.42,
+      "learning_rate": 1.5614035087719297e-05,
+      "loss": 0.0002,
+      "step": 9609
+    },
+    {
+      "epoch": 143.43,
+      "learning_rate": 1.557894736842105e-05,
+      "loss": 0.0002,
+      "step": 9610
+    },
+    {
+      "epoch": 143.45,
+      "learning_rate": 1.5543859649122808e-05,
+      "loss": 0.1441,
+      "step": 9611
+    },
+    {
+      "epoch": 143.46,
+      "learning_rate": 1.550877192982456e-05,
+      "loss": 0.0002,
+      "step": 9612
+    },
+    {
+      "epoch": 143.48,
+      "learning_rate": 1.5473684210526312e-05,
+      "loss": 0.0002,
+      "step": 9613
+    },
+    {
+      "epoch": 143.49,
+      "learning_rate": 1.543859649122807e-05,
+      "loss": 0.0002,
+      "step": 9614
+    },
+    {
+      "epoch": 143.51,
+      "learning_rate": 1.5403508771929823e-05,
+      "loss": 0.0002,
+      "step": 9615
+    },
+    {
+      "epoch": 143.52,
+      "learning_rate": 1.5368421052631576e-05,
+      "loss": 0.0002,
+      "step": 9616
+    },
+    {
+      "epoch": 143.54,
+      "learning_rate": 1.5333333333333334e-05,
+      "loss": 0.0002,
+      "step": 9617
+    },
+    {
+      "epoch": 143.55,
+      "learning_rate": 1.5298245614035087e-05,
+      "loss": 0.0002,
+      "step": 9618
+    },
+    {
+      "epoch": 143.57,
+      "learning_rate": 1.526315789473684e-05,
+      "loss": 0.0002,
+      "step": 9619
+    },
+    {
+      "epoch": 143.58,
+      "learning_rate": 1.5228070175438596e-05,
+      "loss": 0.0002,
+      "step": 9620
+    },
+    {
+      "epoch": 143.59,
+      "learning_rate": 1.519298245614035e-05,
+      "loss": 0.0003,
+      "step": 9621
+    },
+    {
+      "epoch": 143.61,
+      "learning_rate": 1.5157894736842104e-05,
+      "loss": 0.0002,
+      "step": 9622
+    },
+    {
+      "epoch": 143.62,
+      "learning_rate": 1.5122807017543858e-05,
+      "loss": 0.0003,
+      "step": 9623
+    },
+    {
+      "epoch": 143.64,
+      "learning_rate": 1.5087719298245613e-05,
+      "loss": 0.0002,
+      "step": 9624
+    },
+    {
+      "epoch": 143.65,
+      "learning_rate": 1.5052631578947367e-05,
+      "loss": 0.0002,
+      "step": 9625
+    },
+    {
+      "epoch": 143.67,
+      "learning_rate": 1.5017543859649121e-05,
+      "loss": 0.0002,
+      "step": 9626
+    },
+    {
+      "epoch": 143.68,
+      "learning_rate": 1.4982456140350876e-05,
+      "loss": 0.0007,
+      "step": 9627
+    },
+    {
+      "epoch": 143.7,
+      "learning_rate": 1.494736842105263e-05,
+      "loss": 0.0002,
+      "step": 9628
+    },
+    {
+      "epoch": 143.71,
+      "learning_rate": 1.4912280701754384e-05,
+      "loss": 0.0002,
+      "step": 9629
+    },
+    {
+      "epoch": 143.73,
+      "learning_rate": 1.487719298245614e-05,
+      "loss": 0.0002,
+      "step": 9630
+    },
+    {
+      "epoch": 143.74,
+      "learning_rate": 1.4842105263157894e-05,
+      "loss": 0.0002,
+      "step": 9631
+    },
+    {
+      "epoch": 143.76,
+      "learning_rate": 1.4807017543859647e-05,
+      "loss": 0.0002,
+      "step": 9632
+    },
+    {
+      "epoch": 143.77,
+      "learning_rate": 1.4771929824561403e-05,
+      "loss": 0.0002,
+      "step": 9633
+    },
+    {
+      "epoch": 143.79,
+      "learning_rate": 1.4736842105263157e-05,
+      "loss": 0.0002,
+      "step": 9634
+    },
+    {
+      "epoch": 143.8,
+      "learning_rate": 1.470175438596491e-05,
+      "loss": 0.0003,
+      "step": 9635
+    },
+    {
+      "epoch": 143.82,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 0.1042,
+      "step": 9636
+    },
+    {
+      "epoch": 143.83,
+      "learning_rate": 1.463157894736842e-05,
+      "loss": 0.0003,
+      "step": 9637
+    },
+    {
+      "epoch": 143.85,
+      "learning_rate": 1.4596491228070174e-05,
+      "loss": 0.0002,
+      "step": 9638
+    },
+    {
+      "epoch": 143.86,
+      "learning_rate": 1.456140350877193e-05,
+      "loss": 0.0002,
+      "step": 9639
+    },
+    {
+      "epoch": 143.88,
+      "learning_rate": 1.4526315789473683e-05,
+      "loss": 0.0002,
+      "step": 9640
+    },
+    {
+      "epoch": 143.89,
+      "learning_rate": 1.4491228070175437e-05,
+      "loss": 0.0002,
+      "step": 9641
+    },
+    {
+      "epoch": 143.91,
+      "learning_rate": 1.4456140350877193e-05,
+      "loss": 0.0002,
+      "step": 9642
+    },
+    {
+      "epoch": 143.92,
+      "learning_rate": 1.4421052631578947e-05,
+      "loss": 0.0002,
+      "step": 9643
+    },
+    {
+      "epoch": 143.94,
+      "learning_rate": 1.43859649122807e-05,
+      "loss": 0.0001,
+      "step": 9644
+    },
+    {
+      "epoch": 143.95,
+      "learning_rate": 1.4350877192982456e-05,
+      "loss": 0.0002,
+      "step": 9645
+    },
+    {
+      "epoch": 143.97,
+      "learning_rate": 1.431578947368421e-05,
+      "loss": 0.0002,
+      "step": 9646
+    },
+    {
+      "epoch": 143.98,
+      "learning_rate": 1.4280701754385964e-05,
+      "loss": 0.0945,
+      "step": 9647
+    },
+    {
+      "epoch": 144.0,
+      "learning_rate": 1.4245614035087718e-05,
+      "loss": 0.0022,
+      "step": 9648
+    },
+    {
+      "epoch": 144.01,
+      "learning_rate": 1.4210526315789473e-05,
+      "loss": 0.0151,
+      "step": 9649
+    },
+    {
+      "epoch": 144.03,
+      "learning_rate": 1.4175438596491227e-05,
+      "loss": 0.0002,
+      "step": 9650
+    },
+    {
+      "epoch": 144.04,
+      "learning_rate": 1.4140350877192981e-05,
+      "loss": 0.0002,
+      "step": 9651
+    },
+    {
+      "epoch": 144.06,
+      "learning_rate": 1.4105263157894737e-05,
+      "loss": 0.0002,
+      "step": 9652
+    },
+    {
+      "epoch": 144.07,
+      "learning_rate": 1.407017543859649e-05,
+      "loss": 0.0001,
+      "step": 9653
+    },
+    {
+      "epoch": 144.09,
+      "learning_rate": 1.4035087719298244e-05,
+      "loss": 0.0002,
+      "step": 9654
+    },
+    {
+      "epoch": 144.1,
+      "learning_rate": 1.4e-05,
+      "loss": 0.0003,
+      "step": 9655
+    },
+    {
+      "epoch": 144.12,
+      "learning_rate": 1.3964912280701754e-05,
+      "loss": 0.0001,
+      "step": 9656
+    },
+    {
+      "epoch": 144.13,
+      "learning_rate": 1.3929824561403508e-05,
+      "loss": 0.0002,
+      "step": 9657
+    },
+    {
+      "epoch": 144.15,
+      "learning_rate": 1.3894736842105263e-05,
+      "loss": 0.0001,
+      "step": 9658
+    },
+    {
+      "epoch": 144.16,
+      "learning_rate": 1.3859649122807017e-05,
+      "loss": 0.0002,
+      "step": 9659
+    },
+    {
+      "epoch": 144.18,
+      "learning_rate": 1.382456140350877e-05,
+      "loss": 0.0002,
+      "step": 9660
+    },
+    {
+      "epoch": 144.19,
+      "learning_rate": 1.3789473684210526e-05,
+      "loss": 0.0002,
+      "step": 9661
+    },
+    {
+      "epoch": 144.21,
+      "learning_rate": 1.375438596491228e-05,
+      "loss": 0.0002,
+      "step": 9662
+    },
+    {
+      "epoch": 144.22,
+      "learning_rate": 1.3719298245614034e-05,
+      "loss": 0.0002,
+      "step": 9663
+    },
+    {
+      "epoch": 144.24,
+      "learning_rate": 1.368421052631579e-05,
+      "loss": 0.0002,
+      "step": 9664
+    },
+    {
+      "epoch": 144.25,
+      "learning_rate": 1.3649122807017543e-05,
+      "loss": 0.0002,
+      "step": 9665
+    },
+    {
+      "epoch": 144.27,
+      "learning_rate": 1.3614035087719297e-05,
+      "loss": 0.0002,
+      "step": 9666
+    },
+    {
+      "epoch": 144.28,
+      "learning_rate": 1.3578947368421053e-05,
+      "loss": 0.0002,
+      "step": 9667
+    },
+    {
+      "epoch": 144.3,
+      "learning_rate": 1.3543859649122807e-05,
+      "loss": 0.0003,
+      "step": 9668
+    },
+    {
+      "epoch": 144.31,
+      "learning_rate": 1.350877192982456e-05,
+      "loss": 0.0002,
+      "step": 9669
+    },
+    {
+      "epoch": 144.33,
+      "learning_rate": 1.3473684210526313e-05,
+      "loss": 0.0004,
+      "step": 9670
+    },
+    {
+      "epoch": 144.34,
+      "learning_rate": 1.343859649122807e-05,
+      "loss": 0.0002,
+      "step": 9671
+    },
+    {
+      "epoch": 144.36,
+      "learning_rate": 1.3403508771929824e-05,
+      "loss": 0.0098,
+      "step": 9672
+    },
+    {
+      "epoch": 144.37,
+      "learning_rate": 1.3368421052631576e-05,
+      "loss": 0.0002,
+      "step": 9673
+    },
+    {
+      "epoch": 144.39,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.0002,
+      "step": 9674
+    },
+    {
+      "epoch": 144.4,
+      "learning_rate": 1.3298245614035087e-05,
+      "loss": 0.0002,
+      "step": 9675
+    },
+    {
+      "epoch": 144.42,
+      "learning_rate": 1.326315789473684e-05,
+      "loss": 0.0001,
+      "step": 9676
+    },
+    {
+      "epoch": 144.43,
+      "learning_rate": 1.3228070175438597e-05,
+      "loss": 0.0001,
+      "step": 9677
+    },
+    {
+      "epoch": 144.45,
+      "learning_rate": 1.319298245614035e-05,
+      "loss": 0.0002,
+      "step": 9678
+    },
+    {
+      "epoch": 144.46,
+      "learning_rate": 1.3157894736842103e-05,
+      "loss": 0.0068,
+      "step": 9679
+    },
+    {
+      "epoch": 144.48,
+      "learning_rate": 1.312280701754386e-05,
+      "loss": 0.0002,
+      "step": 9680
+    },
+    {
+      "epoch": 144.49,
+      "learning_rate": 1.3087719298245614e-05,
+      "loss": 0.0005,
+      "step": 9681
+    },
+    {
+      "epoch": 144.51,
+      "learning_rate": 1.3052631578947366e-05,
+      "loss": 0.0002,
+      "step": 9682
+    },
+    {
+      "epoch": 144.52,
+      "learning_rate": 1.3017543859649123e-05,
+      "loss": 0.0002,
+      "step": 9683
+    },
+    {
+      "epoch": 144.54,
+      "learning_rate": 1.2982456140350877e-05,
+      "loss": 0.0002,
+      "step": 9684
+    },
+    {
+      "epoch": 144.55,
+      "learning_rate": 1.2947368421052629e-05,
+      "loss": 0.049,
+      "step": 9685
+    },
+    {
+      "epoch": 144.57,
+      "learning_rate": 1.2912280701754386e-05,
+      "loss": 0.0002,
+      "step": 9686
+    },
+    {
+      "epoch": 144.58,
+      "learning_rate": 1.287719298245614e-05,
+      "loss": 0.3249,
+      "step": 9687
+    },
+    {
+      "epoch": 144.59,
+      "learning_rate": 1.2842105263157892e-05,
+      "loss": 0.0002,
+      "step": 9688
+    },
+    {
+      "epoch": 144.61,
+      "learning_rate": 1.280701754385965e-05,
+      "loss": 0.0002,
+      "step": 9689
+    },
+    {
+      "epoch": 144.62,
+      "learning_rate": 1.2771929824561404e-05,
+      "loss": 0.0002,
+      "step": 9690
+    },
+    {
+      "epoch": 144.64,
+      "learning_rate": 1.2736842105263156e-05,
+      "loss": 0.0002,
+      "step": 9691
+    },
+    {
+      "epoch": 144.65,
+      "learning_rate": 1.270175438596491e-05,
+      "loss": 0.0001,
+      "step": 9692
+    },
+    {
+      "epoch": 144.67,
+      "learning_rate": 1.2666666666666665e-05,
+      "loss": 0.0002,
+      "step": 9693
+    },
+    {
+      "epoch": 144.68,
+      "learning_rate": 1.2631578947368419e-05,
+      "loss": 0.0002,
+      "step": 9694
+    },
+    {
+      "epoch": 144.7,
+      "learning_rate": 1.2596491228070173e-05,
+      "loss": 0.0002,
+      "step": 9695
+    },
+    {
+      "epoch": 144.71,
+      "learning_rate": 1.2561403508771928e-05,
+      "loss": 0.0002,
+      "step": 9696
+    },
+    {
+      "epoch": 144.73,
+      "learning_rate": 1.2526315789473682e-05,
+      "loss": 0.0002,
+      "step": 9697
+    },
+    {
+      "epoch": 144.74,
+      "learning_rate": 1.2491228070175436e-05,
+      "loss": 0.0001,
+      "step": 9698
+    },
+    {
+      "epoch": 144.76,
+      "learning_rate": 1.2456140350877192e-05,
+      "loss": 0.0012,
+      "step": 9699
+    },
+    {
+      "epoch": 144.77,
+      "learning_rate": 1.2421052631578946e-05,
+      "loss": 0.0013,
+      "step": 9700
+    },
+    {
+      "epoch": 144.79,
+      "learning_rate": 1.23859649122807e-05,
+      "loss": 0.0001,
+      "step": 9701
+    },
+    {
+      "epoch": 144.8,
+      "learning_rate": 1.2350877192982455e-05,
+      "loss": 0.0005,
+      "step": 9702
+    },
+    {
+      "epoch": 144.82,
+      "learning_rate": 1.2315789473684209e-05,
+      "loss": 0.0002,
+      "step": 9703
+    },
+    {
+      "epoch": 144.83,
+      "learning_rate": 1.2280701754385963e-05,
+      "loss": 0.0002,
+      "step": 9704
+    },
+    {
+      "epoch": 144.85,
+      "learning_rate": 1.2245614035087718e-05,
+      "loss": 0.0002,
+      "step": 9705
+    },
+    {
+      "epoch": 144.86,
+      "learning_rate": 1.2210526315789472e-05,
+      "loss": 0.0003,
+      "step": 9706
+    },
+    {
+      "epoch": 144.88,
+      "learning_rate": 1.2175438596491226e-05,
+      "loss": 0.0002,
+      "step": 9707
+    },
+    {
+      "epoch": 144.89,
+      "learning_rate": 1.2140350877192981e-05,
+      "loss": 0.0002,
+      "step": 9708
+    },
+    {
+      "epoch": 144.91,
+      "learning_rate": 1.2105263157894735e-05,
+      "loss": 0.0002,
+      "step": 9709
+    },
+    {
+      "epoch": 144.92,
+      "learning_rate": 1.207017543859649e-05,
+      "loss": 0.0002,
+      "step": 9710
+    },
+    {
+      "epoch": 144.94,
+      "learning_rate": 1.2035087719298245e-05,
+      "loss": 0.0002,
+      "step": 9711
+    },
+    {
+      "epoch": 144.95,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 0.0002,
+      "step": 9712
+    },
+    {
+      "epoch": 144.97,
+      "learning_rate": 1.1964912280701752e-05,
+      "loss": 0.0002,
+      "step": 9713
+    },
+    {
+      "epoch": 144.98,
+      "learning_rate": 1.1929824561403508e-05,
+      "loss": 0.0002,
+      "step": 9714
+    },
+    {
+      "epoch": 145.0,
+      "learning_rate": 1.1894736842105262e-05,
+      "loss": 0.0002,
+      "step": 9715
+    },
+    {
+      "epoch": 145.01,
+      "learning_rate": 1.1859649122807016e-05,
+      "loss": 0.0002,
+      "step": 9716
+    },
+    {
+      "epoch": 145.03,
+      "learning_rate": 1.182456140350877e-05,
+      "loss": 0.0542,
+      "step": 9717
+    },
+    {
+      "epoch": 145.04,
+      "learning_rate": 1.1789473684210525e-05,
+      "loss": 0.0002,
+      "step": 9718
+    },
+    {
+      "epoch": 145.06,
+      "learning_rate": 1.1754385964912279e-05,
+      "loss": 0.0002,
+      "step": 9719
+    },
+    {
+      "epoch": 145.07,
+      "learning_rate": 1.1719298245614033e-05,
+      "loss": 0.0002,
+      "step": 9720
+    },
+    {
+      "epoch": 145.09,
+      "learning_rate": 1.1684210526315788e-05,
+      "loss": 0.0002,
+      "step": 9721
+    },
+    {
+      "epoch": 145.1,
+      "learning_rate": 1.1649122807017542e-05,
+      "loss": 0.0002,
+      "step": 9722
+    },
+    {
+      "epoch": 145.12,
+      "learning_rate": 1.1614035087719296e-05,
+      "loss": 0.0002,
+      "step": 9723
+    },
+    {
+      "epoch": 145.13,
+      "learning_rate": 1.1578947368421052e-05,
+      "loss": 0.0002,
+      "step": 9724
+    },
+    {
+      "epoch": 145.15,
+      "learning_rate": 1.1543859649122806e-05,
+      "loss": 0.0002,
+      "step": 9725
+    },
+    {
+      "epoch": 145.16,
+      "learning_rate": 1.150877192982456e-05,
+      "loss": 0.0002,
+      "step": 9726
+    },
+    {
+      "epoch": 145.18,
+      "learning_rate": 1.1473684210526315e-05,
+      "loss": 0.0002,
+      "step": 9727
+    },
+    {
+      "epoch": 145.19,
+      "learning_rate": 1.1438596491228069e-05,
+      "loss": 0.0002,
+      "step": 9728
+    },
+    {
+      "epoch": 145.21,
+      "learning_rate": 1.1403508771929823e-05,
+      "loss": 0.0002,
+      "step": 9729
+    },
+    {
+      "epoch": 145.22,
+      "learning_rate": 1.1368421052631578e-05,
+      "loss": 0.0002,
+      "step": 9730
+    },
+    {
+      "epoch": 145.24,
+      "learning_rate": 1.1333333333333332e-05,
+      "loss": 0.0002,
+      "step": 9731
+    },
+    {
+      "epoch": 145.25,
+      "learning_rate": 1.1298245614035086e-05,
+      "loss": 0.0002,
+      "step": 9732
+    },
+    {
+      "epoch": 145.27,
+      "learning_rate": 1.1263157894736842e-05,
+      "loss": 0.0004,
+      "step": 9733
+    },
+    {
+      "epoch": 145.28,
+      "learning_rate": 1.1228070175438595e-05,
+      "loss": 0.0002,
+      "step": 9734
+    },
+    {
+      "epoch": 145.3,
+      "learning_rate": 1.119298245614035e-05,
+      "loss": 0.0002,
+      "step": 9735
+    },
+    {
+      "epoch": 145.31,
+      "learning_rate": 1.1157894736842105e-05,
+      "loss": 0.0009,
+      "step": 9736
+    },
+    {
+      "epoch": 145.33,
+      "learning_rate": 1.1122807017543859e-05,
+      "loss": 0.0002,
+      "step": 9737
+    },
+    {
+      "epoch": 145.34,
+      "learning_rate": 1.1087719298245613e-05,
+      "loss": 0.0002,
+      "step": 9738
+    },
+    {
+      "epoch": 145.36,
+      "learning_rate": 1.1052631578947366e-05,
+      "loss": 0.0002,
+      "step": 9739
+    },
+    {
+      "epoch": 145.37,
+      "learning_rate": 1.1017543859649122e-05,
+      "loss": 0.0002,
+      "step": 9740
+    },
+    {
+      "epoch": 145.39,
+      "learning_rate": 1.0982456140350876e-05,
+      "loss": 0.0002,
+      "step": 9741
+    },
+    {
+      "epoch": 145.4,
+      "learning_rate": 1.094736842105263e-05,
+      "loss": 0.0002,
+      "step": 9742
+    },
+    {
+      "epoch": 145.42,
+      "learning_rate": 1.0912280701754385e-05,
+      "loss": 0.0002,
+      "step": 9743
+    },
+    {
+      "epoch": 145.43,
+      "learning_rate": 1.0877192982456139e-05,
+      "loss": 0.0002,
+      "step": 9744
+    },
+    {
+      "epoch": 145.45,
+      "learning_rate": 1.0842105263157893e-05,
+      "loss": 0.0002,
+      "step": 9745
+    },
+    {
+      "epoch": 145.46,
+      "learning_rate": 1.0807017543859648e-05,
+      "loss": 0.0002,
+      "step": 9746
+    },
+    {
+      "epoch": 145.48,
+      "learning_rate": 1.0771929824561402e-05,
+      "loss": 0.0002,
+      "step": 9747
+    },
+    {
+      "epoch": 145.49,
+      "learning_rate": 1.0736842105263156e-05,
+      "loss": 0.0002,
+      "step": 9748
+    },
+    {
+      "epoch": 145.51,
+      "learning_rate": 1.0701754385964912e-05,
+      "loss": 0.0002,
+      "step": 9749
+    },
+    {
+      "epoch": 145.52,
+      "learning_rate": 1.0666666666666666e-05,
+      "loss": 0.0006,
+      "step": 9750
+    },
+    {
+      "epoch": 145.54,
+      "learning_rate": 1.063157894736842e-05,
+      "loss": 0.0002,
+      "step": 9751
+    },
+    {
+      "epoch": 145.55,
+      "learning_rate": 1.0596491228070175e-05,
+      "loss": 0.0002,
+      "step": 9752
+    },
+    {
+      "epoch": 145.57,
+      "learning_rate": 1.0561403508771929e-05,
+      "loss": 0.0002,
+      "step": 9753
+    },
+    {
+      "epoch": 145.58,
+      "learning_rate": 1.0526315789473683e-05,
+      "loss": 0.0002,
+      "step": 9754
+    },
+    {
+      "epoch": 145.59,
+      "learning_rate": 1.0491228070175438e-05,
+      "loss": 0.0002,
+      "step": 9755
+    },
+    {
+      "epoch": 145.61,
+      "learning_rate": 1.0456140350877192e-05,
+      "loss": 0.0002,
+      "step": 9756
+    },
+    {
+      "epoch": 145.62,
+      "learning_rate": 1.0421052631578946e-05,
+      "loss": 0.0002,
+      "step": 9757
+    },
+    {
+      "epoch": 145.64,
+      "learning_rate": 1.0385964912280702e-05,
+      "loss": 0.0002,
+      "step": 9758
+    },
+    {
+      "epoch": 145.65,
+      "learning_rate": 1.0350877192982455e-05,
+      "loss": 0.0002,
+      "step": 9759
+    },
+    {
+      "epoch": 145.67,
+      "learning_rate": 1.031578947368421e-05,
+      "loss": 0.0001,
+      "step": 9760
+    },
+    {
+      "epoch": 145.68,
+      "learning_rate": 1.0280701754385965e-05,
+      "loss": 0.0002,
+      "step": 9761
+    },
+    {
+      "epoch": 145.7,
+      "learning_rate": 1.0245614035087719e-05,
+      "loss": 0.0004,
+      "step": 9762
+    },
+    {
+      "epoch": 145.71,
+      "learning_rate": 1.0210526315789473e-05,
+      "loss": 0.0005,
+      "step": 9763
+    },
+    {
+      "epoch": 145.73,
+      "learning_rate": 1.0175438596491226e-05,
+      "loss": 0.0002,
+      "step": 9764
+    },
+    {
+      "epoch": 145.74,
+      "learning_rate": 1.0140350877192982e-05,
+      "loss": 0.0002,
+      "step": 9765
+    },
+    {
+      "epoch": 145.76,
+      "learning_rate": 1.0105263157894736e-05,
+      "loss": 0.0002,
+      "step": 9766
+    },
+    {
+      "epoch": 145.77,
+      "learning_rate": 1.007017543859649e-05,
+      "loss": 0.0002,
+      "step": 9767
+    },
+    {
+      "epoch": 145.79,
+      "learning_rate": 1.0035087719298245e-05,
+      "loss": 0.0001,
+      "step": 9768
+    },
+    {
+      "epoch": 145.8,
+      "learning_rate": 9.999999999999999e-06,
+      "loss": 0.0002,
+      "step": 9769
+    },
+    {
+      "epoch": 145.82,
+      "learning_rate": 9.964912280701753e-06,
+      "loss": 0.0003,
+      "step": 9770
+    },
+    {
+      "epoch": 145.83,
+      "learning_rate": 9.929824561403509e-06,
+      "loss": 0.0002,
+      "step": 9771
+    },
+    {
+      "epoch": 145.85,
+      "learning_rate": 9.894736842105262e-06,
+      "loss": 0.0002,
+      "step": 9772
+    },
+    {
+      "epoch": 145.86,
+      "learning_rate": 9.859649122807016e-06,
+      "loss": 0.0009,
+      "step": 9773
+    },
+    {
+      "epoch": 145.88,
+      "learning_rate": 9.824561403508772e-06,
+      "loss": 0.0002,
+      "step": 9774
+    },
+    {
+      "epoch": 145.89,
+      "learning_rate": 9.789473684210526e-06,
+      "loss": 0.0002,
+      "step": 9775
+    },
+    {
+      "epoch": 145.91,
+      "learning_rate": 9.75438596491228e-06,
+      "loss": 0.0802,
+      "step": 9776
+    },
+    {
+      "epoch": 145.92,
+      "learning_rate": 9.719298245614035e-06,
+      "loss": 0.029,
+      "step": 9777
+    },
+    {
+      "epoch": 145.94,
+      "learning_rate": 9.684210526315789e-06,
+      "loss": 0.0002,
+      "step": 9778
+    },
+    {
+      "epoch": 145.95,
+      "learning_rate": 9.649122807017543e-06,
+      "loss": 0.0002,
+      "step": 9779
+    },
+    {
+      "epoch": 145.97,
+      "learning_rate": 9.614035087719298e-06,
+      "loss": 0.0002,
+      "step": 9780
+    },
+    {
+      "epoch": 145.98,
+      "learning_rate": 9.578947368421052e-06,
+      "loss": 0.0001,
+      "step": 9781
+    },
+    {
+      "epoch": 146.0,
+      "learning_rate": 9.543859649122806e-06,
+      "loss": 0.0002,
+      "step": 9782
+    },
+    {
+      "epoch": 146.01,
+      "learning_rate": 9.508771929824562e-06,
+      "loss": 0.0002,
+      "step": 9783
+    },
+    {
+      "epoch": 146.03,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 0.0002,
+      "step": 9784
+    },
+    {
+      "epoch": 146.04,
+      "learning_rate": 9.43859649122807e-06,
+      "loss": 0.0003,
+      "step": 9785
+    },
+    {
+      "epoch": 146.06,
+      "learning_rate": 9.403508771929823e-06,
+      "loss": 0.0002,
+      "step": 9786
+    },
+    {
+      "epoch": 146.07,
+      "learning_rate": 9.368421052631579e-06,
+      "loss": 0.0002,
+      "step": 9787
+    },
+    {
+      "epoch": 146.09,
+      "learning_rate": 9.333333333333333e-06,
+      "loss": 0.0002,
+      "step": 9788
+    },
+    {
+      "epoch": 146.1,
+      "learning_rate": 9.298245614035086e-06,
+      "loss": 0.0002,
+      "step": 9789
+    },
+    {
+      "epoch": 146.12,
+      "learning_rate": 9.263157894736842e-06,
+      "loss": 0.0002,
+      "step": 9790
+    },
+    {
+      "epoch": 146.13,
+      "learning_rate": 9.228070175438596e-06,
+      "loss": 0.0002,
+      "step": 9791
+    },
+    {
+      "epoch": 146.15,
+      "learning_rate": 9.19298245614035e-06,
+      "loss": 0.0002,
+      "step": 9792
+    },
+    {
+      "epoch": 146.16,
+      "learning_rate": 9.157894736842104e-06,
+      "loss": 0.0002,
+      "step": 9793
+    },
+    {
+      "epoch": 146.18,
+      "learning_rate": 9.12280701754386e-06,
+      "loss": 0.0002,
+      "step": 9794
+    },
+    {
+      "epoch": 146.19,
+      "learning_rate": 9.087719298245613e-06,
+      "loss": 0.0002,
+      "step": 9795
+    },
+    {
+      "epoch": 146.21,
+      "learning_rate": 9.052631578947367e-06,
+      "loss": 0.0002,
+      "step": 9796
+    },
+    {
+      "epoch": 146.22,
+      "learning_rate": 9.017543859649122e-06,
+      "loss": 0.0004,
+      "step": 9797
+    },
+    {
+      "epoch": 146.24,
+      "learning_rate": 8.982456140350876e-06,
+      "loss": 0.0002,
+      "step": 9798
+    },
+    {
+      "epoch": 146.25,
+      "learning_rate": 8.94736842105263e-06,
+      "loss": 0.0001,
+      "step": 9799
+    },
+    {
+      "epoch": 146.27,
+      "learning_rate": 8.912280701754386e-06,
+      "loss": 0.0002,
+      "step": 9800
+    },
+    {
+      "epoch": 146.27,
+      "eval_accuracy": 0.8913362701908958,
+      "eval_f1": 0.8909876245610147,
+      "eval_loss": 0.6056233644485474,
+      "eval_runtime": 345.3935,
+      "eval_samples_per_second": 11.83,
+      "eval_steps_per_second": 0.741,
+      "step": 9800
+    },
+    {
+      "epoch": 146.28,
+      "learning_rate": 8.87719298245614e-06,
+      "loss": 0.0002,
+      "step": 9801
+    },
+    {
+      "epoch": 146.3,
+      "learning_rate": 8.842105263157893e-06,
+      "loss": 0.0002,
+      "step": 9802
+    },
+    {
+      "epoch": 146.31,
+      "learning_rate": 8.807017543859649e-06,
+      "loss": 0.0002,
+      "step": 9803
+    },
+    {
+      "epoch": 146.33,
+      "learning_rate": 8.771929824561403e-06,
+      "loss": 0.0002,
+      "step": 9804
+    },
+    {
+      "epoch": 146.34,
+      "learning_rate": 8.736842105263157e-06,
+      "loss": 0.0002,
+      "step": 9805
+    },
+    {
+      "epoch": 146.36,
+      "learning_rate": 8.701754385964912e-06,
+      "loss": 0.0696,
+      "step": 9806
+    },
+    {
+      "epoch": 146.37,
+      "learning_rate": 8.666666666666666e-06,
+      "loss": 0.0002,
+      "step": 9807
+    },
+    {
+      "epoch": 146.39,
+      "learning_rate": 8.63157894736842e-06,
+      "loss": 0.0002,
+      "step": 9808
+    },
+    {
+      "epoch": 146.4,
+      "learning_rate": 8.596491228070176e-06,
+      "loss": 0.0002,
+      "step": 9809
+    },
+    {
+      "epoch": 146.42,
+      "learning_rate": 8.56140350877193e-06,
+      "loss": 0.0002,
+      "step": 9810
+    },
+    {
+      "epoch": 146.43,
+      "learning_rate": 8.526315789473683e-06,
+      "loss": 0.0002,
+      "step": 9811
+    },
+    {
+      "epoch": 146.45,
+      "learning_rate": 8.491228070175439e-06,
+      "loss": 0.0002,
+      "step": 9812
+    },
+    {
+      "epoch": 146.46,
+      "learning_rate": 8.456140350877193e-06,
+      "loss": 0.0002,
+      "step": 9813
+    },
+    {
+      "epoch": 146.48,
+      "learning_rate": 8.421052631578947e-06,
+      "loss": 0.0002,
+      "step": 9814
+    },
+    {
+      "epoch": 146.49,
+      "learning_rate": 8.3859649122807e-06,
+      "loss": 0.1118,
+      "step": 9815
+    },
+    {
+      "epoch": 146.51,
+      "learning_rate": 8.350877192982456e-06,
+      "loss": 0.0002,
+      "step": 9816
+    },
+    {
+      "epoch": 146.52,
+      "learning_rate": 8.31578947368421e-06,
+      "loss": 0.0014,
+      "step": 9817
+    },
+    {
+      "epoch": 146.54,
+      "learning_rate": 8.280701754385964e-06,
+      "loss": 0.0001,
+      "step": 9818
+    },
+    {
+      "epoch": 146.55,
+      "learning_rate": 8.24561403508772e-06,
+      "loss": 0.0002,
+      "step": 9819
+    },
+    {
+      "epoch": 146.57,
+      "learning_rate": 8.210526315789473e-06,
+      "loss": 0.0002,
+      "step": 9820
+    },
+    {
+      "epoch": 146.58,
+      "learning_rate": 8.175438596491227e-06,
+      "loss": 0.0005,
+      "step": 9821
+    },
+    {
+      "epoch": 146.59,
+      "learning_rate": 8.140350877192983e-06,
+      "loss": 0.0003,
+      "step": 9822
+    },
+    {
+      "epoch": 146.61,
+      "learning_rate": 8.105263157894736e-06,
+      "loss": 0.0002,
+      "step": 9823
+    },
+    {
+      "epoch": 146.62,
+      "learning_rate": 8.07017543859649e-06,
+      "loss": 0.0002,
+      "step": 9824
+    },
+    {
+      "epoch": 146.64,
+      "learning_rate": 8.035087719298246e-06,
+      "loss": 0.0002,
+      "step": 9825
+    },
+    {
+      "epoch": 146.65,
+      "learning_rate": 8e-06,
+      "loss": 0.0002,
+      "step": 9826
+    },
+    {
+      "epoch": 146.67,
+      "learning_rate": 7.964912280701753e-06,
+      "loss": 0.0012,
+      "step": 9827
+    },
+    {
+      "epoch": 146.68,
+      "learning_rate": 7.929824561403509e-06,
+      "loss": 0.0002,
+      "step": 9828
+    },
+    {
+      "epoch": 146.7,
+      "learning_rate": 7.894736842105261e-06,
+      "loss": 0.0003,
+      "step": 9829
+    },
+    {
+      "epoch": 146.71,
+      "learning_rate": 7.859649122807017e-06,
+      "loss": 0.0002,
+      "step": 9830
+    },
+    {
+      "epoch": 146.73,
+      "learning_rate": 7.824561403508772e-06,
+      "loss": 0.0002,
+      "step": 9831
+    },
+    {
+      "epoch": 146.74,
+      "learning_rate": 7.789473684210524e-06,
+      "loss": 0.0002,
+      "step": 9832
+    },
+    {
+      "epoch": 146.76,
+      "learning_rate": 7.75438596491228e-06,
+      "loss": 0.0002,
+      "step": 9833
+    },
+    {
+      "epoch": 146.77,
+      "learning_rate": 7.719298245614036e-06,
+      "loss": 0.0002,
+      "step": 9834
+    },
+    {
+      "epoch": 146.79,
+      "learning_rate": 7.684210526315788e-06,
+      "loss": 0.0002,
+      "step": 9835
+    },
+    {
+      "epoch": 146.8,
+      "learning_rate": 7.649122807017543e-06,
+      "loss": 0.0002,
+      "step": 9836
+    },
+    {
+      "epoch": 146.82,
+      "learning_rate": 7.614035087719298e-06,
+      "loss": 0.0002,
+      "step": 9837
+    },
+    {
+      "epoch": 146.83,
+      "learning_rate": 7.578947368421052e-06,
+      "loss": 0.0004,
+      "step": 9838
+    },
+    {
+      "epoch": 146.85,
+      "learning_rate": 7.543859649122807e-06,
+      "loss": 0.0002,
+      "step": 9839
+    },
+    {
+      "epoch": 146.86,
+      "learning_rate": 7.5087719298245605e-06,
+      "loss": 0.0002,
+      "step": 9840
+    },
+    {
+      "epoch": 146.88,
+      "learning_rate": 7.473684210526315e-06,
+      "loss": 0.0002,
+      "step": 9841
+    },
+    {
+      "epoch": 146.89,
+      "learning_rate": 7.43859649122807e-06,
+      "loss": 0.0002,
+      "step": 9842
+    },
+    {
+      "epoch": 146.91,
+      "learning_rate": 7.403508771929824e-06,
+      "loss": 0.0001,
+      "step": 9843
+    },
+    {
+      "epoch": 146.92,
+      "learning_rate": 7.3684210526315784e-06,
+      "loss": 0.0001,
+      "step": 9844
+    },
+    {
+      "epoch": 146.94,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 0.0002,
+      "step": 9845
+    },
+    {
+      "epoch": 146.95,
+      "learning_rate": 7.298245614035087e-06,
+      "loss": 0.0004,
+      "step": 9846
+    },
+    {
+      "epoch": 146.97,
+      "learning_rate": 7.263157894736842e-06,
+      "loss": 0.0003,
+      "step": 9847
+    },
+    {
+      "epoch": 146.98,
+      "learning_rate": 7.228070175438596e-06,
+      "loss": 0.0001,
+      "step": 9848
+    },
+    {
+      "epoch": 147.0,
+      "learning_rate": 7.19298245614035e-06,
+      "loss": 0.0002,
+      "step": 9849
+    },
+    {
+      "epoch": 147.01,
+      "learning_rate": 7.157894736842105e-06,
+      "loss": 0.0002,
+      "step": 9850
+    },
+    {
+      "epoch": 147.03,
+      "learning_rate": 7.122807017543859e-06,
+      "loss": 0.0002,
+      "step": 9851
+    },
+    {
+      "epoch": 147.04,
+      "learning_rate": 7.0877192982456136e-06,
+      "loss": 0.0002,
+      "step": 9852
+    },
+    {
+      "epoch": 147.06,
+      "learning_rate": 7.052631578947368e-06,
+      "loss": 0.0002,
+      "step": 9853
+    },
+    {
+      "epoch": 147.07,
+      "learning_rate": 7.017543859649122e-06,
+      "loss": 0.0002,
+      "step": 9854
+    },
+    {
+      "epoch": 147.09,
+      "learning_rate": 6.982456140350877e-06,
+      "loss": 0.0002,
+      "step": 9855
+    },
+    {
+      "epoch": 147.1,
+      "learning_rate": 6.9473684210526315e-06,
+      "loss": 0.0002,
+      "step": 9856
+    },
+    {
+      "epoch": 147.12,
+      "learning_rate": 6.912280701754385e-06,
+      "loss": 0.0002,
+      "step": 9857
+    },
+    {
+      "epoch": 147.13,
+      "learning_rate": 6.87719298245614e-06,
+      "loss": 0.0002,
+      "step": 9858
+    },
+    {
+      "epoch": 147.15,
+      "learning_rate": 6.842105263157895e-06,
+      "loss": 0.0002,
+      "step": 9859
+    },
+    {
+      "epoch": 147.16,
+      "learning_rate": 6.807017543859649e-06,
+      "loss": 0.0002,
+      "step": 9860
+    },
+    {
+      "epoch": 147.18,
+      "learning_rate": 6.771929824561403e-06,
+      "loss": 0.0002,
+      "step": 9861
+    },
+    {
+      "epoch": 147.19,
+      "learning_rate": 6.736842105263156e-06,
+      "loss": 0.0002,
+      "step": 9862
+    },
+    {
+      "epoch": 147.21,
+      "learning_rate": 6.701754385964912e-06,
+      "loss": 0.0002,
+      "step": 9863
+    },
+    {
+      "epoch": 147.22,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.0002,
+      "step": 9864
+    },
+    {
+      "epoch": 147.24,
+      "learning_rate": 6.63157894736842e-06,
+      "loss": 0.0004,
+      "step": 9865
+    },
+    {
+      "epoch": 147.25,
+      "learning_rate": 6.596491228070175e-06,
+      "loss": 0.0002,
+      "step": 9866
+    },
+    {
+      "epoch": 147.27,
+      "learning_rate": 6.56140350877193e-06,
+      "loss": 0.0002,
+      "step": 9867
+    },
+    {
+      "epoch": 147.28,
+      "learning_rate": 6.526315789473683e-06,
+      "loss": 0.0002,
+      "step": 9868
+    },
+    {
+      "epoch": 147.3,
+      "learning_rate": 6.4912280701754385e-06,
+      "loss": 0.0001,
+      "step": 9869
+    },
+    {
+      "epoch": 147.31,
+      "learning_rate": 6.456140350877193e-06,
+      "loss": 0.0002,
+      "step": 9870
+    },
+    {
+      "epoch": 147.33,
+      "learning_rate": 6.421052631578946e-06,
+      "loss": 0.0002,
+      "step": 9871
+    },
+    {
+      "epoch": 147.34,
+      "learning_rate": 6.385964912280702e-06,
+      "loss": 0.0004,
+      "step": 9872
+    },
+    {
+      "epoch": 147.36,
+      "learning_rate": 6.350877192982455e-06,
+      "loss": 0.0002,
+      "step": 9873
+    },
+    {
+      "epoch": 147.37,
+      "learning_rate": 6.3157894736842095e-06,
+      "loss": 0.0002,
+      "step": 9874
+    },
+    {
+      "epoch": 147.39,
+      "learning_rate": 6.280701754385964e-06,
+      "loss": 0.0002,
+      "step": 9875
+    },
+    {
+      "epoch": 147.4,
+      "learning_rate": 6.245614035087718e-06,
+      "loss": 0.0002,
+      "step": 9876
+    },
+    {
+      "epoch": 147.42,
+      "learning_rate": 6.210526315789473e-06,
+      "loss": 0.0002,
+      "step": 9877
+    },
+    {
+      "epoch": 147.43,
+      "learning_rate": 6.1754385964912275e-06,
+      "loss": 0.0002,
+      "step": 9878
+    },
+    {
+      "epoch": 147.45,
+      "learning_rate": 6.140350877192981e-06,
+      "loss": 0.0002,
+      "step": 9879
+    },
+    {
+      "epoch": 147.46,
+      "learning_rate": 6.105263157894736e-06,
+      "loss": 0.0002,
+      "step": 9880
+    },
+    {
+      "epoch": 147.48,
+      "learning_rate": 6.070175438596491e-06,
+      "loss": 0.0002,
+      "step": 9881
+    },
+    {
+      "epoch": 147.49,
+      "learning_rate": 6.035087719298245e-06,
+      "loss": 0.0002,
+      "step": 9882
+    },
+    {
+      "epoch": 147.51,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 0.0217,
+      "step": 9883
+    },
+    {
+      "epoch": 147.52,
+      "learning_rate": 5.964912280701754e-06,
+      "loss": 0.0003,
+      "step": 9884
+    },
+    {
+      "epoch": 147.54,
+      "learning_rate": 5.929824561403508e-06,
+      "loss": 0.0002,
+      "step": 9885
+    },
+    {
+      "epoch": 147.55,
+      "learning_rate": 5.894736842105263e-06,
+      "loss": 0.0002,
+      "step": 9886
+    },
+    {
+      "epoch": 147.57,
+      "learning_rate": 5.8596491228070164e-06,
+      "loss": 0.0001,
+      "step": 9887
+    },
+    {
+      "epoch": 147.58,
+      "learning_rate": 5.824561403508771e-06,
+      "loss": 0.0003,
+      "step": 9888
+    },
+    {
+      "epoch": 147.59,
+      "learning_rate": 5.789473684210526e-06,
+      "loss": 0.0002,
+      "step": 9889
+    },
+    {
+      "epoch": 147.61,
+      "learning_rate": 5.75438596491228e-06,
+      "loss": 0.0005,
+      "step": 9890
+    },
+    {
+      "epoch": 147.62,
+      "learning_rate": 5.719298245614034e-06,
+      "loss": 0.0002,
+      "step": 9891
+    },
+    {
+      "epoch": 147.64,
+      "learning_rate": 5.684210526315789e-06,
+      "loss": 0.0004,
+      "step": 9892
+    },
+    {
+      "epoch": 147.65,
+      "learning_rate": 5.649122807017543e-06,
+      "loss": 0.0002,
+      "step": 9893
+    },
+    {
+      "epoch": 147.67,
+      "learning_rate": 5.614035087719298e-06,
+      "loss": 0.0001,
+      "step": 9894
+    },
+    {
+      "epoch": 147.68,
+      "learning_rate": 5.578947368421052e-06,
+      "loss": 0.0002,
+      "step": 9895
+    },
+    {
+      "epoch": 147.7,
+      "learning_rate": 5.543859649122806e-06,
+      "loss": 0.0002,
+      "step": 9896
+    },
+    {
+      "epoch": 147.71,
+      "learning_rate": 5.508771929824561e-06,
+      "loss": 0.0002,
+      "step": 9897
+    },
+    {
+      "epoch": 147.73,
+      "learning_rate": 5.473684210526315e-06,
+      "loss": 0.0001,
+      "step": 9898
+    },
+    {
+      "epoch": 147.74,
+      "learning_rate": 5.4385964912280695e-06,
+      "loss": 0.0002,
+      "step": 9899
+    },
+    {
+      "epoch": 147.76,
+      "learning_rate": 5.403508771929824e-06,
+      "loss": 0.0002,
+      "step": 9900
+    },
+    {
+      "epoch": 147.77,
+      "learning_rate": 5.368421052631578e-06,
+      "loss": 0.0004,
+      "step": 9901
+    },
+    {
+      "epoch": 147.79,
+      "learning_rate": 5.333333333333333e-06,
+      "loss": 0.0039,
+      "step": 9902
+    },
+    {
+      "epoch": 147.8,
+      "learning_rate": 5.2982456140350875e-06,
+      "loss": 0.0002,
+      "step": 9903
+    },
+    {
+      "epoch": 147.82,
+      "learning_rate": 5.263157894736841e-06,
+      "loss": 0.0002,
+      "step": 9904
+    },
+    {
+      "epoch": 147.83,
+      "learning_rate": 5.228070175438596e-06,
+      "loss": 0.0002,
+      "step": 9905
+    },
+    {
+      "epoch": 147.85,
+      "learning_rate": 5.192982456140351e-06,
+      "loss": 0.0002,
+      "step": 9906
+    },
+    {
+      "epoch": 147.86,
+      "learning_rate": 5.157894736842105e-06,
+      "loss": 0.0001,
+      "step": 9907
+    },
+    {
+      "epoch": 147.88,
+      "learning_rate": 5.122807017543859e-06,
+      "loss": 0.0092,
+      "step": 9908
+    },
+    {
+      "epoch": 147.89,
+      "learning_rate": 5.087719298245613e-06,
+      "loss": 0.0002,
+      "step": 9909
+    },
+    {
+      "epoch": 147.91,
+      "learning_rate": 5.052631578947368e-06,
+      "loss": 0.0002,
+      "step": 9910
+    },
+    {
+      "epoch": 147.92,
+      "learning_rate": 5.017543859649123e-06,
+      "loss": 0.0002,
+      "step": 9911
+    },
+    {
+      "epoch": 147.94,
+      "learning_rate": 4.9824561403508765e-06,
+      "loss": 0.0002,
+      "step": 9912
+    },
+    {
+      "epoch": 147.95,
+      "learning_rate": 4.947368421052631e-06,
+      "loss": 0.0002,
+      "step": 9913
+    },
+    {
+      "epoch": 147.97,
+      "learning_rate": 4.912280701754386e-06,
+      "loss": 0.0002,
+      "step": 9914
+    },
+    {
+      "epoch": 147.98,
+      "learning_rate": 4.87719298245614e-06,
+      "loss": 0.0002,
+      "step": 9915
+    },
+    {
+      "epoch": 148.0,
+      "learning_rate": 4.8421052631578945e-06,
+      "loss": 0.0002,
+      "step": 9916
+    },
+    {
+      "epoch": 148.01,
+      "learning_rate": 4.807017543859649e-06,
+      "loss": 0.0004,
+      "step": 9917
+    },
+    {
+      "epoch": 148.03,
+      "learning_rate": 4.771929824561403e-06,
+      "loss": 0.0002,
+      "step": 9918
+    },
+    {
+      "epoch": 148.04,
+      "learning_rate": 4.736842105263158e-06,
+      "loss": 0.0002,
+      "step": 9919
+    },
+    {
+      "epoch": 148.06,
+      "learning_rate": 4.701754385964912e-06,
+      "loss": 0.0002,
+      "step": 9920
+    },
+    {
+      "epoch": 148.07,
+      "learning_rate": 4.666666666666666e-06,
+      "loss": 0.0002,
+      "step": 9921
+    },
+    {
+      "epoch": 148.09,
+      "learning_rate": 4.631578947368421e-06,
+      "loss": 0.0002,
+      "step": 9922
+    },
+    {
+      "epoch": 148.1,
+      "learning_rate": 4.596491228070175e-06,
+      "loss": 0.0002,
+      "step": 9923
+    },
+    {
+      "epoch": 148.12,
+      "learning_rate": 4.56140350877193e-06,
+      "loss": 0.0003,
+      "step": 9924
+    },
+    {
+      "epoch": 148.13,
+      "learning_rate": 4.5263157894736834e-06,
+      "loss": 0.0001,
+      "step": 9925
+    },
+    {
+      "epoch": 148.15,
+      "learning_rate": 4.491228070175438e-06,
+      "loss": 0.0003,
+      "step": 9926
+    },
+    {
+      "epoch": 148.16,
+      "learning_rate": 4.456140350877193e-06,
+      "loss": 0.0002,
+      "step": 9927
+    },
+    {
+      "epoch": 148.18,
+      "learning_rate": 4.421052631578947e-06,
+      "loss": 0.0002,
+      "step": 9928
+    },
+    {
+      "epoch": 148.19,
+      "learning_rate": 4.3859649122807014e-06,
+      "loss": 0.0002,
+      "step": 9929
+    },
+    {
+      "epoch": 148.21,
+      "learning_rate": 4.350877192982456e-06,
+      "loss": 0.0006,
+      "step": 9930
+    },
+    {
+      "epoch": 148.22,
+      "learning_rate": 4.31578947368421e-06,
+      "loss": 0.0001,
+      "step": 9931
+    },
+    {
+      "epoch": 148.24,
+      "learning_rate": 4.280701754385965e-06,
+      "loss": 0.0002,
+      "step": 9932
+    },
+    {
+      "epoch": 148.25,
+      "learning_rate": 4.280701754385965e-06,
+      "loss": 0.1554,
+      "step": 9933
+    },
+    {
+      "epoch": 148.27,
+      "learning_rate": 4.245614035087719e-06,
+      "loss": 0.0004,
+      "step": 9934
+    },
+    {
+      "epoch": 148.28,
+      "learning_rate": 4.210526315789473e-06,
+      "loss": 0.0002,
+      "step": 9935
+    },
+    {
+      "epoch": 148.3,
+      "learning_rate": 4.175438596491228e-06,
+      "loss": 0.0001,
+      "step": 9936
+    },
+    {
+      "epoch": 148.31,
+      "learning_rate": 4.140350877192982e-06,
+      "loss": 0.0002,
+      "step": 9937
+    },
+    {
+      "epoch": 148.33,
+      "learning_rate": 4.1052631578947365e-06,
+      "loss": 0.0002,
+      "step": 9938
+    },
+    {
+      "epoch": 148.34,
+      "learning_rate": 4.070175438596491e-06,
+      "loss": 0.0002,
+      "step": 9939
+    },
+    {
+      "epoch": 148.36,
+      "learning_rate": 4.035087719298245e-06,
+      "loss": 0.0002,
+      "step": 9940
+    },
+    {
+      "epoch": 148.37,
+      "learning_rate": 4e-06,
+      "loss": 0.0003,
+      "step": 9941
+    },
+    {
+      "epoch": 148.39,
+      "learning_rate": 3.9649122807017545e-06,
+      "loss": 0.0002,
+      "step": 9942
+    },
+    {
+      "epoch": 148.4,
+      "learning_rate": 3.929824561403508e-06,
+      "loss": 0.0002,
+      "step": 9943
+    },
+    {
+      "epoch": 148.42,
+      "learning_rate": 3.894736842105262e-06,
+      "loss": 0.0001,
+      "step": 9944
+    },
+    {
+      "epoch": 148.43,
+      "learning_rate": 3.859649122807018e-06,
+      "loss": 0.0014,
+      "step": 9945
+    },
+    {
+      "epoch": 148.45,
+      "learning_rate": 3.824561403508772e-06,
+      "loss": 0.0005,
+      "step": 9946
+    },
+    {
+      "epoch": 148.46,
+      "learning_rate": 3.789473684210526e-06,
+      "loss": 0.0002,
+      "step": 9947
+    },
+    {
+      "epoch": 148.48,
+      "learning_rate": 3.7543859649122802e-06,
+      "loss": 0.0002,
+      "step": 9948
+    },
+    {
+      "epoch": 148.49,
+      "learning_rate": 3.719298245614035e-06,
+      "loss": 0.0002,
+      "step": 9949
+    },
+    {
+      "epoch": 148.51,
+      "learning_rate": 3.6842105263157892e-06,
+      "loss": 0.0002,
+      "step": 9950
+    },
+    {
+      "epoch": 148.52,
+      "learning_rate": 3.6491228070175435e-06,
+      "loss": 0.0002,
+      "step": 9951
+    },
+    {
+      "epoch": 148.54,
+      "learning_rate": 3.614035087719298e-06,
+      "loss": 0.0001,
+      "step": 9952
+    },
+    {
+      "epoch": 148.55,
+      "learning_rate": 3.5789473684210525e-06,
+      "loss": 0.0002,
+      "step": 9953
+    },
+    {
+      "epoch": 148.57,
+      "learning_rate": 3.5438596491228068e-06,
+      "loss": 0.003,
+      "step": 9954
+    },
+    {
+      "epoch": 148.58,
+      "learning_rate": 3.508771929824561e-06,
+      "loss": 0.0002,
+      "step": 9955
+    },
+    {
+      "epoch": 148.59,
+      "learning_rate": 3.4736842105263158e-06,
+      "loss": 0.0002,
+      "step": 9956
+    },
+    {
+      "epoch": 148.61,
+      "learning_rate": 3.43859649122807e-06,
+      "loss": 0.0182,
+      "step": 9957
+    },
+    {
+      "epoch": 148.62,
+      "learning_rate": 3.4035087719298243e-06,
+      "loss": 0.0002,
+      "step": 9958
+    },
+    {
+      "epoch": 148.64,
+      "learning_rate": 3.368421052631578e-06,
+      "loss": 0.0002,
+      "step": 9959
+    },
+    {
+      "epoch": 148.65,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.0002,
+      "step": 9960
+    },
+    {
+      "epoch": 148.67,
+      "learning_rate": 3.2982456140350876e-06,
+      "loss": 0.0002,
+      "step": 9961
+    },
+    {
+      "epoch": 148.68,
+      "learning_rate": 3.2631578947368415e-06,
+      "loss": 0.0002,
+      "step": 9962
+    },
+    {
+      "epoch": 148.7,
+      "learning_rate": 3.2280701754385966e-06,
+      "loss": 0.0002,
+      "step": 9963
+    },
+    {
+      "epoch": 148.71,
+      "learning_rate": 3.192982456140351e-06,
+      "loss": 0.0002,
+      "step": 9964
+    },
+    {
+      "epoch": 148.73,
+      "learning_rate": 3.1578947368421047e-06,
+      "loss": 0.0002,
+      "step": 9965
+    },
+    {
+      "epoch": 148.74,
+      "learning_rate": 3.122807017543859e-06,
+      "loss": 0.0002,
+      "step": 9966
+    },
+    {
+      "epoch": 148.76,
+      "learning_rate": 3.0877192982456137e-06,
+      "loss": 0.0002,
+      "step": 9967
+    },
+    {
+      "epoch": 148.77,
+      "learning_rate": 3.052631578947368e-06,
+      "loss": 0.0002,
+      "step": 9968
+    },
+    {
+      "epoch": 148.79,
+      "learning_rate": 3.0175438596491223e-06,
+      "loss": 0.0002,
+      "step": 9969
+    },
+    {
+      "epoch": 148.8,
+      "learning_rate": 2.982456140350877e-06,
+      "loss": 0.0002,
+      "step": 9970
+    },
+    {
+      "epoch": 148.82,
+      "learning_rate": 2.9473684210526313e-06,
+      "loss": 0.0002,
+      "step": 9971
+    },
+    {
+      "epoch": 148.83,
+      "learning_rate": 2.9122807017543856e-06,
+      "loss": 0.0002,
+      "step": 9972
+    },
+    {
+      "epoch": 148.85,
+      "learning_rate": 2.87719298245614e-06,
+      "loss": 0.0002,
+      "step": 9973
+    },
+    {
+      "epoch": 148.86,
+      "learning_rate": 2.8421052631578946e-06,
+      "loss": 0.0002,
+      "step": 9974
+    },
+    {
+      "epoch": 148.88,
+      "learning_rate": 2.807017543859649e-06,
+      "loss": 0.0002,
+      "step": 9975
+    },
+    {
+      "epoch": 148.89,
+      "learning_rate": 2.771929824561403e-06,
+      "loss": 0.0002,
+      "step": 9976
+    },
+    {
+      "epoch": 148.91,
+      "learning_rate": 2.7368421052631574e-06,
+      "loss": 0.0001,
+      "step": 9977
+    },
+    {
+      "epoch": 148.92,
+      "learning_rate": 2.701754385964912e-06,
+      "loss": 0.0002,
+      "step": 9978
+    },
+    {
+      "epoch": 148.94,
+      "learning_rate": 2.6666666666666664e-06,
+      "loss": 0.0019,
+      "step": 9979
+    },
+    {
+      "epoch": 148.95,
+      "learning_rate": 2.6315789473684207e-06,
+      "loss": 0.0003,
+      "step": 9980
+    },
+    {
+      "epoch": 148.97,
+      "learning_rate": 2.5964912280701754e-06,
+      "loss": 0.0002,
+      "step": 9981
+    },
+    {
+      "epoch": 148.98,
+      "learning_rate": 2.5614035087719297e-06,
+      "loss": 0.0002,
+      "step": 9982
+    },
+    {
+      "epoch": 149.0,
+      "learning_rate": 2.526315789473684e-06,
+      "loss": 0.0001,
+      "step": 9983
+    },
+    {
+      "epoch": 149.01,
+      "learning_rate": 2.4912280701754382e-06,
+      "loss": 0.0002,
+      "step": 9984
+    },
+    {
+      "epoch": 149.03,
+      "learning_rate": 2.456140350877193e-06,
+      "loss": 0.0977,
+      "step": 9985
+    },
+    {
+      "epoch": 149.04,
+      "learning_rate": 2.4210526315789472e-06,
+      "loss": 0.0002,
+      "step": 9986
+    },
+    {
+      "epoch": 149.06,
+      "learning_rate": 2.3859649122807015e-06,
+      "loss": 0.0002,
+      "step": 9987
+    },
+    {
+      "epoch": 149.07,
+      "learning_rate": 2.350877192982456e-06,
+      "loss": 0.0002,
+      "step": 9988
+    },
+    {
+      "epoch": 149.09,
+      "learning_rate": 2.3157894736842105e-06,
+      "loss": 0.0002,
+      "step": 9989
+    },
+    {
+      "epoch": 149.1,
+      "learning_rate": 2.280701754385965e-06,
+      "loss": 0.0002,
+      "step": 9990
+    },
+    {
+      "epoch": 149.12,
+      "learning_rate": 2.245614035087719e-06,
+      "loss": 0.0002,
+      "step": 9991
+    },
+    {
+      "epoch": 149.13,
+      "learning_rate": 2.2105263157894734e-06,
+      "loss": 0.0002,
+      "step": 9992
+    },
+    {
+      "epoch": 149.15,
+      "learning_rate": 2.175438596491228e-06,
+      "loss": 0.0002,
+      "step": 9993
+    },
+    {
+      "epoch": 149.16,
+      "learning_rate": 2.1403508771929824e-06,
+      "loss": 0.1007,
+      "step": 9994
+    },
+    {
+      "epoch": 149.18,
+      "learning_rate": 2.1052631578947366e-06,
+      "loss": 0.0002,
+      "step": 9995
+    },
+    {
+      "epoch": 149.19,
+      "learning_rate": 2.070175438596491e-06,
+      "loss": 0.0002,
+      "step": 9996
+    },
+    {
+      "epoch": 149.21,
+      "learning_rate": 2.0350877192982456e-06,
+      "loss": 0.0002,
+      "step": 9997
+    },
+    {
+      "epoch": 149.22,
+      "learning_rate": 2e-06,
+      "loss": 0.0002,
+      "step": 9998
+    },
+    {
+      "epoch": 149.24,
+      "learning_rate": 1.964912280701754e-06,
+      "loss": 0.0002,
+      "step": 9999
+    },
+    {
+      "epoch": 149.25,
+      "learning_rate": 1.929824561403509e-06,
+      "loss": 0.0002,
+      "step": 10000
+    },
+    {
+      "epoch": 149.25,
+      "eval_accuracy": 0.8920704845814978,
+      "eval_f1": 0.8917851618940525,
+      "eval_loss": 0.5978561639785767,
+      "eval_runtime": 345.0023,
+      "eval_samples_per_second": 11.843,
+      "eval_steps_per_second": 0.742,
+      "step": 10000
+    },
+    {
+      "epoch": 149.27,
+      "learning_rate": 1.894736842105263e-06,
+      "loss": 0.0002,
+      "step": 10001
+    },
+    {
+      "epoch": 149.28,
+      "learning_rate": 1.8596491228070175e-06,
+      "loss": 0.0001,
+      "step": 10002
+    },
+    {
+      "epoch": 149.3,
+      "learning_rate": 1.8245614035087718e-06,
+      "loss": 0.0001,
+      "step": 10003
+    },
+    {
+      "epoch": 149.31,
+      "learning_rate": 1.7894736842105262e-06,
+      "loss": 0.0002,
+      "step": 10004
+    },
+    {
+      "epoch": 149.33,
+      "learning_rate": 1.7543859649122805e-06,
+      "loss": 0.0002,
+      "step": 10005
+    },
+    {
+      "epoch": 149.34,
+      "learning_rate": 1.719298245614035e-06,
+      "loss": 0.0002,
+      "step": 10006
+    },
+    {
+      "epoch": 149.36,
+      "learning_rate": 1.684210526315789e-06,
+      "loss": 0.0002,
+      "step": 10007
+    },
+    {
+      "epoch": 149.37,
+      "learning_rate": 1.6491228070175438e-06,
+      "loss": 0.0002,
+      "step": 10008
+    },
+    {
+      "epoch": 149.39,
+      "learning_rate": 1.6140350877192983e-06,
+      "loss": 0.0001,
+      "step": 10009
+    },
+    {
+      "epoch": 149.4,
+      "learning_rate": 1.5789473684210524e-06,
+      "loss": 0.0002,
+      "step": 10010
+    },
+    {
+      "epoch": 149.42,
+      "learning_rate": 1.5438596491228069e-06,
+      "loss": 0.0003,
+      "step": 10011
+    },
+    {
+      "epoch": 149.43,
+      "learning_rate": 1.5087719298245611e-06,
+      "loss": 0.0002,
+      "step": 10012
+    },
+    {
+      "epoch": 149.45,
+      "learning_rate": 1.4736842105263156e-06,
+      "loss": 0.0002,
+      "step": 10013
+    },
+    {
+      "epoch": 149.46,
+      "learning_rate": 1.43859649122807e-06,
+      "loss": 0.0027,
+      "step": 10014
+    },
+    {
+      "epoch": 149.48,
+      "learning_rate": 1.4035087719298244e-06,
+      "loss": 0.0002,
+      "step": 10015
+    },
+    {
+      "epoch": 149.49,
+      "learning_rate": 1.3684210526315787e-06,
+      "loss": 0.0001,
+      "step": 10016
+    },
+    {
+      "epoch": 149.51,
+      "learning_rate": 1.3333333333333332e-06,
+      "loss": 0.0002,
+      "step": 10017
+    },
+    {
+      "epoch": 149.52,
+      "learning_rate": 1.2982456140350877e-06,
+      "loss": 0.0002,
+      "step": 10018
+    },
+    {
+      "epoch": 149.54,
+      "learning_rate": 1.263157894736842e-06,
+      "loss": 0.0002,
+      "step": 10019
+    },
+    {
+      "epoch": 149.55,
+      "learning_rate": 1.2280701754385965e-06,
+      "loss": 0.0002,
+      "step": 10020
+    },
+    {
+      "epoch": 149.57,
+      "learning_rate": 1.1929824561403508e-06,
+      "loss": 0.0002,
+      "step": 10021
+    },
+    {
+      "epoch": 149.58,
+      "learning_rate": 1.1578947368421053e-06,
+      "loss": 0.0002,
+      "step": 10022
+    },
+    {
+      "epoch": 149.59,
+      "learning_rate": 1.1228070175438595e-06,
+      "loss": 0.0002,
+      "step": 10023
+    },
+    {
+      "epoch": 149.61,
+      "learning_rate": 1.087719298245614e-06,
+      "loss": 0.0002,
+      "step": 10024
+    },
+    {
+      "epoch": 149.62,
+      "learning_rate": 1.0526315789473683e-06,
+      "loss": 0.0002,
+      "step": 10025
+    },
+    {
+      "epoch": 149.64,
+      "learning_rate": 1.0175438596491228e-06,
+      "loss": 0.0002,
+      "step": 10026
+    },
+    {
+      "epoch": 149.65,
+      "learning_rate": 9.82456140350877e-07,
+      "loss": 0.0002,
+      "step": 10027
+    },
+    {
+      "epoch": 149.67,
+      "learning_rate": 9.473684210526315e-07,
+      "loss": 0.0002,
+      "step": 10028
+    },
+    {
+      "epoch": 149.68,
+      "learning_rate": 9.122807017543859e-07,
+      "loss": 0.0003,
+      "step": 10029
+    },
+    {
+      "epoch": 149.7,
+      "learning_rate": 8.771929824561403e-07,
+      "loss": 0.0025,
+      "step": 10030
+    },
+    {
+      "epoch": 149.71,
+      "learning_rate": 8.421052631578945e-07,
+      "loss": 0.0769,
+      "step": 10031
+    },
+    {
+      "epoch": 149.73,
+      "learning_rate": 8.070175438596491e-07,
+      "loss": 0.0002,
+      "step": 10032
+    },
+    {
+      "epoch": 149.74,
+      "learning_rate": 7.719298245614034e-07,
+      "loss": 0.0001,
+      "step": 10033
+    },
+    {
+      "epoch": 149.76,
+      "learning_rate": 7.368421052631578e-07,
+      "loss": 0.0003,
+      "step": 10034
+    },
+    {
+      "epoch": 149.77,
+      "learning_rate": 7.017543859649122e-07,
+      "loss": 0.0002,
+      "step": 10035
+    },
+    {
+      "epoch": 149.79,
+      "learning_rate": 6.666666666666666e-07,
+      "loss": 0.0001,
+      "step": 10036
+    },
+    {
+      "epoch": 149.8,
+      "learning_rate": 6.31578947368421e-07,
+      "loss": 0.0002,
+      "step": 10037
+    },
+    {
+      "epoch": 149.82,
+      "learning_rate": 5.964912280701754e-07,
+      "loss": 0.0002,
+      "step": 10038
+    },
+    {
+      "epoch": 149.83,
+      "learning_rate": 5.614035087719298e-07,
+      "loss": 0.0002,
+      "step": 10039
+    },
+    {
+      "epoch": 149.85,
+      "learning_rate": 5.263157894736842e-07,
+      "loss": 0.0002,
+      "step": 10040
+    },
+    {
+      "epoch": 149.86,
+      "learning_rate": 4.912280701754385e-07,
+      "loss": 0.0002,
+      "step": 10041
+    },
+    {
+      "epoch": 149.88,
+      "learning_rate": 4.5614035087719294e-07,
+      "loss": 0.0003,
+      "step": 10042
+    },
+    {
+      "epoch": 149.89,
+      "learning_rate": 4.210526315789473e-07,
+      "loss": 0.0002,
+      "step": 10043
+    },
+    {
+      "epoch": 149.91,
+      "learning_rate": 3.859649122807017e-07,
+      "loss": 0.0002,
+      "step": 10044
+    },
+    {
+      "epoch": 149.92,
+      "learning_rate": 3.508771929824561e-07,
+      "loss": 0.0145,
+      "step": 10045
+    },
+    {
+      "epoch": 149.94,
+      "learning_rate": 3.157894736842105e-07,
+      "loss": 0.0002,
+      "step": 10046
+    },
+    {
+      "epoch": 149.95,
+      "learning_rate": 2.807017543859649e-07,
+      "loss": 0.2392,
+      "step": 10047
+    },
+    {
+      "epoch": 149.97,
+      "learning_rate": 2.456140350877193e-07,
+      "loss": 0.0002,
+      "step": 10048
+    },
+    {
+      "epoch": 149.98,
+      "learning_rate": 2.1052631578947364e-07,
+      "loss": 0.0002,
+      "step": 10049
+    },
+    {
+      "epoch": 150.0,
+      "learning_rate": 1.7543859649122805e-07,
+      "loss": 0.0001,
+      "step": 10050
+    },
+    {
+      "epoch": 150.0,
+      "step": 10050,
+      "total_flos": 1.3732582771233128e+20,
+      "train_loss": 0.2665309856276014,
+      "train_runtime": 44149.9371,
+      "train_samples_per_second": 7.308,
+      "train_steps_per_second": 0.228
+    }
+  ],
+  "max_steps": 10050,
+  "num_train_epochs": 150,
+  "total_flos": 1.3732582771233128e+20,
+  "trial_name": null,
+  "trial_params": null
+}