diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,20059 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 29.995515695067265,
+  "global_step": 3330,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 18.5859,
+      "step": 1
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0,
+      "loss": 18.626,
+      "step": 2
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0,
+      "loss": 18.6052,
+      "step": 3
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0,
+      "loss": 18.6673,
+      "step": 4
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0,
+      "loss": 18.6777,
+      "step": 5
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6e-07,
+      "loss": 18.5482,
+      "step": 6
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6e-07,
+      "loss": 18.6601,
+      "step": 7
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2e-06,
+      "loss": 18.6215,
+      "step": 8
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8e-06,
+      "loss": 18.4706,
+      "step": 9
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8e-06,
+      "loss": 18.5834,
+      "step": 10
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4e-06,
+      "loss": 18.5967,
+      "step": 11
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9999999999999997e-06,
+      "loss": 17.6808,
+      "step": 12
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.6e-06,
+      "loss": 16.3916,
+      "step": 13
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2e-06,
+      "loss": 14.3146,
+      "step": 14
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8e-06,
+      "loss": 13.7872,
+      "step": 15
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.399999999999999e-06,
+      "loss": 12.8243,
+      "step": 16
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 12.7046,
+      "step": 17
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.599999999999999e-06,
+      "loss": 12.1837,
+      "step": 18
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.2e-06,
+      "loss": 11.7183,
+      "step": 19
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.799999999999998e-06,
+      "loss": 11.3459,
+      "step": 20
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.4e-06,
+      "loss": 10.9967,
+      "step": 21
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.999999999999999e-06,
+      "loss": 10.8096,
+      "step": 22
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.6e-06,
+      "loss": 10.403,
+      "step": 23
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.02e-05,
+      "loss": 10.1043,
+      "step": 24
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0799999999999998e-05,
+      "loss": 9.9771,
+      "step": 25
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.14e-05,
+      "loss": 9.5833,
+      "step": 26
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 9.332,
+      "step": 27
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.26e-05,
+      "loss": 9.1017,
+      "step": 28
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.3199999999999997e-05,
+      "loss": 8.9242,
+      "step": 29
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.3799999999999998e-05,
+      "loss": 8.695,
+      "step": 30
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.44e-05,
+      "loss": 8.4996,
+      "step": 31
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 8.3903,
+      "step": 32
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5599999999999996e-05,
+      "loss": 8.2444,
+      "step": 33
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6199999999999997e-05,
+      "loss": 8.0591,
+      "step": 34
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.68e-05,
+      "loss": 7.9648,
+      "step": 35
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.74e-05,
+      "loss": 7.885,
+      "step": 36
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 7.8086,
+      "step": 37
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8599999999999998e-05,
+      "loss": 7.7388,
+      "step": 38
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.92e-05,
+      "loss": 7.5805,
+      "step": 39
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.98e-05,
+      "loss": 7.5902,
+      "step": 40
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.04e-05,
+      "loss": 7.5181,
+      "step": 41
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.1e-05,
+      "loss": 7.4488,
+      "step": 42
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.1599999999999996e-05,
+      "loss": 7.4521,
+      "step": 43
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.2199999999999998e-05,
+      "loss": 7.3848,
+      "step": 44
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.28e-05,
+      "loss": 7.3917,
+      "step": 45
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.34e-05,
+      "loss": 7.3672,
+      "step": 46
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 7.4301,
+      "step": 47
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 7.3792,
+      "step": 48
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.52e-05,
+      "loss": 7.3408,
+      "step": 49
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.5799999999999997e-05,
+      "loss": 7.3113,
+      "step": 50
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6399999999999995e-05,
+      "loss": 7.3015,
+      "step": 51
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 7.281,
+      "step": 52
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7599999999999997e-05,
+      "loss": 7.2185,
+      "step": 53
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 7.2576,
+      "step": 54
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.88e-05,
+      "loss": 7.2705,
+      "step": 55
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.94e-05,
+      "loss": 7.2566,
+      "step": 56
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 7.2337,
+      "step": 57
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.06e-05,
+      "loss": 7.1842,
+      "step": 58
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.119999999999999e-05,
+      "loss": 7.1656,
+      "step": 59
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.1799999999999994e-05,
+      "loss": 7.205,
+      "step": 60
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.2399999999999995e-05,
+      "loss": 7.1949,
+      "step": 61
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 7.1695,
+      "step": 62
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.36e-05,
+      "loss": 7.1858,
+      "step": 63
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.42e-05,
+      "loss": 7.1774,
+      "step": 64
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.48e-05,
+      "loss": 7.2494,
+      "step": 65
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.539999999999999e-05,
+      "loss": 7.1137,
+      "step": 66
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 7.1948,
+      "step": 67
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6599999999999995e-05,
+      "loss": 7.1358,
+      "step": 68
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7199999999999996e-05,
+      "loss": 7.2376,
+      "step": 69
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.78e-05,
+      "loss": 7.1785,
+      "step": 70
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.84e-05,
+      "loss": 7.1986,
+      "step": 71
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9e-05,
+      "loss": 7.1598,
+      "step": 72
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.96e-05,
+      "loss": 7.1845,
+      "step": 73
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.02e-05,
+      "loss": 7.1086,
+      "step": 74
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.08e-05,
+      "loss": 7.1857,
+      "step": 75
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.14e-05,
+      "loss": 7.1101,
+      "step": 76
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2e-05,
+      "loss": 7.145,
+      "step": 77
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.259999999999999e-05,
+      "loss": 7.0492,
+      "step": 78
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.319999999999999e-05,
+      "loss": 7.1061,
+      "step": 79
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3799999999999994e-05,
+      "loss": 7.1139,
+      "step": 80
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.4399999999999995e-05,
+      "loss": 7.1271,
+      "step": 81
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 7.1473,
+      "step": 82
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.56e-05,
+      "loss": 7.1408,
+      "step": 83
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.62e-05,
+      "loss": 7.0381,
+      "step": 84
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.68e-05,
+      "loss": 7.1561,
+      "step": 85
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7399999999999993e-05,
+      "loss": 6.9977,
+      "step": 86
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 7.1057,
+      "step": 87
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8599999999999995e-05,
+      "loss": 7.0334,
+      "step": 88
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9199999999999997e-05,
+      "loss": 7.0516,
+      "step": 89
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.98e-05,
+      "loss": 7.0825,
+      "step": 90
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.04e-05,
+      "loss": 7.0076,
+      "step": 91
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.1e-05,
+      "loss": 7.0604,
+      "step": 92
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.1599999999999994e-05,
+      "loss": 7.0735,
+      "step": 93
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.2199999999999995e-05,
+      "loss": 7.0408,
+      "step": 94
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.279999999999999e-05,
+      "loss": 7.0496,
+      "step": 95
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.339999999999999e-05,
+      "loss": 7.0986,
+      "step": 96
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 7.0262,
+      "step": 97
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.459999999999999e-05,
+      "loss": 7.0272,
+      "step": 98
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.519999999999999e-05,
+      "loss": 7.0175,
+      "step": 99
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5799999999999994e-05,
+      "loss": 7.048,
+      "step": 100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 6.9689,
+      "step": 101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 7.0322,
+      "step": 102
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.76e-05,
+      "loss": 7.012,
+      "step": 103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.82e-05,
+      "loss": 7.0402,
+      "step": 104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.88e-05,
+      "loss": 7.03,
+      "step": 105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.94e-05,
+      "loss": 7.0407,
+      "step": 106
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 6.9622,
+      "step": 107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.0599999999999996e-05,
+      "loss": 7.0176,
+      "step": 108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.12e-05,
+      "loss": 7.0248,
+      "step": 109
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.18e-05,
+      "loss": 7.0791,
+      "step": 110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.239999999999999e-05,
+      "loss": 6.9573,
+      "step": 111
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.299999999999999e-05,
+      "loss": 10.5971,
+      "step": 112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.359999999999999e-05,
+      "loss": 7.0102,
+      "step": 113
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 6.419999999999999e-05,
+      "loss": 6.986,
+      "step": 114
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 6.479999999999999e-05,
+      "loss": 6.9864,
+      "step": 115
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 6.539999999999999e-05,
+      "loss": 6.9585,
+      "step": 116
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 7.002,
+      "step": 117
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.659999999999999e-05,
+      "loss": 7.0419,
+      "step": 118
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.72e-05,
+      "loss": 7.01,
+      "step": 119
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.78e-05,
+      "loss": 6.9781,
+      "step": 120
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.84e-05,
+      "loss": 7.034,
+      "step": 121
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.9e-05,
+      "loss": 7.0387,
+      "step": 122
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.96e-05,
+      "loss": 7.0022,
+      "step": 123
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.02e-05,
+      "loss": 7.025,
+      "step": 124
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.079999999999999e-05,
+      "loss": 6.959,
+      "step": 125
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.139999999999999e-05,
+      "loss": 6.9289,
+      "step": 126
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 6.9987,
+      "step": 127
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.259999999999999e-05,
+      "loss": 6.9594,
+      "step": 128
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.319999999999999e-05,
+      "loss": 6.9025,
+      "step": 129
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.379999999999999e-05,
+      "loss": 6.9451,
+      "step": 130
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.439999999999999e-05,
+      "loss": 6.9658,
+      "step": 131
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.5e-05,
+      "loss": 6.9764,
+      "step": 132
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.56e-05,
+      "loss": 6.975,
+      "step": 133
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.62e-05,
+      "loss": 6.9628,
+      "step": 134
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.68e-05,
+      "loss": 6.9924,
+      "step": 135
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.74e-05,
+      "loss": 6.9381,
+      "step": 136
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.8e-05,
+      "loss": 6.9508,
+      "step": 137
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.86e-05,
+      "loss": 6.9656,
+      "step": 138
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.92e-05,
+      "loss": 6.9351,
+      "step": 139
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.98e-05,
+      "loss": 6.9466,
+      "step": 140
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.04e-05,
+      "loss": 6.932,
+      "step": 141
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.1e-05,
+      "loss": 6.9823,
+      "step": 142
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.16e-05,
+      "loss": 6.9757,
+      "step": 143
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.22e-05,
+      "loss": 6.9562,
+      "step": 144
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.28e-05,
+      "loss": 6.901,
+      "step": 145
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.34e-05,
+      "loss": 6.9871,
+      "step": 146
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.4e-05,
+      "loss": 6.9451,
+      "step": 147
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.459999999999998e-05,
+      "loss": 6.9154,
+      "step": 148
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.519999999999998e-05,
+      "loss": 7.0029,
+      "step": 149
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.579999999999998e-05,
+      "loss": 6.9551,
+      "step": 150
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.639999999999999e-05,
+      "loss": 6.9384,
+      "step": 151
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 6.9115,
+      "step": 152
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.759999999999999e-05,
+      "loss": 6.881,
+      "step": 153
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.819999999999999e-05,
+      "loss": 6.8946,
+      "step": 154
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.879999999999999e-05,
+      "loss": 6.8937,
+      "step": 155
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.939999999999999e-05,
+      "loss": 6.9437,
+      "step": 156
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 6.969,
+      "step": 157
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.059999999999999e-05,
+      "loss": 6.9027,
+      "step": 158
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.12e-05,
+      "loss": 6.8711,
+      "step": 159
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.18e-05,
+      "loss": 6.9052,
+      "step": 160
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.24e-05,
+      "loss": 6.8826,
+      "step": 161
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.3e-05,
+      "loss": 6.8508,
+      "step": 162
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.36e-05,
+      "loss": 6.924,
+      "step": 163
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.419999999999999e-05,
+      "loss": 6.8763,
+      "step": 164
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.479999999999999e-05,
+      "loss": 6.8776,
+      "step": 165
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.539999999999999e-05,
+      "loss": 6.861,
+      "step": 166
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 6.9209,
+      "step": 167
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.659999999999999e-05,
+      "loss": 6.8519,
+      "step": 168
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.719999999999999e-05,
+      "loss": 6.9076,
+      "step": 169
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.779999999999999e-05,
+      "loss": 6.866,
+      "step": 170
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.839999999999999e-05,
+      "loss": 6.8382,
+      "step": 171
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.9e-05,
+      "loss": 6.8583,
+      "step": 172
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.96e-05,
+      "loss": 6.8698,
+      "step": 173
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001002,
+      "loss": 6.8608,
+      "step": 174
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001008,
+      "loss": 6.8679,
+      "step": 175
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001014,
+      "loss": 6.8501,
+      "step": 176
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000102,
+      "loss": 6.8267,
+      "step": 177
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001026,
+      "loss": 6.8313,
+      "step": 178
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00010319999999999999,
+      "loss": 6.8548,
+      "step": 179
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00010379999999999999,
+      "loss": 6.8352,
+      "step": 180
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00010439999999999999,
+      "loss": 6.8261,
+      "step": 181
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00010499999999999999,
+      "loss": 6.778,
+      "step": 182
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00010559999999999998,
+      "loss": 6.8287,
+      "step": 183
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00010619999999999998,
+      "loss": 6.8383,
+      "step": 184
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00010679999999999998,
+      "loss": 6.8232,
+      "step": 185
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00010739999999999998,
+      "loss": 6.7399,
+      "step": 186
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 6.8253,
+      "step": 187
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00010859999999999998,
+      "loss": 6.8254,
+      "step": 188
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00010919999999999998,
+      "loss": 6.8719,
+      "step": 189
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00010979999999999999,
+      "loss": 6.8076,
+      "step": 190
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00011039999999999999,
+      "loss": 6.857,
+      "step": 191
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011099999999999999,
+      "loss": 6.809,
+      "step": 192
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011159999999999999,
+      "loss": 6.8369,
+      "step": 193
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011219999999999999,
+      "loss": 6.7654,
+      "step": 194
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011279999999999999,
+      "loss": 6.7757,
+      "step": 195
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011339999999999999,
+      "loss": 6.7536,
+      "step": 196
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 6.7707,
+      "step": 197
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001146,
+      "loss": 6.7859,
+      "step": 198
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001152,
+      "loss": 6.8059,
+      "step": 199
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001158,
+      "loss": 6.7489,
+      "step": 200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001164,
+      "loss": 6.7833,
+      "step": 201
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000117,
+      "loss": 6.7636,
+      "step": 202
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001176,
+      "loss": 6.7943,
+      "step": 203
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001182,
+      "loss": 6.7078,
+      "step": 204
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001188,
+      "loss": 6.7846,
+      "step": 205
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001194,
+      "loss": 6.6932,
+      "step": 206
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 6.7944,
+      "step": 207
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012059999999999999,
+      "loss": 6.7097,
+      "step": 208
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012119999999999999,
+      "loss": 6.7232,
+      "step": 209
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012179999999999999,
+      "loss": 6.7088,
+      "step": 210
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001224,
+      "loss": 6.7366,
+      "step": 211
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012299999999999998,
+      "loss": 6.7241,
+      "step": 212
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001236,
+      "loss": 6.7373,
+      "step": 213
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012419999999999998,
+      "loss": 6.7283,
+      "step": 214
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00012479999999999997,
+      "loss": 6.7195,
+      "step": 215
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00012539999999999999,
+      "loss": 6.7538,
+      "step": 216
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 6.7389,
+      "step": 217
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001266,
+      "loss": 6.6306,
+      "step": 218
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00012719999999999997,
+      "loss": 6.7153,
+      "step": 219
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001278,
+      "loss": 6.7033,
+      "step": 220
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00012839999999999998,
+      "loss": 6.7307,
+      "step": 221
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.000129,
+      "loss": 6.709,
+      "step": 222
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00012959999999999998,
+      "loss": 10.0804,
+      "step": 223
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001302,
+      "loss": 6.7021,
+      "step": 224
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013079999999999998,
+      "loss": 6.6866,
+      "step": 225
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001314,
+      "loss": 6.7554,
+      "step": 226
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 6.6395,
+      "step": 227
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001326,
+      "loss": 6.7163,
+      "step": 228
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013319999999999999,
+      "loss": 6.6904,
+      "step": 229
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001338,
+      "loss": 6.639,
+      "step": 230
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001344,
+      "loss": 6.682,
+      "step": 231
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.000135,
+      "loss": 6.6265,
+      "step": 232
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001356,
+      "loss": 6.6766,
+      "step": 233
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001362,
+      "loss": 6.6938,
+      "step": 234
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001368,
+      "loss": 6.5945,
+      "step": 235
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001374,
+      "loss": 6.5863,
+      "step": 236
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.000138,
+      "loss": 6.6788,
+      "step": 237
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001386,
+      "loss": 6.6209,
+      "step": 238
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001392,
+      "loss": 6.6078,
+      "step": 239
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00013979999999999998,
+      "loss": 6.6071,
+      "step": 240
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001404,
+      "loss": 6.6026,
+      "step": 241
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00014099999999999998,
+      "loss": 6.5646,
+      "step": 242
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00014159999999999997,
+      "loss": 6.6153,
+      "step": 243
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001422,
+      "loss": 6.6037,
+      "step": 244
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00014279999999999997,
+      "loss": 6.586,
+      "step": 245
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001434,
+      "loss": 6.6125,
+      "step": 246
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 6.6516,
+      "step": 247
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001446,
+      "loss": 6.6051,
+      "step": 248
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00014519999999999998,
+      "loss": 6.6234,
+      "step": 249
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001458,
+      "loss": 6.6312,
+      "step": 250
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00014639999999999998,
+      "loss": 6.535,
+      "step": 251
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.000147,
+      "loss": 6.6321,
+      "step": 252
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00014759999999999998,
+      "loss": 6.6405,
+      "step": 253
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001482,
+      "loss": 6.6096,
+      "step": 254
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00014879999999999998,
+      "loss": 6.5587,
+      "step": 255
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001494,
+      "loss": 6.624,
+      "step": 256
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015,
+      "loss": 6.5688,
+      "step": 257
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015059999999999997,
+      "loss": 6.5877,
+      "step": 258
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001512,
+      "loss": 6.6013,
+      "step": 259
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015179999999999998,
+      "loss": 6.5939,
+      "step": 260
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001524,
+      "loss": 6.5417,
+      "step": 261
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00015299999999999998,
+      "loss": 6.5277,
+      "step": 262
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001536,
+      "loss": 6.5757,
+      "step": 263
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00015419999999999998,
+      "loss": 6.541,
+      "step": 264
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001548,
+      "loss": 6.5708,
+      "step": 265
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00015539999999999998,
+      "loss": 6.5291,
+      "step": 266
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.000156,
+      "loss": 6.5018,
+      "step": 267
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015659999999999998,
+      "loss": 6.5522,
+      "step": 268
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001572,
+      "loss": 6.495,
+      "step": 269
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001578,
+      "loss": 6.5808,
+      "step": 270
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001584,
+      "loss": 6.512,
+      "step": 271
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.000159,
+      "loss": 6.5489,
+      "step": 272
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001596,
+      "loss": 6.5238,
+      "step": 273
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001602,
+      "loss": 6.5494,
+      "step": 274
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001608,
+      "loss": 6.5043,
+      "step": 275
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001614,
+      "loss": 6.516,
+      "step": 276
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.000162,
+      "loss": 6.5504,
+      "step": 277
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001626,
+      "loss": 6.5198,
+      "step": 278
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001632,
+      "loss": 6.3816,
+      "step": 279
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001638,
+      "loss": 6.5435,
+      "step": 280
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001644,
+      "loss": 6.6009,
+      "step": 281
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.000165,
+      "loss": 6.4185,
+      "step": 282
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001656,
+      "loss": 6.5391,
+      "step": 283
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001662,
+      "loss": 6.5006,
+      "step": 284
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001668,
+      "loss": 6.4928,
+      "step": 285
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001674,
+      "loss": 6.3818,
+      "step": 286
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000168,
+      "loss": 6.3806,
+      "step": 287
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001686,
+      "loss": 6.3834,
+      "step": 288
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00016919999999999997,
+      "loss": 6.3448,
+      "step": 289
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00016979999999999998,
+      "loss": 6.3585,
+      "step": 290
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00017039999999999997,
+      "loss": 6.4841,
+      "step": 291
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00017099999999999998,
+      "loss": 6.3267,
+      "step": 292
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00017159999999999997,
+      "loss": 6.4136,
+      "step": 293
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00017219999999999998,
+      "loss": 6.4625,
+      "step": 294
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00017279999999999997,
+      "loss": 6.3798,
+      "step": 295
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00017339999999999996,
+      "loss": 6.3827,
+      "step": 296
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 6.2869,
+      "step": 297
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00017459999999999996,
+      "loss": 6.4024,
+      "step": 298
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00017519999999999998,
+      "loss": 6.3518,
+      "step": 299
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00017579999999999996,
+      "loss": 6.3007,
+      "step": 300
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00017639999999999998,
+      "loss": 6.3029,
+      "step": 301
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00017699999999999997,
+      "loss": 6.2631,
+      "step": 302
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00017759999999999998,
+      "loss": 6.3604,
+      "step": 303
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00017819999999999997,
+      "loss": 6.2004,
+      "step": 304
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00017879999999999998,
+      "loss": 6.2803,
+      "step": 305
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00017939999999999997,
+      "loss": 6.2629,
+      "step": 306
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 6.2384,
+      "step": 307
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00018059999999999997,
+      "loss": 6.1998,
+      "step": 308
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00018119999999999999,
+      "loss": 6.2329,
+      "step": 309
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00018179999999999997,
+      "loss": 6.2285,
+      "step": 310
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001824,
+      "loss": 6.1805,
+      "step": 311
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00018299999999999998,
+      "loss": 6.2105,
+      "step": 312
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0001836,
+      "loss": 6.2011,
+      "step": 313
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00018419999999999998,
+      "loss": 6.1308,
+      "step": 314
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001848,
+      "loss": 6.1239,
+      "step": 315
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00018539999999999998,
+      "loss": 6.0931,
+      "step": 316
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000186,
+      "loss": 6.2032,
+      "step": 317
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00018659999999999998,
+      "loss": 6.0973,
+      "step": 318
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001872,
+      "loss": 6.0579,
+      "step": 319
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00018779999999999998,
+      "loss": 6.2122,
+      "step": 320
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00018839999999999997,
+      "loss": 6.2831,
+      "step": 321
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00018899999999999999,
+      "loss": 6.117,
+      "step": 322
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00018959999999999997,
+      "loss": 6.0904,
+      "step": 323
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0001902,
+      "loss": 6.0899,
+      "step": 324
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00019079999999999998,
+      "loss": 6.0549,
+      "step": 325
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0001914,
+      "loss": 6.022,
+      "step": 326
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 6.066,
+      "step": 327
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0001926,
+      "loss": 6.039,
+      "step": 328
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00019319999999999998,
+      "loss": 5.8918,
+      "step": 329
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001938,
+      "loss": 6.0871,
+      "step": 330
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00019439999999999998,
+      "loss": 6.1173,
+      "step": 331
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.000195,
+      "loss": 6.0359,
+      "step": 332
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00019559999999999998,
+      "loss": 6.1046,
+      "step": 333
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001962,
+      "loss": 9.1506,
+      "step": 334
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00019679999999999999,
+      "loss": 6.0684,
+      "step": 335
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0001974,
+      "loss": 6.0031,
+      "step": 336
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.000198,
+      "loss": 5.9665,
+      "step": 337
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0001986,
+      "loss": 5.9677,
+      "step": 338
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0001992,
+      "loss": 5.9844,
+      "step": 339
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0001998,
+      "loss": 5.9563,
+      "step": 340
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0002004,
+      "loss": 5.8874,
+      "step": 341
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.000201,
+      "loss": 5.877,
+      "step": 342
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0002016,
+      "loss": 5.8356,
+      "step": 343
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0002022,
+      "loss": 5.8978,
+      "step": 344
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0002028,
+      "loss": 5.9132,
+      "step": 345
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00020339999999999998,
+      "loss": 5.8663,
+      "step": 346
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.000204,
+      "loss": 5.842,
+      "step": 347
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00020459999999999999,
+      "loss": 5.8438,
+      "step": 348
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0002052,
+      "loss": 5.8179,
+      "step": 349
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0002058,
+      "loss": 5.8131,
+      "step": 350
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00020639999999999998,
+      "loss": 5.7429,
+      "step": 351
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00020699999999999996,
+      "loss": 5.8356,
+      "step": 352
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00020759999999999998,
+      "loss": 5.7869,
+      "step": 353
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00020819999999999996,
+      "loss": 5.799,
+      "step": 354
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00020879999999999998,
+      "loss": 5.8168,
+      "step": 355
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00020939999999999997,
+      "loss": 5.7863,
+      "step": 356
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 5.7791,
+      "step": 357
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00021059999999999997,
+      "loss": 5.7531,
+      "step": 358
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00021119999999999996,
+      "loss": 5.7887,
+      "step": 359
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021179999999999997,
+      "loss": 5.7684,
+      "step": 360
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021239999999999996,
+      "loss": 5.8428,
+      "step": 361
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 5.7234,
+      "step": 362
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021359999999999996,
+      "loss": 5.851,
+      "step": 363
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021419999999999998,
+      "loss": 5.7622,
+      "step": 364
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00021479999999999996,
+      "loss": 5.7535,
+      "step": 365
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021539999999999998,
+      "loss": 5.7839,
+      "step": 366
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 5.7288,
+      "step": 367
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021659999999999998,
+      "loss": 5.7774,
+      "step": 368
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021719999999999997,
+      "loss": 5.7072,
+      "step": 369
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021779999999999998,
+      "loss": 5.8048,
+      "step": 370
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021839999999999997,
+      "loss": 5.7507,
+      "step": 371
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 5.6932,
+      "step": 372
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021959999999999997,
+      "loss": 5.7731,
+      "step": 373
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00022019999999999999,
+      "loss": 5.7148,
+      "step": 374
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00022079999999999997,
+      "loss": 5.8572,
+      "step": 375
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0002214,
+      "loss": 5.7136,
+      "step": 376
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 5.775,
+      "step": 377
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0002226,
+      "loss": 5.7633,
+      "step": 378
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00022319999999999998,
+      "loss": 5.7403,
+      "step": 379
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0002238,
+      "loss": 5.7671,
+      "step": 380
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00022439999999999998,
+      "loss": 5.7795,
+      "step": 381
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.000225,
+      "loss": 5.8191,
+      "step": 382
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00022559999999999998,
+      "loss": 5.7037,
+      "step": 383
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00022619999999999997,
+      "loss": 5.785,
+      "step": 384
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00022679999999999998,
+      "loss": 5.871,
+      "step": 385
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00022739999999999997,
+      "loss": 5.8056,
+      "step": 386
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 5.7663,
+      "step": 387
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00022859999999999997,
+      "loss": 5.8173,
+      "step": 388
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0002292,
+      "loss": 5.8114,
+      "step": 389
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00022979999999999997,
+      "loss": 5.747,
+      "step": 390
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0002304,
+      "loss": 5.7499,
+      "step": 391
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00023099999999999998,
+      "loss": 5.7537,
+      "step": 392
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0002316,
+      "loss": 5.7054,
+      "step": 393
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00023219999999999998,
+      "loss": 5.7481,
+      "step": 394
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0002328,
+      "loss": 5.7803,
+      "step": 395
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00023339999999999998,
+      "loss": 5.7818,
+      "step": 396
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.000234,
+      "loss": 5.7315,
+      "step": 397
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00023459999999999998,
+      "loss": 5.7955,
+      "step": 398
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0002352,
+      "loss": 5.7742,
+      "step": 399
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00023579999999999999,
+      "loss": 5.7424,
+      "step": 400
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0002364,
+      "loss": 5.7091,
+      "step": 401
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.000237,
+      "loss": 5.7537,
+      "step": 402
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0002376,
+      "loss": 5.8381,
+      "step": 403
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0002382,
+      "loss": 5.6786,
+      "step": 404
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0002388,
+      "loss": 5.6061,
+      "step": 405
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0002394,
+      "loss": 5.706,
+      "step": 406
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 5.7225,
+      "step": 407
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0002406,
+      "loss": 5.7046,
+      "step": 408
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00024119999999999998,
+      "loss": 5.7265,
+      "step": 409
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0002418,
+      "loss": 5.8282,
+      "step": 410
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00024239999999999998,
+      "loss": 5.7929,
+      "step": 411
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.000243,
+      "loss": 5.7428,
+      "step": 412
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00024359999999999999,
+      "loss": 5.7645,
+      "step": 413
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00024419999999999997,
+      "loss": 5.8391,
+      "step": 414
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0002448,
+      "loss": 5.6972,
+      "step": 415
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00024539999999999995,
+      "loss": 5.7646,
+      "step": 416
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 5.6206,
+      "step": 417
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0002466,
+      "loss": 5.6798,
+      "step": 418
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002472,
+      "loss": 5.6642,
+      "step": 419
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00024779999999999995,
+      "loss": 5.6462,
+      "step": 420
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00024839999999999997,
+      "loss": 5.659,
+      "step": 421
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.000249,
+      "loss": 5.6215,
+      "step": 422
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00024959999999999994,
+      "loss": 5.6075,
+      "step": 423
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00025019999999999996,
+      "loss": 5.6371,
+      "step": 424
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00025079999999999997,
+      "loss": 5.784,
+      "step": 425
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002514,
+      "loss": 5.6776,
+      "step": 426
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 5.6431,
+      "step": 427
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00025259999999999996,
+      "loss": 5.6445,
+      "step": 428
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002532,
+      "loss": 5.5539,
+      "step": 429
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002538,
+      "loss": 5.6444,
+      "step": 430
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025439999999999995,
+      "loss": 5.6156,
+      "step": 431
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00025499999999999996,
+      "loss": 5.6254,
+      "step": 432
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002556,
+      "loss": 5.5693,
+      "step": 433
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0002562,
+      "loss": 5.5847,
+      "step": 434
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00025679999999999995,
+      "loss": 5.6204,
+      "step": 435
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00025739999999999997,
+      "loss": 5.6144,
+      "step": 436
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.000258,
+      "loss": 5.5248,
+      "step": 437
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0002586,
+      "loss": 5.6255,
+      "step": 438
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00025919999999999996,
+      "loss": 5.5507,
+      "step": 439
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00025979999999999997,
+      "loss": 5.4663,
+      "step": 440
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0002604,
+      "loss": 5.5826,
+      "step": 441
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.000261,
+      "loss": 5.61,
+      "step": 442
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00026159999999999996,
+      "loss": 5.6032,
+      "step": 443
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0002622,
+      "loss": 5.5795,
+      "step": 444
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0002628,
+      "loss": 8.3446,
+      "step": 445
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00026339999999999995,
+      "loss": 5.5092,
+      "step": 446
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 5.3557,
+      "step": 447
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0002646,
+      "loss": 5.4065,
+      "step": 448
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0002652,
+      "loss": 5.4681,
+      "step": 449
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00026579999999999996,
+      "loss": 5.5193,
+      "step": 450
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00026639999999999997,
+      "loss": 5.4022,
+      "step": 451
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.000267,
+      "loss": 5.6223,
+      "step": 452
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0002676,
+      "loss": 5.4754,
+      "step": 453
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00026819999999999996,
+      "loss": 5.539,
+      "step": 454
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0002688,
+      "loss": 5.5043,
+      "step": 455
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0002694,
+      "loss": 5.5368,
+      "step": 456
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00027,
+      "loss": 5.496,
+      "step": 457
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00027059999999999996,
+      "loss": 5.4204,
+      "step": 458
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0002712,
+      "loss": 5.4042,
+      "step": 459
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0002718,
+      "loss": 5.4032,
+      "step": 460
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0002724,
+      "loss": 5.4651,
+      "step": 461
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00027299999999999997,
+      "loss": 5.5531,
+      "step": 462
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0002736,
+      "loss": 5.5038,
+      "step": 463
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0002742,
+      "loss": 5.4859,
+      "step": 464
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0002748,
+      "loss": 5.5218,
+      "step": 465
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00027539999999999997,
+      "loss": 5.3667,
+      "step": 466
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.000276,
+      "loss": 5.473,
+      "step": 467
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0002766,
+      "loss": 5.5945,
+      "step": 468
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0002772,
+      "loss": 5.5028,
+      "step": 469
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0002778,
+      "loss": 5.6346,
+      "step": 470
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0002784,
+      "loss": 5.7832,
+      "step": 471
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.000279,
+      "loss": 5.719,
+      "step": 472
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00027959999999999997,
+      "loss": 5.6302,
+      "step": 473
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0002802,
+      "loss": 5.6425,
+      "step": 474
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0002808,
+      "loss": 5.5588,
+      "step": 475
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00028139999999999996,
+      "loss": 5.4399,
+      "step": 476
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 5.523,
+      "step": 477
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0002826,
+      "loss": 5.5261,
+      "step": 478
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00028319999999999994,
+      "loss": 5.464,
+      "step": 479
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00028379999999999996,
+      "loss": 5.4577,
+      "step": 480
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0002844,
+      "loss": 5.5357,
+      "step": 481
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.000285,
+      "loss": 5.5401,
+      "step": 482
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00028559999999999995,
+      "loss": 5.4237,
+      "step": 483
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00028619999999999996,
+      "loss": 5.3906,
+      "step": 484
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0002868,
+      "loss": 5.5005,
+      "step": 485
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00028739999999999994,
+      "loss": 5.4272,
+      "step": 486
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 5.4783,
+      "step": 487
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00028859999999999997,
+      "loss": 5.4033,
+      "step": 488
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0002892,
+      "loss": 5.4951,
+      "step": 489
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00028979999999999994,
+      "loss": 5.4467,
+      "step": 490
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00029039999999999996,
+      "loss": 5.3867,
+      "step": 491
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00029099999999999997,
+      "loss": 5.5488,
+      "step": 492
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0002916,
+      "loss": 5.6137,
+      "step": 493
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00029219999999999995,
+      "loss": 5.4724,
+      "step": 494
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00029279999999999996,
+      "loss": 5.537,
+      "step": 495
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0002934,
+      "loss": 5.4694,
+      "step": 496
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.000294,
+      "loss": 5.511,
+      "step": 497
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00029459999999999995,
+      "loss": 5.473,
+      "step": 498
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00029519999999999997,
+      "loss": 5.4759,
+      "step": 499
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0002958,
+      "loss": 5.3852,
+      "step": 500
+    },
+    {
+      "epoch": 4.5,
+      "eval_loss": 5.2345147132873535,
+      "eval_runtime": 103.4402,
+      "eval_samples_per_second": 25.541,
+      "eval_steps_per_second": 0.802,
+      "eval_wer": 0.8426418088060293,
+      "step": 500
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0002964,
+      "loss": 5.3022,
+      "step": 501
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00029699999999999996,
+      "loss": 5.4158,
+      "step": 502
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00029759999999999997,
+      "loss": 5.3723,
+      "step": 503
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0002982,
+      "loss": 5.2877,
+      "step": 504
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0002988,
+      "loss": 5.3059,
+      "step": 505
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00029939999999999996,
+      "loss": 5.3536,
+      "step": 506
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0003,
+      "loss": 5.3552,
+      "step": 507
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00029989399293286215,
+      "loss": 5.2612,
+      "step": 508
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0002997879858657244,
+      "loss": 5.4167,
+      "step": 509
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00029968197879858655,
+      "loss": 5.3359,
+      "step": 510
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00029957597173144873,
+      "loss": 5.4127,
+      "step": 511
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0002994699646643109,
+      "loss": 5.3649,
+      "step": 512
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00029936395759717313,
+      "loss": 5.2999,
+      "step": 513
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0002992579505300353,
+      "loss": 5.3059,
+      "step": 514
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00029915194346289754,
+      "loss": 5.4164,
+      "step": 515
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00029904593639575966,
+      "loss": 5.98,
+      "step": 516
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0002989399293286219,
+      "loss": 7.2467,
+      "step": 517
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00029883392226148406,
+      "loss": 7.0267,
+      "step": 518
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0002987279151943463,
+      "loss": 6.8798,
+      "step": 519
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00029862190812720847,
+      "loss": 6.6034,
+      "step": 520
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00029851590106007064,
+      "loss": 6.499,
+      "step": 521
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0002984098939929328,
+      "loss": 6.4003,
+      "step": 522
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00029830388692579505,
+      "loss": 6.3149,
+      "step": 523
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0002981978798586572,
+      "loss": 6.2935,
+      "step": 524
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0002980918727915194,
+      "loss": 6.2604,
+      "step": 525
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00029798586572438163,
+      "loss": 6.1509,
+      "step": 526
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00029787985865724375,
+      "loss": 6.2713,
+      "step": 527
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.000297773851590106,
+      "loss": 6.1769,
+      "step": 528
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00029766784452296815,
+      "loss": 6.1752,
+      "step": 529
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0002975618374558304,
+      "loss": 6.0983,
+      "step": 530
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00029745583038869256,
+      "loss": 6.1157,
+      "step": 531
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00029734982332155473,
+      "loss": 6.0675,
+      "step": 532
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0002972438162544169,
+      "loss": 6.0723,
+      "step": 533
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00029713780918727914,
+      "loss": 6.1012,
+      "step": 534
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0002970318021201413,
+      "loss": 6.0315,
+      "step": 535
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0002969257950530035,
+      "loss": 6.1201,
+      "step": 536
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0002968197879858657,
+      "loss": 5.9976,
+      "step": 537
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0002967137809187279,
+      "loss": 5.9866,
+      "step": 538
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00029660777385159007,
+      "loss": 6.089,
+      "step": 539
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00029650176678445224,
+      "loss": 6.0477,
+      "step": 540
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0002963957597173145,
+      "loss": 5.9833,
+      "step": 541
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00029628975265017665,
+      "loss": 6.0144,
+      "step": 542
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0002961837455830389,
+      "loss": 6.0408,
+      "step": 543
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.000296077738515901,
+      "loss": 5.9154,
+      "step": 544
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00029597173144876323,
+      "loss": 6.0494,
+      "step": 545
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0002958657243816254,
+      "loss": 5.9958,
+      "step": 546
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00029575971731448763,
+      "loss": 5.9682,
+      "step": 547
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0002956537102473498,
+      "loss": 5.8824,
+      "step": 548
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.000295547703180212,
+      "loss": 6.0269,
+      "step": 549
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00029544169611307416,
+      "loss": 6.0007,
+      "step": 550
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0002953356890459364,
+      "loss": 5.8982,
+      "step": 551
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00029522968197879856,
+      "loss": 5.9696,
+      "step": 552
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00029512367491166074,
+      "loss": 6.0027,
+      "step": 553
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00029501766784452297,
+      "loss": 5.8897,
+      "step": 554
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00029491166077738514,
+      "loss": 5.9003,
+      "step": 555
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0002948056537102473,
+      "loss": 8.7668,
+      "step": 556
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002946996466431095,
+      "loss": 5.8374,
+      "step": 557
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002945936395759717,
+      "loss": 5.7949,
+      "step": 558
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002944876325088339,
+      "loss": 5.7621,
+      "step": 559
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00029438162544169613,
+      "loss": 5.7757,
+      "step": 560
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029427561837455825,
+      "loss": 5.8545,
+      "step": 561
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002941696113074205,
+      "loss": 5.6965,
+      "step": 562
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029406360424028265,
+      "loss": 5.8485,
+      "step": 563
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00029395759717314483,
+      "loss": 5.7802,
+      "step": 564
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00029385159010600706,
+      "loss": 5.8165,
+      "step": 565
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029374558303886923,
+      "loss": 5.7408,
+      "step": 566
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002936395759717314,
+      "loss": 5.7879,
+      "step": 567
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002935335689045936,
+      "loss": 5.7206,
+      "step": 568
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002934275618374558,
+      "loss": 5.7492,
+      "step": 569
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.000293321554770318,
+      "loss": 5.7532,
+      "step": 570
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002932155477031802,
+      "loss": 5.7004,
+      "step": 571
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00029310954063604234,
+      "loss": 5.6947,
+      "step": 572
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00029300353356890457,
+      "loss": 5.693,
+      "step": 573
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00029289752650176674,
+      "loss": 5.6636,
+      "step": 574
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00029279151943462897,
+      "loss": 5.6422,
+      "step": 575
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00029268551236749115,
+      "loss": 5.7337,
+      "step": 576
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002925795053003533,
+      "loss": 5.6347,
+      "step": 577
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002924734982332155,
+      "loss": 5.6895,
+      "step": 578
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00029236749116607773,
+      "loss": 5.702,
+      "step": 579
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002922614840989399,
+      "loss": 5.6125,
+      "step": 580
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002921554770318021,
+      "loss": 5.7309,
+      "step": 581
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002920494699646643,
+      "loss": 5.7588,
+      "step": 582
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002919434628975265,
+      "loss": 5.6899,
+      "step": 583
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00029183745583038866,
+      "loss": 5.6627,
+      "step": 584
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00029173144876325083,
+      "loss": 5.7932,
+      "step": 585
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00029162544169611306,
+      "loss": 5.6853,
+      "step": 586
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00029151943462897524,
+      "loss": 5.5287,
+      "step": 587
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00029141342756183747,
+      "loss": 5.6856,
+      "step": 588
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002913074204946996,
+      "loss": 5.6104,
+      "step": 589
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002912014134275618,
+      "loss": 5.6015,
+      "step": 590
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.000291095406360424,
+      "loss": 5.6012,
+      "step": 591
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002909893992932862,
+      "loss": 5.7916,
+      "step": 592
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002908833922261484,
+      "loss": 5.6054,
+      "step": 593
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00029077738515901057,
+      "loss": 5.6235,
+      "step": 594
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00029067137809187275,
+      "loss": 5.6769,
+      "step": 595
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.000290565371024735,
+      "loss": 5.6043,
+      "step": 596
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00029045936395759715,
+      "loss": 5.562,
+      "step": 597
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00029035335689045933,
+      "loss": 5.6625,
+      "step": 598
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00029024734982332156,
+      "loss": 5.6077,
+      "step": 599
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00029014134275618373,
+      "loss": 5.6001,
+      "step": 600
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002900353356890459,
+      "loss": 5.5974,
+      "step": 601
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002899293286219081,
+      "loss": 5.4134,
+      "step": 602
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002898233215547703,
+      "loss": 5.5982,
+      "step": 603
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002897173144876325,
+      "loss": 5.573,
+      "step": 604
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00028961130742049466,
+      "loss": 5.3671,
+      "step": 605
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00028950530035335684,
+      "loss": 5.6459,
+      "step": 606
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00028939929328621907,
+      "loss": 5.5461,
+      "step": 607
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00028929328621908124,
+      "loss": 5.5624,
+      "step": 608
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002891872791519434,
+      "loss": 5.3497,
+      "step": 609
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00028908127208480565,
+      "loss": 5.6089,
+      "step": 610
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002889752650176678,
+      "loss": 5.5181,
+      "step": 611
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00028886925795053,
+      "loss": 5.2985,
+      "step": 612
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00028876325088339217,
+      "loss": 5.5233,
+      "step": 613
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002886572438162544,
+      "loss": 5.5152,
+      "step": 614
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002885512367491166,
+      "loss": 5.5146,
+      "step": 615
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002884452296819788,
+      "loss": 5.4393,
+      "step": 616
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00028833922261484093,
+      "loss": 5.6929,
+      "step": 617
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00028823321554770316,
+      "loss": 5.5239,
+      "step": 618
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00028812720848056533,
+      "loss": 5.5568,
+      "step": 619
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00028802120141342756,
+      "loss": 5.5452,
+      "step": 620
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00028791519434628974,
+      "loss": 5.5266,
+      "step": 621
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0002878091872791519,
+      "loss": 5.5177,
+      "step": 622
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0002877031802120141,
+      "loss": 5.4003,
+      "step": 623
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0002875971731448763,
+      "loss": 5.466,
+      "step": 624
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0002874911660777385,
+      "loss": 5.5367,
+      "step": 625
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00028738515901060067,
+      "loss": 5.4316,
+      "step": 626
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0002872791519434629,
+      "loss": 5.3539,
+      "step": 627
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00028717314487632507,
+      "loss": 5.4431,
+      "step": 628
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00028706713780918725,
+      "loss": 5.4221,
+      "step": 629
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002869611307420494,
+      "loss": 5.255,
+      "step": 630
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00028685512367491165,
+      "loss": 5.4126,
+      "step": 631
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0002867491166077738,
+      "loss": 5.4404,
+      "step": 632
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00028664310954063606,
+      "loss": 5.4202,
+      "step": 633
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0002865371024734982,
+      "loss": 5.4341,
+      "step": 634
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0002864310954063604,
+      "loss": 5.4088,
+      "step": 635
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002863250883392226,
+      "loss": 5.5343,
+      "step": 636
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00028621908127208476,
+      "loss": 5.35,
+      "step": 637
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.000286113074204947,
+      "loss": 5.5146,
+      "step": 638
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00028600706713780916,
+      "loss": 5.4257,
+      "step": 639
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00028590106007067134,
+      "loss": 5.3155,
+      "step": 640
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0002857950530035335,
+      "loss": 5.3158,
+      "step": 641
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00028568904593639574,
+      "loss": 5.4106,
+      "step": 642
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002855830388692579,
+      "loss": 5.4283,
+      "step": 643
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00028547703180212015,
+      "loss": 5.2962,
+      "step": 644
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0002853710247349823,
+      "loss": 5.491,
+      "step": 645
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0002852650176678445,
+      "loss": 5.4984,
+      "step": 646
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00028515901060070667,
+      "loss": 5.3189,
+      "step": 647
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0002850530035335689,
+      "loss": 5.3799,
+      "step": 648
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0002849469964664311,
+      "loss": 5.4903,
+      "step": 649
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00028484098939929325,
+      "loss": 5.4052,
+      "step": 650
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0002847349823321554,
+      "loss": 5.3978,
+      "step": 651
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00028462897526501766,
+      "loss": 5.3221,
+      "step": 652
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00028452296819787983,
+      "loss": 5.4221,
+      "step": 653
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.000284416961130742,
+      "loss": 5.4889,
+      "step": 654
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00028431095406360424,
+      "loss": 5.3777,
+      "step": 655
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0002842049469964664,
+      "loss": 5.2941,
+      "step": 656
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0002840989399293286,
+      "loss": 5.5401,
+      "step": 657
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00028399293286219076,
+      "loss": 5.5146,
+      "step": 658
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.000283886925795053,
+      "loss": 5.384,
+      "step": 659
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00028378091872791517,
+      "loss": 5.4563,
+      "step": 660
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0002836749116607774,
+      "loss": 5.5953,
+      "step": 661
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00028356890459363957,
+      "loss": 5.2605,
+      "step": 662
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00028346289752650175,
+      "loss": 5.4869,
+      "step": 663
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0002833568904593639,
+      "loss": 5.4814,
+      "step": 664
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00028325088339222615,
+      "loss": 5.2614,
+      "step": 665
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0002831448763250883,
+      "loss": 5.2598,
+      "step": 666
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0002830388692579505,
+      "loss": 8.0586,
+      "step": 667
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0002829328621908127,
+      "loss": 5.2129,
+      "step": 668
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00028282685512367485,
+      "loss": 5.206,
+      "step": 669
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0002827208480565371,
+      "loss": 5.0363,
+      "step": 670
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00028261484098939926,
+      "loss": 5.135,
+      "step": 671
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0002825088339222615,
+      "loss": 5.1632,
+      "step": 672
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00028240282685512366,
+      "loss": 4.9917,
+      "step": 673
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00028229681978798584,
+      "loss": 5.1641,
+      "step": 674
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.000282190812720848,
+      "loss": 5.1345,
+      "step": 675
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00028208480565371024,
+      "loss": 5.0138,
+      "step": 676
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0002819787985865724,
+      "loss": 4.9309,
+      "step": 677
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0002818727915194346,
+      "loss": 5.0972,
+      "step": 678
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00028176678445229677,
+      "loss": 4.982,
+      "step": 679
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.000281660777385159,
+      "loss": 4.9516,
+      "step": 680
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00028155477031802117,
+      "loss": 5.0129,
+      "step": 681
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00028144876325088335,
+      "loss": 5.0735,
+      "step": 682
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0002813427561837456,
+      "loss": 5.1696,
+      "step": 683
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00028123674911660775,
+      "loss": 5.0266,
+      "step": 684
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0002811307420494699,
+      "loss": 5.1083,
+      "step": 685
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0002810247349823321,
+      "loss": 5.0749,
+      "step": 686
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00028091872791519433,
+      "loss": 5.0072,
+      "step": 687
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0002808127208480565,
+      "loss": 4.9214,
+      "step": 688
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00028070671378091874,
+      "loss": 5.0497,
+      "step": 689
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0002806007067137809,
+      "loss": 5.0876,
+      "step": 690
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0002804946996466431,
+      "loss": 4.8687,
+      "step": 691
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00028038869257950526,
+      "loss": 5.0763,
+      "step": 692
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0002802826855123675,
+      "loss": 5.0954,
+      "step": 693
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00028017667844522967,
+      "loss": 5.004,
+      "step": 694
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00028007067137809184,
+      "loss": 5.0077,
+      "step": 695
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.000279964664310954,
+      "loss": 5.1099,
+      "step": 696
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00027985865724381625,
+      "loss": 5.0522,
+      "step": 697
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0002797526501766784,
+      "loss": 4.7406,
+      "step": 698
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.0002796466431095406,
+      "loss": 5.0909,
+      "step": 699
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.0002795406360424028,
+      "loss": 5.0943,
+      "step": 700
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.000279434628975265,
+      "loss": 4.8696,
+      "step": 701
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0002793286219081272,
+      "loss": 4.9255,
+      "step": 702
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00027922261484098935,
+      "loss": 5.0925,
+      "step": 703
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0002791166077738516,
+      "loss": 4.9585,
+      "step": 704
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00027901060070671376,
+      "loss": 4.9068,
+      "step": 705
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00027890459363957593,
+      "loss": 4.9544,
+      "step": 706
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00027879858657243816,
+      "loss": 5.0369,
+      "step": 707
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00027869257950530034,
+      "loss": 5.0351,
+      "step": 708
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0002785865724381625,
+      "loss": 4.8495,
+      "step": 709
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0002784805653710247,
+      "loss": 5.1366,
+      "step": 710
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0002783745583038869,
+      "loss": 5.102,
+      "step": 711
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0002782685512367491,
+      "loss": 5.3177,
+      "step": 712
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00027816254416961127,
+      "loss": 5.377,
+      "step": 713
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00027805653710247344,
+      "loss": 5.4019,
+      "step": 714
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00027795053003533567,
+      "loss": 5.0786,
+      "step": 715
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00027784452296819785,
+      "loss": 5.0237,
+      "step": 716
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0002777385159010601,
+      "loss": 5.0305,
+      "step": 717
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00027763250883392225,
+      "loss": 5.136,
+      "step": 718
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0002775265017667844,
+      "loss": 5.1054,
+      "step": 719
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0002774204946996466,
+      "loss": 4.9914,
+      "step": 720
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00027731448763250883,
+      "loss": 5.2612,
+      "step": 721
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.000277208480565371,
+      "loss": 5.3295,
+      "step": 722
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0002771024734982332,
+      "loss": 5.1276,
+      "step": 723
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0002769964664310954,
+      "loss": 5.2177,
+      "step": 724
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0002768904593639576,
+      "loss": 5.0446,
+      "step": 725
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00027678445229681976,
+      "loss": 5.0489,
+      "step": 726
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00027667844522968194,
+      "loss": 5.135,
+      "step": 727
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00027657243816254416,
+      "loss": 5.121,
+      "step": 728
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00027646643109540634,
+      "loss": 5.0093,
+      "step": 729
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.0002763604240282685,
+      "loss": 4.9,
+      "step": 730
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0002762544169611307,
+      "loss": 5.0399,
+      "step": 731
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0002761484098939929,
+      "loss": 4.8973,
+      "step": 732
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0002760424028268551,
+      "loss": 5.0075,
+      "step": 733
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.0002759363957597173,
+      "loss": 4.9165,
+      "step": 734
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0002758303886925795,
+      "loss": 5.03,
+      "step": 735
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0002757243816254417,
+      "loss": 5.0202,
+      "step": 736
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00027561837455830385,
+      "loss": 4.917,
+      "step": 737
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0002755123674911661,
+      "loss": 4.9499,
+      "step": 738
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00027540636042402825,
+      "loss": 5.0531,
+      "step": 739
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00027530035335689043,
+      "loss": 4.951,
+      "step": 740
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0002751943462897526,
+      "loss": 4.7329,
+      "step": 741
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0002750883392226148,
+      "loss": 4.9726,
+      "step": 742
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.000274982332155477,
+      "loss": 4.9969,
+      "step": 743
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0002748763250883392,
+      "loss": 4.8898,
+      "step": 744
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0002747703180212014,
+      "loss": 4.8392,
+      "step": 745
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.0002746643109540636,
+      "loss": 5.0914,
+      "step": 746
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00027455830388692576,
+      "loss": 5.0423,
+      "step": 747
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00027445229681978794,
+      "loss": 4.7109,
+      "step": 748
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00027434628975265017,
+      "loss": 5.1205,
+      "step": 749
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00027424028268551234,
+      "loss": 4.9849,
+      "step": 750
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0002741342756183745,
+      "loss": 4.8591,
+      "step": 751
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00027402826855123675,
+      "loss": 4.9539,
+      "step": 752
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0002739222614840989,
+      "loss": 5.0014,
+      "step": 753
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0002738162544169611,
+      "loss": 4.8776,
+      "step": 754
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0002737102473498233,
+      "loss": 4.8251,
+      "step": 755
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.0002736042402826855,
+      "loss": 4.9026,
+      "step": 756
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0002734982332155477,
+      "loss": 5.0377,
+      "step": 757
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00027339222614840985,
+      "loss": 4.992,
+      "step": 758
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00027328621908127203,
+      "loss": 4.8511,
+      "step": 759
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00027318021201413426,
+      "loss": 4.8725,
+      "step": 760
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00027307420494699643,
+      "loss": 4.9185,
+      "step": 761
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00027296819787985866,
+      "loss": 4.855,
+      "step": 762
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00027286219081272084,
+      "loss": 4.7058,
+      "step": 763
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.000272756183745583,
+      "loss": 4.951,
+      "step": 764
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0002726501766784452,
+      "loss": 4.8087,
+      "step": 765
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0002725441696113074,
+      "loss": 4.786,
+      "step": 766
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.0002724381625441696,
+      "loss": 5.2038,
+      "step": 767
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00027233215547703177,
+      "loss": 5.0199,
+      "step": 768
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.000272226148409894,
+      "loss": 5.1245,
+      "step": 769
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.0002721201413427562,
+      "loss": 5.164,
+      "step": 770
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00027201413427561835,
+      "loss": 5.0617,
+      "step": 771
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0002719081272084805,
+      "loss": 4.9169,
+      "step": 772
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00027180212014134275,
+      "loss": 4.7902,
+      "step": 773
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00027169611307420493,
+      "loss": 4.8823,
+      "step": 774
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.0002715901060070671,
+      "loss": 4.8453,
+      "step": 775
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0002714840989399293,
+      "loss": 4.8212,
+      "step": 776
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.0002713780918727915,
+      "loss": 4.8423,
+      "step": 777
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.0002712720848056537,
+      "loss": 7.2163,
+      "step": 778
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.00027116607773851586,
+      "loss": 4.7852,
+      "step": 779
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.0002710600706713781,
+      "loss": 4.6514,
+      "step": 780
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.00027095406360424026,
+      "loss": 4.6681,
+      "step": 781
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.00027084805653710244,
+      "loss": 4.7892,
+      "step": 782
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.0002707420494699646,
+      "loss": 4.7226,
+      "step": 783
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.00027063604240282684,
+      "loss": 4.625,
+      "step": 784
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.000270530035335689,
+      "loss": 4.5961,
+      "step": 785
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.00027042402826855125,
+      "loss": 4.633,
+      "step": 786
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.00027031802120141337,
+      "loss": 4.664,
+      "step": 787
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.0002702120141342756,
+      "loss": 4.5812,
+      "step": 788
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.0002701060070671378,
+      "loss": 4.8005,
+      "step": 789
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.00027,
+      "loss": 4.6664,
+      "step": 790
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.0002698939929328622,
+      "loss": 4.5628,
+      "step": 791
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.00026978798586572435,
+      "loss": 4.8547,
+      "step": 792
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00026968197879858653,
+      "loss": 4.6844,
+      "step": 793
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.00026957597173144876,
+      "loss": 4.7206,
+      "step": 794
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.00026946996466431093,
+      "loss": 4.694,
+      "step": 795
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.0002693639575971731,
+      "loss": 4.6707,
+      "step": 796
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.00026925795053003534,
+      "loss": 4.7875,
+      "step": 797
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.0002691519434628975,
+      "loss": 4.7269,
+      "step": 798
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.0002690459363957597,
+      "loss": 4.6514,
+      "step": 799
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00026893992932862186,
+      "loss": 4.6829,
+      "step": 800
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.0002688339222614841,
+      "loss": 4.6732,
+      "step": 801
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.00026872791519434627,
+      "loss": 4.5352,
+      "step": 802
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.00026862190812720844,
+      "loss": 4.7352,
+      "step": 803
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.0002685159010600706,
+      "loss": 4.7328,
+      "step": 804
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.00026840989399293285,
+      "loss": 4.7364,
+      "step": 805
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.000268303886925795,
+      "loss": 4.7362,
+      "step": 806
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.00026819787985865725,
+      "loss": 4.696,
+      "step": 807
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.00026809187279151943,
+      "loss": 4.76,
+      "step": 808
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.0002679858657243816,
+      "loss": 4.4898,
+      "step": 809
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.0002678798586572438,
+      "loss": 4.8896,
+      "step": 810
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.000267773851590106,
+      "loss": 4.8916,
+      "step": 811
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.0002676678445229682,
+      "loss": 4.6619,
+      "step": 812
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00026756183745583036,
+      "loss": 4.8336,
+      "step": 813
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.0002674558303886926,
+      "loss": 5.0522,
+      "step": 814
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.0002673498233215547,
+      "loss": 4.8387,
+      "step": 815
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.00026724381625441694,
+      "loss": 4.7157,
+      "step": 816
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.0002671378091872791,
+      "loss": 4.9384,
+      "step": 817
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.00026703180212014134,
+      "loss": 4.8173,
+      "step": 818
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.0002669257950530035,
+      "loss": 4.6691,
+      "step": 819
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.0002668197879858657,
+      "loss": 4.6355,
+      "step": 820
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.00026671378091872787,
+      "loss": 4.6347,
+      "step": 821
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.0002666077738515901,
+      "loss": 4.7467,
+      "step": 822
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.0002665017667844523,
+      "loss": 4.6809,
+      "step": 823
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.00026639575971731445,
+      "loss": 4.6578,
+      "step": 824
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.0002662897526501767,
+      "loss": 4.8273,
+      "step": 825
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00026618374558303885,
+      "loss": 4.7274,
+      "step": 826
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.00026607773851590103,
+      "loss": 4.451,
+      "step": 827
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.0002659717314487632,
+      "loss": 4.804,
+      "step": 828
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00026586572438162543,
+      "loss": 4.732,
+      "step": 829
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.0002657597173144876,
+      "loss": 4.864,
+      "step": 830
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00026565371024734984,
+      "loss": 4.8633,
+      "step": 831
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.00026554770318021196,
+      "loss": 4.8346,
+      "step": 832
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.0002654416961130742,
+      "loss": 4.7613,
+      "step": 833
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.00026533568904593636,
+      "loss": 4.6769,
+      "step": 834
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.0002652296819787986,
+      "loss": 4.734,
+      "step": 835
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.00026512367491166077,
+      "loss": 4.8394,
+      "step": 836
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.00026501766784452294,
+      "loss": 4.797,
+      "step": 837
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.0002649116607773851,
+      "loss": 4.7597,
+      "step": 838
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.00026480565371024735,
+      "loss": 4.7198,
+      "step": 839
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.0002646996466431095,
+      "loss": 4.7431,
+      "step": 840
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.0002645936395759717,
+      "loss": 4.6374,
+      "step": 841
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.00026448763250883393,
+      "loss": 4.7204,
+      "step": 842
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.0002643816254416961,
+      "loss": 4.8418,
+      "step": 843
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.0002642756183745583,
+      "loss": 4.7306,
+      "step": 844
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.00026416961130742045,
+      "loss": 4.6944,
+      "step": 845
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.0002640636042402827,
+      "loss": 4.7453,
+      "step": 846
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.00026395759717314486,
+      "loss": 4.8704,
+      "step": 847
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 0.0002638515901060071,
+      "loss": 4.7941,
+      "step": 848
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.0002637455830388692,
+      "loss": 4.7031,
+      "step": 849
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.00026363957597173144,
+      "loss": 4.7867,
+      "step": 850
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.0002635335689045936,
+      "loss": 4.8137,
+      "step": 851
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 0.0002634275618374558,
+      "loss": 4.7122,
+      "step": 852
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 0.000263321554770318,
+      "loss": 4.876,
+      "step": 853
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.0002632155477031802,
+      "loss": 4.8111,
+      "step": 854
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.00026310954063604237,
+      "loss": 4.6445,
+      "step": 855
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 0.00026300353356890454,
+      "loss": 4.6689,
+      "step": 856
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.00026289752650176677,
+      "loss": 4.7872,
+      "step": 857
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.00026279151943462895,
+      "loss": 4.8233,
+      "step": 858
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 0.0002626855123674912,
+      "loss": 4.8814,
+      "step": 859
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 0.0002625795053003533,
+      "loss": 4.992,
+      "step": 860
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.00026247349823321553,
+      "loss": 4.8053,
+      "step": 861
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.0002623674911660777,
+      "loss": 4.7105,
+      "step": 862
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 0.00026226148409893993,
+      "loss": 4.7013,
+      "step": 863
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.0002621554770318021,
+      "loss": 4.8119,
+      "step": 864
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.0002620494699646643,
+      "loss": 4.6678,
+      "step": 865
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.00026194346289752646,
+      "loss": 4.6054,
+      "step": 866
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 0.0002618374558303887,
+      "loss": 4.8377,
+      "step": 867
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.00026173144876325086,
+      "loss": 4.6252,
+      "step": 868
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.00026162544169611304,
+      "loss": 4.6608,
+      "step": 869
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.00026151943462897527,
+      "loss": 4.5992,
+      "step": 870
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.00026141342756183744,
+      "loss": 4.7453,
+      "step": 871
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.0002613074204946996,
+      "loss": 4.6745,
+      "step": 872
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 0.0002612014134275618,
+      "loss": 4.5886,
+      "step": 873
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.000261095406360424,
+      "loss": 4.6283,
+      "step": 874
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.0002609893992932862,
+      "loss": 4.6828,
+      "step": 875
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 0.0002608833922261484,
+      "loss": 4.6668,
+      "step": 876
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.00026077738515901055,
+      "loss": 4.6704,
+      "step": 877
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.0002606713780918728,
+      "loss": 4.9387,
+      "step": 878
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.00026056537102473495,
+      "loss": 4.8809,
+      "step": 879
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 0.0002604593639575972,
+      "loss": 4.6931,
+      "step": 880
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.00026035335689045936,
+      "loss": 4.6084,
+      "step": 881
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.00026024734982332153,
+      "loss": 4.7524,
+      "step": 882
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 0.0002601413427561837,
+      "loss": 4.7022,
+      "step": 883
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.0002600353356890459,
+      "loss": 4.4785,
+      "step": 884
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.0002599293286219081,
+      "loss": 4.7606,
+      "step": 885
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.0002598233215547703,
+      "loss": 4.6873,
+      "step": 886
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.0002597173144876325,
+      "loss": 4.5922,
+      "step": 887
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.00025961130742049464,
+      "loss": 4.5589,
+      "step": 888
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 0.00025950530035335687,
+      "loss": 6.8355,
+      "step": 889
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 0.00025939929328621904,
+      "loss": 4.5329,
+      "step": 890
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 0.00025929328621908127,
+      "loss": 4.4104,
+      "step": 891
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 0.00025918727915194345,
+      "loss": 4.3879,
+      "step": 892
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 0.0002590812720848056,
+      "loss": 4.6089,
+      "step": 893
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 0.0002589752650176678,
+      "loss": 4.5567,
+      "step": 894
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 0.00025886925795053,
+      "loss": 4.4093,
+      "step": 895
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 0.0002587632508833922,
+      "loss": 4.6055,
+      "step": 896
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 0.0002586572438162544,
+      "loss": 4.5585,
+      "step": 897
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 0.0002585512367491166,
+      "loss": 4.4368,
+      "step": 898
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 0.0002584452296819788,
+      "loss": 4.5846,
+      "step": 899
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 0.00025833922261484096,
+      "loss": 4.6313,
+      "step": 900
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 0.00025823321554770313,
+      "loss": 4.6879,
+      "step": 901
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 0.00025812720848056536,
+      "loss": 4.5497,
+      "step": 902
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 0.00025802120141342754,
+      "loss": 4.6019,
+      "step": 903
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 0.00025791519434628977,
+      "loss": 4.5156,
+      "step": 904
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 0.0002578091872791519,
+      "loss": 4.4476,
+      "step": 905
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 0.0002577031802120141,
+      "loss": 4.4108,
+      "step": 906
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 0.0002575971731448763,
+      "loss": 4.5996,
+      "step": 907
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 0.0002574911660777385,
+      "loss": 4.4934,
+      "step": 908
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 0.0002573851590106007,
+      "loss": 4.4645,
+      "step": 909
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 0.00025727915194346287,
+      "loss": 4.4766,
+      "step": 910
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 0.00025717314487632505,
+      "loss": 4.4728,
+      "step": 911
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 0.0002570671378091873,
+      "loss": 4.4158,
+      "step": 912
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 0.00025696113074204945,
+      "loss": 4.2816,
+      "step": 913
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 0.0002568551236749116,
+      "loss": 4.5644,
+      "step": 914
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 0.00025674911660777386,
+      "loss": 4.5487,
+      "step": 915
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 0.00025664310954063603,
+      "loss": 4.754,
+      "step": 916
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 0.0002565371024734982,
+      "loss": 4.6099,
+      "step": 917
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 0.0002564310954063604,
+      "loss": 4.7047,
+      "step": 918
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 0.0002563250883392226,
+      "loss": 4.5998,
+      "step": 919
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 0.0002562190812720848,
+      "loss": 4.4305,
+      "step": 920
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 0.00025611307420494696,
+      "loss": 4.7719,
+      "step": 921
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 0.00025600706713780914,
+      "loss": 4.6597,
+      "step": 922
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 0.00025590106007067137,
+      "loss": 4.4868,
+      "step": 923
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 0.00025579505300353354,
+      "loss": 4.4035,
+      "step": 924
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 0.0002556890459363957,
+      "loss": 4.578,
+      "step": 925
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 0.00025558303886925795,
+      "loss": 4.5363,
+      "step": 926
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 0.0002554770318021201,
+      "loss": 4.4661,
+      "step": 927
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 0.0002553710247349823,
+      "loss": 4.5416,
+      "step": 928
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 0.00025526501766784447,
+      "loss": 4.5615,
+      "step": 929
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 0.0002551590106007067,
+      "loss": 4.556,
+      "step": 930
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 0.0002550530035335689,
+      "loss": 4.4032,
+      "step": 931
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 0.0002549469964664311,
+      "loss": 4.5457,
+      "step": 932
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 0.0002548409893992932,
+      "loss": 4.4969,
+      "step": 933
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 0.00025473498233215546,
+      "loss": 4.7067,
+      "step": 934
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 0.00025462897526501763,
+      "loss": 4.6858,
+      "step": 935
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 0.00025452296819787986,
+      "loss": 4.5573,
+      "step": 936
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 0.00025441696113074204,
+      "loss": 4.8281,
+      "step": 937
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 0.0002543109540636042,
+      "loss": 4.7376,
+      "step": 938
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 0.0002542049469964664,
+      "loss": 4.6729,
+      "step": 939
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 0.0002540989399293286,
+      "loss": 4.9028,
+      "step": 940
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 0.0002539929328621908,
+      "loss": 4.9849,
+      "step": 941
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 0.00025388692579505297,
+      "loss": 4.6394,
+      "step": 942
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 0.0002537809187279152,
+      "loss": 4.4856,
+      "step": 943
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 0.00025367491166077737,
+      "loss": 4.645,
+      "step": 944
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 0.00025356890459363955,
+      "loss": 4.5072,
+      "step": 945
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 0.0002534628975265017,
+      "loss": 4.5711,
+      "step": 946
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 0.00025335689045936395,
+      "loss": 4.4907,
+      "step": 947
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 0.0002532508833922261,
+      "loss": 4.5544,
+      "step": 948
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 0.00025314487632508836,
+      "loss": 4.4275,
+      "step": 949
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 0.0002530388692579505,
+      "loss": 4.5716,
+      "step": 950
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 0.0002529328621908127,
+      "loss": 4.4881,
+      "step": 951
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 0.0002528268551236749,
+      "loss": 4.3292,
+      "step": 952
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 0.0002527208480565371,
+      "loss": 4.5126,
+      "step": 953
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 0.0002526148409893993,
+      "loss": 4.6459,
+      "step": 954
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 0.00025250883392226146,
+      "loss": 4.5761,
+      "step": 955
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 0.00025240282685512364,
+      "loss": 4.5056,
+      "step": 956
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 0.0002522968197879858,
+      "loss": 4.8061,
+      "step": 957
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 0.00025219081272084804,
+      "loss": 4.6038,
+      "step": 958
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 0.0002520848056537102,
+      "loss": 4.6265,
+      "step": 959
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 0.00025197879858657245,
+      "loss": 4.4573,
+      "step": 960
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 0.00025187279151943457,
+      "loss": 4.4897,
+      "step": 961
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 0.0002517667844522968,
+      "loss": 4.5708,
+      "step": 962
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 0.00025166077738515897,
+      "loss": 4.4722,
+      "step": 963
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 0.0002515547703180212,
+      "loss": 4.772,
+      "step": 964
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 0.0002514487632508834,
+      "loss": 4.5794,
+      "step": 965
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 0.00025134275618374555,
+      "loss": 4.5364,
+      "step": 966
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 0.0002512367491166077,
+      "loss": 4.4184,
+      "step": 967
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 0.00025113074204946996,
+      "loss": 4.5323,
+      "step": 968
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 0.00025102473498233213,
+      "loss": 4.4833,
+      "step": 969
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 0.0002509187279151943,
+      "loss": 4.26,
+      "step": 970
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 0.00025081272084805654,
+      "loss": 4.5104,
+      "step": 971
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 0.0002507067137809187,
+      "loss": 4.6244,
+      "step": 972
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 0.0002506007067137809,
+      "loss": 4.6356,
+      "step": 973
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 0.00025049469964664306,
+      "loss": 4.5347,
+      "step": 974
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 0.0002503886925795053,
+      "loss": 4.7853,
+      "step": 975
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 0.00025028268551236747,
+      "loss": 4.9244,
+      "step": 976
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 0.0002501766784452297,
+      "loss": 4.6614,
+      "step": 977
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 0.0002500706713780918,
+      "loss": 4.4815,
+      "step": 978
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 0.00024996466431095405,
+      "loss": 4.6658,
+      "step": 979
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 0.0002498586572438162,
+      "loss": 4.6891,
+      "step": 980
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 0.00024975265017667845,
+      "loss": 4.4108,
+      "step": 981
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 0.0002496466431095406,
+      "loss": 4.4338,
+      "step": 982
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 0.0002495406360424028,
+      "loss": 4.5634,
+      "step": 983
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 0.000249434628975265,
+      "loss": 4.4104,
+      "step": 984
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 0.0002493286219081272,
+      "loss": 4.394,
+      "step": 985
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 0.0002492226148409894,
+      "loss": 4.4736,
+      "step": 986
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 0.00024911660777385156,
+      "loss": 4.4291,
+      "step": 987
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 0.0002490106007067138,
+      "loss": 4.2817,
+      "step": 988
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 0.00024890459363957596,
+      "loss": 4.6034,
+      "step": 989
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 0.00024879858657243814,
+      "loss": 4.6585,
+      "step": 990
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 0.0002486925795053003,
+      "loss": 4.5998,
+      "step": 991
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 0.00024858657243816254,
+      "loss": 4.5369,
+      "step": 992
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 0.0002484805653710247,
+      "loss": 4.6979,
+      "step": 993
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 0.0002483745583038869,
+      "loss": 4.6805,
+      "step": 994
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 0.00024826855123674907,
+      "loss": 4.4464,
+      "step": 995
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 0.0002481625441696113,
+      "loss": 4.6343,
+      "step": 996
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 0.00024805653710247347,
+      "loss": 4.5612,
+      "step": 997
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 0.00024795053003533565,
+      "loss": 4.4845,
+      "step": 998
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 0.0002478445229681979,
+      "loss": 4.6566,
+      "step": 999
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 0.00024773851590106005,
+      "loss": 7.0888,
+      "step": 1000
+    },
+    {
+      "epoch": 9.01,
+      "eval_loss": 12.838645935058594,
+      "eval_runtime": 119.6743,
+      "eval_samples_per_second": 22.077,
+      "eval_steps_per_second": 0.694,
+      "eval_wer": 1.1262594208647363,
+      "step": 1000
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 0.0002476325088339222,
+      "loss": 4.5649,
+      "step": 1001
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 0.0002475265017667844,
+      "loss": 4.5105,
+      "step": 1002
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 0.00024742049469964663,
+      "loss": 4.5627,
+      "step": 1003
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 0.0002473144876325088,
+      "loss": 4.6048,
+      "step": 1004
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 0.00024720848056537103,
+      "loss": 4.4699,
+      "step": 1005
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 0.00024710247349823316,
+      "loss": 4.4121,
+      "step": 1006
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 0.0002469964664310954,
+      "loss": 4.5244,
+      "step": 1007
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 0.00024689045936395756,
+      "loss": 4.6568,
+      "step": 1008
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 0.0002467844522968198,
+      "loss": 4.5614,
+      "step": 1009
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 0.00024667844522968196,
+      "loss": 4.3736,
+      "step": 1010
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 0.00024657243816254414,
+      "loss": 4.5541,
+      "step": 1011
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 0.0002464664310954063,
+      "loss": 4.7002,
+      "step": 1012
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 0.00024636042402826854,
+      "loss": 4.5315,
+      "step": 1013
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 0.0002462544169611307,
+      "loss": 4.4573,
+      "step": 1014
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 0.0002461484098939929,
+      "loss": 4.6468,
+      "step": 1015
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 0.0002460424028268551,
+      "loss": 4.9792,
+      "step": 1016
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 0.0002459363957597173,
+      "loss": 4.8133,
+      "step": 1017
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 0.0002458303886925795,
+      "loss": 4.5505,
+      "step": 1018
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 0.00024572438162544165,
+      "loss": 4.4603,
+      "step": 1019
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 0.0002456183745583039,
+      "loss": 4.4901,
+      "step": 1020
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 0.00024551236749116605,
+      "loss": 4.2694,
+      "step": 1021
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 0.0002454063604240283,
+      "loss": 4.5635,
+      "step": 1022
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 0.0002453003533568904,
+      "loss": 4.6001,
+      "step": 1023
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 0.00024519434628975263,
+      "loss": 4.403,
+      "step": 1024
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 0.0002450883392226148,
+      "loss": 4.4781,
+      "step": 1025
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 0.00024498233215547704,
+      "loss": 4.6113,
+      "step": 1026
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 0.0002448763250883392,
+      "loss": 4.3365,
+      "step": 1027
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 0.0002447703180212014,
+      "loss": 4.308,
+      "step": 1028
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 0.00024466431095406356,
+      "loss": 4.5436,
+      "step": 1029
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 0.00024455830388692574,
+      "loss": 4.4194,
+      "step": 1030
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 0.00024445229681978797,
+      "loss": 4.4236,
+      "step": 1031
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 0.00024434628975265014,
+      "loss": 4.6864,
+      "step": 1032
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 0.0002442402826855124,
+      "loss": 4.5107,
+      "step": 1033
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 0.00024413427561837455,
+      "loss": 4.2188,
+      "step": 1034
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 0.00024402826855123672,
+      "loss": 4.2031,
+      "step": 1035
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 0.0002439222614840989,
+      "loss": 4.3665,
+      "step": 1036
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 0.0002438162544169611,
+      "loss": 4.393,
+      "step": 1037
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 0.0002437102473498233,
+      "loss": 4.4485,
+      "step": 1038
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 0.0002436042402826855,
+      "loss": 4.4073,
+      "step": 1039
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 0.00024349823321554765,
+      "loss": 4.5809,
+      "step": 1040
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 0.00024339222614840986,
+      "loss": 4.8375,
+      "step": 1041
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 0.00024328621908127206,
+      "loss": 4.6541,
+      "step": 1042
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 0.00024318021201413426,
+      "loss": 4.315,
+      "step": 1043
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 0.00024307420494699646,
+      "loss": 4.8332,
+      "step": 1044
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 0.00024296819787985864,
+      "loss": 5.1275,
+      "step": 1045
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 0.00024286219081272081,
+      "loss": 4.6698,
+      "step": 1046
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 0.00024275618374558302,
+      "loss": 4.3889,
+      "step": 1047
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 0.00024265017667844522,
+      "loss": 4.5939,
+      "step": 1048
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 0.0002425441696113074,
+      "loss": 4.7274,
+      "step": 1049
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 0.0002424381625441696,
+      "loss": 4.6868,
+      "step": 1050
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 0.0002423321554770318,
+      "loss": 4.5352,
+      "step": 1051
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 0.00024222614840989397,
+      "loss": 4.4876,
+      "step": 1052
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 0.00024212014134275615,
+      "loss": 4.3894,
+      "step": 1053
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 0.00024201413427561835,
+      "loss": 4.4757,
+      "step": 1054
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 0.00024190812720848055,
+      "loss": 4.376,
+      "step": 1055
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 0.00024180212014134276,
+      "loss": 4.1473,
+      "step": 1056
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 0.0002416961130742049,
+      "loss": 4.4559,
+      "step": 1057
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 0.0002415901060070671,
+      "loss": 4.5006,
+      "step": 1058
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 0.0002414840989399293,
+      "loss": 4.5808,
+      "step": 1059
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 0.0002413780918727915,
+      "loss": 4.3527,
+      "step": 1060
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 0.0002412720848056537,
+      "loss": 4.3143,
+      "step": 1061
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 0.0002411660777385159,
+      "loss": 4.3772,
+      "step": 1062
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 0.00024106007067137806,
+      "loss": 4.2439,
+      "step": 1063
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 0.00024095406360424027,
+      "loss": 4.3815,
+      "step": 1064
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 0.00024084805653710244,
+      "loss": 4.3464,
+      "step": 1065
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 0.00024074204946996464,
+      "loss": 4.3673,
+      "step": 1066
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 0.00024063604240282685,
+      "loss": 4.283,
+      "step": 1067
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 0.00024053003533568902,
+      "loss": 4.4524,
+      "step": 1068
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 0.0002404240282685512,
+      "loss": 4.6646,
+      "step": 1069
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 0.0002403180212014134,
+      "loss": 4.4214,
+      "step": 1070
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 0.0002402120141342756,
+      "loss": 4.3124,
+      "step": 1071
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 0.0002401060070671378,
+      "loss": 4.4943,
+      "step": 1072
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 4.4357,
+      "step": 1073
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 0.00023989399293286215,
+      "loss": 4.1677,
+      "step": 1074
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 0.00023978798586572436,
+      "loss": 4.3869,
+      "step": 1075
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 0.00023968197879858656,
+      "loss": 4.2312,
+      "step": 1076
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 0.00023957597173144873,
+      "loss": 4.2381,
+      "step": 1077
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 0.00023946996466431094,
+      "loss": 4.1415,
+      "step": 1078
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 0.00023936395759717314,
+      "loss": 4.3891,
+      "step": 1079
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 0.00023925795053003531,
+      "loss": 4.2006,
+      "step": 1080
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 0.0002391519434628975,
+      "loss": 4.0746,
+      "step": 1081
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 0.0002390459363957597,
+      "loss": 4.3251,
+      "step": 1082
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 0.0002389399293286219,
+      "loss": 4.2759,
+      "step": 1083
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 0.0002388339222614841,
+      "loss": 4.2216,
+      "step": 1084
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 0.00023872791519434624,
+      "loss": 4.3261,
+      "step": 1085
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 0.00023862190812720845,
+      "loss": 4.599,
+      "step": 1086
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 0.00023851590106007065,
+      "loss": 4.4763,
+      "step": 1087
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 0.00023840989399293285,
+      "loss": 4.6232,
+      "step": 1088
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 0.00023830388692579505,
+      "loss": 4.71,
+      "step": 1089
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 0.00023819787985865723,
+      "loss": 4.5713,
+      "step": 1090
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 0.0002380918727915194,
+      "loss": 4.6182,
+      "step": 1091
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 0.0002379858657243816,
+      "loss": 4.6535,
+      "step": 1092
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 0.00023787985865724378,
+      "loss": 4.7202,
+      "step": 1093
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 0.00023777385159010598,
+      "loss": 4.5593,
+      "step": 1094
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 0.00023766784452296819,
+      "loss": 4.4061,
+      "step": 1095
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 0.0002375618374558304,
+      "loss": 4.4052,
+      "step": 1096
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 0.00023745583038869254,
+      "loss": 4.5536,
+      "step": 1097
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 0.00023734982332155474,
+      "loss": 4.4435,
+      "step": 1098
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 0.00023724381625441694,
+      "loss": 4.2357,
+      "step": 1099
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 0.00023713780918727914,
+      "loss": 4.5167,
+      "step": 1100
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 0.00023703180212014135,
+      "loss": 4.4608,
+      "step": 1101
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 0.0002369257950530035,
+      "loss": 4.3149,
+      "step": 1102
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 0.0002368197879858657,
+      "loss": 4.2513,
+      "step": 1103
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 0.0002367137809187279,
+      "loss": 4.5624,
+      "step": 1104
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 0.0002366077738515901,
+      "loss": 4.5932,
+      "step": 1105
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 0.00023650176678445228,
+      "loss": 4.3363,
+      "step": 1106
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 0.00023639575971731448,
+      "loss": 4.514,
+      "step": 1107
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 0.00023628975265017665,
+      "loss": 4.4721,
+      "step": 1108
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 0.00023618374558303883,
+      "loss": 4.3818,
+      "step": 1109
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 0.00023607773851590103,
+      "loss": 4.2223,
+      "step": 1110
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 0.00023597173144876323,
+      "loss": 6.8651,
+      "step": 1111
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 0.00023586572438162544,
+      "loss": 4.6816,
+      "step": 1112
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 0.00023575971731448764,
+      "loss": 4.6562,
+      "step": 1113
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 0.00023565371024734979,
+      "loss": 4.5242,
+      "step": 1114
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 0.000235547703180212,
+      "loss": 4.709,
+      "step": 1115
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 0.0002354416961130742,
+      "loss": 4.8391,
+      "step": 1116
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 0.0002353356890459364,
+      "loss": 4.5993,
+      "step": 1117
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 0.00023522968197879857,
+      "loss": 4.534,
+      "step": 1118
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 0.00023512367491166074,
+      "loss": 4.8081,
+      "step": 1119
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 0.00023501766784452295,
+      "loss": 5.1667,
+      "step": 1120
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 0.00023491166077738515,
+      "loss": 5.116,
+      "step": 1121
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 0.00023480565371024732,
+      "loss": 4.9184,
+      "step": 1122
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 0.00023469964664310953,
+      "loss": 4.4577,
+      "step": 1123
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 0.00023459363957597173,
+      "loss": 4.4842,
+      "step": 1124
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 0.00023448763250883388,
+      "loss": 4.6714,
+      "step": 1125
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 0.00023438162544169608,
+      "loss": 4.6268,
+      "step": 1126
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 0.00023427561837455828,
+      "loss": 4.5181,
+      "step": 1127
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 0.00023416961130742048,
+      "loss": 4.3477,
+      "step": 1128
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 0.00023406360424028269,
+      "loss": 4.3961,
+      "step": 1129
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 0.00023395759717314483,
+      "loss": 4.4938,
+      "step": 1130
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 0.00023385159010600704,
+      "loss": 4.386,
+      "step": 1131
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 0.00023374558303886924,
+      "loss": 4.3226,
+      "step": 1132
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 0.00023363957597173144,
+      "loss": 4.4667,
+      "step": 1133
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 0.00023353356890459362,
+      "loss": 4.6154,
+      "step": 1134
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 0.00023342756183745582,
+      "loss": 4.3583,
+      "step": 1135
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 0.000233321554770318,
+      "loss": 4.4218,
+      "step": 1136
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 0.0002332155477031802,
+      "loss": 4.6267,
+      "step": 1137
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 0.00023310954063604237,
+      "loss": 4.8069,
+      "step": 1138
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 0.00023300353356890457,
+      "loss": 4.6079,
+      "step": 1139
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 0.00023289752650176678,
+      "loss": 4.6172,
+      "step": 1140
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 0.00023279151943462898,
+      "loss": 4.3636,
+      "step": 1141
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 0.00023268551236749113,
+      "loss": 4.2117,
+      "step": 1142
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 0.00023257950530035333,
+      "loss": 4.4285,
+      "step": 1143
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 0.00023247349823321553,
+      "loss": 4.3642,
+      "step": 1144
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 0.00023236749116607773,
+      "loss": 4.2758,
+      "step": 1145
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 0.0002322614840989399,
+      "loss": 4.3997,
+      "step": 1146
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 0.00023215547703180208,
+      "loss": 4.3858,
+      "step": 1147
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 0.00023204946996466429,
+      "loss": 4.3441,
+      "step": 1148
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 0.0002319434628975265,
+      "loss": 4.281,
+      "step": 1149
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 0.00023183745583038866,
+      "loss": 4.378,
+      "step": 1150
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 0.00023173144876325087,
+      "loss": 4.3031,
+      "step": 1151
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 0.00023162544169611307,
+      "loss": 4.2594,
+      "step": 1152
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 0.00023151943462897524,
+      "loss": 4.1592,
+      "step": 1153
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 0.00023141342756183742,
+      "loss": 4.3867,
+      "step": 1154
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 0.00023130742049469962,
+      "loss": 4.3716,
+      "step": 1155
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 0.00023120141342756182,
+      "loss": 4.1703,
+      "step": 1156
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 0.00023109540636042402,
+      "loss": 4.2299,
+      "step": 1157
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 0.00023098939929328623,
+      "loss": 4.4647,
+      "step": 1158
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 0.00023088339222614838,
+      "loss": 4.362,
+      "step": 1159
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 0.00023077738515901058,
+      "loss": 4.2338,
+      "step": 1160
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 0.00023067137809187278,
+      "loss": 4.4642,
+      "step": 1161
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 0.00023056537102473496,
+      "loss": 4.3834,
+      "step": 1162
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 0.00023045936395759716,
+      "loss": 4.254,
+      "step": 1163
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 0.00023035335689045933,
+      "loss": 4.2586,
+      "step": 1164
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 0.00023024734982332153,
+      "loss": 4.4135,
+      "step": 1165
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 0.0002301413427561837,
+      "loss": 4.3503,
+      "step": 1166
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 0.0002300353356890459,
+      "loss": 4.1564,
+      "step": 1167
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 0.00022992932862190811,
+      "loss": 4.3131,
+      "step": 1168
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 0.00022982332155477032,
+      "loss": 4.3957,
+      "step": 1169
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 0.00022971731448763247,
+      "loss": 4.2336,
+      "step": 1170
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 0.00022961130742049467,
+      "loss": 4.2662,
+      "step": 1171
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 0.00022950530035335687,
+      "loss": 4.3295,
+      "step": 1172
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 0.00022939929328621907,
+      "loss": 4.1605,
+      "step": 1173
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 0.00022929328621908127,
+      "loss": 4.3085,
+      "step": 1174
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 0.00022918727915194345,
+      "loss": 4.4979,
+      "step": 1175
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 0.00022908127208480562,
+      "loss": 4.2964,
+      "step": 1176
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 0.00022897526501766783,
+      "loss": 4.2695,
+      "step": 1177
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 0.00022886925795053003,
+      "loss": 4.4426,
+      "step": 1178
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 0.0002287632508833922,
+      "loss": 4.3213,
+      "step": 1179
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 0.0002286572438162544,
+      "loss": 4.2728,
+      "step": 1180
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 0.00022855123674911658,
+      "loss": 4.2961,
+      "step": 1181
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 0.00022844522968197876,
+      "loss": 4.289,
+      "step": 1182
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 0.00022833922261484096,
+      "loss": 4.2025,
+      "step": 1183
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 0.00022823321554770316,
+      "loss": 4.2726,
+      "step": 1184
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 0.00022812720848056536,
+      "loss": 4.062,
+      "step": 1185
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 0.00022802120141342757,
+      "loss": 4.3434,
+      "step": 1186
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 0.00022791519434628971,
+      "loss": 4.2801,
+      "step": 1187
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 0.00022780918727915192,
+      "loss": 4.2141,
+      "step": 1188
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 0.00022770318021201412,
+      "loss": 4.1444,
+      "step": 1189
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 0.00022759717314487632,
+      "loss": 4.5184,
+      "step": 1190
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 0.0002274911660777385,
+      "loss": 4.3208,
+      "step": 1191
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 0.00022738515901060067,
+      "loss": 4.1566,
+      "step": 1192
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 0.00022727915194346287,
+      "loss": 4.349,
+      "step": 1193
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 0.00022717314487632508,
+      "loss": 4.3859,
+      "step": 1194
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 0.00022706713780918725,
+      "loss": 4.0947,
+      "step": 1195
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 0.00022696113074204945,
+      "loss": 4.2082,
+      "step": 1196
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 0.00022685512367491166,
+      "loss": 4.3792,
+      "step": 1197
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 0.0002267491166077738,
+      "loss": 4.3292,
+      "step": 1198
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 0.000226643109540636,
+      "loss": 4.2348,
+      "step": 1199
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 0.0002265371024734982,
+      "loss": 4.4521,
+      "step": 1200
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 0.0002264310954063604,
+      "loss": 4.3444,
+      "step": 1201
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 0.00022632508833922261,
+      "loss": 4.2818,
+      "step": 1202
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 0.0002262190812720848,
+      "loss": 4.188,
+      "step": 1203
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 0.00022611307420494696,
+      "loss": 4.438,
+      "step": 1204
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 0.00022600706713780917,
+      "loss": 4.5514,
+      "step": 1205
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 0.00022590106007067137,
+      "loss": 4.4007,
+      "step": 1206
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 0.00022579505300353354,
+      "loss": 4.4092,
+      "step": 1207
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 0.00022568904593639575,
+      "loss": 4.5005,
+      "step": 1208
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 0.00022558303886925792,
+      "loss": 4.3859,
+      "step": 1209
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 0.00022547703180212012,
+      "loss": 4.1798,
+      "step": 1210
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 0.0002253710247349823,
+      "loss": 4.3648,
+      "step": 1211
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 0.0002252650176678445,
+      "loss": 4.5077,
+      "step": 1212
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 0.0002251590106007067,
+      "loss": 4.5449,
+      "step": 1213
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 0.0002250530035335689,
+      "loss": 4.3445,
+      "step": 1214
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 0.00022494699646643105,
+      "loss": 4.5186,
+      "step": 1215
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 0.00022484098939929326,
+      "loss": 4.5537,
+      "step": 1216
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 0.00022473498233215546,
+      "loss": 4.2659,
+      "step": 1217
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 0.00022462897526501766,
+      "loss": 4.6695,
+      "step": 1218
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 0.00022452296819787984,
+      "loss": 4.9624,
+      "step": 1219
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 0.00022441696113074204,
+      "loss": 4.691,
+      "step": 1220
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 0.00022431095406360421,
+      "loss": 4.4038,
+      "step": 1221
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 0.00022420494699646642,
+      "loss": 6.8829,
+      "step": 1222
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 0.0002240989399293286,
+      "loss": 4.7514,
+      "step": 1223
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 0.0002239929328621908,
+      "loss": 4.6721,
+      "step": 1224
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 0.000223886925795053,
+      "loss": 4.2423,
+      "step": 1225
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 0.00022378091872791517,
+      "loss": 4.7188,
+      "step": 1226
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 0.00022367491166077735,
+      "loss": 5.1182,
+      "step": 1227
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 0.00022356890459363955,
+      "loss": 4.9893,
+      "step": 1228
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 0.00022346289752650175,
+      "loss": 4.7754,
+      "step": 1229
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 0.00022335689045936395,
+      "loss": 4.6313,
+      "step": 1230
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 0.00022325088339222616,
+      "loss": 4.2774,
+      "step": 1231
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 0.0002231448763250883,
+      "loss": 4.3309,
+      "step": 1232
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 0.0002230388692579505,
+      "loss": 4.3576,
+      "step": 1233
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 0.0002229328621908127,
+      "loss": 4.2891,
+      "step": 1234
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 0.00022282685512367488,
+      "loss": 4.3101,
+      "step": 1235
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 0.00022272084805653709,
+      "loss": 4.4138,
+      "step": 1236
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 0.0002226148409893993,
+      "loss": 4.3298,
+      "step": 1237
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 0.00022250883392226146,
+      "loss": 4.3801,
+      "step": 1238
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 0.00022240282685512364,
+      "loss": 4.2883,
+      "step": 1239
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 0.00022229681978798584,
+      "loss": 4.4369,
+      "step": 1240
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 0.00022219081272084804,
+      "loss": 4.3181,
+      "step": 1241
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 0.00022208480565371025,
+      "loss": 4.2781,
+      "step": 1242
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 0.0002219787985865724,
+      "loss": 4.3219,
+      "step": 1243
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 0.0002218727915194346,
+      "loss": 4.3848,
+      "step": 1244
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 0.0002217667844522968,
+      "loss": 4.3109,
+      "step": 1245
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 0.000221660777385159,
+      "loss": 4.0722,
+      "step": 1246
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 0.0002215547703180212,
+      "loss": 4.3123,
+      "step": 1247
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 0.00022144876325088338,
+      "loss": 4.2236,
+      "step": 1248
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 0.00022134275618374555,
+      "loss": 4.1664,
+      "step": 1249
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 0.00022123674911660776,
+      "loss": 4.1325,
+      "step": 1250
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 0.00022113074204946993,
+      "loss": 4.2846,
+      "step": 1251
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 0.00022102473498233213,
+      "loss": 4.1984,
+      "step": 1252
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 0.00022091872791519434,
+      "loss": 3.9815,
+      "step": 1253
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 0.0002208127208480565,
+      "loss": 4.1603,
+      "step": 1254
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 0.00022070671378091869,
+      "loss": 4.1629,
+      "step": 1255
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 0.0002206007067137809,
+      "loss": 4.149,
+      "step": 1256
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 0.0002204946996466431,
+      "loss": 4.2019,
+      "step": 1257
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 0.0002203886925795053,
+      "loss": 4.2029,
+      "step": 1258
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 0.0002202826855123675,
+      "loss": 4.1307,
+      "step": 1259
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 0.00022017667844522964,
+      "loss": 4.0319,
+      "step": 1260
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 0.00022007067137809185,
+      "loss": 4.0844,
+      "step": 1261
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 0.00021996466431095405,
+      "loss": 4.1634,
+      "step": 1262
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 0.00021985865724381625,
+      "loss": 4.0733,
+      "step": 1263
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 0.00021975265017667843,
+      "loss": 4.0324,
+      "step": 1264
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 0.00021964664310954063,
+      "loss": 4.1555,
+      "step": 1265
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 0.0002195406360424028,
+      "loss": 4.3528,
+      "step": 1266
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 0.000219434628975265,
+      "loss": 4.2472,
+      "step": 1267
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 0.00021932862190812718,
+      "loss": 4.194,
+      "step": 1268
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 0.00021922261484098938,
+      "loss": 4.3294,
+      "step": 1269
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 0.00021911660777385159,
+      "loss": 4.2611,
+      "step": 1270
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 0.00021901060070671373,
+      "loss": 3.9961,
+      "step": 1271
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 0.00021890459363957594,
+      "loss": 4.1082,
+      "step": 1272
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 0.00021879858657243814,
+      "loss": 4.2423,
+      "step": 1273
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 0.00021869257950530034,
+      "loss": 4.075,
+      "step": 1274
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 0.00021858657243816254,
+      "loss": 3.9514,
+      "step": 1275
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 0.00021848056537102472,
+      "loss": 4.1946,
+      "step": 1276
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 0.0002183745583038869,
+      "loss": 4.2992,
+      "step": 1277
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 0.0002182685512367491,
+      "loss": 4.2331,
+      "step": 1278
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 0.0002181625441696113,
+      "loss": 4.3793,
+      "step": 1279
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 0.00021805653710247347,
+      "loss": 4.4308,
+      "step": 1280
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 0.00021795053003533568,
+      "loss": 4.3995,
+      "step": 1281
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 0.00021784452296819788,
+      "loss": 4.2638,
+      "step": 1282
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 0.00021773851590106005,
+      "loss": 4.3435,
+      "step": 1283
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 0.00021763250883392223,
+      "loss": 4.2549,
+      "step": 1284
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 0.00021752650176678443,
+      "loss": 4.263,
+      "step": 1285
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 0.00021742049469964663,
+      "loss": 4.3268,
+      "step": 1286
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 0.00021731448763250883,
+      "loss": 4.3477,
+      "step": 1287
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 0.00021720848056537098,
+      "loss": 4.2929,
+      "step": 1288
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 0.00021710247349823319,
+      "loss": 4.1919,
+      "step": 1289
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 0.0002169964664310954,
+      "loss": 4.2555,
+      "step": 1290
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 0.0002168904593639576,
+      "loss": 4.0891,
+      "step": 1291
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 0.00021678445229681977,
+      "loss": 4.0748,
+      "step": 1292
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 0.00021667844522968197,
+      "loss": 4.1233,
+      "step": 1293
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 0.00021657243816254414,
+      "loss": 4.0584,
+      "step": 1294
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 0.00021646643109540635,
+      "loss": 4.09,
+      "step": 1295
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 0.00021636042402826852,
+      "loss": 3.9525,
+      "step": 1296
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 0.00021625441696113072,
+      "loss": 4.1247,
+      "step": 1297
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 0.00021614840989399292,
+      "loss": 4.1117,
+      "step": 1298
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 0.00021604240282685513,
+      "loss": 4.0782,
+      "step": 1299
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 0.00021593639575971728,
+      "loss": 4.0973,
+      "step": 1300
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 0.00021583038869257948,
+      "loss": 4.3596,
+      "step": 1301
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 0.00021572438162544168,
+      "loss": 4.1288,
+      "step": 1302
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 0.00021561837455830388,
+      "loss": 4.0973,
+      "step": 1303
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 0.00021551236749116608,
+      "loss": 4.3969,
+      "step": 1304
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 0.00021540636042402823,
+      "loss": 4.3851,
+      "step": 1305
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 0.00021530035335689044,
+      "loss": 4.0062,
+      "step": 1306
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 0.00021519434628975264,
+      "loss": 4.2947,
+      "step": 1307
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 0.0002150883392226148,
+      "loss": 4.5652,
+      "step": 1308
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 0.00021498233215547701,
+      "loss": 4.4227,
+      "step": 1309
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 0.00021487632508833922,
+      "loss": 4.0281,
+      "step": 1310
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 0.0002147703180212014,
+      "loss": 4.0604,
+      "step": 1311
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 0.00021466431095406357,
+      "loss": 4.3729,
+      "step": 1312
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 0.00021455830388692577,
+      "loss": 4.1629,
+      "step": 1313
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 0.00021445229681978797,
+      "loss": 4.0121,
+      "step": 1314
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 0.00021434628975265017,
+      "loss": 4.2544,
+      "step": 1315
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 0.00021424028268551232,
+      "loss": 4.6315,
+      "step": 1316
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 0.00021413427561837452,
+      "loss": 4.3106,
+      "step": 1317
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 0.00021402826855123673,
+      "loss": 4.1065,
+      "step": 1318
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 0.00021392226148409893,
+      "loss": 4.1656,
+      "step": 1319
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 0.00021381625441696113,
+      "loss": 4.5068,
+      "step": 1320
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 0.0002137102473498233,
+      "loss": 4.2359,
+      "step": 1321
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 0.00021360424028268548,
+      "loss": 4.2212,
+      "step": 1322
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 0.00021349823321554768,
+      "loss": 4.2604,
+      "step": 1323
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 0.00021339222614840986,
+      "loss": 4.2048,
+      "step": 1324
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 0.00021328621908127206,
+      "loss": 4.1068,
+      "step": 1325
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 0.00021318021201413426,
+      "loss": 4.1267,
+      "step": 1326
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 0.00021307420494699647,
+      "loss": 4.267,
+      "step": 1327
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 0.00021296819787985861,
+      "loss": 4.2387,
+      "step": 1328
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 0.00021286219081272082,
+      "loss": 4.3972,
+      "step": 1329
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 0.00021275618374558302,
+      "loss": 4.1103,
+      "step": 1330
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 0.00021265017667844522,
+      "loss": 4.0963,
+      "step": 1331
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 0.00021254416961130742,
+      "loss": 4.1952,
+      "step": 1332
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 0.00021243816254416957,
+      "loss": 6.3638,
+      "step": 1333
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 0.00021233215547703177,
+      "loss": 4.1216,
+      "step": 1334
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 0.00021222614840989398,
+      "loss": 4.1049,
+      "step": 1335
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 0.00021212014134275618,
+      "loss": 3.9374,
+      "step": 1336
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 0.00021201413427561835,
+      "loss": 4.0152,
+      "step": 1337
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 0.00021190812720848056,
+      "loss": 4.12,
+      "step": 1338
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 0.00021180212014134273,
+      "loss": 3.9621,
+      "step": 1339
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 0.0002116961130742049,
+      "loss": 4.1374,
+      "step": 1340
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 0.0002115901060070671,
+      "loss": 4.0703,
+      "step": 1341
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 0.0002114840989399293,
+      "loss": 4.0614,
+      "step": 1342
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 0.00021137809187279151,
+      "loss": 4.0909,
+      "step": 1343
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 0.00021127208480565372,
+      "loss": 4.0816,
+      "step": 1344
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 0.00021116607773851586,
+      "loss": 3.8822,
+      "step": 1345
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 0.00021106007067137807,
+      "loss": 3.9136,
+      "step": 1346
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 0.00021095406360424027,
+      "loss": 4.1319,
+      "step": 1347
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 0.00021084805653710247,
+      "loss": 4.2832,
+      "step": 1348
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 0.00021074204946996465,
+      "loss": 4.181,
+      "step": 1349
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 0.00021063604240282682,
+      "loss": 4.1775,
+      "step": 1350
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 0.00021053003533568902,
+      "loss": 4.3617,
+      "step": 1351
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 0.00021042402826855123,
+      "loss": 4.3408,
+      "step": 1352
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 0.0002103180212014134,
+      "loss": 4.1075,
+      "step": 1353
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 0.0002102120141342756,
+      "loss": 4.0313,
+      "step": 1354
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 0.0002101060070671378,
+      "loss": 4.2706,
+      "step": 1355
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 4.2503,
+      "step": 1356
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 0.00020989399293286216,
+      "loss": 4.1128,
+      "step": 1357
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 0.00020978798586572436,
+      "loss": 4.2083,
+      "step": 1358
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 0.00020968197879858656,
+      "loss": 4.2275,
+      "step": 1359
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 0.00020957597173144876,
+      "loss": 4.2133,
+      "step": 1360
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 0.00020946996466431094,
+      "loss": 4.2419,
+      "step": 1361
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 0.00020936395759717311,
+      "loss": 4.3774,
+      "step": 1362
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 0.00020925795053003532,
+      "loss": 4.4968,
+      "step": 1363
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 0.00020915194346289752,
+      "loss": 4.1719,
+      "step": 1364
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 0.0002090459363957597,
+      "loss": 4.5106,
+      "step": 1365
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 0.0002089399293286219,
+      "loss": 4.6668,
+      "step": 1366
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 0.00020883392226148407,
+      "loss": 4.5911,
+      "step": 1367
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 0.00020872791519434627,
+      "loss": 4.3638,
+      "step": 1368
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 0.00020862190812720845,
+      "loss": 4.6713,
+      "step": 1369
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 0.00020851590106007065,
+      "loss": 4.673,
+      "step": 1370
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 0.00020840989399293285,
+      "loss": 4.5564,
+      "step": 1371
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 0.00020830388692579506,
+      "loss": 4.628,
+      "step": 1372
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 0.0002081978798586572,
+      "loss": 4.569,
+      "step": 1373
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 0.0002080918727915194,
+      "loss": 4.7107,
+      "step": 1374
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 0.0002079858657243816,
+      "loss": 4.6676,
+      "step": 1375
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 0.0002078798586572438,
+      "loss": 4.674,
+      "step": 1376
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 0.00020777385159010599,
+      "loss": 4.5545,
+      "step": 1377
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 0.00020766784452296816,
+      "loss": 4.4173,
+      "step": 1378
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 0.00020756183745583036,
+      "loss": 4.496,
+      "step": 1379
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 0.00020745583038869257,
+      "loss": 4.5265,
+      "step": 1380
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 0.00020734982332155474,
+      "loss": 4.4294,
+      "step": 1381
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 0.00020724381625441694,
+      "loss": 4.2498,
+      "step": 1382
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 0.00020713780918727915,
+      "loss": 4.3994,
+      "step": 1383
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 0.00020703180212014132,
+      "loss": 4.4102,
+      "step": 1384
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 0.0002069257950530035,
+      "loss": 4.3271,
+      "step": 1385
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 0.0002068197879858657,
+      "loss": 4.1863,
+      "step": 1386
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 0.0002067137809187279,
+      "loss": 4.3781,
+      "step": 1387
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 0.0002066077738515901,
+      "loss": 4.4389,
+      "step": 1388
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 0.0002065017667844523,
+      "loss": 4.278,
+      "step": 1389
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 0.00020639575971731445,
+      "loss": 4.2342,
+      "step": 1390
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 0.00020628975265017666,
+      "loss": 4.316,
+      "step": 1391
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 0.00020618374558303886,
+      "loss": 4.3014,
+      "step": 1392
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 0.00020607773851590106,
+      "loss": 4.1806,
+      "step": 1393
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 0.00020597173144876324,
+      "loss": 4.2268,
+      "step": 1394
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 0.0002058657243816254,
+      "loss": 4.1739,
+      "step": 1395
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 0.0002057597173144876,
+      "loss": 4.1154,
+      "step": 1396
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 0.0002056537102473498,
+      "loss": 4.1201,
+      "step": 1397
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 0.000205547703180212,
+      "loss": 4.2346,
+      "step": 1398
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 0.0002054416961130742,
+      "loss": 4.078,
+      "step": 1399
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 0.0002053356890459364,
+      "loss": 4.1334,
+      "step": 1400
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 0.00020522968197879854,
+      "loss": 4.2767,
+      "step": 1401
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 0.00020512367491166075,
+      "loss": 4.084,
+      "step": 1402
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 0.00020501766784452295,
+      "loss": 4.0987,
+      "step": 1403
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 0.00020491166077738515,
+      "loss": 4.3766,
+      "step": 1404
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 0.00020480565371024735,
+      "loss": 4.3535,
+      "step": 1405
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 0.00020469964664310953,
+      "loss": 4.1464,
+      "step": 1406
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 0.0002045936395759717,
+      "loss": 4.1395,
+      "step": 1407
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 0.0002044876325088339,
+      "loss": 4.5542,
+      "step": 1408
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 0.0002043816254416961,
+      "loss": 4.5186,
+      "step": 1409
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 0.00020427561837455828,
+      "loss": 4.2614,
+      "step": 1410
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 0.00020416961130742049,
+      "loss": 4.0264,
+      "step": 1411
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 0.00020406360424028266,
+      "loss": 4.2734,
+      "step": 1412
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 0.00020395759717314484,
+      "loss": 4.2837,
+      "step": 1413
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 0.00020385159010600704,
+      "loss": 4.0578,
+      "step": 1414
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 0.00020374558303886924,
+      "loss": 4.2627,
+      "step": 1415
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 0.00020363957597173144,
+      "loss": 4.2924,
+      "step": 1416
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 0.00020353356890459365,
+      "loss": 4.0606,
+      "step": 1417
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 0.0002034275618374558,
+      "loss": 4.106,
+      "step": 1418
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 0.000203321554770318,
+      "loss": 4.1314,
+      "step": 1419
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 0.0002032155477031802,
+      "loss": 4.0711,
+      "step": 1420
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 0.0002031095406360424,
+      "loss": 4.0915,
+      "step": 1421
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 0.00020300353356890458,
+      "loss": 4.3066,
+      "step": 1422
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 0.00020289752650176678,
+      "loss": 4.1466,
+      "step": 1423
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 0.00020279151943462895,
+      "loss": 4.2903,
+      "step": 1424
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 0.00020268551236749116,
+      "loss": 4.359,
+      "step": 1425
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 0.00020257950530035333,
+      "loss": 4.2133,
+      "step": 1426
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 0.00020247349823321553,
+      "loss": 4.1337,
+      "step": 1427
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 0.00020236749116607774,
+      "loss": 4.4211,
+      "step": 1428
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 0.00020226148409893988,
+      "loss": 4.8453,
+      "step": 1429
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 0.00020215547703180209,
+      "loss": 5.0693,
+      "step": 1430
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 0.0002020494699646643,
+      "loss": 4.7392,
+      "step": 1431
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 0.0002019434628975265,
+      "loss": 4.2157,
+      "step": 1432
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 0.0002018374558303887,
+      "loss": 4.3292,
+      "step": 1433
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 0.00020173144876325087,
+      "loss": 4.334,
+      "step": 1434
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 0.00020162544169611304,
+      "loss": 4.2974,
+      "step": 1435
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 0.00020151943462897525,
+      "loss": 4.2368,
+      "step": 1436
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 0.00020141342756183745,
+      "loss": 4.2408,
+      "step": 1437
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 0.00020130742049469962,
+      "loss": 4.2412,
+      "step": 1438
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 0.00020120141342756183,
+      "loss": 3.9215,
+      "step": 1439
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 0.000201095406360424,
+      "loss": 4.3949,
+      "step": 1440
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 0.0002009893992932862,
+      "loss": 4.4693,
+      "step": 1441
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 0.00020088339222614838,
+      "loss": 4.2493,
+      "step": 1442
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 0.00020077738515901058,
+      "loss": 4.1324,
+      "step": 1443
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 0.00020067137809187278,
+      "loss": 6.5789,
+      "step": 1444
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 0.00020056537102473498,
+      "loss": 4.669,
+      "step": 1445
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 0.00020045936395759713,
+      "loss": 4.5186,
+      "step": 1446
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 0.00020035335689045934,
+      "loss": 4.1561,
+      "step": 1447
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 0.00020024734982332154,
+      "loss": 4.1633,
+      "step": 1448
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 0.00020014134275618374,
+      "loss": 4.2904,
+      "step": 1449
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 0.00020003533568904592,
+      "loss": 3.9881,
+      "step": 1450
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 0.00019992932862190812,
+      "loss": 4.0905,
+      "step": 1451
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 0.0001998233215547703,
+      "loss": 4.1359,
+      "step": 1452
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 0.0001997173144876325,
+      "loss": 4.2493,
+      "step": 1453
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 0.00019961130742049467,
+      "loss": 3.9967,
+      "step": 1454
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 0.00019950530035335687,
+      "loss": 4.1014,
+      "step": 1455
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 0.00019939929328621907,
+      "loss": 4.2154,
+      "step": 1456
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 0.00019929328621908125,
+      "loss": 3.9963,
+      "step": 1457
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 0.00019918727915194343,
+      "loss": 4.1449,
+      "step": 1458
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 0.00019908127208480563,
+      "loss": 4.1978,
+      "step": 1459
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 0.00019897526501766783,
+      "loss": 4.1542,
+      "step": 1460
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 0.00019886925795053003,
+      "loss": 3.9659,
+      "step": 1461
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 0.00019876325088339223,
+      "loss": 4.1012,
+      "step": 1462
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 0.00019865724381625438,
+      "loss": 4.0386,
+      "step": 1463
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 0.00019855123674911658,
+      "loss": 3.9022,
+      "step": 1464
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 0.0001984452296819788,
+      "loss": 3.9313,
+      "step": 1465
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 0.00019833922261484096,
+      "loss": 4.1079,
+      "step": 1466
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 0.00019823321554770316,
+      "loss": 4.0969,
+      "step": 1467
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 0.00019812720848056537,
+      "loss": 3.933,
+      "step": 1468
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 0.00019802120141342754,
+      "loss": 4.1313,
+      "step": 1469
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 0.00019791519434628972,
+      "loss": 3.931,
+      "step": 1470
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 0.00019780918727915192,
+      "loss": 4.0305,
+      "step": 1471
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 0.00019770318021201412,
+      "loss": 3.8834,
+      "step": 1472
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 0.00019759717314487632,
+      "loss": 3.9791,
+      "step": 1473
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 0.00019749116607773847,
+      "loss": 3.9516,
+      "step": 1474
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 0.00019738515901060067,
+      "loss": 3.8197,
+      "step": 1475
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 0.00019727915194346288,
+      "loss": 4.125,
+      "step": 1476
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 0.00019717314487632508,
+      "loss": 3.9858,
+      "step": 1477
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 0.00019706713780918728,
+      "loss": 4.2096,
+      "step": 1478
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 0.00019696113074204946,
+      "loss": 4.4145,
+      "step": 1479
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 0.00019685512367491163,
+      "loss": 4.5517,
+      "step": 1480
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 0.00019674911660777383,
+      "loss": 4.0303,
+      "step": 1481
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 0.00019664310954063604,
+      "loss": 4.0767,
+      "step": 1482
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 0.0001965371024734982,
+      "loss": 4.4176,
+      "step": 1483
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 0.00019643109540636041,
+      "loss": 4.3015,
+      "step": 1484
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 0.00019632508833922262,
+      "loss": 4.1761,
+      "step": 1485
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 0.00019621908127208476,
+      "loss": 3.8936,
+      "step": 1486
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 0.00019611307420494697,
+      "loss": 4.0972,
+      "step": 1487
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 0.00019600706713780917,
+      "loss": 3.9676,
+      "step": 1488
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 0.00019590106007067137,
+      "loss": 3.9527,
+      "step": 1489
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 0.00019579505300353357,
+      "loss": 3.8698,
+      "step": 1490
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 0.00019568904593639572,
+      "loss": 3.9877,
+      "step": 1491
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 0.00019558303886925792,
+      "loss": 3.9973,
+      "step": 1492
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 0.00019547703180212013,
+      "loss": 3.7693,
+      "step": 1493
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 0.00019537102473498233,
+      "loss": 3.9921,
+      "step": 1494
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 0.0001952650176678445,
+      "loss": 4.0378,
+      "step": 1495
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 0.0001951590106007067,
+      "loss": 3.8833,
+      "step": 1496
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 0.00019505300353356888,
+      "loss": 3.8157,
+      "step": 1497
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 0.00019494699646643108,
+      "loss": 3.9983,
+      "step": 1498
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 0.00019484098939929326,
+      "loss": 4.0309,
+      "step": 1499
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 0.00019473498233215546,
+      "loss": 3.8539,
+      "step": 1500
+    },
+    {
+      "epoch": 13.51,
+      "eval_loss": 17.95220184326172,
+      "eval_runtime": 120.5708,
+      "eval_samples_per_second": 21.912,
+      "eval_steps_per_second": 0.688,
+      "eval_wer": 1.0014875049583498,
+      "step": 1500
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 0.00019462897526501766,
+      "loss": 4.0906,
+      "step": 1501
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 0.00019452296819787987,
+      "loss": 3.9744,
+      "step": 1502
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 0.00019441696113074201,
+      "loss": 3.9933,
+      "step": 1503
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 0.00019431095406360422,
+      "loss": 3.9988,
+      "step": 1504
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 0.00019420494699646642,
+      "loss": 4.1301,
+      "step": 1505
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 0.00019409893992932862,
+      "loss": 4.1331,
+      "step": 1506
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 0.0001939929328621908,
+      "loss": 4.0318,
+      "step": 1507
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 0.00019388692579505297,
+      "loss": 4.0255,
+      "step": 1508
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 0.00019378091872791517,
+      "loss": 3.9745,
+      "step": 1509
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 0.00019367491166077738,
+      "loss": 4.0409,
+      "step": 1510
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 0.00019356890459363955,
+      "loss": 3.9426,
+      "step": 1511
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 0.00019346289752650175,
+      "loss": 4.035,
+      "step": 1512
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 0.00019335689045936396,
+      "loss": 4.0852,
+      "step": 1513
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 0.00019325088339222613,
+      "loss": 4.1002,
+      "step": 1514
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 0.0001931448763250883,
+      "loss": 4.3594,
+      "step": 1515
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 0.0001930388692579505,
+      "loss": 4.5277,
+      "step": 1516
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 0.0001929328621908127,
+      "loss": 4.1814,
+      "step": 1517
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 0.0001928268551236749,
+      "loss": 3.9,
+      "step": 1518
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 0.00019272084805653706,
+      "loss": 4.0462,
+      "step": 1519
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 0.00019261484098939926,
+      "loss": 4.0675,
+      "step": 1520
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 0.00019250883392226147,
+      "loss": 3.8554,
+      "step": 1521
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 0.00019240282685512367,
+      "loss": 4.0719,
+      "step": 1522
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 0.00019229681978798584,
+      "loss": 4.356,
+      "step": 1523
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 0.00019219081272084805,
+      "loss": 4.1353,
+      "step": 1524
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 0.00019208480565371022,
+      "loss": 3.7793,
+      "step": 1525
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 0.00019197879858657242,
+      "loss": 4.1189,
+      "step": 1526
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 0.0001918727915194346,
+      "loss": 4.4203,
+      "step": 1527
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 0.0001917667844522968,
+      "loss": 4.1081,
+      "step": 1528
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 0.000191660777385159,
+      "loss": 3.9067,
+      "step": 1529
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 0.0001915547703180212,
+      "loss": 4.0498,
+      "step": 1530
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 0.00019144876325088335,
+      "loss": 4.1662,
+      "step": 1531
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 0.00019134275618374556,
+      "loss": 4.0257,
+      "step": 1532
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 0.00019123674911660776,
+      "loss": 3.9296,
+      "step": 1533
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 0.00019113074204946996,
+      "loss": 3.9762,
+      "step": 1534
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 0.00019102473498233216,
+      "loss": 4.0418,
+      "step": 1535
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 0.0001909187279151943,
+      "loss": 3.869,
+      "step": 1536
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 0.00019081272084805651,
+      "loss": 3.8489,
+      "step": 1537
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 0.00019070671378091872,
+      "loss": 3.978,
+      "step": 1538
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 0.0001906007067137809,
+      "loss": 4.101,
+      "step": 1539
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 0.0001904946996466431,
+      "loss": 4.1656,
+      "step": 1540
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 0.0001903886925795053,
+      "loss": 3.9853,
+      "step": 1541
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 0.00019028268551236747,
+      "loss": 3.9247,
+      "step": 1542
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 0.00019017667844522965,
+      "loss": 3.8897,
+      "step": 1543
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 0.00019007067137809185,
+      "loss": 4.126,
+      "step": 1544
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 0.00018996466431095405,
+      "loss": 3.9216,
+      "step": 1545
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 0.00018985865724381625,
+      "loss": 3.998,
+      "step": 1546
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 0.00018975265017667846,
+      "loss": 4.0799,
+      "step": 1547
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 0.0001896466431095406,
+      "loss": 4.1233,
+      "step": 1548
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 0.0001895406360424028,
+      "loss": 3.9164,
+      "step": 1549
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 0.000189434628975265,
+      "loss": 3.7542,
+      "step": 1550
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 0.0001893286219081272,
+      "loss": 4.0908,
+      "step": 1551
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 0.00018922261484098939,
+      "loss": 4.2209,
+      "step": 1552
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 0.00018911660777385156,
+      "loss": 3.8687,
+      "step": 1553
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 0.00018901060070671376,
+      "loss": 3.8151,
+      "step": 1554
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 0.00018890459363957594,
+      "loss": 6.0242,
+      "step": 1555
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 0.00018879858657243814,
+      "loss": 4.0986,
+      "step": 1556
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 0.00018869257950530034,
+      "loss": 3.7692,
+      "step": 1557
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 0.00018858657243816255,
+      "loss": 3.812,
+      "step": 1558
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 0.0001884805653710247,
+      "loss": 4.0286,
+      "step": 1559
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 0.0001883745583038869,
+      "loss": 3.959,
+      "step": 1560
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 0.0001882685512367491,
+      "loss": 3.7078,
+      "step": 1561
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 0.0001881625441696113,
+      "loss": 3.9728,
+      "step": 1562
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 0.0001880565371024735,
+      "loss": 4.1141,
+      "step": 1563
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 0.00018795053003533568,
+      "loss": 3.8846,
+      "step": 1564
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 0.00018784452296819785,
+      "loss": 3.7652,
+      "step": 1565
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 0.00018773851590106006,
+      "loss": 3.8781,
+      "step": 1566
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 0.00018763250883392226,
+      "loss": 3.9402,
+      "step": 1567
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 0.00018752650176678443,
+      "loss": 3.8678,
+      "step": 1568
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 0.00018742049469964664,
+      "loss": 3.84,
+      "step": 1569
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 0.0001873144876325088,
+      "loss": 3.8698,
+      "step": 1570
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 0.000187208480565371,
+      "loss": 3.9105,
+      "step": 1571
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 0.0001871024734982332,
+      "loss": 3.8262,
+      "step": 1572
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 0.0001869964664310954,
+      "loss": 3.8873,
+      "step": 1573
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 0.0001868904593639576,
+      "loss": 3.8678,
+      "step": 1574
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 0.0001867844522968198,
+      "loss": 4.0736,
+      "step": 1575
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 0.00018667844522968194,
+      "loss": 3.9051,
+      "step": 1576
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 0.00018657243816254415,
+      "loss": 3.9239,
+      "step": 1577
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 0.00018646643109540635,
+      "loss": 3.7392,
+      "step": 1578
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 0.00018636042402826855,
+      "loss": 3.7801,
+      "step": 1579
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 0.00018625441696113073,
+      "loss": 3.9344,
+      "step": 1580
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 0.0001861484098939929,
+      "loss": 3.752,
+      "step": 1581
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 0.0001860424028268551,
+      "loss": 3.8429,
+      "step": 1582
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 0.0001859363957597173,
+      "loss": 3.7931,
+      "step": 1583
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 0.00018583038869257948,
+      "loss": 3.8605,
+      "step": 1584
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 0.00018572438162544168,
+      "loss": 3.9925,
+      "step": 1585
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 0.00018561837455830388,
+      "loss": 3.9503,
+      "step": 1586
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 0.00018551236749116606,
+      "loss": 3.9796,
+      "step": 1587
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 0.00018540636042402824,
+      "loss": 3.9014,
+      "step": 1588
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 0.00018530035335689044,
+      "loss": 3.9769,
+      "step": 1589
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 0.00018519434628975264,
+      "loss": 4.0285,
+      "step": 1590
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 0.00018508833922261484,
+      "loss": 3.9446,
+      "step": 1591
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 0.00018498233215547702,
+      "loss": 3.9636,
+      "step": 1592
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 0.0001848763250883392,
+      "loss": 4.1253,
+      "step": 1593
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 0.0001847703180212014,
+      "loss": 4.0358,
+      "step": 1594
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 0.0001846643109540636,
+      "loss": 3.7455,
+      "step": 1595
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 0.00018455830388692577,
+      "loss": 3.9122,
+      "step": 1596
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 0.00018445229681978797,
+      "loss": 3.7123,
+      "step": 1597
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 0.00018434628975265015,
+      "loss": 3.8581,
+      "step": 1598
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 0.00018424028268551235,
+      "loss": 3.8411,
+      "step": 1599
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 0.00018413427561837453,
+      "loss": 3.9593,
+      "step": 1600
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 0.00018402826855123673,
+      "loss": 3.9123,
+      "step": 1601
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 0.00018392226148409893,
+      "loss": 3.8934,
+      "step": 1602
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 0.00018381625441696113,
+      "loss": 3.9914,
+      "step": 1603
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 0.00018371024734982328,
+      "loss": 3.6967,
+      "step": 1604
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 0.00018360424028268548,
+      "loss": 3.9719,
+      "step": 1605
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 0.0001834982332155477,
+      "loss": 4.0004,
+      "step": 1606
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 0.0001833922261484099,
+      "loss": 4.1345,
+      "step": 1607
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 0.0001832862190812721,
+      "loss": 3.9624,
+      "step": 1608
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 0.00018318021201413427,
+      "loss": 4.0265,
+      "step": 1609
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 0.00018307420494699644,
+      "loss": 3.927,
+      "step": 1610
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 0.00018296819787985864,
+      "loss": 3.8849,
+      "step": 1611
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 0.00018286219081272082,
+      "loss": 3.9754,
+      "step": 1612
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 0.00018275618374558302,
+      "loss": 4.0859,
+      "step": 1613
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 0.00018265017667844522,
+      "loss": 3.9888,
+      "step": 1614
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 0.0001825441696113074,
+      "loss": 3.969,
+      "step": 1615
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 0.00018243816254416957,
+      "loss": 4.1039,
+      "step": 1616
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 0.00018233215547703178,
+      "loss": 4.0758,
+      "step": 1617
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 0.00018222614840989398,
+      "loss": 4.0426,
+      "step": 1618
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 0.00018212014134275618,
+      "loss": 4.0928,
+      "step": 1619
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 0.00018201413427561838,
+      "loss": 4.2383,
+      "step": 1620
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 0.00018190812720848053,
+      "loss": 4.6135,
+      "step": 1621
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 0.00018180212014134273,
+      "loss": 4.3368,
+      "step": 1622
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 0.00018169611307420494,
+      "loss": 4.0562,
+      "step": 1623
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 0.00018159010600706714,
+      "loss": 4.2344,
+      "step": 1624
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 0.00018148409893992931,
+      "loss": 4.1691,
+      "step": 1625
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 0.00018137809187279152,
+      "loss": 4.1175,
+      "step": 1626
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 0.0001812720848056537,
+      "loss": 4.2339,
+      "step": 1627
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 0.00018116607773851587,
+      "loss": 4.0677,
+      "step": 1628
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 0.00018106007067137807,
+      "loss": 3.8828,
+      "step": 1629
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 0.00018095406360424027,
+      "loss": 4.0864,
+      "step": 1630
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 0.00018084805653710247,
+      "loss": 4.1445,
+      "step": 1631
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 0.00018074204946996462,
+      "loss": 4.1358,
+      "step": 1632
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 0.00018063604240282682,
+      "loss": 3.9856,
+      "step": 1633
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 0.00018053003533568903,
+      "loss": 4.0951,
+      "step": 1634
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 0.00018042402826855123,
+      "loss": 3.985,
+      "step": 1635
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 0.00018031802120141343,
+      "loss": 4.0259,
+      "step": 1636
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 0.0001802120141342756,
+      "loss": 4.3511,
+      "step": 1637
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 0.00018010600706713778,
+      "loss": 4.2884,
+      "step": 1638
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 4.2167,
+      "step": 1639
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 0.0001798939929328622,
+      "loss": 4.0923,
+      "step": 1640
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 0.00017978798586572436,
+      "loss": 4.5421,
+      "step": 1641
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 0.00017968197879858656,
+      "loss": 4.3893,
+      "step": 1642
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 0.00017957597173144874,
+      "loss": 4.1976,
+      "step": 1643
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 0.00017946996466431091,
+      "loss": 4.1085,
+      "step": 1644
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 0.00017936395759717312,
+      "loss": 4.0882,
+      "step": 1645
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 0.00017925795053003532,
+      "loss": 4.1506,
+      "step": 1646
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 0.00017915194346289752,
+      "loss": 4.0026,
+      "step": 1647
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 0.00017904593639575972,
+      "loss": 4.2335,
+      "step": 1648
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 0.00017893992932862187,
+      "loss": 4.2255,
+      "step": 1649
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 0.00017883392226148407,
+      "loss": 4.024,
+      "step": 1650
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 0.00017872791519434628,
+      "loss": 3.9624,
+      "step": 1651
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 0.00017862190812720848,
+      "loss": 4.0863,
+      "step": 1652
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 0.00017851590106007065,
+      "loss": 4.0114,
+      "step": 1653
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 0.00017840989399293286,
+      "loss": 3.8535,
+      "step": 1654
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 0.00017830388692579503,
+      "loss": 4.139,
+      "step": 1655
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 0.00017819787985865723,
+      "loss": 4.1369,
+      "step": 1656
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 0.0001780918727915194,
+      "loss": 4.0223,
+      "step": 1657
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 0.0001779858657243816,
+      "loss": 3.9563,
+      "step": 1658
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 0.00017787985865724381,
+      "loss": 4.1606,
+      "step": 1659
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 0.000177773851590106,
+      "loss": 3.9419,
+      "step": 1660
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 0.00017766784452296816,
+      "loss": 3.8788,
+      "step": 1661
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 0.00017756183745583037,
+      "loss": 4.0314,
+      "step": 1662
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 0.00017745583038869257,
+      "loss": 4.0599,
+      "step": 1663
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 0.00017734982332155477,
+      "loss": 4.0213,
+      "step": 1664
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 0.00017724381625441695,
+      "loss": 3.9593,
+      "step": 1665
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 0.00017713780918727912,
+      "loss": 6.0833,
+      "step": 1666
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 0.00017703180212014132,
+      "loss": 3.9667,
+      "step": 1667
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 0.00017692579505300353,
+      "loss": 4.0082,
+      "step": 1668
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 0.0001768197879858657,
+      "loss": 3.887,
+      "step": 1669
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 0.0001767137809187279,
+      "loss": 3.957,
+      "step": 1670
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 0.0001766077738515901,
+      "loss": 4.0235,
+      "step": 1671
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 0.00017650176678445228,
+      "loss": 3.9065,
+      "step": 1672
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 0.00017639575971731446,
+      "loss": 4.2944,
+      "step": 1673
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 0.00017628975265017666,
+      "loss": 3.9772,
+      "step": 1674
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 0.00017618374558303886,
+      "loss": 4.1171,
+      "step": 1675
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 0.00017607773851590106,
+      "loss": 4.4715,
+      "step": 1676
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 0.0001759717314487632,
+      "loss": 4.739,
+      "step": 1677
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 0.00017586572438162541,
+      "loss": 4.2065,
+      "step": 1678
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 0.00017575971731448762,
+      "loss": 3.9474,
+      "step": 1679
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 0.00017565371024734982,
+      "loss": 4.0496,
+      "step": 1680
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 0.000175547703180212,
+      "loss": 4.0034,
+      "step": 1681
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 0.0001754416961130742,
+      "loss": 4.0683,
+      "step": 1682
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 0.00017533568904593637,
+      "loss": 3.9301,
+      "step": 1683
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 0.00017522968197879857,
+      "loss": 3.9261,
+      "step": 1684
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 0.00017512367491166075,
+      "loss": 4.0139,
+      "step": 1685
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 0.00017501766784452295,
+      "loss": 3.9105,
+      "step": 1686
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 0.00017491166077738515,
+      "loss": 3.9157,
+      "step": 1687
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 0.00017480565371024736,
+      "loss": 4.0963,
+      "step": 1688
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 0.0001746996466431095,
+      "loss": 3.9046,
+      "step": 1689
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 0.0001745936395759717,
+      "loss": 3.6704,
+      "step": 1690
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 0.0001744876325088339,
+      "loss": 3.9918,
+      "step": 1691
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 0.0001743816254416961,
+      "loss": 3.9386,
+      "step": 1692
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 0.0001742756183745583,
+      "loss": 3.8342,
+      "step": 1693
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 0.00017416961130742046,
+      "loss": 3.7513,
+      "step": 1694
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 0.00017406360424028266,
+      "loss": 3.9413,
+      "step": 1695
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 0.00017395759717314487,
+      "loss": 3.8877,
+      "step": 1696
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 0.00017385159010600707,
+      "loss": 3.7488,
+      "step": 1697
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 0.00017374558303886924,
+      "loss": 3.9076,
+      "step": 1698
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 0.00017363957597173145,
+      "loss": 3.972,
+      "step": 1699
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 0.00017353356890459362,
+      "loss": 3.854,
+      "step": 1700
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 0.0001734275618374558,
+      "loss": 3.9133,
+      "step": 1701
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 0.000173321554770318,
+      "loss": 4.1887,
+      "step": 1702
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 0.0001732155477031802,
+      "loss": 4.0737,
+      "step": 1703
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 0.0001731095406360424,
+      "loss": 3.7979,
+      "step": 1704
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 0.00017300353356890455,
+      "loss": 4.1076,
+      "step": 1705
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 0.00017289752650176675,
+      "loss": 4.3067,
+      "step": 1706
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 0.00017279151943462896,
+      "loss": 4.232,
+      "step": 1707
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 0.00017268551236749116,
+      "loss": 4.1182,
+      "step": 1708
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 0.00017257950530035336,
+      "loss": 3.882,
+      "step": 1709
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 0.00017247349823321554,
+      "loss": 4.2368,
+      "step": 1710
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 0.0001723674911660777,
+      "loss": 4.387,
+      "step": 1711
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 0.0001722614840989399,
+      "loss": 4.3922,
+      "step": 1712
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 0.00017215547703180212,
+      "loss": 4.2325,
+      "step": 1713
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 0.0001720494699646643,
+      "loss": 3.9292,
+      "step": 1714
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 0.0001719434628975265,
+      "loss": 3.823,
+      "step": 1715
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 0.0001718374558303887,
+      "loss": 4.3287,
+      "step": 1716
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 0.00017173144876325084,
+      "loss": 4.4728,
+      "step": 1717
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 0.00017162544169611305,
+      "loss": 4.1196,
+      "step": 1718
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 0.00017151943462897525,
+      "loss": 3.8097,
+      "step": 1719
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 0.00017141342756183745,
+      "loss": 3.9834,
+      "step": 1720
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 0.00017130742049469965,
+      "loss": 4.0624,
+      "step": 1721
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 0.0001712014134275618,
+      "loss": 4.0535,
+      "step": 1722
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 0.000171095406360424,
+      "loss": 3.9676,
+      "step": 1723
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 0.0001709893992932862,
+      "loss": 3.8299,
+      "step": 1724
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 0.0001708833922261484,
+      "loss": 3.9946,
+      "step": 1725
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 0.00017077738515901058,
+      "loss": 4.227,
+      "step": 1726
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 0.00017067137809187279,
+      "loss": 4.5175,
+      "step": 1727
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 0.00017056537102473496,
+      "loss": 4.2755,
+      "step": 1728
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 0.00017045936395759716,
+      "loss": 3.8813,
+      "step": 1729
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 0.00017035335689045934,
+      "loss": 3.8888,
+      "step": 1730
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 0.00017024734982332154,
+      "loss": 3.8543,
+      "step": 1731
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 0.00017014134275618374,
+      "loss": 3.9493,
+      "step": 1732
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 0.00017003533568904594,
+      "loss": 3.7343,
+      "step": 1733
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 0.0001699293286219081,
+      "loss": 3.7266,
+      "step": 1734
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 0.0001698233215547703,
+      "loss": 3.8349,
+      "step": 1735
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 0.0001697173144876325,
+      "loss": 3.8405,
+      "step": 1736
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 0.0001696113074204947,
+      "loss": 3.8129,
+      "step": 1737
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 0.00016950530035335688,
+      "loss": 3.8652,
+      "step": 1738
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 0.00016939929328621905,
+      "loss": 3.8465,
+      "step": 1739
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 0.00016929328621908125,
+      "loss": 3.6405,
+      "step": 1740
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 0.00016918727915194345,
+      "loss": 3.9461,
+      "step": 1741
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 0.00016908127208480563,
+      "loss": 3.8749,
+      "step": 1742
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 0.00016897526501766783,
+      "loss": 3.8359,
+      "step": 1743
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 0.00016886925795053003,
+      "loss": 3.7829,
+      "step": 1744
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 0.0001687632508833922,
+      "loss": 3.8598,
+      "step": 1745
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 0.00016865724381625439,
+      "loss": 3.8899,
+      "step": 1746
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 0.0001685512367491166,
+      "loss": 3.6177,
+      "step": 1747
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 0.0001684452296819788,
+      "loss": 3.8476,
+      "step": 1748
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 0.000168339222614841,
+      "loss": 3.8215,
+      "step": 1749
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 0.0001682332155477032,
+      "loss": 3.669,
+      "step": 1750
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 0.00016812720848056534,
+      "loss": 3.71,
+      "step": 1751
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 0.00016802120141342754,
+      "loss": 3.8914,
+      "step": 1752
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 0.00016791519434628975,
+      "loss": 3.7652,
+      "step": 1753
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 0.00016780918727915192,
+      "loss": 3.7089,
+      "step": 1754
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 0.00016770318021201412,
+      "loss": 3.8427,
+      "step": 1755
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 0.0001675971731448763,
+      "loss": 3.7216,
+      "step": 1756
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 0.0001674911660777385,
+      "loss": 3.741,
+      "step": 1757
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 0.00016738515901060068,
+      "loss": 3.6875,
+      "step": 1758
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 0.00016727915194346288,
+      "loss": 3.8192,
+      "step": 1759
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 0.00016717314487632508,
+      "loss": 3.8987,
+      "step": 1760
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 0.00016706713780918728,
+      "loss": 3.8901,
+      "step": 1761
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 0.00016696113074204943,
+      "loss": 3.7543,
+      "step": 1762
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 0.00016685512367491163,
+      "loss": 3.7647,
+      "step": 1763
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 0.00016674911660777384,
+      "loss": 3.9275,
+      "step": 1764
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 0.00016664310954063604,
+      "loss": 3.8569,
+      "step": 1765
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 0.00016653710247349824,
+      "loss": 4.1918,
+      "step": 1766
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 0.0001664310954063604,
+      "loss": 3.9241,
+      "step": 1767
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 0.0001663250883392226,
+      "loss": 3.7838,
+      "step": 1768
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 0.0001662190812720848,
+      "loss": 3.7223,
+      "step": 1769
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 0.00016611307420494697,
+      "loss": 3.93,
+      "step": 1770
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 0.00016600706713780917,
+      "loss": 3.884,
+      "step": 1771
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 0.00016590106007067137,
+      "loss": 3.8688,
+      "step": 1772
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 0.00016579505300353355,
+      "loss": 4.37,
+      "step": 1773
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 0.00016568904593639572,
+      "loss": 4.1545,
+      "step": 1774
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 0.00016558303886925793,
+      "loss": 3.9627,
+      "step": 1775
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 0.00016547703180212013,
+      "loss": 4.0032,
+      "step": 1776
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 0.00016537102473498233,
+      "loss": 6.1629,
+      "step": 1777
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 0.00016526501766784453,
+      "loss": 4.0004,
+      "step": 1778
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 0.00016515901060070668,
+      "loss": 3.7599,
+      "step": 1779
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 0.00016505300353356888,
+      "loss": 3.9641,
+      "step": 1780
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 0.0001649469964664311,
+      "loss": 4.1383,
+      "step": 1781
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 0.0001648409893992933,
+      "loss": 3.936,
+      "step": 1782
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 0.00016473498233215546,
+      "loss": 3.7666,
+      "step": 1783
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 0.00016462897526501764,
+      "loss": 3.9638,
+      "step": 1784
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 0.00016452296819787984,
+      "loss": 3.8662,
+      "step": 1785
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 0.00016441696113074204,
+      "loss": 3.793,
+      "step": 1786
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 0.00016431095406360422,
+      "loss": 3.7715,
+      "step": 1787
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 0.00016420494699646642,
+      "loss": 3.8317,
+      "step": 1788
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 0.00016409893992932862,
+      "loss": 3.7159,
+      "step": 1789
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 0.00016399293286219077,
+      "loss": 3.7016,
+      "step": 1790
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 0.00016388692579505297,
+      "loss": 3.7729,
+      "step": 1791
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 0.00016378091872791518,
+      "loss": 3.7946,
+      "step": 1792
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 0.00016367491166077738,
+      "loss": 3.6454,
+      "step": 1793
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 0.00016356890459363958,
+      "loss": 3.6207,
+      "step": 1794
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 0.00016346289752650176,
+      "loss": 3.8222,
+      "step": 1795
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 0.00016335689045936393,
+      "loss": 3.7732,
+      "step": 1796
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 0.00016325088339222613,
+      "loss": 3.6452,
+      "step": 1797
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 0.00016314487632508834,
+      "loss": 3.703,
+      "step": 1798
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 0.0001630388692579505,
+      "loss": 3.7528,
+      "step": 1799
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 0.00016293286219081271,
+      "loss": 3.9122,
+      "step": 1800
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 0.0001628268551236749,
+      "loss": 3.7687,
+      "step": 1801
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 0.0001627208480565371,
+      "loss": 3.8898,
+      "step": 1802
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 0.00016261484098939927,
+      "loss": 3.6946,
+      "step": 1803
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 0.00016250883392226147,
+      "loss": 3.9187,
+      "step": 1804
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 0.00016240282685512367,
+      "loss": 4.1077,
+      "step": 1805
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 0.00016229681978798587,
+      "loss": 4.2483,
+      "step": 1806
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 0.00016219081272084802,
+      "loss": 3.926,
+      "step": 1807
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 0.00016208480565371022,
+      "loss": 3.5455,
+      "step": 1808
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 0.00016197879858657243,
+      "loss": 3.8617,
+      "step": 1809
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 0.00016187279151943463,
+      "loss": 3.907,
+      "step": 1810
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 0.0001617667844522968,
+      "loss": 3.7368,
+      "step": 1811
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 0.000161660777385159,
+      "loss": 3.6622,
+      "step": 1812
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 0.00016155477031802118,
+      "loss": 3.7315,
+      "step": 1813
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 0.00016144876325088338,
+      "loss": 3.7375,
+      "step": 1814
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 0.00016134275618374556,
+      "loss": 3.5825,
+      "step": 1815
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 0.00016123674911660776,
+      "loss": 3.6879,
+      "step": 1816
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 0.00016113074204946996,
+      "loss": 3.6675,
+      "step": 1817
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 0.00016102473498233214,
+      "loss": 3.658,
+      "step": 1818
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 0.00016091872791519431,
+      "loss": 3.6502,
+      "step": 1819
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 0.00016081272084805652,
+      "loss": 3.7766,
+      "step": 1820
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 0.00016070671378091872,
+      "loss": 3.7297,
+      "step": 1821
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 0.00016060070671378092,
+      "loss": 3.7071,
+      "step": 1822
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 0.00016049469964664312,
+      "loss": 3.7674,
+      "step": 1823
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 0.00016038869257950527,
+      "loss": 3.7621,
+      "step": 1824
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 0.00016028268551236747,
+      "loss": 3.8053,
+      "step": 1825
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 0.00016017667844522968,
+      "loss": 3.6331,
+      "step": 1826
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 0.00016007067137809185,
+      "loss": 3.8991,
+      "step": 1827
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 0.00015996466431095405,
+      "loss": 3.8141,
+      "step": 1828
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 0.00015985865724381623,
+      "loss": 3.7893,
+      "step": 1829
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 0.00015975265017667843,
+      "loss": 3.7852,
+      "step": 1830
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 0.0001596466431095406,
+      "loss": 3.7598,
+      "step": 1831
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 0.0001595406360424028,
+      "loss": 3.7695,
+      "step": 1832
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 0.000159434628975265,
+      "loss": 3.8943,
+      "step": 1833
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 0.0001593286219081272,
+      "loss": 3.8255,
+      "step": 1834
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 0.00015922261484098936,
+      "loss": 3.7562,
+      "step": 1835
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 0.00015911660777385156,
+      "loss": 3.7891,
+      "step": 1836
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 0.00015901060070671377,
+      "loss": 4.0645,
+      "step": 1837
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 0.00015890459363957597,
+      "loss": 4.0207,
+      "step": 1838
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 0.00015879858657243817,
+      "loss": 3.6955,
+      "step": 1839
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 0.00015869257950530035,
+      "loss": 3.7636,
+      "step": 1840
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 0.00015858657243816252,
+      "loss": 4.1412,
+      "step": 1841
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 0.00015848056537102472,
+      "loss": 4.052,
+      "step": 1842
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 0.0001583745583038869,
+      "loss": 3.808,
+      "step": 1843
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 0.0001582685512367491,
+      "loss": 3.6491,
+      "step": 1844
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 0.0001581625441696113,
+      "loss": 3.8017,
+      "step": 1845
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 0.00015805653710247348,
+      "loss": 3.8646,
+      "step": 1846
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 0.00015795053003533565,
+      "loss": 3.6845,
+      "step": 1847
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 0.00015784452296819786,
+      "loss": 3.6605,
+      "step": 1848
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 0.00015773851590106006,
+      "loss": 3.9289,
+      "step": 1849
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 0.00015763250883392226,
+      "loss": 3.8414,
+      "step": 1850
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 0.00015752650176678446,
+      "loss": 3.4809,
+      "step": 1851
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 0.0001574204946996466,
+      "loss": 3.8971,
+      "step": 1852
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 0.0001573144876325088,
+      "loss": 4.1246,
+      "step": 1853
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 0.00015720848056537102,
+      "loss": 4.0907,
+      "step": 1854
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 0.00015710247349823322,
+      "loss": 3.8334,
+      "step": 1855
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 0.0001569964664310954,
+      "loss": 3.7705,
+      "step": 1856
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 0.0001568904593639576,
+      "loss": 3.7674,
+      "step": 1857
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 0.00015678445229681977,
+      "loss": 3.5527,
+      "step": 1858
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 0.00015667844522968195,
+      "loss": 3.7244,
+      "step": 1859
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 0.00015657243816254415,
+      "loss": 3.8212,
+      "step": 1860
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 0.00015646643109540635,
+      "loss": 3.8396,
+      "step": 1861
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 0.00015636042402826855,
+      "loss": 3.8278,
+      "step": 1862
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 0.0001562544169611307,
+      "loss": 3.8262,
+      "step": 1863
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 0.0001561484098939929,
+      "loss": 3.7127,
+      "step": 1864
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 0.0001560424028268551,
+      "loss": 3.6556,
+      "step": 1865
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 0.0001559363957597173,
+      "loss": 3.7359,
+      "step": 1866
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 0.0001558303886925795,
+      "loss": 3.6976,
+      "step": 1867
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 0.00015572438162544169,
+      "loss": 3.7941,
+      "step": 1868
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 0.00015561837455830386,
+      "loss": 3.6849,
+      "step": 1869
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 0.00015551236749116606,
+      "loss": 3.7115,
+      "step": 1870
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 0.00015540636042402827,
+      "loss": 3.8301,
+      "step": 1871
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 0.00015530035335689044,
+      "loss": 3.8389,
+      "step": 1872
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 0.00015519434628975264,
+      "loss": 3.9283,
+      "step": 1873
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 0.00015508833922261484,
+      "loss": 3.7525,
+      "step": 1874
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 0.00015498233215547702,
+      "loss": 3.7493,
+      "step": 1875
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 0.0001548763250883392,
+      "loss": 3.7537,
+      "step": 1876
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 0.0001547703180212014,
+      "loss": 3.8914,
+      "step": 1877
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 0.0001546643109540636,
+      "loss": 3.8688,
+      "step": 1878
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 0.0001545583038869258,
+      "loss": 3.8909,
+      "step": 1879
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 0.00015445229681978795,
+      "loss": 3.8518,
+      "step": 1880
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 0.00015445229681978795,
+      "loss": 3.8488,
+      "step": 1881
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 0.00015434628975265015,
+      "loss": 3.7993,
+      "step": 1882
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 0.00015424028268551236,
+      "loss": 3.6359,
+      "step": 1883
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 0.00015413427561837456,
+      "loss": 3.9729,
+      "step": 1884
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 0.00015402826855123673,
+      "loss": 3.8442,
+      "step": 1885
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 0.00015392226148409893,
+      "loss": 3.723,
+      "step": 1886
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 0.0001538162544169611,
+      "loss": 3.8214,
+      "step": 1887
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 0.0001537102473498233,
+      "loss": 5.6508,
+      "step": 1888
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 0.0001536042402826855,
+      "loss": 3.8756,
+      "step": 1889
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 0.0001534982332155477,
+      "loss": 3.7948,
+      "step": 1890
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 0.0001533922261484099,
+      "loss": 3.6964,
+      "step": 1891
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 0.00015328621908127207,
+      "loss": 3.8715,
+      "step": 1892
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 0.00015318021201413424,
+      "loss": 3.8678,
+      "step": 1893
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 0.00015307420494699644,
+      "loss": 3.6011,
+      "step": 1894
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 0.00015296819787985865,
+      "loss": 4.0025,
+      "step": 1895
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 0.00015286219081272085,
+      "loss": 3.8034,
+      "step": 1896
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 0.00015275618374558305,
+      "loss": 3.7675,
+      "step": 1897
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 0.0001526501766784452,
+      "loss": 3.7133,
+      "step": 1898
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 0.0001525441696113074,
+      "loss": 3.7985,
+      "step": 1899
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 0.0001524381625441696,
+      "loss": 3.7787,
+      "step": 1900
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 0.00015233215547703178,
+      "loss": 3.6582,
+      "step": 1901
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 0.00015222614840989398,
+      "loss": 3.7979,
+      "step": 1902
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 0.00015212014134275618,
+      "loss": 3.7163,
+      "step": 1903
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 0.00015201413427561836,
+      "loss": 3.7654,
+      "step": 1904
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 0.00015190812720848053,
+      "loss": 3.6955,
+      "step": 1905
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 0.00015180212014134274,
+      "loss": 3.7787,
+      "step": 1906
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 0.00015169611307420494,
+      "loss": 3.6833,
+      "step": 1907
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 0.00015159010600706714,
+      "loss": 3.6126,
+      "step": 1908
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 0.0001514840989399293,
+      "loss": 3.7051,
+      "step": 1909
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 0.0001513780918727915,
+      "loss": 3.7688,
+      "step": 1910
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 0.0001512720848056537,
+      "loss": 3.735,
+      "step": 1911
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 0.0001511660777385159,
+      "loss": 3.678,
+      "step": 1912
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 0.0001510600706713781,
+      "loss": 3.9441,
+      "step": 1913
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 0.00015095406360424027,
+      "loss": 3.9667,
+      "step": 1914
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 0.00015084805653710245,
+      "loss": 3.9685,
+      "step": 1915
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 0.00015074204946996465,
+      "loss": 3.9074,
+      "step": 1916
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 0.00015063604240282683,
+      "loss": 4.1573,
+      "step": 1917
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 0.00015053003533568903,
+      "loss": 3.9623,
+      "step": 1918
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 0.00015042402826855123,
+      "loss": 3.9057,
+      "step": 1919
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 0.00015031802120141343,
+      "loss": 4.0516,
+      "step": 1920
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 0.00015021201413427558,
+      "loss": 3.846,
+      "step": 1921
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 0.00015010600706713778,
+      "loss": 3.7817,
+      "step": 1922
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 0.00015,
+      "loss": 3.759,
+      "step": 1923
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 0.0001498939929328622,
+      "loss": 3.8316,
+      "step": 1924
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 0.00014978798586572436,
+      "loss": 3.7931,
+      "step": 1925
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 0.00014968197879858657,
+      "loss": 3.7878,
+      "step": 1926
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 0.00014957597173144877,
+      "loss": 4.0014,
+      "step": 1927
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 0.00014946996466431094,
+      "loss": 3.8599,
+      "step": 1928
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 0.00014936395759717315,
+      "loss": 3.7481,
+      "step": 1929
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 0.00014925795053003532,
+      "loss": 3.7613,
+      "step": 1930
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 0.00014915194346289752,
+      "loss": 3.8165,
+      "step": 1931
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 0.0001490459363957597,
+      "loss": 3.862,
+      "step": 1932
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 0.00014893992932862187,
+      "loss": 3.8483,
+      "step": 1933
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 0.00014883392226148408,
+      "loss": 3.7779,
+      "step": 1934
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 0.00014872791519434628,
+      "loss": 3.8563,
+      "step": 1935
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 0.00014862190812720845,
+      "loss": 3.9046,
+      "step": 1936
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 0.00014851590106007066,
+      "loss": 3.6964,
+      "step": 1937
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 0.00014840989399293286,
+      "loss": 3.9483,
+      "step": 1938
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 0.00014830388692579503,
+      "loss": 4.138,
+      "step": 1939
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 0.00014819787985865724,
+      "loss": 3.8247,
+      "step": 1940
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 0.00014809187279151944,
+      "loss": 3.8185,
+      "step": 1941
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 0.00014798586572438161,
+      "loss": 4.0784,
+      "step": 1942
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 0.00014787985865724382,
+      "loss": 4.0586,
+      "step": 1943
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 0.000147773851590106,
+      "loss": 3.8954,
+      "step": 1944
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 0.0001476678445229682,
+      "loss": 3.9753,
+      "step": 1945
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 0.00014756183745583037,
+      "loss": 3.9262,
+      "step": 1946
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 0.00014745583038869257,
+      "loss": 3.7441,
+      "step": 1947
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 0.00014734982332155475,
+      "loss": 3.9567,
+      "step": 1948
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 0.00014724381625441695,
+      "loss": 4.1266,
+      "step": 1949
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 0.00014713780918727912,
+      "loss": 4.1689,
+      "step": 1950
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 0.00014703180212014133,
+      "loss": 4.0496,
+      "step": 1951
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 0.00014692579505300353,
+      "loss": 4.1732,
+      "step": 1952
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 0.0001468197879858657,
+      "loss": 4.2311,
+      "step": 1953
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 0.0001467137809187279,
+      "loss": 4.164,
+      "step": 1954
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 0.0001466077738515901,
+      "loss": 4.116,
+      "step": 1955
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 0.00014650176678445228,
+      "loss": 4.1724,
+      "step": 1956
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 0.00014639575971731449,
+      "loss": 4.1283,
+      "step": 1957
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 0.00014628975265017666,
+      "loss": 3.897,
+      "step": 1958
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 0.00014618374558303886,
+      "loss": 3.9323,
+      "step": 1959
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 0.00014607773851590104,
+      "loss": 4.0214,
+      "step": 1960
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 0.00014597173144876324,
+      "loss": 3.9487,
+      "step": 1961
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 0.00014586572438162542,
+      "loss": 3.5332,
+      "step": 1962
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 0.00014575971731448762,
+      "loss": 4.0542,
+      "step": 1963
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 0.0001456537102473498,
+      "loss": 3.8636,
+      "step": 1964
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 0.000145547703180212,
+      "loss": 3.8076,
+      "step": 1965
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 0.0001454416961130742,
+      "loss": 3.7267,
+      "step": 1966
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 0.00014533568904593637,
+      "loss": 3.8872,
+      "step": 1967
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 0.00014522968197879858,
+      "loss": 3.834,
+      "step": 1968
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 0.00014512367491166078,
+      "loss": 3.6647,
+      "step": 1969
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 0.00014501766784452295,
+      "loss": 3.852,
+      "step": 1970
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 0.00014491166077738516,
+      "loss": 3.853,
+      "step": 1971
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 0.00014480565371024733,
+      "loss": 3.7009,
+      "step": 1972
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 0.00014469964664310953,
+      "loss": 3.7108,
+      "step": 1973
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 0.0001445936395759717,
+      "loss": 3.929,
+      "step": 1974
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 0.0001444876325088339,
+      "loss": 3.848,
+      "step": 1975
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 0.00014438162544169609,
+      "loss": 3.6901,
+      "step": 1976
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 0.0001442756183745583,
+      "loss": 3.8599,
+      "step": 1977
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 0.00014416961130742046,
+      "loss": 3.8723,
+      "step": 1978
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 0.00014406360424028267,
+      "loss": 3.7905,
+      "step": 1979
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 0.00014395759717314487,
+      "loss": 3.7136,
+      "step": 1980
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 0.00014385159010600704,
+      "loss": 3.9335,
+      "step": 1981
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 0.00014374558303886925,
+      "loss": 3.7591,
+      "step": 1982
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 0.00014363957597173145,
+      "loss": 3.7664,
+      "step": 1983
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 0.00014353356890459362,
+      "loss": 3.7171,
+      "step": 1984
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 0.00014342756183745583,
+      "loss": 3.8075,
+      "step": 1985
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 0.00014332155477031803,
+      "loss": 3.7729,
+      "step": 1986
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 0.0001432155477031802,
+      "loss": 3.794,
+      "step": 1987
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 0.00014310954063604238,
+      "loss": 4.1934,
+      "step": 1988
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 0.00014300353356890458,
+      "loss": 4.0699,
+      "step": 1989
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 0.00014289752650176676,
+      "loss": 3.8371,
+      "step": 1990
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 0.00014279151943462896,
+      "loss": 3.708,
+      "step": 1991
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 0.00014268551236749116,
+      "loss": 3.9465,
+      "step": 1992
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 0.00014257950530035334,
+      "loss": 3.941,
+      "step": 1993
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 0.00014247349823321554,
+      "loss": 3.7272,
+      "step": 1994
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 0.0001423674911660777,
+      "loss": 3.9462,
+      "step": 1995
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 0.00014226148409893992,
+      "loss": 3.9622,
+      "step": 1996
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 0.00014215547703180212,
+      "loss": 3.8176,
+      "step": 1997
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 0.0001420494699646643,
+      "loss": 3.8018,
+      "step": 1998
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 0.0001419434628975265,
+      "loss": 5.765,
+      "step": 1999
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 0.0001418374558303887,
+      "loss": 3.7747,
+      "step": 2000
+    },
+    {
+      "epoch": 18.02,
+      "eval_loss": 16.53677749633789,
+      "eval_runtime": 121.2184,
+      "eval_samples_per_second": 21.795,
+      "eval_steps_per_second": 0.685,
+      "eval_wer": 1.0061880206267355,
+      "step": 2000
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 0.00014173144876325087,
+      "loss": 3.7247,
+      "step": 2001
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 0.00014162544169611308,
+      "loss": 3.7924,
+      "step": 2002
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 0.00014151943462897525,
+      "loss": 3.8912,
+      "step": 2003
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 0.00014141342756183743,
+      "loss": 3.8391,
+      "step": 2004
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 0.00014130742049469963,
+      "loss": 3.7322,
+      "step": 2005
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 0.00014120141342756183,
+      "loss": 3.8854,
+      "step": 2006
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 0.000141095406360424,
+      "loss": 3.9298,
+      "step": 2007
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 0.0001409893992932862,
+      "loss": 3.6001,
+      "step": 2008
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 0.00014088339222614838,
+      "loss": 3.7053,
+      "step": 2009
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 0.00014077738515901059,
+      "loss": 3.9675,
+      "step": 2010
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 0.0001406713780918728,
+      "loss": 3.7489,
+      "step": 2011
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 0.00014056537102473496,
+      "loss": 3.7522,
+      "step": 2012
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 0.00014045936395759717,
+      "loss": 3.7345,
+      "step": 2013
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 0.00014035335689045937,
+      "loss": 3.7331,
+      "step": 2014
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 0.00014024734982332154,
+      "loss": 3.7652,
+      "step": 2015
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 0.00014014134275618375,
+      "loss": 3.7761,
+      "step": 2016
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 0.00014003533568904592,
+      "loss": 3.7261,
+      "step": 2017
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 0.00013992932862190812,
+      "loss": 3.7262,
+      "step": 2018
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 0.0001398233215547703,
+      "loss": 3.6453,
+      "step": 2019
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 0.0001397173144876325,
+      "loss": 3.6403,
+      "step": 2020
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 0.00013961130742049468,
+      "loss": 3.6382,
+      "step": 2021
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 0.00013950530035335688,
+      "loss": 3.8417,
+      "step": 2022
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 0.00013939929328621908,
+      "loss": 3.5619,
+      "step": 2023
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 0.00013929328621908126,
+      "loss": 3.748,
+      "step": 2024
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 0.00013918727915194346,
+      "loss": 3.7215,
+      "step": 2025
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 0.00013908127208480563,
+      "loss": 3.7693,
+      "step": 2026
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 0.00013897526501766784,
+      "loss": 3.6456,
+      "step": 2027
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 0.00013886925795053004,
+      "loss": 3.833,
+      "step": 2028
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 0.0001387632508833922,
+      "loss": 3.9113,
+      "step": 2029
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 0.00013865724381625441,
+      "loss": 3.7128,
+      "step": 2030
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 0.0001385512367491166,
+      "loss": 3.7871,
+      "step": 2031
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 0.0001384452296819788,
+      "loss": 3.9871,
+      "step": 2032
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 0.00013833922261484097,
+      "loss": 4.0819,
+      "step": 2033
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 0.00013823321554770317,
+      "loss": 3.8545,
+      "step": 2034
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 0.00013812720848056535,
+      "loss": 3.9603,
+      "step": 2035
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 0.00013802120141342755,
+      "loss": 3.7316,
+      "step": 2036
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 0.00013791519434628975,
+      "loss": 3.7517,
+      "step": 2037
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 0.00013780918727915192,
+      "loss": 4.0877,
+      "step": 2038
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 0.00013770318021201413,
+      "loss": 4.0222,
+      "step": 2039
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 0.0001375971731448763,
+      "loss": 3.7692,
+      "step": 2040
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 0.0001374911660777385,
+      "loss": 3.5642,
+      "step": 2041
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 0.0001373851590106007,
+      "loss": 3.6833,
+      "step": 2042
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 0.00013727915194346288,
+      "loss": 3.8452,
+      "step": 2043
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 0.00013717314487632508,
+      "loss": 3.6445,
+      "step": 2044
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 0.00013706713780918726,
+      "loss": 3.6416,
+      "step": 2045
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 0.00013696113074204946,
+      "loss": 3.8673,
+      "step": 2046
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 0.00013685512367491164,
+      "loss": 3.8712,
+      "step": 2047
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 0.00013674911660777384,
+      "loss": 3.5018,
+      "step": 2048
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 0.00013664310954063601,
+      "loss": 3.7289,
+      "step": 2049
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 0.00013653710247349822,
+      "loss": 3.8741,
+      "step": 2050
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 0.00013643109540636042,
+      "loss": 3.8859,
+      "step": 2051
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 0.0001363250883392226,
+      "loss": 3.9279,
+      "step": 2052
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 0.0001362190812720848,
+      "loss": 3.9727,
+      "step": 2053
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 0.000136113074204947,
+      "loss": 3.7868,
+      "step": 2054
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 0.00013600706713780917,
+      "loss": 3.5979,
+      "step": 2055
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 0.00013590106007067138,
+      "loss": 3.8353,
+      "step": 2056
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 0.00013579505300353355,
+      "loss": 3.7851,
+      "step": 2057
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 0.00013568904593639575,
+      "loss": 3.6329,
+      "step": 2058
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 0.00013558303886925793,
+      "loss": 3.6302,
+      "step": 2059
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 0.00013547703180212013,
+      "loss": 3.675,
+      "step": 2060
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 0.0001353710247349823,
+      "loss": 3.6072,
+      "step": 2061
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 0.0001352650176678445,
+      "loss": 3.5323,
+      "step": 2062
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 0.00013515901060070668,
+      "loss": 3.685,
+      "step": 2063
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 0.0001350530035335689,
+      "loss": 3.6791,
+      "step": 2064
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 0.0001349469964664311,
+      "loss": 3.5639,
+      "step": 2065
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 0.00013484098939929326,
+      "loss": 3.6225,
+      "step": 2066
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 0.00013473498233215547,
+      "loss": 3.6739,
+      "step": 2067
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 0.00013462897526501767,
+      "loss": 3.4845,
+      "step": 2068
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 0.00013452296819787984,
+      "loss": 3.5545,
+      "step": 2069
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 0.00013441696113074205,
+      "loss": 3.5152,
+      "step": 2070
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 0.00013431095406360422,
+      "loss": 3.6296,
+      "step": 2071
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 0.00013420494699646642,
+      "loss": 3.6692,
+      "step": 2072
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 0.00013409893992932863,
+      "loss": 3.5679,
+      "step": 2073
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 0.0001339929328621908,
+      "loss": 3.7405,
+      "step": 2074
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 0.000133886925795053,
+      "loss": 3.6891,
+      "step": 2075
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 0.00013378091872791518,
+      "loss": 3.5639,
+      "step": 2076
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 0.00013367491166077735,
+      "loss": 3.5194,
+      "step": 2077
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 0.00013356890459363956,
+      "loss": 3.6333,
+      "step": 2078
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 0.00013346289752650176,
+      "loss": 3.6379,
+      "step": 2079
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 0.00013335689045936393,
+      "loss": 3.6557,
+      "step": 2080
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 0.00013325088339222614,
+      "loss": 3.6125,
+      "step": 2081
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 0.00013314487632508834,
+      "loss": 3.6943,
+      "step": 2082
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 0.00013303886925795051,
+      "loss": 3.5908,
+      "step": 2083
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 0.00013293286219081272,
+      "loss": 3.6679,
+      "step": 2084
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 0.00013282685512367492,
+      "loss": 3.8352,
+      "step": 2085
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 0.0001327208480565371,
+      "loss": 3.6775,
+      "step": 2086
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 0.0001326148409893993,
+      "loss": 3.5964,
+      "step": 2087
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 0.00013250883392226147,
+      "loss": 3.7621,
+      "step": 2088
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 0.00013240282685512367,
+      "loss": 3.7037,
+      "step": 2089
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 0.00013229681978798585,
+      "loss": 3.6341,
+      "step": 2090
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 0.00013219081272084805,
+      "loss": 3.5761,
+      "step": 2091
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 0.00013208480565371023,
+      "loss": 3.6039,
+      "step": 2092
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 0.00013197879858657243,
+      "loss": 3.6912,
+      "step": 2093
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 0.0001318727915194346,
+      "loss": 3.5617,
+      "step": 2094
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 0.0001317667844522968,
+      "loss": 3.5584,
+      "step": 2095
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 0.000131660777385159,
+      "loss": 3.7374,
+      "step": 2096
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 0.00013155477031802118,
+      "loss": 3.7749,
+      "step": 2097
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 0.00013144876325088339,
+      "loss": 3.5983,
+      "step": 2098
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 0.0001313427561837456,
+      "loss": 3.836,
+      "step": 2099
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 0.00013123674911660776,
+      "loss": 3.8993,
+      "step": 2100
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 0.00013113074204946997,
+      "loss": 3.7685,
+      "step": 2101
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 0.00013102473498233214,
+      "loss": 3.512,
+      "step": 2102
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 0.00013091872791519434,
+      "loss": 3.6436,
+      "step": 2103
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 0.00013081272084805652,
+      "loss": 3.7529,
+      "step": 2104
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 0.00013070671378091872,
+      "loss": 3.5505,
+      "step": 2105
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 0.0001306007067137809,
+      "loss": 3.6481,
+      "step": 2106
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 0.0001304946996466431,
+      "loss": 3.6733,
+      "step": 2107
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 0.00013038869257950527,
+      "loss": 3.5211,
+      "step": 2108
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 0.00013028268551236748,
+      "loss": 3.6182,
+      "step": 2109
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 0.00013017667844522968,
+      "loss": 5.664,
+      "step": 2110
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 0.00013007067137809185,
+      "loss": 3.6426,
+      "step": 2111
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 0.00012996466431095406,
+      "loss": 3.59,
+      "step": 2112
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 0.00012985865724381626,
+      "loss": 3.4775,
+      "step": 2113
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 0.00012975265017667843,
+      "loss": 3.6828,
+      "step": 2114
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 0.00012964664310954064,
+      "loss": 3.675,
+      "step": 2115
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 0.0001295406360424028,
+      "loss": 3.445,
+      "step": 2116
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 0.000129434628975265,
+      "loss": 3.965,
+      "step": 2117
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 0.0001293286219081272,
+      "loss": 3.6896,
+      "step": 2118
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 0.0001292226148409894,
+      "loss": 3.734,
+      "step": 2119
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 0.00012911660777385157,
+      "loss": 3.8772,
+      "step": 2120
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 0.00012901060070671377,
+      "loss": 3.9423,
+      "step": 2121
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 0.00012890459363957594,
+      "loss": 3.6762,
+      "step": 2122
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 0.00012879858657243815,
+      "loss": 3.6096,
+      "step": 2123
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 0.00012869257950530035,
+      "loss": 3.9851,
+      "step": 2124
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 0.00012858657243816252,
+      "loss": 3.8202,
+      "step": 2125
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 0.00012848056537102473,
+      "loss": 3.5826,
+      "step": 2126
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 0.00012837455830388693,
+      "loss": 3.6971,
+      "step": 2127
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 0.0001282685512367491,
+      "loss": 4.0922,
+      "step": 2128
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 0.0001281625441696113,
+      "loss": 4.1582,
+      "step": 2129
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 0.00012805653710247348,
+      "loss": 3.6985,
+      "step": 2130
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 0.00012795053003533568,
+      "loss": 3.745,
+      "step": 2131
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 0.00012784452296819786,
+      "loss": 3.9401,
+      "step": 2132
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 0.00012773851590106006,
+      "loss": 3.8654,
+      "step": 2133
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 0.00012763250883392224,
+      "loss": 3.5528,
+      "step": 2134
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 0.00012752650176678444,
+      "loss": 3.7509,
+      "step": 2135
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 0.0001274204946996466,
+      "loss": 3.7992,
+      "step": 2136
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 0.00012731448763250882,
+      "loss": 3.6802,
+      "step": 2137
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 0.00012720848056537102,
+      "loss": 3.4174,
+      "step": 2138
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 0.0001271024734982332,
+      "loss": 3.7384,
+      "step": 2139
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 0.0001269964664310954,
+      "loss": 3.6338,
+      "step": 2140
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 0.0001268904593639576,
+      "loss": 3.3903,
+      "step": 2141
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 0.00012678445229681977,
+      "loss": 3.695,
+      "step": 2142
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 0.00012667844522968198,
+      "loss": 3.6497,
+      "step": 2143
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 0.00012657243816254418,
+      "loss": 3.5723,
+      "step": 2144
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 0.00012646643109540635,
+      "loss": 3.5882,
+      "step": 2145
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 0.00012636042402826856,
+      "loss": 3.7139,
+      "step": 2146
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 0.00012625441696113073,
+      "loss": 3.6394,
+      "step": 2147
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 0.0001261484098939929,
+      "loss": 3.7566,
+      "step": 2148
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 0.0001260424028268551,
+      "loss": 3.6369,
+      "step": 2149
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 0.00012593639575971728,
+      "loss": 3.6847,
+      "step": 2150
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 0.00012583038869257949,
+      "loss": 3.6181,
+      "step": 2151
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 0.0001257243816254417,
+      "loss": 3.5587,
+      "step": 2152
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 0.00012561837455830386,
+      "loss": 3.6609,
+      "step": 2153
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 0.00012551236749116607,
+      "loss": 3.6574,
+      "step": 2154
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 0.00012540636042402827,
+      "loss": 3.4865,
+      "step": 2155
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 0.00012530035335689044,
+      "loss": 3.5804,
+      "step": 2156
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 0.00012519434628975265,
+      "loss": 3.8173,
+      "step": 2157
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 0.00012508833922261485,
+      "loss": 3.6135,
+      "step": 2158
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 0.00012498233215547702,
+      "loss": 3.4987,
+      "step": 2159
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 0.00012487632508833923,
+      "loss": 3.8645,
+      "step": 2160
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 0.0001247703180212014,
+      "loss": 3.7177,
+      "step": 2161
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 0.0001246643109540636,
+      "loss": 3.5133,
+      "step": 2162
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 0.00012455830388692578,
+      "loss": 3.4219,
+      "step": 2163
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 0.00012445229681978798,
+      "loss": 3.6381,
+      "step": 2164
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 0.00012434628975265016,
+      "loss": 3.5244,
+      "step": 2165
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 0.00012424028268551236,
+      "loss": 3.3501,
+      "step": 2166
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 0.00012413427561837453,
+      "loss": 3.6943,
+      "step": 2167
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 0.00012402826855123674,
+      "loss": 3.7505,
+      "step": 2168
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 0.00012392226148409894,
+      "loss": 3.6407,
+      "step": 2169
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 0.0001238162544169611,
+      "loss": 3.6566,
+      "step": 2170
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 0.00012371024734982332,
+      "loss": 3.7903,
+      "step": 2171
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 0.00012360424028268552,
+      "loss": 3.5534,
+      "step": 2172
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 0.0001234982332155477,
+      "loss": 3.3831,
+      "step": 2173
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 0.0001233922261484099,
+      "loss": 3.5046,
+      "step": 2174
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 0.00012328621908127207,
+      "loss": 3.6842,
+      "step": 2175
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 0.00012318021201413427,
+      "loss": 3.631,
+      "step": 2176
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 0.00012307420494699645,
+      "loss": 3.5851,
+      "step": 2177
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 0.00012296819787985865,
+      "loss": 3.6313,
+      "step": 2178
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 0.00012286219081272083,
+      "loss": 3.5975,
+      "step": 2179
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 0.00012275618374558303,
+      "loss": 3.3813,
+      "step": 2180
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 0.0001226501766784452,
+      "loss": 3.5749,
+      "step": 2181
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 0.0001225441696113074,
+      "loss": 3.7847,
+      "step": 2182
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 0.0001224381625441696,
+      "loss": 3.6124,
+      "step": 2183
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 0.00012233215547703178,
+      "loss": 3.4983,
+      "step": 2184
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 0.00012222614840989398,
+      "loss": 3.812,
+      "step": 2185
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 0.0001221201413427562,
+      "loss": 3.8044,
+      "step": 2186
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 0.00012201413427561836,
+      "loss": 3.5041,
+      "step": 2187
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 0.00012190812720848055,
+      "loss": 3.4181,
+      "step": 2188
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 0.00012180212014134275,
+      "loss": 3.7739,
+      "step": 2189
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 0.00012169611307420493,
+      "loss": 3.8667,
+      "step": 2190
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 0.00012159010600706713,
+      "loss": 3.7432,
+      "step": 2191
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 0.00012148409893992932,
+      "loss": 3.6941,
+      "step": 2192
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 0.00012137809187279151,
+      "loss": 3.7047,
+      "step": 2193
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 0.0001212720848056537,
+      "loss": 3.8801,
+      "step": 2194
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 0.0001211660777385159,
+      "loss": 3.8466,
+      "step": 2195
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 0.00012106007067137807,
+      "loss": 3.7767,
+      "step": 2196
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 0.00012095406360424028,
+      "loss": 3.7232,
+      "step": 2197
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 0.00012084805653710245,
+      "loss": 3.5524,
+      "step": 2198
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 0.00012074204946996465,
+      "loss": 3.697,
+      "step": 2199
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 0.00012063604240282684,
+      "loss": 3.5497,
+      "step": 2200
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 0.00012053003533568903,
+      "loss": 3.7435,
+      "step": 2201
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 0.00012042402826855122,
+      "loss": 3.6187,
+      "step": 2202
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 0.00012031802120141342,
+      "loss": 3.7487,
+      "step": 2203
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 0.0001202120141342756,
+      "loss": 3.5109,
+      "step": 2204
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 0.0001201060070671378,
+      "loss": 3.5953,
+      "step": 2205
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 3.618,
+      "step": 2206
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 0.00011989399293286218,
+      "loss": 3.6204,
+      "step": 2207
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 0.00011978798586572437,
+      "loss": 3.6971,
+      "step": 2208
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 0.00011968197879858657,
+      "loss": 3.4236,
+      "step": 2209
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 0.00011957597173144874,
+      "loss": 3.8139,
+      "step": 2210
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 0.00011946996466431095,
+      "loss": 3.6818,
+      "step": 2211
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 0.00011936395759717312,
+      "loss": 3.5731,
+      "step": 2212
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 0.00011925795053003532,
+      "loss": 3.5196,
+      "step": 2213
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 0.00011915194346289753,
+      "loss": 3.7253,
+      "step": 2214
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 0.0001190459363957597,
+      "loss": 3.5921,
+      "step": 2215
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 0.00011893992932862189,
+      "loss": 3.5107,
+      "step": 2216
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 0.00011883392226148409,
+      "loss": 3.7826,
+      "step": 2217
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 0.00011872791519434627,
+      "loss": 3.5882,
+      "step": 2218
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 0.00011862190812720847,
+      "loss": 3.6352,
+      "step": 2219
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 0.00011851590106007067,
+      "loss": 3.6242,
+      "step": 2220
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 0.00011840989399293285,
+      "loss": 5.4751,
+      "step": 2221
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 0.00011830388692579505,
+      "loss": 3.5508,
+      "step": 2222
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 0.00011819787985865724,
+      "loss": 3.4648,
+      "step": 2223
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 0.00011809187279151941,
+      "loss": 3.4863,
+      "step": 2224
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 0.00011798586572438162,
+      "loss": 3.6706,
+      "step": 2225
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 0.00011787985865724382,
+      "loss": 3.5964,
+      "step": 2226
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 0.000117773851590106,
+      "loss": 3.4024,
+      "step": 2227
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 0.0001176678445229682,
+      "loss": 3.4889,
+      "step": 2228
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 0.00011756183745583037,
+      "loss": 3.5056,
+      "step": 2229
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 0.00011745583038869257,
+      "loss": 3.5258,
+      "step": 2230
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 0.00011734982332155476,
+      "loss": 3.5862,
+      "step": 2231
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 0.00011724381625441694,
+      "loss": 3.8836,
+      "step": 2232
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 0.00011713780918727914,
+      "loss": 3.6284,
+      "step": 2233
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 0.00011703180212014134,
+      "loss": 3.4849,
+      "step": 2234
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 0.00011692579505300352,
+      "loss": 3.5079,
+      "step": 2235
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 0.00011681978798586572,
+      "loss": 3.6083,
+      "step": 2236
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 0.00011671378091872791,
+      "loss": 3.6973,
+      "step": 2237
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 0.0001166077738515901,
+      "loss": 3.6862,
+      "step": 2238
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 0.00011650176678445229,
+      "loss": 3.6256,
+      "step": 2239
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 0.00011639575971731449,
+      "loss": 3.9385,
+      "step": 2240
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 0.00011628975265017666,
+      "loss": 4.2145,
+      "step": 2241
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 0.00011618374558303887,
+      "loss": 4.4114,
+      "step": 2242
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 0.00011607773851590104,
+      "loss": 4.5874,
+      "step": 2243
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 0.00011597173144876324,
+      "loss": 4.5355,
+      "step": 2244
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 0.00011586572438162543,
+      "loss": 4.4304,
+      "step": 2245
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 0.00011575971731448762,
+      "loss": 4.5293,
+      "step": 2246
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 0.00011565371024734981,
+      "loss": 4.2665,
+      "step": 2247
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 0.00011554770318021201,
+      "loss": 3.9049,
+      "step": 2248
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 0.00011544169611307419,
+      "loss": 3.7486,
+      "step": 2249
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 0.00011533568904593639,
+      "loss": 3.7358,
+      "step": 2250
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 0.00011522968197879858,
+      "loss": 3.794,
+      "step": 2251
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 0.00011512367491166077,
+      "loss": 3.7376,
+      "step": 2252
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 0.00011501766784452296,
+      "loss": 4.2575,
+      "step": 2253
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 0.00011491166077738516,
+      "loss": 4.2207,
+      "step": 2254
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 0.00011480565371024733,
+      "loss": 3.996,
+      "step": 2255
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 0.00011469964664310954,
+      "loss": 3.8353,
+      "step": 2256
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 0.00011459363957597172,
+      "loss": 3.785,
+      "step": 2257
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 0.00011448763250883391,
+      "loss": 3.513,
+      "step": 2258
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 0.0001143816254416961,
+      "loss": 3.5092,
+      "step": 2259
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 0.00011427561837455829,
+      "loss": 3.7837,
+      "step": 2260
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 0.00011416961130742048,
+      "loss": 3.8923,
+      "step": 2261
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 0.00011406360424028268,
+      "loss": 3.945,
+      "step": 2262
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 0.00011395759717314486,
+      "loss": 3.9879,
+      "step": 2263
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 0.00011385159010600706,
+      "loss": 3.9537,
+      "step": 2264
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 0.00011374558303886925,
+      "loss": 3.8886,
+      "step": 2265
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 0.00011363957597173144,
+      "loss": 3.6998,
+      "step": 2266
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 0.00011353356890459363,
+      "loss": 3.7704,
+      "step": 2267
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 0.00011342756183745583,
+      "loss": 3.7017,
+      "step": 2268
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 0.000113321554770318,
+      "loss": 3.612,
+      "step": 2269
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 0.0001132155477031802,
+      "loss": 3.4241,
+      "step": 2270
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 0.0001131095406360424,
+      "loss": 3.7184,
+      "step": 2271
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 0.00011300353356890458,
+      "loss": 3.6394,
+      "step": 2272
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 0.00011289752650176677,
+      "loss": 3.6652,
+      "step": 2273
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 0.00011279151943462896,
+      "loss": 3.4712,
+      "step": 2274
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 0.00011268551236749115,
+      "loss": 3.6534,
+      "step": 2275
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 0.00011257950530035335,
+      "loss": 3.5765,
+      "step": 2276
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 0.00011247349823321553,
+      "loss": 3.4525,
+      "step": 2277
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 0.00011236749116607773,
+      "loss": 3.6226,
+      "step": 2278
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 0.00011226148409893992,
+      "loss": 3.6927,
+      "step": 2279
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 0.00011215547703180211,
+      "loss": 3.4812,
+      "step": 2280
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 0.0001120494699646643,
+      "loss": 3.4172,
+      "step": 2281
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 0.0001119434628975265,
+      "loss": 3.5212,
+      "step": 2282
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 0.00011183745583038867,
+      "loss": 3.5621,
+      "step": 2283
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 0.00011173144876325088,
+      "loss": 3.4777,
+      "step": 2284
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 0.00011162544169611308,
+      "loss": 3.5146,
+      "step": 2285
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 0.00011151943462897525,
+      "loss": 3.3702,
+      "step": 2286
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 0.00011141342756183744,
+      "loss": 3.4904,
+      "step": 2287
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 0.00011130742049469964,
+      "loss": 3.4483,
+      "step": 2288
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 0.00011120141342756182,
+      "loss": 3.5496,
+      "step": 2289
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 0.00011109540636042402,
+      "loss": 3.5011,
+      "step": 2290
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 0.0001109893992932862,
+      "loss": 3.4317,
+      "step": 2291
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 0.0001108833922261484,
+      "loss": 3.4748,
+      "step": 2292
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 0.0001107773851590106,
+      "loss": 3.451,
+      "step": 2293
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 0.00011067137809187278,
+      "loss": 3.4897,
+      "step": 2294
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 0.00011056537102473497,
+      "loss": 3.5218,
+      "step": 2295
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 0.00011045936395759717,
+      "loss": 3.6431,
+      "step": 2296
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 0.00011035335689045934,
+      "loss": 3.4171,
+      "step": 2297
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 0.00011024734982332155,
+      "loss": 3.4319,
+      "step": 2298
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 0.00011014134275618375,
+      "loss": 3.4677,
+      "step": 2299
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 0.00011003533568904592,
+      "loss": 3.6951,
+      "step": 2300
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 0.00010992932862190813,
+      "loss": 3.5091,
+      "step": 2301
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 0.00010982332155477031,
+      "loss": 3.5109,
+      "step": 2302
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 0.0001097173144876325,
+      "loss": 3.7188,
+      "step": 2303
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 0.00010961130742049469,
+      "loss": 3.6691,
+      "step": 2304
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 0.00010950530035335687,
+      "loss": 3.4075,
+      "step": 2305
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 0.00010939929328621907,
+      "loss": 3.5636,
+      "step": 2306
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 0.00010929328621908127,
+      "loss": 3.7745,
+      "step": 2307
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 0.00010918727915194345,
+      "loss": 3.5367,
+      "step": 2308
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 0.00010908127208480565,
+      "loss": 3.4175,
+      "step": 2309
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 0.00010897526501766784,
+      "loss": 3.4514,
+      "step": 2310
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 0.00010886925795053003,
+      "loss": 3.5519,
+      "step": 2311
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 0.00010876325088339222,
+      "loss": 3.5187,
+      "step": 2312
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 0.00010865724381625442,
+      "loss": 3.3783,
+      "step": 2313
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 0.00010855123674911659,
+      "loss": 3.4596,
+      "step": 2314
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 0.0001084452296819788,
+      "loss": 3.5169,
+      "step": 2315
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 0.00010833922261484098,
+      "loss": 3.3969,
+      "step": 2316
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 0.00010823321554770317,
+      "loss": 3.457,
+      "step": 2317
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 0.00010812720848056536,
+      "loss": 3.507,
+      "step": 2318
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 0.00010802120141342756,
+      "loss": 3.6131,
+      "step": 2319
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 0.00010791519434628974,
+      "loss": 3.4487,
+      "step": 2320
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 0.00010780918727915194,
+      "loss": 3.6299,
+      "step": 2321
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 0.00010770318021201412,
+      "loss": 3.6049,
+      "step": 2322
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 0.00010759717314487632,
+      "loss": 3.566,
+      "step": 2323
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 0.00010749116607773851,
+      "loss": 3.3669,
+      "step": 2324
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 0.0001073851590106007,
+      "loss": 3.5169,
+      "step": 2325
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 0.00010727915194346288,
+      "loss": 3.5634,
+      "step": 2326
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 0.00010717314487632509,
+      "loss": 3.4137,
+      "step": 2327
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 0.00010706713780918726,
+      "loss": 3.601,
+      "step": 2328
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 0.00010696113074204946,
+      "loss": 3.7138,
+      "step": 2329
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 0.00010685512367491165,
+      "loss": 3.7492,
+      "step": 2330
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 0.00010674911660777384,
+      "loss": 3.6442,
+      "step": 2331
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 0.00010664310954063603,
+      "loss": 5.4639,
+      "step": 2332
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 0.00010653710247349823,
+      "loss": 3.8246,
+      "step": 2333
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 0.00010643109540636041,
+      "loss": 3.7054,
+      "step": 2334
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 0.00010632508833922261,
+      "loss": 3.4854,
+      "step": 2335
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 0.00010621908127208479,
+      "loss": 3.5843,
+      "step": 2336
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 0.00010611307420494699,
+      "loss": 3.6206,
+      "step": 2337
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 0.00010600706713780918,
+      "loss": 3.5001,
+      "step": 2338
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 0.00010590106007067137,
+      "loss": 3.6885,
+      "step": 2339
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 0.00010579505300353355,
+      "loss": 3.6619,
+      "step": 2340
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 0.00010568904593639576,
+      "loss": 3.4638,
+      "step": 2341
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 0.00010558303886925793,
+      "loss": 3.4569,
+      "step": 2342
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 0.00010547703180212013,
+      "loss": 3.5367,
+      "step": 2343
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 0.00010537102473498232,
+      "loss": 3.4955,
+      "step": 2344
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 0.00010526501766784451,
+      "loss": 3.4079,
+      "step": 2345
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 0.0001051590106007067,
+      "loss": 3.5645,
+      "step": 2346
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 0.0001050530035335689,
+      "loss": 3.4405,
+      "step": 2347
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 0.00010494699646643108,
+      "loss": 3.5055,
+      "step": 2348
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 0.00010484098939929328,
+      "loss": 3.4259,
+      "step": 2349
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 0.00010473498233215547,
+      "loss": 3.5817,
+      "step": 2350
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 0.00010462897526501766,
+      "loss": 3.5182,
+      "step": 2351
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 0.00010452296819787985,
+      "loss": 3.5307,
+      "step": 2352
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 0.00010441696113074204,
+      "loss": 3.5494,
+      "step": 2353
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 0.00010431095406360422,
+      "loss": 3.6419,
+      "step": 2354
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 0.00010420494699646643,
+      "loss": 3.5574,
+      "step": 2355
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 0.0001040989399293286,
+      "loss": 3.3826,
+      "step": 2356
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 0.0001039929328621908,
+      "loss": 3.5383,
+      "step": 2357
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 0.00010388692579505299,
+      "loss": 3.5216,
+      "step": 2358
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 0.00010378091872791518,
+      "loss": 3.572,
+      "step": 2359
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 0.00010367491166077737,
+      "loss": 3.4075,
+      "step": 2360
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 0.00010356890459363957,
+      "loss": 3.4723,
+      "step": 2361
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 0.00010346289752650175,
+      "loss": 3.5112,
+      "step": 2362
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 0.00010335689045936395,
+      "loss": 3.322,
+      "step": 2363
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 0.00010325088339222615,
+      "loss": 3.4522,
+      "step": 2364
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 0.00010314487632508833,
+      "loss": 3.4068,
+      "step": 2365
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 0.00010303886925795053,
+      "loss": 3.2576,
+      "step": 2366
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 0.0001029328621908127,
+      "loss": 3.3591,
+      "step": 2367
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 0.0001028268551236749,
+      "loss": 3.5324,
+      "step": 2368
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 0.0001027208480565371,
+      "loss": 3.4486,
+      "step": 2369
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 0.00010261484098939927,
+      "loss": 3.3956,
+      "step": 2370
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 0.00010250883392226147,
+      "loss": 3.5315,
+      "step": 2371
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 0.00010240282685512368,
+      "loss": 3.6265,
+      "step": 2372
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 0.00010229681978798585,
+      "loss": 3.6234,
+      "step": 2373
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 0.00010219081272084805,
+      "loss": 3.5643,
+      "step": 2374
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 0.00010208480565371024,
+      "loss": 3.5658,
+      "step": 2375
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 0.00010197879858657242,
+      "loss": 3.498,
+      "step": 2376
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 0.00010187279151943462,
+      "loss": 3.6673,
+      "step": 2377
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 0.00010176678445229682,
+      "loss": 3.4835,
+      "step": 2378
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 0.000101660777385159,
+      "loss": 3.6293,
+      "step": 2379
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 0.0001015547703180212,
+      "loss": 3.5273,
+      "step": 2380
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 0.00010144876325088339,
+      "loss": 3.2644,
+      "step": 2381
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 0.00010134275618374558,
+      "loss": 3.4973,
+      "step": 2382
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 0.00010123674911660777,
+      "loss": 3.617,
+      "step": 2383
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 0.00010113074204946994,
+      "loss": 3.3324,
+      "step": 2384
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 0.00010102473498233214,
+      "loss": 3.2646,
+      "step": 2385
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 0.00010091872791519435,
+      "loss": 3.4719,
+      "step": 2386
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 0.00010081272084805652,
+      "loss": 3.4427,
+      "step": 2387
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 0.00010070671378091872,
+      "loss": 3.3607,
+      "step": 2388
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 0.00010060070671378091,
+      "loss": 3.5445,
+      "step": 2389
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 0.0001004946996466431,
+      "loss": 3.4501,
+      "step": 2390
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 0.00010038869257950529,
+      "loss": 3.2923,
+      "step": 2391
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 0.00010028268551236749,
+      "loss": 3.3831,
+      "step": 2392
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 0.00010017667844522967,
+      "loss": 3.4337,
+      "step": 2393
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 0.00010017667844522967,
+      "loss": 3.3636,
+      "step": 2394
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 0.00010007067137809187,
+      "loss": 3.2106,
+      "step": 2395
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 9.996466431095406e-05,
+      "loss": 3.4435,
+      "step": 2396
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 9.985865724381625e-05,
+      "loss": 3.3602,
+      "step": 2397
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 9.975265017667844e-05,
+      "loss": 3.3354,
+      "step": 2398
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 9.964664310954062e-05,
+      "loss": 3.3494,
+      "step": 2399
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 9.954063604240281e-05,
+      "loss": 3.5057,
+      "step": 2400
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 9.943462897526502e-05,
+      "loss": 3.5507,
+      "step": 2401
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 9.932862190812719e-05,
+      "loss": 3.5971,
+      "step": 2402
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 9.92226148409894e-05,
+      "loss": 3.4922,
+      "step": 2403
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 9.911660777385158e-05,
+      "loss": 3.3786,
+      "step": 2404
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 9.901060070671377e-05,
+      "loss": 3.4674,
+      "step": 2405
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 9.890459363957596e-05,
+      "loss": 3.5318,
+      "step": 2406
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 9.879858657243816e-05,
+      "loss": 3.7037,
+      "step": 2407
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 9.869257950530034e-05,
+      "loss": 3.5074,
+      "step": 2408
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 9.858657243816254e-05,
+      "loss": 3.277,
+      "step": 2409
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 9.848056537102473e-05,
+      "loss": 3.2541,
+      "step": 2410
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 9.837455830388692e-05,
+      "loss": 3.4699,
+      "step": 2411
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 9.82685512367491e-05,
+      "loss": 3.3953,
+      "step": 2412
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 9.816254416961131e-05,
+      "loss": 3.1565,
+      "step": 2413
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 9.805653710247348e-05,
+      "loss": 3.3731,
+      "step": 2414
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 9.795053003533569e-05,
+      "loss": 3.3907,
+      "step": 2415
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 9.784452296819786e-05,
+      "loss": 3.2527,
+      "step": 2416
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 9.773851590106006e-05,
+      "loss": 3.3428,
+      "step": 2417
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 9.763250883392225e-05,
+      "loss": 3.4096,
+      "step": 2418
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 9.752650176678444e-05,
+      "loss": 3.2278,
+      "step": 2419
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 9.742049469964663e-05,
+      "loss": 3.351,
+      "step": 2420
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 9.731448763250883e-05,
+      "loss": 3.3205,
+      "step": 2421
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 9.720848056537101e-05,
+      "loss": 3.4092,
+      "step": 2422
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 9.710247349823321e-05,
+      "loss": 3.407,
+      "step": 2423
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 9.69964664310954e-05,
+      "loss": 3.2834,
+      "step": 2424
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 9.689045936395759e-05,
+      "loss": 3.393,
+      "step": 2425
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 9.678445229681978e-05,
+      "loss": 3.3954,
+      "step": 2426
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 9.667844522968198e-05,
+      "loss": 3.322,
+      "step": 2427
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 9.657243816254415e-05,
+      "loss": 3.3319,
+      "step": 2428
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 9.646643109540636e-05,
+      "loss": 3.4689,
+      "step": 2429
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 9.636042402826853e-05,
+      "loss": 3.4359,
+      "step": 2430
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 9.625441696113073e-05,
+      "loss": 3.2629,
+      "step": 2431
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 9.614840989399292e-05,
+      "loss": 3.4475,
+      "step": 2432
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 9.604240282685511e-05,
+      "loss": 3.4295,
+      "step": 2433
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 9.59363957597173e-05,
+      "loss": 3.3758,
+      "step": 2434
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 9.58303886925795e-05,
+      "loss": 3.2312,
+      "step": 2435
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 9.572438162544168e-05,
+      "loss": 3.4642,
+      "step": 2436
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 9.561837455830388e-05,
+      "loss": 3.386,
+      "step": 2437
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 9.551236749116608e-05,
+      "loss": 3.2666,
+      "step": 2438
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 9.540636042402826e-05,
+      "loss": 3.3672,
+      "step": 2439
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 9.530035335689045e-05,
+      "loss": 3.3185,
+      "step": 2440
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 9.519434628975265e-05,
+      "loss": 3.3347,
+      "step": 2441
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 9.508833922261482e-05,
+      "loss": 3.3542,
+      "step": 2442
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 9.498233215547703e-05,
+      "loss": 5.1151,
+      "step": 2443
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 9.487632508833923e-05,
+      "loss": 3.3377,
+      "step": 2444
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 9.47703180212014e-05,
+      "loss": 3.4289,
+      "step": 2445
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 9.46643109540636e-05,
+      "loss": 3.7556,
+      "step": 2446
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 9.455830388692578e-05,
+      "loss": 4.0287,
+      "step": 2447
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 9.445229681978797e-05,
+      "loss": 3.8539,
+      "step": 2448
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 9.434628975265017e-05,
+      "loss": 3.8799,
+      "step": 2449
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 9.424028268551235e-05,
+      "loss": 4.265,
+      "step": 2450
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 9.413427561837455e-05,
+      "loss": 4.3413,
+      "step": 2451
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 9.402826855123675e-05,
+      "loss": 3.8177,
+      "step": 2452
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 9.392226148409893e-05,
+      "loss": 3.8502,
+      "step": 2453
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 9.381625441696113e-05,
+      "loss": 3.9777,
+      "step": 2454
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 9.371024734982332e-05,
+      "loss": 3.6483,
+      "step": 2455
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 9.36042402826855e-05,
+      "loss": 3.646,
+      "step": 2456
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 9.34982332155477e-05,
+      "loss": 3.5971,
+      "step": 2457
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 9.33922261484099e-05,
+      "loss": 3.6541,
+      "step": 2458
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 9.328621908127207e-05,
+      "loss": 3.7147,
+      "step": 2459
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 9.318021201413428e-05,
+      "loss": 3.5919,
+      "step": 2460
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 9.307420494699645e-05,
+      "loss": 3.5029,
+      "step": 2461
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 9.296819787985865e-05,
+      "loss": 3.6148,
+      "step": 2462
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 9.286219081272084e-05,
+      "loss": 3.6723,
+      "step": 2463
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 9.275618374558303e-05,
+      "loss": 3.5135,
+      "step": 2464
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 9.265017667844522e-05,
+      "loss": 3.5415,
+      "step": 2465
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 9.254416961130742e-05,
+      "loss": 3.5228,
+      "step": 2466
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 9.24381625441696e-05,
+      "loss": 3.487,
+      "step": 2467
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 9.23321554770318e-05,
+      "loss": 3.7543,
+      "step": 2468
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 9.222614840989399e-05,
+      "loss": 3.5086,
+      "step": 2469
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 9.212014134275618e-05,
+      "loss": 3.4425,
+      "step": 2470
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 9.201413427561836e-05,
+      "loss": 3.3642,
+      "step": 2471
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 9.190812720848057e-05,
+      "loss": 3.4293,
+      "step": 2472
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 9.180212014134274e-05,
+      "loss": 3.3868,
+      "step": 2473
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 9.169611307420494e-05,
+      "loss": 3.134,
+      "step": 2474
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 9.159010600706713e-05,
+      "loss": 3.4804,
+      "step": 2475
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 9.148409893992932e-05,
+      "loss": 3.5193,
+      "step": 2476
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 9.137809187279151e-05,
+      "loss": 3.3093,
+      "step": 2477
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 9.12720848056537e-05,
+      "loss": 3.3653,
+      "step": 2478
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 9.116607773851589e-05,
+      "loss": 3.6353,
+      "step": 2479
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 9.106007067137809e-05,
+      "loss": 3.6237,
+      "step": 2480
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 9.095406360424027e-05,
+      "loss": 3.5669,
+      "step": 2481
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 9.084805653710247e-05,
+      "loss": 3.6346,
+      "step": 2482
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 9.074204946996466e-05,
+      "loss": 3.5151,
+      "step": 2483
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 9.063604240282685e-05,
+      "loss": 3.4987,
+      "step": 2484
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 9.053003533568903e-05,
+      "loss": 3.4181,
+      "step": 2485
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 9.042402826855124e-05,
+      "loss": 3.4106,
+      "step": 2486
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 9.031802120141341e-05,
+      "loss": 3.5021,
+      "step": 2487
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 9.021201413427561e-05,
+      "loss": 3.378,
+      "step": 2488
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 9.01060070671378e-05,
+      "loss": 3.5737,
+      "step": 2489
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 3.7904,
+      "step": 2490
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 8.989399293286218e-05,
+      "loss": 3.8039,
+      "step": 2491
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 8.978798586572437e-05,
+      "loss": 3.4625,
+      "step": 2492
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 8.968197879858656e-05,
+      "loss": 3.6144,
+      "step": 2493
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 8.957597173144876e-05,
+      "loss": 3.6288,
+      "step": 2494
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 8.946996466431094e-05,
+      "loss": 3.569,
+      "step": 2495
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 8.936395759717314e-05,
+      "loss": 3.4985,
+      "step": 2496
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 8.925795053003533e-05,
+      "loss": 3.5635,
+      "step": 2497
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 8.915194346289752e-05,
+      "loss": 3.5544,
+      "step": 2498
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 8.90459363957597e-05,
+      "loss": 3.3307,
+      "step": 2499
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 8.893992932862191e-05,
+      "loss": 3.5509,
+      "step": 2500
+    },
+    {
+      "epoch": 22.52,
+      "eval_loss": 18.95335578918457,
+      "eval_runtime": 120.7896,
+      "eval_samples_per_second": 21.873,
+      "eval_steps_per_second": 0.687,
+      "eval_wer": 1.0015073383577946,
+      "step": 2500
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 8.883392226148408e-05,
+      "loss": 3.4995,
+      "step": 2501
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 8.872791519434628e-05,
+      "loss": 3.3347,
+      "step": 2502
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 8.862190812720847e-05,
+      "loss": 3.3088,
+      "step": 2503
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 8.851590106007066e-05,
+      "loss": 3.505,
+      "step": 2504
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 8.840989399293285e-05,
+      "loss": 3.3422,
+      "step": 2505
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 8.830388692579505e-05,
+      "loss": 3.1682,
+      "step": 2506
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 8.819787985865723e-05,
+      "loss": 3.2806,
+      "step": 2507
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 8.809187279151943e-05,
+      "loss": 3.2457,
+      "step": 2508
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 8.79858657243816e-05,
+      "loss": 3.3456,
+      "step": 2509
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 8.787985865724381e-05,
+      "loss": 3.2396,
+      "step": 2510
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 8.7773851590106e-05,
+      "loss": 3.4484,
+      "step": 2511
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 8.766784452296819e-05,
+      "loss": 3.3925,
+      "step": 2512
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 8.756183745583037e-05,
+      "loss": 3.2913,
+      "step": 2513
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 8.745583038869258e-05,
+      "loss": 3.3969,
+      "step": 2514
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 8.734982332155475e-05,
+      "loss": 3.3608,
+      "step": 2515
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 8.724381625441695e-05,
+      "loss": 3.4167,
+      "step": 2516
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 8.713780918727916e-05,
+      "loss": 3.233,
+      "step": 2517
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 8.703180212014133e-05,
+      "loss": 3.4533,
+      "step": 2518
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 8.692579505300353e-05,
+      "loss": 3.3646,
+      "step": 2519
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 8.681978798586572e-05,
+      "loss": 3.3803,
+      "step": 2520
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 8.67137809187279e-05,
+      "loss": 3.2697,
+      "step": 2521
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 8.66077738515901e-05,
+      "loss": 3.393,
+      "step": 2522
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 8.650176678445228e-05,
+      "loss": 3.2894,
+      "step": 2523
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 8.639575971731448e-05,
+      "loss": 3.2549,
+      "step": 2524
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 8.628975265017668e-05,
+      "loss": 3.4293,
+      "step": 2525
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 8.618374558303886e-05,
+      "loss": 3.4446,
+      "step": 2526
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 8.607773851590106e-05,
+      "loss": 3.2302,
+      "step": 2527
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 8.597173144876325e-05,
+      "loss": 3.3397,
+      "step": 2528
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 8.586572438162542e-05,
+      "loss": 3.4881,
+      "step": 2529
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 8.575971731448762e-05,
+      "loss": 3.3561,
+      "step": 2530
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 8.565371024734983e-05,
+      "loss": 3.1869,
+      "step": 2531
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 8.5547703180212e-05,
+      "loss": 3.4495,
+      "step": 2532
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 8.54416961130742e-05,
+      "loss": 3.3732,
+      "step": 2533
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 8.533568904593639e-05,
+      "loss": 3.3652,
+      "step": 2534
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 8.522968197879858e-05,
+      "loss": 3.2977,
+      "step": 2535
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 8.512367491166077e-05,
+      "loss": 3.3213,
+      "step": 2536
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 8.501766784452297e-05,
+      "loss": 3.3925,
+      "step": 2537
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 8.491166077738515e-05,
+      "loss": 3.3511,
+      "step": 2538
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 8.480565371024735e-05,
+      "loss": 3.2017,
+      "step": 2539
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 8.469964664310953e-05,
+      "loss": 3.3013,
+      "step": 2540
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 8.459363957597173e-05,
+      "loss": 3.4297,
+      "step": 2541
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 8.448763250883392e-05,
+      "loss": 3.1154,
+      "step": 2542
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 8.43816254416961e-05,
+      "loss": 3.4351,
+      "step": 2543
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 8.42756183745583e-05,
+      "loss": 3.3482,
+      "step": 2544
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 8.41696113074205e-05,
+      "loss": 3.2934,
+      "step": 2545
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 8.406360424028267e-05,
+      "loss": 3.3558,
+      "step": 2546
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 8.395759717314487e-05,
+      "loss": 3.4131,
+      "step": 2547
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 8.385159010600706e-05,
+      "loss": 3.3604,
+      "step": 2548
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 8.374558303886925e-05,
+      "loss": 3.2015,
+      "step": 2549
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 8.363957597173144e-05,
+      "loss": 3.5673,
+      "step": 2550
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 8.353356890459364e-05,
+      "loss": 3.4289,
+      "step": 2551
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 8.342756183745582e-05,
+      "loss": 3.2931,
+      "step": 2552
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 8.332155477031802e-05,
+      "loss": 3.4033,
+      "step": 2553
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 8.32155477031802e-05,
+      "loss": 5.2697,
+      "step": 2554
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 8.31095406360424e-05,
+      "loss": 3.5435,
+      "step": 2555
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 8.300353356890459e-05,
+      "loss": 3.4113,
+      "step": 2556
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 8.289752650176677e-05,
+      "loss": 3.1419,
+      "step": 2557
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 8.279151943462896e-05,
+      "loss": 3.3949,
+      "step": 2558
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 8.268551236749117e-05,
+      "loss": 3.3532,
+      "step": 2559
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 8.257950530035334e-05,
+      "loss": 3.2101,
+      "step": 2560
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 8.247349823321554e-05,
+      "loss": 3.303,
+      "step": 2561
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 8.236749116607773e-05,
+      "loss": 3.3314,
+      "step": 2562
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 8.226148409893992e-05,
+      "loss": 3.1836,
+      "step": 2563
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 8.215547703180211e-05,
+      "loss": 3.1847,
+      "step": 2564
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 8.204946996466431e-05,
+      "loss": 3.3492,
+      "step": 2565
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 8.194346289752649e-05,
+      "loss": 3.3284,
+      "step": 2566
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 8.183745583038869e-05,
+      "loss": 3.1301,
+      "step": 2567
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 8.173144876325088e-05,
+      "loss": 3.4179,
+      "step": 2568
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 8.162544169611307e-05,
+      "loss": 3.5106,
+      "step": 2569
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 8.151943462897526e-05,
+      "loss": 3.3794,
+      "step": 2570
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 8.141342756183744e-05,
+      "loss": 3.2055,
+      "step": 2571
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 8.130742049469963e-05,
+      "loss": 3.3457,
+      "step": 2572
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 8.120141342756184e-05,
+      "loss": 3.2678,
+      "step": 2573
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 8.109540636042401e-05,
+      "loss": 3.3894,
+      "step": 2574
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 8.098939929328621e-05,
+      "loss": 3.3244,
+      "step": 2575
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 8.08833922261484e-05,
+      "loss": 3.4436,
+      "step": 2576
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 8.077738515901059e-05,
+      "loss": 3.3274,
+      "step": 2577
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 8.067137809187278e-05,
+      "loss": 3.1896,
+      "step": 2578
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 8.056537102473498e-05,
+      "loss": 3.5052,
+      "step": 2579
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 8.045936395759716e-05,
+      "loss": 3.4334,
+      "step": 2580
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 8.035335689045936e-05,
+      "loss": 3.2772,
+      "step": 2581
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 8.024734982332156e-05,
+      "loss": 3.3929,
+      "step": 2582
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 8.014134275618374e-05,
+      "loss": 3.9248,
+      "step": 2583
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 8.003533568904593e-05,
+      "loss": 3.9689,
+      "step": 2584
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 7.992932862190811e-05,
+      "loss": 3.5695,
+      "step": 2585
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 7.98233215547703e-05,
+      "loss": 3.6465,
+      "step": 2586
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 7.97173144876325e-05,
+      "loss": 3.4606,
+      "step": 2587
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 7.961130742049468e-05,
+      "loss": 3.7275,
+      "step": 2588
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 7.950530035335688e-05,
+      "loss": 3.7264,
+      "step": 2589
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 7.939929328621909e-05,
+      "loss": 3.7157,
+      "step": 2590
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 7.929328621908126e-05,
+      "loss": 3.369,
+      "step": 2591
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 7.918727915194345e-05,
+      "loss": 3.1828,
+      "step": 2592
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 7.908127208480565e-05,
+      "loss": 3.5765,
+      "step": 2593
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 7.897526501766783e-05,
+      "loss": 3.621,
+      "step": 2594
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 7.886925795053003e-05,
+      "loss": 3.3948,
+      "step": 2595
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 7.876325088339223e-05,
+      "loss": 3.2082,
+      "step": 2596
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 7.86572438162544e-05,
+      "loss": 3.4725,
+      "step": 2597
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 7.855123674911661e-05,
+      "loss": 3.5789,
+      "step": 2598
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 7.84452296819788e-05,
+      "loss": 3.5201,
+      "step": 2599
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 7.833922261484097e-05,
+      "loss": 3.3934,
+      "step": 2600
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 7.823321554770318e-05,
+      "loss": 3.4097,
+      "step": 2601
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 7.812720848056535e-05,
+      "loss": 3.3803,
+      "step": 2602
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 7.802120141342755e-05,
+      "loss": 3.4528,
+      "step": 2603
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 7.791519434628976e-05,
+      "loss": 3.7904,
+      "step": 2604
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 7.780918727915193e-05,
+      "loss": 3.6547,
+      "step": 2605
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 7.770318021201413e-05,
+      "loss": 3.3026,
+      "step": 2606
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 7.759717314487632e-05,
+      "loss": 3.0945,
+      "step": 2607
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 7.749116607773851e-05,
+      "loss": 3.4397,
+      "step": 2608
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 7.73851590106007e-05,
+      "loss": 3.3998,
+      "step": 2609
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 7.72791519434629e-05,
+      "loss": 3.2703,
+      "step": 2610
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 7.717314487632508e-05,
+      "loss": 3.41,
+      "step": 2611
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 7.706713780918728e-05,
+      "loss": 3.3141,
+      "step": 2612
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 7.696113074204947e-05,
+      "loss": 3.2612,
+      "step": 2613
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 7.685512367491166e-05,
+      "loss": 3.1168,
+      "step": 2614
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 7.674911660777384e-05,
+      "loss": 3.3201,
+      "step": 2615
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 7.664310954063603e-05,
+      "loss": 3.3007,
+      "step": 2616
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 7.653710247349822e-05,
+      "loss": 3.0391,
+      "step": 2617
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 7.643109540636042e-05,
+      "loss": 3.3371,
+      "step": 2618
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 7.63250883392226e-05,
+      "loss": 3.3152,
+      "step": 2619
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 7.62190812720848e-05,
+      "loss": 3.4218,
+      "step": 2620
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 7.611307420494699e-05,
+      "loss": 3.349,
+      "step": 2621
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 7.600706713780918e-05,
+      "loss": 3.4161,
+      "step": 2622
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 7.590106007067137e-05,
+      "loss": 3.3654,
+      "step": 2623
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 7.579505300353357e-05,
+      "loss": 3.2101,
+      "step": 2624
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 7.568904593639575e-05,
+      "loss": 3.1571,
+      "step": 2625
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 7.558303886925795e-05,
+      "loss": 3.4123,
+      "step": 2626
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 7.547703180212014e-05,
+      "loss": 3.3919,
+      "step": 2627
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 7.537102473498233e-05,
+      "loss": 3.3312,
+      "step": 2628
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 7.526501766784451e-05,
+      "loss": 3.4957,
+      "step": 2629
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 7.515901060070672e-05,
+      "loss": 3.4523,
+      "step": 2630
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 7.505300353356889e-05,
+      "loss": 3.3968,
+      "step": 2631
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 7.49469964664311e-05,
+      "loss": 3.3779,
+      "step": 2632
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 7.484098939929328e-05,
+      "loss": 3.4397,
+      "step": 2633
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 7.473498233215547e-05,
+      "loss": 3.3517,
+      "step": 2634
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 7.462897526501766e-05,
+      "loss": 3.2391,
+      "step": 2635
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 7.452296819787985e-05,
+      "loss": 3.3995,
+      "step": 2636
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 7.441696113074204e-05,
+      "loss": 3.197,
+      "step": 2637
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 7.431095406360423e-05,
+      "loss": 3.2491,
+      "step": 2638
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 7.420494699646643e-05,
+      "loss": 3.2247,
+      "step": 2639
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 7.409893992932862e-05,
+      "loss": 3.3318,
+      "step": 2640
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 7.399293286219081e-05,
+      "loss": 3.2718,
+      "step": 2641
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 7.3886925795053e-05,
+      "loss": 3.1723,
+      "step": 2642
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 7.378091872791518e-05,
+      "loss": 3.3551,
+      "step": 2643
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 7.367491166077737e-05,
+      "loss": 3.233,
+      "step": 2644
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 7.356890459363956e-05,
+      "loss": 3.3061,
+      "step": 2645
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 7.346289752650176e-05,
+      "loss": 3.1779,
+      "step": 2646
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 7.335689045936395e-05,
+      "loss": 3.3175,
+      "step": 2647
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 7.325088339222614e-05,
+      "loss": 3.3348,
+      "step": 2648
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 7.314487632508833e-05,
+      "loss": 3.2137,
+      "step": 2649
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 7.303886925795052e-05,
+      "loss": 3.2091,
+      "step": 2650
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 7.293286219081271e-05,
+      "loss": 3.2898,
+      "step": 2651
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 7.28268551236749e-05,
+      "loss": 3.1929,
+      "step": 2652
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 7.27208480565371e-05,
+      "loss": 3.1241,
+      "step": 2653
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 7.261484098939929e-05,
+      "loss": 3.3168,
+      "step": 2654
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 7.250883392226148e-05,
+      "loss": 3.3407,
+      "step": 2655
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 7.240282685512367e-05,
+      "loss": 3.16,
+      "step": 2656
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 7.229681978798585e-05,
+      "loss": 3.1582,
+      "step": 2657
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 7.219081272084804e-05,
+      "loss": 3.3416,
+      "step": 2658
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 7.208480565371023e-05,
+      "loss": 3.2037,
+      "step": 2659
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 7.197879858657243e-05,
+      "loss": 3.1594,
+      "step": 2660
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 7.187279151943462e-05,
+      "loss": 3.4741,
+      "step": 2661
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 7.176678445229681e-05,
+      "loss": 3.4241,
+      "step": 2662
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 7.166077738515901e-05,
+      "loss": 3.1129,
+      "step": 2663
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 7.155477031802119e-05,
+      "loss": 3.3097,
+      "step": 2664
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 7.144876325088338e-05,
+      "loss": 5.3238,
+      "step": 2665
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 7.134275618374558e-05,
+      "loss": 3.4678,
+      "step": 2666
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 7.123674911660777e-05,
+      "loss": 3.2241,
+      "step": 2667
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 7.113074204946996e-05,
+      "loss": 3.0745,
+      "step": 2668
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 7.102473498233215e-05,
+      "loss": 3.489,
+      "step": 2669
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 7.091872791519435e-05,
+      "loss": 3.7114,
+      "step": 2670
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 7.081272084805654e-05,
+      "loss": 3.6504,
+      "step": 2671
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 7.070671378091871e-05,
+      "loss": 3.7932,
+      "step": 2672
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 7.060070671378092e-05,
+      "loss": 3.4734,
+      "step": 2673
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 7.04946996466431e-05,
+      "loss": 3.1534,
+      "step": 2674
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 7.038869257950529e-05,
+      "loss": 3.2182,
+      "step": 2675
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 7.028268551236748e-05,
+      "loss": 3.3727,
+      "step": 2676
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 7.017667844522968e-05,
+      "loss": 3.2396,
+      "step": 2677
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 7.007067137809187e-05,
+      "loss": 3.0961,
+      "step": 2678
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 6.996466431095406e-05,
+      "loss": 3.3054,
+      "step": 2679
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 6.985865724381625e-05,
+      "loss": 3.2903,
+      "step": 2680
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 6.975265017667844e-05,
+      "loss": 3.2829,
+      "step": 2681
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 6.964664310954063e-05,
+      "loss": 3.2173,
+      "step": 2682
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 6.954063604240282e-05,
+      "loss": 3.3626,
+      "step": 2683
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 6.943462897526502e-05,
+      "loss": 3.3205,
+      "step": 2684
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 6.932862190812721e-05,
+      "loss": 3.1959,
+      "step": 2685
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 6.92226148409894e-05,
+      "loss": 3.1208,
+      "step": 2686
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 6.911660777385158e-05,
+      "loss": 3.2243,
+      "step": 2687
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 6.901060070671377e-05,
+      "loss": 3.1361,
+      "step": 2688
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 6.890459363957596e-05,
+      "loss": 3.0347,
+      "step": 2689
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 6.879858657243815e-05,
+      "loss": 3.3378,
+      "step": 2690
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 6.869257950530035e-05,
+      "loss": 3.3169,
+      "step": 2691
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 6.858657243816254e-05,
+      "loss": 3.2938,
+      "step": 2692
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 6.848056537102473e-05,
+      "loss": 3.1955,
+      "step": 2693
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 6.837455830388692e-05,
+      "loss": 3.2994,
+      "step": 2694
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 6.826855123674911e-05,
+      "loss": 3.2438,
+      "step": 2695
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 6.81625441696113e-05,
+      "loss": 3.1657,
+      "step": 2696
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 6.80565371024735e-05,
+      "loss": 3.3737,
+      "step": 2697
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 6.795053003533569e-05,
+      "loss": 3.2318,
+      "step": 2698
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 6.784452296819788e-05,
+      "loss": 3.3236,
+      "step": 2699
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 6.773851590106007e-05,
+      "loss": 3.2385,
+      "step": 2700
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 6.763250883392225e-05,
+      "loss": 3.3256,
+      "step": 2701
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 6.752650176678444e-05,
+      "loss": 3.2368,
+      "step": 2702
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 6.742049469964663e-05,
+      "loss": 3.1774,
+      "step": 2703
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 6.731448763250883e-05,
+      "loss": 3.2132,
+      "step": 2704
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 6.720848056537102e-05,
+      "loss": 3.2475,
+      "step": 2705
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 6.710247349823321e-05,
+      "loss": 3.2372,
+      "step": 2706
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 6.69964664310954e-05,
+      "loss": 3.2302,
+      "step": 2707
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 6.689045936395759e-05,
+      "loss": 3.3177,
+      "step": 2708
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 6.678445229681978e-05,
+      "loss": 3.2637,
+      "step": 2709
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 6.667844522968197e-05,
+      "loss": 3.2034,
+      "step": 2710
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 6.657243816254417e-05,
+      "loss": 3.1673,
+      "step": 2711
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 6.646643109540636e-05,
+      "loss": 3.2739,
+      "step": 2712
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 6.636042402826855e-05,
+      "loss": 3.2241,
+      "step": 2713
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 6.625441696113074e-05,
+      "loss": 3.0768,
+      "step": 2714
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 6.614840989399292e-05,
+      "loss": 3.2059,
+      "step": 2715
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 6.604240282685511e-05,
+      "loss": 3.321,
+      "step": 2716
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 6.59363957597173e-05,
+      "loss": 3.2387,
+      "step": 2717
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 6.58303886925795e-05,
+      "loss": 3.0936,
+      "step": 2718
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 6.572438162544169e-05,
+      "loss": 3.2883,
+      "step": 2719
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 6.561837455830388e-05,
+      "loss": 3.194,
+      "step": 2720
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 6.551236749116607e-05,
+      "loss": 3.1249,
+      "step": 2721
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 6.540636042402826e-05,
+      "loss": 3.2514,
+      "step": 2722
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 6.530035335689045e-05,
+      "loss": 3.2464,
+      "step": 2723
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 6.519434628975264e-05,
+      "loss": 3.1439,
+      "step": 2724
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 6.508833922261484e-05,
+      "loss": 3.2376,
+      "step": 2725
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 6.498233215547703e-05,
+      "loss": 3.3524,
+      "step": 2726
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 6.487632508833922e-05,
+      "loss": 3.2084,
+      "step": 2727
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 6.47703180212014e-05,
+      "loss": 3.2201,
+      "step": 2728
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 6.46643109540636e-05,
+      "loss": 3.2867,
+      "step": 2729
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 6.455830388692578e-05,
+      "loss": 3.2504,
+      "step": 2730
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 6.445229681978797e-05,
+      "loss": 3.2566,
+      "step": 2731
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 6.434628975265017e-05,
+      "loss": 3.1564,
+      "step": 2732
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 6.424028268551236e-05,
+      "loss": 3.2063,
+      "step": 2733
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 6.413427561837455e-05,
+      "loss": 3.2544,
+      "step": 2734
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 6.402826855123674e-05,
+      "loss": 3.2446,
+      "step": 2735
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 6.392226148409893e-05,
+      "loss": 3.1915,
+      "step": 2736
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 6.381625441696112e-05,
+      "loss": 3.3171,
+      "step": 2737
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 6.37102473498233e-05,
+      "loss": 3.2577,
+      "step": 2738
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 6.360424028268551e-05,
+      "loss": 3.1228,
+      "step": 2739
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 6.34982332155477e-05,
+      "loss": 3.3514,
+      "step": 2740
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 6.339222614840989e-05,
+      "loss": 3.3631,
+      "step": 2741
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 6.328621908127209e-05,
+      "loss": 3.2781,
+      "step": 2742
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 6.318021201413428e-05,
+      "loss": 3.2043,
+      "step": 2743
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 6.307420494699645e-05,
+      "loss": 3.3263,
+      "step": 2744
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 6.296819787985864e-05,
+      "loss": 3.3082,
+      "step": 2745
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 6.286219081272084e-05,
+      "loss": 3.105,
+      "step": 2746
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 6.275618374558303e-05,
+      "loss": 3.3432,
+      "step": 2747
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 6.265017667844522e-05,
+      "loss": 3.2724,
+      "step": 2748
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 6.254416961130742e-05,
+      "loss": 3.1429,
+      "step": 2749
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 6.243816254416961e-05,
+      "loss": 3.0788,
+      "step": 2750
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 6.23321554770318e-05,
+      "loss": 3.3065,
+      "step": 2751
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 6.222614840989399e-05,
+      "loss": 3.1474,
+      "step": 2752
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 6.212014134275618e-05,
+      "loss": 3.0788,
+      "step": 2753
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 6.201413427561837e-05,
+      "loss": 3.2663,
+      "step": 2754
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 6.190812720848056e-05,
+      "loss": 3.1718,
+      "step": 2755
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 6.180212014134276e-05,
+      "loss": 3.1953,
+      "step": 2756
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 6.169611307420495e-05,
+      "loss": 3.1206,
+      "step": 2757
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 6.159010600706714e-05,
+      "loss": 3.2402,
+      "step": 2758
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 6.148409893992932e-05,
+      "loss": 3.1651,
+      "step": 2759
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 6.137809187279151e-05,
+      "loss": 3.0973,
+      "step": 2760
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 6.12720848056537e-05,
+      "loss": 3.1885,
+      "step": 2761
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 6.116607773851589e-05,
+      "loss": 3.3498,
+      "step": 2762
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 6.10600706713781e-05,
+      "loss": 3.2852,
+      "step": 2763
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 6.0954063604240276e-05,
+      "loss": 3.1437,
+      "step": 2764
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 6.0848056537102464e-05,
+      "loss": 3.2014,
+      "step": 2765
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 6.074204946996466e-05,
+      "loss": 3.2083,
+      "step": 2766
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 6.063604240282685e-05,
+      "loss": 3.2174,
+      "step": 2767
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 6.053003533568904e-05,
+      "loss": 3.0556,
+      "step": 2768
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 6.0424028268551226e-05,
+      "loss": 3.2411,
+      "step": 2769
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 6.031802120141342e-05,
+      "loss": 3.1865,
+      "step": 2770
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 6.021201413427561e-05,
+      "loss": 3.0244,
+      "step": 2771
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 6.01060070671378e-05,
+      "loss": 3.2811,
+      "step": 2772
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 3.2401,
+      "step": 2773
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 5.9893992932862184e-05,
+      "loss": 3.1229,
+      "step": 2774
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 5.978798586572437e-05,
+      "loss": 3.1064,
+      "step": 2775
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 5.968197879858656e-05,
+      "loss": 4.8402,
+      "step": 2776
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 5.957597173144876e-05,
+      "loss": 3.2128,
+      "step": 2777
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 5.9469964664310945e-05,
+      "loss": 3.0691,
+      "step": 2778
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 5.9363957597173134e-05,
+      "loss": 2.8839,
+      "step": 2779
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 5.9257950530035336e-05,
+      "loss": 3.1906,
+      "step": 2780
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 5.9151943462897525e-05,
+      "loss": 3.0978,
+      "step": 2781
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 5.904593639575971e-05,
+      "loss": 2.9866,
+      "step": 2782
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 5.893992932862191e-05,
+      "loss": 3.2107,
+      "step": 2783
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 5.88339222614841e-05,
+      "loss": 3.0948,
+      "step": 2784
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 5.872791519434629e-05,
+      "loss": 3.0814,
+      "step": 2785
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 5.862190812720847e-05,
+      "loss": 3.086,
+      "step": 2786
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 5.851590106007067e-05,
+      "loss": 3.2263,
+      "step": 2787
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 5.840989399293286e-05,
+      "loss": 3.0205,
+      "step": 2788
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 5.830388692579505e-05,
+      "loss": 3.0586,
+      "step": 2789
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 5.8197879858657244e-05,
+      "loss": 3.2935,
+      "step": 2790
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 5.809187279151943e-05,
+      "loss": 3.1869,
+      "step": 2791
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 5.798586572438162e-05,
+      "loss": 3.1318,
+      "step": 2792
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 5.787985865724381e-05,
+      "loss": 3.063,
+      "step": 2793
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 5.7773851590106006e-05,
+      "loss": 3.1966,
+      "step": 2794
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 5.7667844522968195e-05,
+      "loss": 3.2103,
+      "step": 2795
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 5.7561837455830384e-05,
+      "loss": 3.1881,
+      "step": 2796
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 5.745583038869258e-05,
+      "loss": 3.1205,
+      "step": 2797
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 5.734982332155477e-05,
+      "loss": 3.3192,
+      "step": 2798
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 5.724381625441696e-05,
+      "loss": 3.1529,
+      "step": 2799
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 5.7137809187279146e-05,
+      "loss": 3.0725,
+      "step": 2800
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 5.703180212014134e-05,
+      "loss": 3.2555,
+      "step": 2801
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 5.692579505300353e-05,
+      "loss": 3.2253,
+      "step": 2802
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 5.681978798586572e-05,
+      "loss": 3.1399,
+      "step": 2803
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 5.6713780918727914e-05,
+      "loss": 3.0934,
+      "step": 2804
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 5.66077738515901e-05,
+      "loss": 3.2149,
+      "step": 2805
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 5.650176678445229e-05,
+      "loss": 3.3134,
+      "step": 2806
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 5.639575971731448e-05,
+      "loss": 3.1448,
+      "step": 2807
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 5.6289752650176676e-05,
+      "loss": 3.2531,
+      "step": 2808
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 5.6183745583038865e-05,
+      "loss": 3.1772,
+      "step": 2809
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 5.6077738515901054e-05,
+      "loss": 3.1285,
+      "step": 2810
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 5.597173144876325e-05,
+      "loss": 3.0617,
+      "step": 2811
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 5.586572438162544e-05,
+      "loss": 3.2101,
+      "step": 2812
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 5.5759717314487627e-05,
+      "loss": 3.2604,
+      "step": 2813
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 5.565371024734982e-05,
+      "loss": 3.2682,
+      "step": 2814
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 5.554770318021201e-05,
+      "loss": 3.5159,
+      "step": 2815
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 5.54416961130742e-05,
+      "loss": 3.2922,
+      "step": 2816
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 5.533568904593639e-05,
+      "loss": 3.2221,
+      "step": 2817
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 5.5229681978798584e-05,
+      "loss": 3.1275,
+      "step": 2818
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 5.512367491166077e-05,
+      "loss": 3.2256,
+      "step": 2819
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 5.501766784452296e-05,
+      "loss": 3.2744,
+      "step": 2820
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 5.491166077738516e-05,
+      "loss": 3.2159,
+      "step": 2821
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 5.4805653710247346e-05,
+      "loss": 3.1746,
+      "step": 2822
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 5.4699646643109535e-05,
+      "loss": 3.3393,
+      "step": 2823
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 5.459363957597172e-05,
+      "loss": 3.275,
+      "step": 2824
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 5.448763250883392e-05,
+      "loss": 3.0182,
+      "step": 2825
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 5.438162544169611e-05,
+      "loss": 3.2173,
+      "step": 2826
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 5.4275618374558296e-05,
+      "loss": 3.3218,
+      "step": 2827
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 5.416961130742049e-05,
+      "loss": 3.2305,
+      "step": 2828
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 5.406360424028268e-05,
+      "loss": 3.0851,
+      "step": 2829
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 5.395759717314487e-05,
+      "loss": 3.3582,
+      "step": 2830
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 5.385159010600706e-05,
+      "loss": 3.2673,
+      "step": 2831
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 5.3745583038869254e-05,
+      "loss": 3.0614,
+      "step": 2832
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 5.363957597173144e-05,
+      "loss": 3.2352,
+      "step": 2833
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 5.353356890459363e-05,
+      "loss": 3.2092,
+      "step": 2834
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 5.342756183745583e-05,
+      "loss": 3.1671,
+      "step": 2835
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 5.3321554770318016e-05,
+      "loss": 3.157,
+      "step": 2836
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 5.3215547703180204e-05,
+      "loss": 3.2953,
+      "step": 2837
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 5.310954063604239e-05,
+      "loss": 3.2081,
+      "step": 2838
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 5.300353356890459e-05,
+      "loss": 3.1169,
+      "step": 2839
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 5.289752650176678e-05,
+      "loss": 3.0921,
+      "step": 2840
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 5.2791519434628966e-05,
+      "loss": 3.1325,
+      "step": 2841
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 5.268551236749116e-05,
+      "loss": 3.221,
+      "step": 2842
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 5.257950530035335e-05,
+      "loss": 3.0667,
+      "step": 2843
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 5.247349823321554e-05,
+      "loss": 3.2511,
+      "step": 2844
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 5.2367491166077735e-05,
+      "loss": 3.1404,
+      "step": 2845
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 5.2261484098939924e-05,
+      "loss": 3.241,
+      "step": 2846
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 5.215547703180211e-05,
+      "loss": 3.1956,
+      "step": 2847
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 5.20494699646643e-05,
+      "loss": 3.2935,
+      "step": 2848
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 5.1943462897526497e-05,
+      "loss": 3.1959,
+      "step": 2849
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 5.1837455830388685e-05,
+      "loss": 3.0285,
+      "step": 2850
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 5.1731448763250874e-05,
+      "loss": 3.3444,
+      "step": 2851
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 5.1625441696113076e-05,
+      "loss": 3.1338,
+      "step": 2852
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 5.1519434628975265e-05,
+      "loss": 3.1559,
+      "step": 2853
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 5.141342756183745e-05,
+      "loss": 3.0769,
+      "step": 2854
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 5.1307420494699636e-05,
+      "loss": 3.1053,
+      "step": 2855
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 5.120141342756184e-05,
+      "loss": 3.17,
+      "step": 2856
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 5.109540636042403e-05,
+      "loss": 2.935,
+      "step": 2857
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 5.098939929328621e-05,
+      "loss": 3.2248,
+      "step": 2858
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 5.088339222614841e-05,
+      "loss": 3.2833,
+      "step": 2859
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 5.07773851590106e-05,
+      "loss": 3.1593,
+      "step": 2860
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 5.067137809187279e-05,
+      "loss": 3.2013,
+      "step": 2861
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 5.056537102473497e-05,
+      "loss": 3.2723,
+      "step": 2862
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 5.045936395759717e-05,
+      "loss": 3.1659,
+      "step": 2863
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 5.035335689045936e-05,
+      "loss": 3.0526,
+      "step": 2864
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 5.024734982332155e-05,
+      "loss": 3.2674,
+      "step": 2865
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 5.0141342756183746e-05,
+      "loss": 3.315,
+      "step": 2866
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 5.0035335689045935e-05,
+      "loss": 3.2179,
+      "step": 2867
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 4.9929328621908124e-05,
+      "loss": 3.2427,
+      "step": 2868
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 4.982332155477031e-05,
+      "loss": 3.1847,
+      "step": 2869
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 4.971731448763251e-05,
+      "loss": 3.2339,
+      "step": 2870
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 4.96113074204947e-05,
+      "loss": 3.17,
+      "step": 2871
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 4.9505300353356886e-05,
+      "loss": 3.0885,
+      "step": 2872
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 4.939929328621908e-05,
+      "loss": 3.3062,
+      "step": 2873
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 4.929328621908127e-05,
+      "loss": 3.2545,
+      "step": 2874
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 4.918727915194346e-05,
+      "loss": 3.0476,
+      "step": 2875
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 4.9081272084805654e-05,
+      "loss": 3.2573,
+      "step": 2876
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 4.897526501766784e-05,
+      "loss": 3.3096,
+      "step": 2877
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 4.886925795053003e-05,
+      "loss": 3.0509,
+      "step": 2878
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 4.876325088339222e-05,
+      "loss": 3.0803,
+      "step": 2879
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 4.8657243816254416e-05,
+      "loss": 3.1481,
+      "step": 2880
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 4.8551236749116605e-05,
+      "loss": 3.3024,
+      "step": 2881
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 4.8445229681978794e-05,
+      "loss": 2.9781,
+      "step": 2882
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 4.833922261484099e-05,
+      "loss": 3.2417,
+      "step": 2883
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 4.823321554770318e-05,
+      "loss": 3.2529,
+      "step": 2884
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 4.8127208480565367e-05,
+      "loss": 3.0366,
+      "step": 2885
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 4.8021201413427555e-05,
+      "loss": 3.1848,
+      "step": 2886
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 4.791519434628975e-05,
+      "loss": 4.8187,
+      "step": 2887
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 4.780918727915194e-05,
+      "loss": 3.1334,
+      "step": 2888
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 4.770318021201413e-05,
+      "loss": 3.1754,
+      "step": 2889
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 4.7597173144876324e-05,
+      "loss": 3.0821,
+      "step": 2890
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 4.749116607773851e-05,
+      "loss": 3.2007,
+      "step": 2891
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 4.73851590106007e-05,
+      "loss": 3.1771,
+      "step": 2892
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 4.727915194346289e-05,
+      "loss": 2.942,
+      "step": 2893
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 4.7173144876325086e-05,
+      "loss": 3.156,
+      "step": 2894
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 4.7067137809187275e-05,
+      "loss": 3.0437,
+      "step": 2895
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 4.696113074204946e-05,
+      "loss": 3.0684,
+      "step": 2896
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 4.685512367491166e-05,
+      "loss": 3.1459,
+      "step": 2897
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 4.674911660777385e-05,
+      "loss": 3.2425,
+      "step": 2898
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 4.6643109540636036e-05,
+      "loss": 3.0972,
+      "step": 2899
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 4.6537102473498225e-05,
+      "loss": 3.0363,
+      "step": 2900
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 4.643109540636042e-05,
+      "loss": 3.1171,
+      "step": 2901
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 4.632508833922261e-05,
+      "loss": 3.1445,
+      "step": 2902
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 4.62190812720848e-05,
+      "loss": 3.0878,
+      "step": 2903
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 4.6113074204946994e-05,
+      "loss": 3.141,
+      "step": 2904
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 4.600706713780918e-05,
+      "loss": 3.0691,
+      "step": 2905
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 4.590106007067137e-05,
+      "loss": 3.0938,
+      "step": 2906
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 4.579505300353357e-05,
+      "loss": 3.0097,
+      "step": 2907
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 4.5689045936395756e-05,
+      "loss": 3.1443,
+      "step": 2908
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 4.5583038869257944e-05,
+      "loss": 3.17,
+      "step": 2909
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 4.547703180212013e-05,
+      "loss": 3.1759,
+      "step": 2910
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 4.537102473498233e-05,
+      "loss": 3.0652,
+      "step": 2911
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 4.526501766784452e-05,
+      "loss": 3.2946,
+      "step": 2912
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 4.5159010600706706e-05,
+      "loss": 3.2158,
+      "step": 2913
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 4.50530035335689e-05,
+      "loss": 3.0747,
+      "step": 2914
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 4.494699646643109e-05,
+      "loss": 2.9669,
+      "step": 2915
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 4.484098939929328e-05,
+      "loss": 3.1517,
+      "step": 2916
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 4.473498233215547e-05,
+      "loss": 3.2117,
+      "step": 2917
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 4.4628975265017664e-05,
+      "loss": 2.9735,
+      "step": 2918
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 4.452296819787985e-05,
+      "loss": 3.2502,
+      "step": 2919
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 4.441696113074204e-05,
+      "loss": 3.2192,
+      "step": 2920
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 4.4310954063604237e-05,
+      "loss": 3.0886,
+      "step": 2921
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 4.4204946996466425e-05,
+      "loss": 3.1318,
+      "step": 2922
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 4.4098939929328614e-05,
+      "loss": 3.2258,
+      "step": 2923
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 4.39929328621908e-05,
+      "loss": 3.1193,
+      "step": 2924
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 4.3886925795053e-05,
+      "loss": 3.1336,
+      "step": 2925
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 4.378091872791519e-05,
+      "loss": 3.3267,
+      "step": 2926
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 4.3674911660777376e-05,
+      "loss": 3.1993,
+      "step": 2927
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 4.356890459363958e-05,
+      "loss": 3.1887,
+      "step": 2928
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 4.346289752650177e-05,
+      "loss": 3.244,
+      "step": 2929
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 4.335689045936395e-05,
+      "loss": 3.2169,
+      "step": 2930
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 4.325088339222614e-05,
+      "loss": 3.1733,
+      "step": 2931
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 4.314487632508834e-05,
+      "loss": 3.1932,
+      "step": 2932
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 4.303886925795053e-05,
+      "loss": 3.0813,
+      "step": 2933
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 4.293286219081271e-05,
+      "loss": 3.1865,
+      "step": 2934
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 4.282685512367491e-05,
+      "loss": 3.1733,
+      "step": 2935
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 4.27208480565371e-05,
+      "loss": 2.9429,
+      "step": 2936
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 4.261484098939929e-05,
+      "loss": 3.2706,
+      "step": 2937
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 4.2508833922261486e-05,
+      "loss": 3.2338,
+      "step": 2938
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 4.2402826855123675e-05,
+      "loss": 3.0826,
+      "step": 2939
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 4.2296819787985864e-05,
+      "loss": 3.0408,
+      "step": 2940
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 4.219081272084805e-05,
+      "loss": 3.1684,
+      "step": 2941
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 4.208480565371025e-05,
+      "loss": 3.1051,
+      "step": 2942
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 4.197879858657244e-05,
+      "loss": 2.9122,
+      "step": 2943
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 4.1872791519434626e-05,
+      "loss": 3.1355,
+      "step": 2944
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 4.176678445229682e-05,
+      "loss": 3.1015,
+      "step": 2945
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 4.166077738515901e-05,
+      "loss": 3.0621,
+      "step": 2946
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 4.15547703180212e-05,
+      "loss": 3.0412,
+      "step": 2947
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 4.144876325088339e-05,
+      "loss": 3.1096,
+      "step": 2948
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 4.134275618374558e-05,
+      "loss": 3.1051,
+      "step": 2949
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 4.123674911660777e-05,
+      "loss": 2.9526,
+      "step": 2950
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 4.113074204946996e-05,
+      "loss": 3.1041,
+      "step": 2951
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 4.1024734982332156e-05,
+      "loss": 3.1457,
+      "step": 2952
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 4.0918727915194345e-05,
+      "loss": 3.0362,
+      "step": 2953
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 4.0812720848056534e-05,
+      "loss": 3.0005,
+      "step": 2954
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 4.070671378091872e-05,
+      "loss": 3.1334,
+      "step": 2955
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 4.060070671378092e-05,
+      "loss": 3.0616,
+      "step": 2956
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 4.0494699646643107e-05,
+      "loss": 3.1125,
+      "step": 2957
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 4.0388692579505295e-05,
+      "loss": 2.9742,
+      "step": 2958
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 4.028268551236749e-05,
+      "loss": 3.1242,
+      "step": 2959
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 4.017667844522968e-05,
+      "loss": 3.1327,
+      "step": 2960
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 4.007067137809187e-05,
+      "loss": 2.8366,
+      "step": 2961
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 3.996466431095406e-05,
+      "loss": 3.1364,
+      "step": 2962
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 3.985865724381625e-05,
+      "loss": 3.2288,
+      "step": 2963
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 3.975265017667844e-05,
+      "loss": 3.0212,
+      "step": 2964
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 3.964664310954063e-05,
+      "loss": 2.9684,
+      "step": 2965
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 3.9540636042402826e-05,
+      "loss": 3.1764,
+      "step": 2966
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 3.9434628975265015e-05,
+      "loss": 3.0274,
+      "step": 2967
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 3.93286219081272e-05,
+      "loss": 2.9554,
+      "step": 2968
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 3.92226148409894e-05,
+      "loss": 3.2311,
+      "step": 2969
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 3.911660777385159e-05,
+      "loss": 3.1886,
+      "step": 2970
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 3.9010600706713776e-05,
+      "loss": 3.0399,
+      "step": 2971
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 3.8904593639575965e-05,
+      "loss": 3.1264,
+      "step": 2972
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 3.879858657243816e-05,
+      "loss": 3.1141,
+      "step": 2973
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 3.869257950530035e-05,
+      "loss": 3.083,
+      "step": 2974
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 3.858657243816254e-05,
+      "loss": 3.0329,
+      "step": 2975
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 3.8480565371024734e-05,
+      "loss": 3.1952,
+      "step": 2976
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 3.837455830388692e-05,
+      "loss": 3.1552,
+      "step": 2977
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 3.826855123674911e-05,
+      "loss": 3.0631,
+      "step": 2978
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 3.81625441696113e-05,
+      "loss": 2.9543,
+      "step": 2979
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 3.8056537102473496e-05,
+      "loss": 3.1818,
+      "step": 2980
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 3.7950530035335684e-05,
+      "loss": 3.0588,
+      "step": 2981
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 3.784452296819787e-05,
+      "loss": 3.1343,
+      "step": 2982
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 3.773851590106007e-05,
+      "loss": 3.1379,
+      "step": 2983
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 3.763250883392226e-05,
+      "loss": 3.1373,
+      "step": 2984
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 3.7526501766784446e-05,
+      "loss": 3.2487,
+      "step": 2985
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 3.742049469964664e-05,
+      "loss": 2.9049,
+      "step": 2986
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 3.731448763250883e-05,
+      "loss": 3.1572,
+      "step": 2987
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 3.720848056537102e-05,
+      "loss": 3.1766,
+      "step": 2988
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 3.7102473498233215e-05,
+      "loss": 3.2703,
+      "step": 2989
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 3.6996466431095404e-05,
+      "loss": 3.0703,
+      "step": 2990
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 3.689045936395759e-05,
+      "loss": 3.1087,
+      "step": 2991
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 3.678445229681978e-05,
+      "loss": 3.1053,
+      "step": 2992
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 3.6678445229681977e-05,
+      "loss": 2.864,
+      "step": 2993
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 3.6572438162544165e-05,
+      "loss": 3.0499,
+      "step": 2994
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 3.6466431095406354e-05,
+      "loss": 3.0969,
+      "step": 2995
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 3.636042402826855e-05,
+      "loss": 3.0966,
+      "step": 2996
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 3.625441696113074e-05,
+      "loss": 3.0429,
+      "step": 2997
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 3.614840989399293e-05,
+      "loss": 4.5981,
+      "step": 2998
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 3.6042402826855116e-05,
+      "loss": 3.1553,
+      "step": 2999
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 3.593639575971731e-05,
+      "loss": 2.9985,
+      "step": 3000
+    },
+    {
+      "epoch": 27.03,
+      "eval_loss": 20.310462951660156,
+      "eval_runtime": 121.098,
+      "eval_samples_per_second": 21.817,
+      "eval_steps_per_second": 0.685,
+      "eval_wer": 1.0022808409361366,
+      "step": 3000
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 3.583038869257951e-05,
+      "loss": 2.8454,
+      "step": 3001
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 3.572438162544169e-05,
+      "loss": 3.0389,
+      "step": 3002
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 3.5618374558303885e-05,
+      "loss": 3.096,
+      "step": 3003
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 3.551236749116607e-05,
+      "loss": 2.8286,
+      "step": 3004
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 3.540636042402827e-05,
+      "loss": 3.1693,
+      "step": 3005
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 3.530035335689046e-05,
+      "loss": 2.9865,
+      "step": 3006
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 3.5194346289752646e-05,
+      "loss": 3.0241,
+      "step": 3007
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 3.508833922261484e-05,
+      "loss": 2.988,
+      "step": 3008
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 3.498233215547703e-05,
+      "loss": 3.1487,
+      "step": 3009
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 3.487632508833922e-05,
+      "loss": 2.9634,
+      "step": 3010
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 3.477031802120141e-05,
+      "loss": 3.008,
+      "step": 3011
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 3.4664310954063604e-05,
+      "loss": 3.1137,
+      "step": 3012
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 3.455830388692579e-05,
+      "loss": 3.0338,
+      "step": 3013
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 3.445229681978798e-05,
+      "loss": 2.9772,
+      "step": 3014
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 3.434628975265018e-05,
+      "loss": 3.0267,
+      "step": 3015
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 3.4240282685512366e-05,
+      "loss": 3.0636,
+      "step": 3016
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 3.4134275618374554e-05,
+      "loss": 3.0293,
+      "step": 3017
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 3.402826855123675e-05,
+      "loss": 2.9965,
+      "step": 3018
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 3.392226148409894e-05,
+      "loss": 2.8825,
+      "step": 3019
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 3.381625441696113e-05,
+      "loss": 3.1024,
+      "step": 3020
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 3.3710247349823316e-05,
+      "loss": 2.9456,
+      "step": 3021
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 3.360424028268551e-05,
+      "loss": 2.8154,
+      "step": 3022
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 3.34982332155477e-05,
+      "loss": 3.1612,
+      "step": 3023
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 3.339222614840989e-05,
+      "loss": 3.1105,
+      "step": 3024
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 3.3286219081272085e-05,
+      "loss": 3.0415,
+      "step": 3025
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 3.3180212014134274e-05,
+      "loss": 2.9697,
+      "step": 3026
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 3.307420494699646e-05,
+      "loss": 3.1358,
+      "step": 3027
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 3.296819787985865e-05,
+      "loss": 3.0735,
+      "step": 3028
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 3.2862190812720847e-05,
+      "loss": 2.8724,
+      "step": 3029
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 3.2756183745583035e-05,
+      "loss": 3.0852,
+      "step": 3030
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 3.2650176678445224e-05,
+      "loss": 3.0942,
+      "step": 3031
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 3.254416961130742e-05,
+      "loss": 3.0706,
+      "step": 3032
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 3.243816254416961e-05,
+      "loss": 3.0099,
+      "step": 3033
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 3.23321554770318e-05,
+      "loss": 3.1059,
+      "step": 3034
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 3.2226148409893986e-05,
+      "loss": 3.0452,
+      "step": 3035
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 3.212014134275618e-05,
+      "loss": 3.0549,
+      "step": 3036
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 3.201413427561837e-05,
+      "loss": 2.9988,
+      "step": 3037
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 3.190812720848056e-05,
+      "loss": 2.9597,
+      "step": 3038
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 3.1802120141342755e-05,
+      "loss": 2.9232,
+      "step": 3039
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 3.169611307420494e-05,
+      "loss": 2.9389,
+      "step": 3040
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 3.159010600706714e-05,
+      "loss": 2.9908,
+      "step": 3041
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 3.148409893992932e-05,
+      "loss": 3.1306,
+      "step": 3042
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 3.1378091872791516e-05,
+      "loss": 3.023,
+      "step": 3043
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 3.127208480565371e-05,
+      "loss": 3.0,
+      "step": 3044
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 3.11660777385159e-05,
+      "loss": 3.2194,
+      "step": 3045
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 3.106007067137809e-05,
+      "loss": 3.0307,
+      "step": 3046
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 3.095406360424028e-05,
+      "loss": 2.9635,
+      "step": 3047
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 3.0848056537102474e-05,
+      "loss": 3.1697,
+      "step": 3048
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 3.074204946996466e-05,
+      "loss": 3.1802,
+      "step": 3049
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 3.063604240282685e-05,
+      "loss": 3.0849,
+      "step": 3050
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 3.053003533568905e-05,
+      "loss": 2.9766,
+      "step": 3051
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 3.0424028268551232e-05,
+      "loss": 3.2149,
+      "step": 3052
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 3.0318021201413424e-05,
+      "loss": 3.2048,
+      "step": 3053
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 3.0212014134275613e-05,
+      "loss": 3.0065,
+      "step": 3054
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 3.0106007067137805e-05,
+      "loss": 3.1239,
+      "step": 3055
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 3.0997,
+      "step": 3056
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 2.9893992932862186e-05,
+      "loss": 3.0628,
+      "step": 3057
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 2.978798586572438e-05,
+      "loss": 3.1694,
+      "step": 3058
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 2.9681978798586567e-05,
+      "loss": 3.0683,
+      "step": 3059
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 2.9575971731448763e-05,
+      "loss": 3.0883,
+      "step": 3060
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 2.9469964664310955e-05,
+      "loss": 3.0237,
+      "step": 3061
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 2.9363957597173144e-05,
+      "loss": 3.1606,
+      "step": 3062
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 2.9257950530035336e-05,
+      "loss": 3.1112,
+      "step": 3063
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 2.9151943462897524e-05,
+      "loss": 3.0229,
+      "step": 3064
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 2.9045936395759717e-05,
+      "loss": 2.9163,
+      "step": 3065
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 2.8939929328621905e-05,
+      "loss": 3.0613,
+      "step": 3066
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 2.8833922261484097e-05,
+      "loss": 3.0797,
+      "step": 3067
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 2.872791519434629e-05,
+      "loss": 3.1091,
+      "step": 3068
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 2.862190812720848e-05,
+      "loss": 3.0541,
+      "step": 3069
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 2.851590106007067e-05,
+      "loss": 3.0552,
+      "step": 3070
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 2.840989399293286e-05,
+      "loss": 3.0531,
+      "step": 3071
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 2.830388692579505e-05,
+      "loss": 2.8543,
+      "step": 3072
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 2.819787985865724e-05,
+      "loss": 3.15,
+      "step": 3073
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 2.8091872791519432e-05,
+      "loss": 3.1063,
+      "step": 3074
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 2.7985865724381625e-05,
+      "loss": 3.0732,
+      "step": 3075
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 2.7879858657243813e-05,
+      "loss": 2.8793,
+      "step": 3076
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 2.7773851590106005e-05,
+      "loss": 3.0408,
+      "step": 3077
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 2.7667844522968194e-05,
+      "loss": 2.9924,
+      "step": 3078
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 2.7561837455830386e-05,
+      "loss": 2.9692,
+      "step": 3079
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 2.745583038869258e-05,
+      "loss": 3.0357,
+      "step": 3080
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 2.7349823321554767e-05,
+      "loss": 3.0755,
+      "step": 3081
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 2.724381625441696e-05,
+      "loss": 3.002,
+      "step": 3082
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 2.7137809187279148e-05,
+      "loss": 2.9077,
+      "step": 3083
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 2.703180212014134e-05,
+      "loss": 3.1129,
+      "step": 3084
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 2.692579505300353e-05,
+      "loss": 3.0254,
+      "step": 3085
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 2.681978798586572e-05,
+      "loss": 2.9293,
+      "step": 3086
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 2.6713780918727913e-05,
+      "loss": 3.0853,
+      "step": 3087
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 2.6607773851590102e-05,
+      "loss": 2.9346,
+      "step": 3088
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 2.6501766784452294e-05,
+      "loss": 3.1126,
+      "step": 3089
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 2.6395759717314483e-05,
+      "loss": 2.9547,
+      "step": 3090
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 2.6289752650176675e-05,
+      "loss": 3.036,
+      "step": 3091
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 2.6183745583038867e-05,
+      "loss": 3.0065,
+      "step": 3092
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 2.6077738515901056e-05,
+      "loss": 3.0024,
+      "step": 3093
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 2.5971731448763248e-05,
+      "loss": 3.0273,
+      "step": 3094
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 2.5865724381625437e-05,
+      "loss": 3.1141,
+      "step": 3095
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 2.5759717314487633e-05,
+      "loss": 3.0644,
+      "step": 3096
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 2.5653710247349818e-05,
+      "loss": 2.9754,
+      "step": 3097
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 2.5547703180212014e-05,
+      "loss": 3.2264,
+      "step": 3098
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 2.5441696113074206e-05,
+      "loss": 3.1976,
+      "step": 3099
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 2.5335689045936394e-05,
+      "loss": 3.0197,
+      "step": 3100
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 2.5229681978798587e-05,
+      "loss": 2.8941,
+      "step": 3101
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 2.5123674911660775e-05,
+      "loss": 3.1294,
+      "step": 3102
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 2.5017667844522967e-05,
+      "loss": 3.196,
+      "step": 3103
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 2.4911660777385156e-05,
+      "loss": 2.9556,
+      "step": 3104
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 2.480565371024735e-05,
+      "loss": 3.0865,
+      "step": 3105
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 2.469964664310954e-05,
+      "loss": 3.0376,
+      "step": 3106
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 2.459363957597173e-05,
+      "loss": 3.0103,
+      "step": 3107
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 2.448763250883392e-05,
+      "loss": 3.0901,
+      "step": 3108
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 2.438162544169611e-05,
+      "loss": 4.6456,
+      "step": 3109
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 2.4275618374558302e-05,
+      "loss": 3.0941,
+      "step": 3110
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 2.4169611307420495e-05,
+      "loss": 2.9554,
+      "step": 3111
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 2.4063604240282683e-05,
+      "loss": 2.8577,
+      "step": 3112
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 2.3957597173144875e-05,
+      "loss": 3.1149,
+      "step": 3113
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 2.3851590106007064e-05,
+      "loss": 3.1068,
+      "step": 3114
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 2.3745583038869256e-05,
+      "loss": 2.9608,
+      "step": 3115
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 2.3639575971731445e-05,
+      "loss": 3.1093,
+      "step": 3116
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 2.3533568904593637e-05,
+      "loss": 2.9863,
+      "step": 3117
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 2.342756183745583e-05,
+      "loss": 2.9977,
+      "step": 3118
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 2.3321554770318018e-05,
+      "loss": 2.9239,
+      "step": 3119
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 2.321554770318021e-05,
+      "loss": 3.0137,
+      "step": 3120
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 2.31095406360424e-05,
+      "loss": 3.0736,
+      "step": 3121
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 2.300353356890459e-05,
+      "loss": 2.9382,
+      "step": 3122
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 2.2897526501766783e-05,
+      "loss": 3.0285,
+      "step": 3123
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 2.2791519434628972e-05,
+      "loss": 2.997,
+      "step": 3124
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 2.2685512367491164e-05,
+      "loss": 2.9922,
+      "step": 3125
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 2.2579505300353353e-05,
+      "loss": 2.8958,
+      "step": 3126
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 2.2473498233215545e-05,
+      "loss": 2.9947,
+      "step": 3127
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 2.2367491166077734e-05,
+      "loss": 3.0986,
+      "step": 3128
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 2.2261484098939926e-05,
+      "loss": 2.9888,
+      "step": 3129
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 2.2155477031802118e-05,
+      "loss": 2.9069,
+      "step": 3130
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 2.2049469964664307e-05,
+      "loss": 3.1772,
+      "step": 3131
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 2.19434628975265e-05,
+      "loss": 3.0329,
+      "step": 3132
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 2.1837455830388688e-05,
+      "loss": 2.8865,
+      "step": 3133
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 2.1731448763250883e-05,
+      "loss": 3.1056,
+      "step": 3134
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 2.162544169611307e-05,
+      "loss": 3.0004,
+      "step": 3135
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 2.1519434628975264e-05,
+      "loss": 3.0758,
+      "step": 3136
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 2.1413427561837457e-05,
+      "loss": 2.9413,
+      "step": 3137
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 2.1307420494699645e-05,
+      "loss": 3.0341,
+      "step": 3138
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 2.1201413427561837e-05,
+      "loss": 3.0874,
+      "step": 3139
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 2.1095406360424026e-05,
+      "loss": 2.8475,
+      "step": 3140
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 2.098939929328622e-05,
+      "loss": 3.0307,
+      "step": 3141
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 2.088339222614841e-05,
+      "loss": 3.0402,
+      "step": 3142
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 2.07773851590106e-05,
+      "loss": 2.8843,
+      "step": 3143
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 2.067137809187279e-05,
+      "loss": 2.8826,
+      "step": 3144
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 2.056537102473498e-05,
+      "loss": 3.0764,
+      "step": 3145
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 2.0459363957597172e-05,
+      "loss": 2.9719,
+      "step": 3146
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 2.035335689045936e-05,
+      "loss": 3.0607,
+      "step": 3147
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 2.0247349823321553e-05,
+      "loss": 3.0717,
+      "step": 3148
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 2.0141342756183745e-05,
+      "loss": 3.0239,
+      "step": 3149
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 2.0035335689045934e-05,
+      "loss": 3.0434,
+      "step": 3150
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 1.9929328621908126e-05,
+      "loss": 2.9394,
+      "step": 3151
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 1.9823321554770315e-05,
+      "loss": 2.9773,
+      "step": 3152
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 1.9717314487632507e-05,
+      "loss": 2.9522,
+      "step": 3153
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 1.96113074204947e-05,
+      "loss": 2.875,
+      "step": 3154
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 1.9505300353356888e-05,
+      "loss": 2.8619,
+      "step": 3155
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 1.939929328621908e-05,
+      "loss": 2.9834,
+      "step": 3156
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 1.929328621908127e-05,
+      "loss": 2.9856,
+      "step": 3157
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 1.918727915194346e-05,
+      "loss": 2.7966,
+      "step": 3158
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 1.908127208480565e-05,
+      "loss": 3.0303,
+      "step": 3159
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 1.8975265017667842e-05,
+      "loss": 3.0651,
+      "step": 3160
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 1.8869257950530034e-05,
+      "loss": 2.9677,
+      "step": 3161
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 1.8763250883392223e-05,
+      "loss": 2.8925,
+      "step": 3162
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 1.8657243816254415e-05,
+      "loss": 3.0567,
+      "step": 3163
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 1.8551236749116607e-05,
+      "loss": 2.9366,
+      "step": 3164
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 1.8445229681978796e-05,
+      "loss": 2.9796,
+      "step": 3165
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 1.8339222614840988e-05,
+      "loss": 3.0526,
+      "step": 3166
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 1.8233215547703177e-05,
+      "loss": 3.0476,
+      "step": 3167
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 1.812720848056537e-05,
+      "loss": 2.9343,
+      "step": 3168
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 1.8021201413427558e-05,
+      "loss": 2.9384,
+      "step": 3169
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 1.7915194346289753e-05,
+      "loss": 3.0812,
+      "step": 3170
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 1.7809187279151942e-05,
+      "loss": 2.9255,
+      "step": 3171
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 1.7703180212014134e-05,
+      "loss": 2.9277,
+      "step": 3172
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 1.7597173144876323e-05,
+      "loss": 3.0208,
+      "step": 3173
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 1.7491166077738515e-05,
+      "loss": 2.9575,
+      "step": 3174
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 1.7385159010600704e-05,
+      "loss": 3.0095,
+      "step": 3175
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 1.7279151943462896e-05,
+      "loss": 2.9226,
+      "step": 3176
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 1.717314487632509e-05,
+      "loss": 3.0523,
+      "step": 3177
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 1.7067137809187277e-05,
+      "loss": 2.9905,
+      "step": 3178
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 1.696113074204947e-05,
+      "loss": 2.9764,
+      "step": 3179
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 1.6855123674911658e-05,
+      "loss": 2.9299,
+      "step": 3180
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 1.674911660777385e-05,
+      "loss": 3.1068,
+      "step": 3181
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 1.6643109540636042e-05,
+      "loss": 3.0189,
+      "step": 3182
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 1.653710247349823e-05,
+      "loss": 2.8015,
+      "step": 3183
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 1.6431095406360423e-05,
+      "loss": 3.0141,
+      "step": 3184
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 1.6325088339222612e-05,
+      "loss": 2.9598,
+      "step": 3185
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 1.6219081272084804e-05,
+      "loss": 2.901,
+      "step": 3186
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 1.6113074204946993e-05,
+      "loss": 2.881,
+      "step": 3187
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 1.6007067137809185e-05,
+      "loss": 3.0659,
+      "step": 3188
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 1.5901060070671377e-05,
+      "loss": 2.9852,
+      "step": 3189
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 1.579505300353357e-05,
+      "loss": 2.7604,
+      "step": 3190
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 1.5689045936395758e-05,
+      "loss": 3.084,
+      "step": 3191
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 1.558303886925795e-05,
+      "loss": 3.1153,
+      "step": 3192
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 1.547703180212014e-05,
+      "loss": 2.8912,
+      "step": 3193
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 1.537102473498233e-05,
+      "loss": 2.9513,
+      "step": 3194
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 1.5265017667844523e-05,
+      "loss": 3.0449,
+      "step": 3195
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 1.5159010600706712e-05,
+      "loss": 2.8979,
+      "step": 3196
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 1.5053003533568903e-05,
+      "loss": 2.8967,
+      "step": 3197
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 1.4946996466431093e-05,
+      "loss": 3.0743,
+      "step": 3198
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 1.4840989399293284e-05,
+      "loss": 2.9308,
+      "step": 3199
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 1.4734982332155477e-05,
+      "loss": 2.9344,
+      "step": 3200
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 1.4628975265017668e-05,
+      "loss": 2.8775,
+      "step": 3201
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 1.4522968197879858e-05,
+      "loss": 2.9786,
+      "step": 3202
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 1.4416961130742049e-05,
+      "loss": 2.9942,
+      "step": 3203
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 1.431095406360424e-05,
+      "loss": 2.8866,
+      "step": 3204
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 1.420494699646643e-05,
+      "loss": 2.9844,
+      "step": 3205
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 1.409893992932862e-05,
+      "loss": 3.0987,
+      "step": 3206
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 1.3992932862190812e-05,
+      "loss": 2.9472,
+      "step": 3207
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 1.3886925795053003e-05,
+      "loss": 2.7832,
+      "step": 3208
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 1.3780918727915193e-05,
+      "loss": 2.9734,
+      "step": 3209
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 1.3674911660777384e-05,
+      "loss": 3.0497,
+      "step": 3210
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 1.3568904593639574e-05,
+      "loss": 2.984,
+      "step": 3211
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 1.3462897526501765e-05,
+      "loss": 3.0108,
+      "step": 3212
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 1.3356890459363957e-05,
+      "loss": 3.0669,
+      "step": 3213
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 1.3250883392226147e-05,
+      "loss": 2.9398,
+      "step": 3214
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 1.3144876325088338e-05,
+      "loss": 2.893,
+      "step": 3215
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 1.3038869257950528e-05,
+      "loss": 3.0166,
+      "step": 3216
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 1.2932862190812719e-05,
+      "loss": 3.0122,
+      "step": 3217
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 1.2826855123674909e-05,
+      "loss": 3.0066,
+      "step": 3218
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 1.2720848056537103e-05,
+      "loss": 2.9157,
+      "step": 3219
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 1.2614840989399293e-05,
+      "loss": 4.4977,
+      "step": 3220
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 1.2508833922261484e-05,
+      "loss": 3.0029,
+      "step": 3221
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 1.2402826855123674e-05,
+      "loss": 2.967,
+      "step": 3222
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 1.2296819787985865e-05,
+      "loss": 2.8858,
+      "step": 3223
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 1.2190812720848055e-05,
+      "loss": 3.087,
+      "step": 3224
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 1.2084805653710247e-05,
+      "loss": 3.0952,
+      "step": 3225
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 1.1978798586572438e-05,
+      "loss": 2.8562,
+      "step": 3226
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 1.1872791519434628e-05,
+      "loss": 2.934,
+      "step": 3227
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 1.1766784452296819e-05,
+      "loss": 3.0283,
+      "step": 3228
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 1.1660777385159009e-05,
+      "loss": 2.917,
+      "step": 3229
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 1.15547703180212e-05,
+      "loss": 2.7824,
+      "step": 3230
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 1.1448763250883392e-05,
+      "loss": 2.9767,
+      "step": 3231
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 1.1342756183745582e-05,
+      "loss": 2.8672,
+      "step": 3232
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 1.1236749116607773e-05,
+      "loss": 2.9382,
+      "step": 3233
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 1.1130742049469963e-05,
+      "loss": 2.9956,
+      "step": 3234
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 1.1024734982332154e-05,
+      "loss": 2.9874,
+      "step": 3235
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 1.0918727915194344e-05,
+      "loss": 2.9604,
+      "step": 3236
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 1.0812720848056534e-05,
+      "loss": 2.8458,
+      "step": 3237
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 1.0706713780918728e-05,
+      "loss": 2.9667,
+      "step": 3238
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 1.0600706713780919e-05,
+      "loss": 2.9734,
+      "step": 3239
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 1.049469964664311e-05,
+      "loss": 2.8578,
+      "step": 3240
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 1.03886925795053e-05,
+      "loss": 2.841,
+      "step": 3241
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.028268551236749e-05,
+      "loss": 2.9833,
+      "step": 3242
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.017667844522968e-05,
+      "loss": 3.0091,
+      "step": 3243
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.0070671378091873e-05,
+      "loss": 2.7913,
+      "step": 3244
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 9.964664310954063e-06,
+      "loss": 2.9679,
+      "step": 3245
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 9.858657243816254e-06,
+      "loss": 3.0111,
+      "step": 3246
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 9.752650176678444e-06,
+      "loss": 2.834,
+      "step": 3247
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 9.646643109540635e-06,
+      "loss": 2.8931,
+      "step": 3248
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 9.540636042402825e-06,
+      "loss": 3.0909,
+      "step": 3249
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 9.434628975265017e-06,
+      "loss": 2.9526,
+      "step": 3250
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 9.328621908127208e-06,
+      "loss": 2.7327,
+      "step": 3251
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 9.222614840989398e-06,
+      "loss": 3.0511,
+      "step": 3252
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 9.116607773851589e-06,
+      "loss": 3.0031,
+      "step": 3253
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 9.010600706713779e-06,
+      "loss": 2.8639,
+      "step": 3254
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 8.904593639575971e-06,
+      "loss": 2.9074,
+      "step": 3255
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 8.798586572438162e-06,
+      "loss": 3.0456,
+      "step": 3256
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 8.692579505300352e-06,
+      "loss": 3.0029,
+      "step": 3257
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 8.586572438162544e-06,
+      "loss": 2.8942,
+      "step": 3258
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 8.480565371024735e-06,
+      "loss": 3.1312,
+      "step": 3259
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 8.374558303886925e-06,
+      "loss": 2.9629,
+      "step": 3260
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 8.268551236749116e-06,
+      "loss": 2.9841,
+      "step": 3261
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 8.162544169611306e-06,
+      "loss": 2.865,
+      "step": 3262
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 8.056537102473496e-06,
+      "loss": 2.987,
+      "step": 3263
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 7.950530035335689e-06,
+      "loss": 2.9917,
+      "step": 3264
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 7.844522968197879e-06,
+      "loss": 3.0013,
+      "step": 3265
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 7.73851590106007e-06,
+      "loss": 2.9072,
+      "step": 3266
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 7.632508833922262e-06,
+      "loss": 2.997,
+      "step": 3267
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 7.526501766784451e-06,
+      "loss": 2.955,
+      "step": 3268
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 7.420494699646642e-06,
+      "loss": 2.8441,
+      "step": 3269
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 7.314487632508834e-06,
+      "loss": 3.0565,
+      "step": 3270
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 7.208480565371024e-06,
+      "loss": 2.9889,
+      "step": 3271
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 7.102473498233215e-06,
+      "loss": 2.9812,
+      "step": 3272
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 6.996466431095406e-06,
+      "loss": 2.8178,
+      "step": 3273
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 6.890459363957597e-06,
+      "loss": 2.9941,
+      "step": 3274
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 6.784452296819787e-06,
+      "loss": 3.0175,
+      "step": 3275
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 6.678445229681978e-06,
+      "loss": 2.8629,
+      "step": 3276
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 6.572438162544169e-06,
+      "loss": 2.967,
+      "step": 3277
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 6.466431095406359e-06,
+      "loss": 3.0145,
+      "step": 3278
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 6.360424028268551e-06,
+      "loss": 2.9512,
+      "step": 3279
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 6.254416961130742e-06,
+      "loss": 2.9375,
+      "step": 3280
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 6.148409893992932e-06,
+      "loss": 3.0846,
+      "step": 3281
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 6.042402826855124e-06,
+      "loss": 2.9733,
+      "step": 3282
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 5.936395759717314e-06,
+      "loss": 2.8497,
+      "step": 3283
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 5.8303886925795045e-06,
+      "loss": 3.0611,
+      "step": 3284
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 5.724381625441696e-06,
+      "loss": 2.964,
+      "step": 3285
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 5.618374558303886e-06,
+      "loss": 2.9852,
+      "step": 3286
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 5.512367491166077e-06,
+      "loss": 2.8856,
+      "step": 3287
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 5.406360424028267e-06,
+      "loss": 3.0137,
+      "step": 3288
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 5.300353356890459e-06,
+      "loss": 2.9812,
+      "step": 3289
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 5.19434628975265e-06,
+      "loss": 2.9039,
+      "step": 3290
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 5.08833922261484e-06,
+      "loss": 2.9469,
+      "step": 3291
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 4.982332155477032e-06,
+      "loss": 2.9484,
+      "step": 3292
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 4.876325088339222e-06,
+      "loss": 2.9197,
+      "step": 3293
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 4.7703180212014125e-06,
+      "loss": 2.8554,
+      "step": 3294
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 4.664310954063604e-06,
+      "loss": 3.089,
+      "step": 3295
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 4.558303886925794e-06,
+      "loss": 3.0325,
+      "step": 3296
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 4.4522968197879856e-06,
+      "loss": 2.9911,
+      "step": 3297
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 4.346289752650176e-06,
+      "loss": 2.7905,
+      "step": 3298
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 4.240282685512367e-06,
+      "loss": 3.0218,
+      "step": 3299
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 4.134275618374558e-06,
+      "loss": 2.9269,
+      "step": 3300
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 4.028268551236748e-06,
+      "loss": 2.8031,
+      "step": 3301
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 3.9222614840989395e-06,
+      "loss": 3.0996,
+      "step": 3302
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 3.816254416961131e-06,
+      "loss": 3.0674,
+      "step": 3303
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 3.710247349823321e-06,
+      "loss": 2.9776,
+      "step": 3304
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 3.604240282685512e-06,
+      "loss": 2.8922,
+      "step": 3305
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 3.498233215547703e-06,
+      "loss": 3.0246,
+      "step": 3306
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 3.3922261484098935e-06,
+      "loss": 2.9513,
+      "step": 3307
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 3.2862190812720844e-06,
+      "loss": 2.9434,
+      "step": 3308
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 3.1802120141342757e-06,
+      "loss": 2.962,
+      "step": 3309
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 3.074204946996466e-06,
+      "loss": 2.9459,
+      "step": 3310
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 2.968197879858657e-06,
+      "loss": 3.0799,
+      "step": 3311
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 2.862190812720848e-06,
+      "loss": 2.9247,
+      "step": 3312
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 2.7561837455830384e-06,
+      "loss": 3.0599,
+      "step": 3313
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 2.6501766784452297e-06,
+      "loss": 2.9877,
+      "step": 3314
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 2.54416961130742e-06,
+      "loss": 2.8924,
+      "step": 3315
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 2.438162544169611e-06,
+      "loss": 2.9469,
+      "step": 3316
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 2.332155477031802e-06,
+      "loss": 2.9196,
+      "step": 3317
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.2261484098939928e-06,
+      "loss": 2.9614,
+      "step": 3318
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.1201413427561837e-06,
+      "loss": 2.8117,
+      "step": 3319
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.014134275618374e-06,
+      "loss": 2.9932,
+      "step": 3320
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 1.9081272084805654e-06,
+      "loss": 2.9385,
+      "step": 3321
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 1.802120141342756e-06,
+      "loss": 2.8995,
+      "step": 3322
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.6961130742049468e-06,
+      "loss": 2.8314,
+      "step": 3323
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.5901060070671379e-06,
+      "loss": 2.9494,
+      "step": 3324
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.4840989399293285e-06,
+      "loss": 3.0582,
+      "step": 3325
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 1.3780918727915192e-06,
+      "loss": 2.7562,
+      "step": 3326
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 1.27208480565371e-06,
+      "loss": 2.9327,
+      "step": 3327
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 1.166077738515901e-06,
+      "loss": 3.0293,
+      "step": 3328
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 1.0600706713780918e-06,
+      "loss": 2.9029,
+      "step": 3329
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 9.540636042402827e-07,
+      "loss": 2.938,
+      "step": 3330
+    },
+    {
+      "epoch": 30.0,
+      "step": 3330,
+      "total_flos": 5.990701911701811e+20,
+      "train_loss": 4.37666116677247,
+      "train_runtime": 28536.6396,
+      "train_samples_per_second": 30.001,
+      "train_steps_per_second": 0.117
+    }
+  ],
+  "max_steps": 3330,
+  "num_train_epochs": 30,
+  "total_flos": 5.990701911701811e+20,
+  "trial_name": null,
+  "trial_params": null
+}