diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,29266 @@
+{
+  "best_metric": 1.471304930449711e-13,
+  "best_model_checkpoint": "./checkpoint-1000",
+  "epoch": 2.999922845459455,
+  "global_step": 4860,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 9.5135,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 9.8103,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 9.8101,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-07,
+      "loss": 10.1898,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-06,
+      "loss": 9.8821,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-06,
+      "loss": 9.6124,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-06,
+      "loss": 9.9878,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-06,
+      "loss": 9.1242,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-06,
+      "loss": 9.1298,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-06,
+      "loss": 9.2409,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-06,
+      "loss": 9.132,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999999999999997e-06,
+      "loss": 9.4073,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-06,
+      "loss": 8.771,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-06,
+      "loss": 8.6399,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8e-06,
+      "loss": 7.7917,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.399999999999999e-06,
+      "loss": 7.6343,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 7.1902,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.599999999999999e-06,
+      "loss": 6.8994,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2e-06,
+      "loss": 6.5587,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.799999999999998e-06,
+      "loss": 6.6495,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.4e-06,
+      "loss": 6.097,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.999999999999999e-06,
+      "loss": 5.8177,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.6e-06,
+      "loss": 5.8191,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.02e-05,
+      "loss": 5.7257,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0799999999999998e-05,
+      "loss": 5.6024,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.14e-05,
+      "loss": 5.3748,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 5.379,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.26e-05,
+      "loss": 5.3335,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3199999999999997e-05,
+      "loss": 5.0844,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3799999999999998e-05,
+      "loss": 5.1396,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.44e-05,
+      "loss": 5.1483,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 4.7176,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5599999999999996e-05,
+      "loss": 4.8332,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6199999999999997e-05,
+      "loss": 4.8118,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.68e-05,
+      "loss": 4.6633,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.74e-05,
+      "loss": 4.6537,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 4.5941,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8599999999999998e-05,
+      "loss": 4.6503,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.92e-05,
+      "loss": 4.5184,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.98e-05,
+      "loss": 4.4501,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.04e-05,
+      "loss": 4.3468,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1e-05,
+      "loss": 4.4475,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1599999999999996e-05,
+      "loss": 4.3442,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.2199999999999998e-05,
+      "loss": 4.0437,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.28e-05,
+      "loss": 4.0322,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.34e-05,
+      "loss": 3.8393,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 3.7762,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 3.514,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.52e-05,
+      "loss": 3.4973,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5799999999999997e-05,
+      "loss": 3.3104,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6399999999999995e-05,
+      "loss": 6.5173,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 6.3584,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7599999999999997e-05,
+      "loss": 5.862,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 5.544,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.88e-05,
+      "loss": 5.3065,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.94e-05,
+      "loss": 5.209,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 5.3224,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.06e-05,
+      "loss": 5.1511,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.119999999999999e-05,
+      "loss": 5.1337,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.1799999999999994e-05,
+      "loss": 4.9414,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.2399999999999995e-05,
+      "loss": 4.8724,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 4.7764,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.36e-05,
+      "loss": 4.6878,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.42e-05,
+      "loss": 4.7272,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.48e-05,
+      "loss": 4.7984,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.539999999999999e-05,
+      "loss": 4.5159,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 4.8491,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6599999999999995e-05,
+      "loss": 4.6092,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7199999999999996e-05,
+      "loss": 4.5272,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.78e-05,
+      "loss": 4.5579,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.84e-05,
+      "loss": 4.4117,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9e-05,
+      "loss": 4.5083,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.96e-05,
+      "loss": 4.555,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.02e-05,
+      "loss": 4.2987,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.08e-05,
+      "loss": 4.3312,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.14e-05,
+      "loss": 4.3045,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.2e-05,
+      "loss": 4.379,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.259999999999999e-05,
+      "loss": 4.3704,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.319999999999999e-05,
+      "loss": 4.3909,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.3799999999999994e-05,
+      "loss": 4.2669,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.4399999999999995e-05,
+      "loss": 4.0042,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 4.2425,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.56e-05,
+      "loss": 4.0899,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.62e-05,
+      "loss": 4.1337,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.68e-05,
+      "loss": 4.0863,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7399999999999993e-05,
+      "loss": 4.0835,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 3.9743,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8599999999999995e-05,
+      "loss": 4.0583,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9199999999999997e-05,
+      "loss": 3.9071,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98e-05,
+      "loss": 3.9489,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.04e-05,
+      "loss": 3.97,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.1e-05,
+      "loss": 3.7487,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.1599999999999994e-05,
+      "loss": 3.7244,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.2199999999999995e-05,
+      "loss": 3.6222,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.279999999999999e-05,
+      "loss": 3.6013,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.339999999999999e-05,
+      "loss": 3.4787,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 3.3783,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.459999999999999e-05,
+      "loss": 3.392,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.519999999999999e-05,
+      "loss": 3.0215,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5799999999999994e-05,
+      "loss": 2.9419,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 6.114,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 5.5252,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.76e-05,
+      "loss": 5.0695,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.82e-05,
+      "loss": 4.9737,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.88e-05,
+      "loss": 4.8192,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.94e-05,
+      "loss": 4.6852,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 4.8017,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.0599999999999996e-05,
+      "loss": 4.6657,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.12e-05,
+      "loss": 4.6415,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.18e-05,
+      "loss": 4.4974,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.239999999999999e-05,
+      "loss": 4.4745,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.299999999999999e-05,
+      "loss": 4.4467,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.359999999999999e-05,
+      "loss": 4.4226,
+      "step": 113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.419999999999999e-05,
+      "loss": 4.4979,
+      "step": 114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.479999999999999e-05,
+      "loss": 4.2777,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.539999999999999e-05,
+      "loss": 4.4617,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 4.3205,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.659999999999999e-05,
+      "loss": 4.2775,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.72e-05,
+      "loss": 4.3194,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.78e-05,
+      "loss": 4.319,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.84e-05,
+      "loss": 4.38,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.9e-05,
+      "loss": 4.3715,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.96e-05,
+      "loss": 4.5273,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.02e-05,
+      "loss": 4.3016,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.079999999999999e-05,
+      "loss": 4.2581,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.139999999999999e-05,
+      "loss": 4.2407,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 4.2266,
+      "step": 127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.259999999999999e-05,
+      "loss": 4.1345,
+      "step": 128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.319999999999999e-05,
+      "loss": 4.255,
+      "step": 129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.379999999999999e-05,
+      "loss": 4.1264,
+      "step": 130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.439999999999999e-05,
+      "loss": 4.1036,
+      "step": 131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.5e-05,
+      "loss": 4.0236,
+      "step": 132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.56e-05,
+      "loss": 4.1043,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.62e-05,
+      "loss": 4.0537,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.68e-05,
+      "loss": 4.0848,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.74e-05,
+      "loss": 4.0079,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.8e-05,
+      "loss": 3.9205,
+      "step": 137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.86e-05,
+      "loss": 3.9029,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.92e-05,
+      "loss": 3.918,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.98e-05,
+      "loss": 3.9781,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.04e-05,
+      "loss": 3.7566,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.1e-05,
+      "loss": 3.7063,
+      "step": 142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.16e-05,
+      "loss": 3.5859,
+      "step": 143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.22e-05,
+      "loss": 3.3238,
+      "step": 144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.28e-05,
+      "loss": 3.3694,
+      "step": 145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.34e-05,
+      "loss": 3.3059,
+      "step": 146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.4e-05,
+      "loss": 3.2696,
+      "step": 147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.459999999999998e-05,
+      "loss": 3.1954,
+      "step": 148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.519999999999998e-05,
+      "loss": 2.932,
+      "step": 149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.579999999999998e-05,
+      "loss": 2.9027,
+      "step": 150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.639999999999999e-05,
+      "loss": 5.9382,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 5.2861,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.759999999999999e-05,
+      "loss": 4.9462,
+      "step": 153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.819999999999999e-05,
+      "loss": 4.7416,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.879999999999999e-05,
+      "loss": 4.6738,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.939999999999999e-05,
+      "loss": 4.6159,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 4.661,
+      "step": 157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.059999999999999e-05,
+      "loss": 4.5277,
+      "step": 158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.12e-05,
+      "loss": 4.3633,
+      "step": 159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.18e-05,
+      "loss": 4.3865,
+      "step": 160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.24e-05,
+      "loss": 4.397,
+      "step": 161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.3e-05,
+      "loss": 4.2932,
+      "step": 162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.36e-05,
+      "loss": 4.5437,
+      "step": 163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.419999999999999e-05,
+      "loss": 4.51,
+      "step": 164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.479999999999999e-05,
+      "loss": 4.3269,
+      "step": 165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.539999999999999e-05,
+      "loss": 4.3626,
+      "step": 166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 4.3732,
+      "step": 167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.659999999999999e-05,
+      "loss": 4.3292,
+      "step": 168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.719999999999999e-05,
+      "loss": 4.2986,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.779999999999999e-05,
+      "loss": 4.2012,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839999999999999e-05,
+      "loss": 4.1703,
+      "step": 171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.9e-05,
+      "loss": 4.2443,
+      "step": 172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.96e-05,
+      "loss": 4.2526,
+      "step": 173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001002,
+      "loss": 4.2244,
+      "step": 174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001008,
+      "loss": 4.1979,
+      "step": 175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001014,
+      "loss": 4.1536,
+      "step": 176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000102,
+      "loss": 4.2401,
+      "step": 177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001026,
+      "loss": 4.0987,
+      "step": 178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00010319999999999999,
+      "loss": 4.0979,
+      "step": 179
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00010379999999999999,
+      "loss": 4.0801,
+      "step": 180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00010439999999999999,
+      "loss": 4.1392,
+      "step": 181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00010499999999999999,
+      "loss": 4.0059,
+      "step": 182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00010559999999999998,
+      "loss": 4.051,
+      "step": 183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00010619999999999998,
+      "loss": 3.9285,
+      "step": 184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00010679999999999998,
+      "loss": 3.945,
+      "step": 185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00010739999999999998,
+      "loss": 3.8085,
+      "step": 186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 3.9593,
+      "step": 187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00010859999999999998,
+      "loss": 3.8777,
+      "step": 188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00010919999999999998,
+      "loss": 3.815,
+      "step": 189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00010979999999999999,
+      "loss": 3.7849,
+      "step": 190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00011039999999999999,
+      "loss": 3.6886,
+      "step": 191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00011099999999999999,
+      "loss": 3.5941,
+      "step": 192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00011159999999999999,
+      "loss": 3.4718,
+      "step": 193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00011219999999999999,
+      "loss": 3.355,
+      "step": 194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00011279999999999999,
+      "loss": 3.2773,
+      "step": 195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00011339999999999999,
+      "loss": 3.3025,
+      "step": 196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 3.2324,
+      "step": 197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001146,
+      "loss": 3.2263,
+      "step": 198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001152,
+      "loss": 2.8137,
+      "step": 199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001158,
+      "loss": 2.9228,
+      "step": 200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001164,
+      "loss": 5.8291,
+      "step": 201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000117,
+      "loss": 4.9896,
+      "step": 202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001176,
+      "loss": 4.513,
+      "step": 203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001182,
+      "loss": 4.6057,
+      "step": 204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001188,
+      "loss": 4.7012,
+      "step": 205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001194,
+      "loss": 4.4101,
+      "step": 206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 4.4692,
+      "step": 207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00012059999999999999,
+      "loss": 4.3521,
+      "step": 208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00012119999999999999,
+      "loss": 4.299,
+      "step": 209
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00012179999999999999,
+      "loss": 4.3289,
+      "step": 210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001224,
+      "loss": 4.3255,
+      "step": 211
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00012299999999999998,
+      "loss": 4.2024,
+      "step": 212
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001236,
+      "loss": 4.356,
+      "step": 213
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00012419999999999998,
+      "loss": 4.3464,
+      "step": 214
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00012479999999999997,
+      "loss": 4.2892,
+      "step": 215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00012539999999999999,
+      "loss": 4.2194,
+      "step": 216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 4.3643,
+      "step": 217
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001266,
+      "loss": 4.0856,
+      "step": 218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00012719999999999997,
+      "loss": 4.2346,
+      "step": 219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001278,
+      "loss": 4.2318,
+      "step": 220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00012839999999999998,
+      "loss": 4.1432,
+      "step": 221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000129,
+      "loss": 4.2051,
+      "step": 222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00012959999999999998,
+      "loss": 4.1461,
+      "step": 223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001302,
+      "loss": 4.1237,
+      "step": 224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013079999999999998,
+      "loss": 4.1349,
+      "step": 225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001314,
+      "loss": 4.0425,
+      "step": 226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 4.1628,
+      "step": 227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001326,
+      "loss": 4.0593,
+      "step": 228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00013319999999999999,
+      "loss": 4.0131,
+      "step": 229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001338,
+      "loss": 4.0298,
+      "step": 230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001344,
+      "loss": 3.9874,
+      "step": 231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000135,
+      "loss": 3.9562,
+      "step": 232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001356,
+      "loss": 3.9514,
+      "step": 233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001362,
+      "loss": 3.9638,
+      "step": 234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001368,
+      "loss": 4.055,
+      "step": 235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001374,
+      "loss": 3.9593,
+      "step": 236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000138,
+      "loss": 3.7526,
+      "step": 237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001386,
+      "loss": 3.8853,
+      "step": 238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001392,
+      "loss": 3.6557,
+      "step": 239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00013979999999999998,
+      "loss": 3.8447,
+      "step": 240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001404,
+      "loss": 3.6613,
+      "step": 241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00014099999999999998,
+      "loss": 3.6865,
+      "step": 242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00014159999999999997,
+      "loss": 3.5905,
+      "step": 243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001422,
+      "loss": 3.5122,
+      "step": 244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00014279999999999997,
+      "loss": 3.5215,
+      "step": 245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001434,
+      "loss": 3.2726,
+      "step": 246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 3.4131,
+      "step": 247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001446,
+      "loss": 3.0505,
+      "step": 248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00014519999999999998,
+      "loss": 2.9078,
+      "step": 249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001458,
+      "loss": 2.7024,
+      "step": 250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00014639999999999998,
+      "loss": 5.8578,
+      "step": 251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000147,
+      "loss": 5.3405,
+      "step": 252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00014759999999999998,
+      "loss": 4.6349,
+      "step": 253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001482,
+      "loss": 4.7555,
+      "step": 254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00014879999999999998,
+      "loss": 4.4432,
+      "step": 255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001494,
+      "loss": 4.4003,
+      "step": 256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00015,
+      "loss": 4.5716,
+      "step": 257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00015059999999999997,
+      "loss": 4.4934,
+      "step": 258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001512,
+      "loss": 4.3529,
+      "step": 259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00015179999999999998,
+      "loss": 4.3214,
+      "step": 260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001524,
+      "loss": 4.3262,
+      "step": 261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00015299999999999998,
+      "loss": 4.313,
+      "step": 262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001536,
+      "loss": 4.3297,
+      "step": 263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00015419999999999998,
+      "loss": 4.2544,
+      "step": 264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001548,
+      "loss": 4.1435,
+      "step": 265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00015539999999999998,
+      "loss": 4.1702,
+      "step": 266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000156,
+      "loss": 4.2813,
+      "step": 267
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00015659999999999998,
+      "loss": 4.0714,
+      "step": 268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001572,
+      "loss": 4.3009,
+      "step": 269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001578,
+      "loss": 4.2906,
+      "step": 270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001584,
+      "loss": 4.1794,
+      "step": 271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000159,
+      "loss": 4.0437,
+      "step": 272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001596,
+      "loss": 4.2904,
+      "step": 273
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001602,
+      "loss": 4.4145,
+      "step": 274
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001608,
+      "loss": 4.19,
+      "step": 275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001614,
+      "loss": 4.0709,
+      "step": 276
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000162,
+      "loss": 3.9692,
+      "step": 277
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001626,
+      "loss": 4.0098,
+      "step": 278
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001632,
+      "loss": 4.0831,
+      "step": 279
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001638,
+      "loss": 4.2022,
+      "step": 280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001644,
+      "loss": 4.0885,
+      "step": 281
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000165,
+      "loss": 3.9334,
+      "step": 282
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001656,
+      "loss": 3.838,
+      "step": 283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001662,
+      "loss": 4.0429,
+      "step": 284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001668,
+      "loss": 3.9703,
+      "step": 285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001674,
+      "loss": 4.0123,
+      "step": 286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000168,
+      "loss": 4.0291,
+      "step": 287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001686,
+      "loss": 3.9261,
+      "step": 288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00016919999999999997,
+      "loss": 3.8273,
+      "step": 289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00016979999999999998,
+      "loss": 3.7701,
+      "step": 290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00017039999999999997,
+      "loss": 3.833,
+      "step": 291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00017099999999999998,
+      "loss": 3.5294,
+      "step": 292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00017159999999999997,
+      "loss": 3.5094,
+      "step": 293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00017219999999999998,
+      "loss": 3.4339,
+      "step": 294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00017279999999999997,
+      "loss": 3.5833,
+      "step": 295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00017339999999999996,
+      "loss": 3.2647,
+      "step": 296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 3.2495,
+      "step": 297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00017459999999999996,
+      "loss": 3.2158,
+      "step": 298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00017519999999999998,
+      "loss": 3.2869,
+      "step": 299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017579999999999996,
+      "loss": 3.2838,
+      "step": 300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017639999999999998,
+      "loss": 6.0054,
+      "step": 301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017699999999999997,
+      "loss": 5.665,
+      "step": 302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017759999999999998,
+      "loss": 4.9385,
+      "step": 303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017819999999999997,
+      "loss": 4.7416,
+      "step": 304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017879999999999998,
+      "loss": 4.6585,
+      "step": 305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017939999999999997,
+      "loss": 4.783,
+      "step": 306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 4.4418,
+      "step": 307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018059999999999997,
+      "loss": 4.5525,
+      "step": 308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018119999999999999,
+      "loss": 4.3392,
+      "step": 309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018179999999999997,
+      "loss": 4.3896,
+      "step": 310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001824,
+      "loss": 4.3516,
+      "step": 311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018299999999999998,
+      "loss": 4.4005,
+      "step": 312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001836,
+      "loss": 4.3409,
+      "step": 313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018419999999999998,
+      "loss": 4.3136,
+      "step": 314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001848,
+      "loss": 4.3484,
+      "step": 315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018539999999999998,
+      "loss": 4.214,
+      "step": 316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000186,
+      "loss": 4.2314,
+      "step": 317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018659999999999998,
+      "loss": 4.1683,
+      "step": 318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001872,
+      "loss": 4.1335,
+      "step": 319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018779999999999998,
+      "loss": 4.157,
+      "step": 320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018839999999999997,
+      "loss": 4.3069,
+      "step": 321
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018899999999999999,
+      "loss": 4.1282,
+      "step": 322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018959999999999997,
+      "loss": 4.0788,
+      "step": 323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001902,
+      "loss": 4.2503,
+      "step": 324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019079999999999998,
+      "loss": 4.1146,
+      "step": 325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001914,
+      "loss": 4.0915,
+      "step": 326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 4.1609,
+      "step": 327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001926,
+      "loss": 3.9915,
+      "step": 328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019319999999999998,
+      "loss": 4.0215,
+      "step": 329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001938,
+      "loss": 4.1747,
+      "step": 330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019439999999999998,
+      "loss": 3.9796,
+      "step": 331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000195,
+      "loss": 4.0624,
+      "step": 332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019559999999999998,
+      "loss": 3.9148,
+      "step": 333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001962,
+      "loss": 3.773,
+      "step": 334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019679999999999999,
+      "loss": 3.6838,
+      "step": 335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001974,
+      "loss": 3.9912,
+      "step": 336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000198,
+      "loss": 3.7385,
+      "step": 337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001986,
+      "loss": 3.7082,
+      "step": 338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001992,
+      "loss": 3.6715,
+      "step": 339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001998,
+      "loss": 3.6575,
+      "step": 340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002004,
+      "loss": 3.6502,
+      "step": 341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000201,
+      "loss": 3.6207,
+      "step": 342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002016,
+      "loss": 3.52,
+      "step": 343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002022,
+      "loss": 3.2921,
+      "step": 344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002028,
+      "loss": 3.2503,
+      "step": 345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00020339999999999998,
+      "loss": 3.1721,
+      "step": 346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000204,
+      "loss": 3.1538,
+      "step": 347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00020459999999999999,
+      "loss": 3.0343,
+      "step": 348
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002052,
+      "loss": 2.7939,
+      "step": 349
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002058,
+      "loss": 2.6998,
+      "step": 350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00020639999999999998,
+      "loss": 5.7549,
+      "step": 351
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00020699999999999996,
+      "loss": 5.1232,
+      "step": 352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00020759999999999998,
+      "loss": 4.7916,
+      "step": 353
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00020819999999999996,
+      "loss": 4.7006,
+      "step": 354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00020879999999999998,
+      "loss": 4.5348,
+      "step": 355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00020939999999999997,
+      "loss": 4.5586,
+      "step": 356
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 4.4419,
+      "step": 357
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00021059999999999997,
+      "loss": 4.1398,
+      "step": 358
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00021119999999999996,
+      "loss": 4.3476,
+      "step": 359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00021179999999999997,
+      "loss": 4.2306,
+      "step": 360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00021239999999999996,
+      "loss": 4.2166,
+      "step": 361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 4.2093,
+      "step": 362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00021359999999999996,
+      "loss": 4.2283,
+      "step": 363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00021419999999999998,
+      "loss": 4.3396,
+      "step": 364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00021479999999999996,
+      "loss": 4.2494,
+      "step": 365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00021539999999999998,
+      "loss": 4.1817,
+      "step": 366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 4.2588,
+      "step": 367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00021659999999999998,
+      "loss": 4.2413,
+      "step": 368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00021719999999999997,
+      "loss": 4.1388,
+      "step": 369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00021779999999999998,
+      "loss": 4.2769,
+      "step": 370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00021839999999999997,
+      "loss": 4.1678,
+      "step": 371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 4.0985,
+      "step": 372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00021959999999999997,
+      "loss": 4.0278,
+      "step": 373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00022019999999999999,
+      "loss": 4.0552,
+      "step": 374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00022079999999999997,
+      "loss": 4.0627,
+      "step": 375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002214,
+      "loss": 4.1588,
+      "step": 376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 4.0489,
+      "step": 377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002226,
+      "loss": 3.8041,
+      "step": 378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00022319999999999998,
+      "loss": 3.8734,
+      "step": 379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002238,
+      "loss": 3.9844,
+      "step": 380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00022439999999999998,
+      "loss": 4.0289,
+      "step": 381
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000225,
+      "loss": 3.8937,
+      "step": 382
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00022559999999999998,
+      "loss": 3.7529,
+      "step": 383
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00022619999999999997,
+      "loss": 3.9653,
+      "step": 384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00022679999999999998,
+      "loss": 3.8649,
+      "step": 385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00022739999999999997,
+      "loss": 3.8224,
+      "step": 386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 3.749,
+      "step": 387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00022859999999999997,
+      "loss": 3.9977,
+      "step": 388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002292,
+      "loss": 3.7363,
+      "step": 389
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00022979999999999997,
+      "loss": 3.775,
+      "step": 390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002304,
+      "loss": 3.7372,
+      "step": 391
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00023099999999999998,
+      "loss": 3.5344,
+      "step": 392
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002316,
+      "loss": 3.5338,
+      "step": 393
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00023219999999999998,
+      "loss": 3.3479,
+      "step": 394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002328,
+      "loss": 3.3349,
+      "step": 395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00023339999999999998,
+      "loss": 3.278,
+      "step": 396
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000234,
+      "loss": 3.1448,
+      "step": 397
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00023459999999999998,
+      "loss": 3.1193,
+      "step": 398
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002352,
+      "loss": 2.7586,
+      "step": 399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00023579999999999999,
+      "loss": 2.8001,
+      "step": 400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002364,
+      "loss": 5.8658,
+      "step": 401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000237,
+      "loss": 4.9548,
+      "step": 402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002376,
+      "loss": 4.9834,
+      "step": 403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002382,
+      "loss": 4.5121,
+      "step": 404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002388,
+      "loss": 4.5158,
+      "step": 405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002394,
+      "loss": 4.389,
+      "step": 406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 4.4211,
+      "step": 407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002406,
+      "loss": 4.3436,
+      "step": 408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00024119999999999998,
+      "loss": 4.3404,
+      "step": 409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002418,
+      "loss": 4.4149,
+      "step": 410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00024239999999999998,
+      "loss": 4.1582,
+      "step": 411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000243,
+      "loss": 4.2055,
+      "step": 412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00024359999999999999,
+      "loss": 4.3762,
+      "step": 413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00024419999999999997,
+      "loss": 4.3286,
+      "step": 414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002448,
+      "loss": 4.3115,
+      "step": 415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00024539999999999995,
+      "loss": 4.2365,
+      "step": 416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 4.2317,
+      "step": 417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002466,
+      "loss": 4.225,
+      "step": 418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002472,
+      "loss": 4.2532,
+      "step": 419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00024779999999999995,
+      "loss": 4.1849,
+      "step": 420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00024839999999999997,
+      "loss": 4.1829,
+      "step": 421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000249,
+      "loss": 4.1261,
+      "step": 422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00024959999999999994,
+      "loss": 4.2554,
+      "step": 423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025019999999999996,
+      "loss": 4.2029,
+      "step": 424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025079999999999997,
+      "loss": 4.0218,
+      "step": 425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002514,
+      "loss": 3.9889,
+      "step": 426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 4.0182,
+      "step": 427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025259999999999996,
+      "loss": 3.9755,
+      "step": 428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002532,
+      "loss": 4.0512,
+      "step": 429
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002538,
+      "loss": 3.8566,
+      "step": 430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025439999999999995,
+      "loss": 3.9493,
+      "step": 431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025499999999999996,
+      "loss": 3.8412,
+      "step": 432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002556,
+      "loss": 3.8395,
+      "step": 433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002562,
+      "loss": 3.8858,
+      "step": 434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025679999999999995,
+      "loss": 3.7504,
+      "step": 435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025739999999999997,
+      "loss": 3.7391,
+      "step": 436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000258,
+      "loss": 3.6711,
+      "step": 437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002586,
+      "loss": 3.742,
+      "step": 438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025919999999999996,
+      "loss": 3.6552,
+      "step": 439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025979999999999997,
+      "loss": 3.4729,
+      "step": 440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002604,
+      "loss": 3.5555,
+      "step": 441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000261,
+      "loss": 3.4044,
+      "step": 442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00026159999999999996,
+      "loss": 3.3579,
+      "step": 443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002622,
+      "loss": 3.2758,
+      "step": 444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002628,
+      "loss": 3.3752,
+      "step": 445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00026339999999999995,
+      "loss": 3.0825,
+      "step": 446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 2.949,
+      "step": 447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002646,
+      "loss": 2.8313,
+      "step": 448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002652,
+      "loss": 2.8842,
+      "step": 449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00026579999999999996,
+      "loss": 2.6834,
+      "step": 450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00026639999999999997,
+      "loss": 5.9447,
+      "step": 451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000267,
+      "loss": 5.0768,
+      "step": 452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002676,
+      "loss": 4.7908,
+      "step": 453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00026819999999999996,
+      "loss": 4.722,
+      "step": 454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002688,
+      "loss": 4.4559,
+      "step": 455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002694,
+      "loss": 4.4045,
+      "step": 456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00027,
+      "loss": 4.4917,
+      "step": 457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00027059999999999996,
+      "loss": 4.3895,
+      "step": 458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002712,
+      "loss": 4.4436,
+      "step": 459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002718,
+      "loss": 4.0629,
+      "step": 460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002724,
+      "loss": 4.2174,
+      "step": 461
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00027299999999999997,
+      "loss": 4.2901,
+      "step": 462
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002736,
+      "loss": 4.0733,
+      "step": 463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002742,
+      "loss": 4.2551,
+      "step": 464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002748,
+      "loss": 4.2565,
+      "step": 465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00027539999999999997,
+      "loss": 4.4094,
+      "step": 466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000276,
+      "loss": 4.1442,
+      "step": 467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002766,
+      "loss": 4.2225,
+      "step": 468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002772,
+      "loss": 4.1311,
+      "step": 469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002778,
+      "loss": 4.2722,
+      "step": 470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002784,
+      "loss": 4.0853,
+      "step": 471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000279,
+      "loss": 4.275,
+      "step": 472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00027959999999999997,
+      "loss": 4.1574,
+      "step": 473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002802,
+      "loss": 4.1913,
+      "step": 474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002808,
+      "loss": 4.0092,
+      "step": 475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00028139999999999996,
+      "loss": 3.9728,
+      "step": 476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 4.0581,
+      "step": 477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002826,
+      "loss": 4.1989,
+      "step": 478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00028319999999999994,
+      "loss": 4.0023,
+      "step": 479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00028379999999999996,
+      "loss": 3.9102,
+      "step": 480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002844,
+      "loss": 3.9074,
+      "step": 481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000285,
+      "loss": 4.0167,
+      "step": 482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00028559999999999995,
+      "loss": 3.8945,
+      "step": 483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00028619999999999996,
+      "loss": 3.8795,
+      "step": 484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002868,
+      "loss": 3.8178,
+      "step": 485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00028739999999999994,
+      "loss": 3.6555,
+      "step": 486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 3.763,
+      "step": 487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00028859999999999997,
+      "loss": 3.5611,
+      "step": 488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002892,
+      "loss": 3.7752,
+      "step": 489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00028979999999999994,
+      "loss": 3.5473,
+      "step": 490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029039999999999996,
+      "loss": 3.4998,
+      "step": 491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029099999999999997,
+      "loss": 3.5321,
+      "step": 492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002916,
+      "loss": 3.3559,
+      "step": 493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029219999999999995,
+      "loss": 3.289,
+      "step": 494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029279999999999996,
+      "loss": 3.1925,
+      "step": 495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002934,
+      "loss": 3.0487,
+      "step": 496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000294,
+      "loss": 2.9119,
+      "step": 497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029459999999999995,
+      "loss": 3.026,
+      "step": 498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029519999999999997,
+      "loss": 2.753,
+      "step": 499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002958,
+      "loss": 2.5511,
+      "step": 500
+    },
+    {
+      "epoch": 0.31,
+      "eval_bleu": 0.0,
+      "eval_loss": 5.10392951965332,
+      "eval_runtime": 2560.6032,
+      "eval_samples_per_second": 5.764,
+      "eval_steps_per_second": 0.721,
+      "step": 500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002964,
+      "loss": 6.4484,
+      "step": 501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029699999999999996,
+      "loss": 5.5086,
+      "step": 502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029759999999999997,
+      "loss": 4.8983,
+      "step": 503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002982,
+      "loss": 4.9282,
+      "step": 504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002988,
+      "loss": 4.6444,
+      "step": 505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029939999999999996,
+      "loss": 4.4085,
+      "step": 506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003,
+      "loss": 4.4217,
+      "step": 507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002999311926605504,
+      "loss": 4.2745,
+      "step": 508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002998623853211009,
+      "loss": 4.4082,
+      "step": 509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029979357798165133,
+      "loss": 4.2605,
+      "step": 510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002997247706422018,
+      "loss": 4.4833,
+      "step": 511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029965596330275225,
+      "loss": 4.0582,
+      "step": 512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029958715596330274,
+      "loss": 4.0002,
+      "step": 513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002995183486238532,
+      "loss": 4.2744,
+      "step": 514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029944954128440366,
+      "loss": 4.1895,
+      "step": 515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002993807339449541,
+      "loss": 4.2262,
+      "step": 516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029931192660550453,
+      "loss": 4.2125,
+      "step": 517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000299243119266055,
+      "loss": 4.1036,
+      "step": 518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029917431192660545,
+      "loss": 4.202,
+      "step": 519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029910550458715594,
+      "loss": 4.1888,
+      "step": 520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002990366972477064,
+      "loss": 4.1997,
+      "step": 521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029896788990825687,
+      "loss": 4.1112,
+      "step": 522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002988990825688073,
+      "loss": 4.1197,
+      "step": 523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002988302752293578,
+      "loss": 4.061,
+      "step": 524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002987614678899082,
+      "loss": 3.8829,
+      "step": 525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002986926605504587,
+      "loss": 3.8193,
+      "step": 526
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029862385321100914,
+      "loss": 3.9218,
+      "step": 527
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029855504587155963,
+      "loss": 4.0122,
+      "step": 528
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029848623853211007,
+      "loss": 3.9433,
+      "step": 529
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029841743119266056,
+      "loss": 4.0117,
+      "step": 530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000298348623853211,
+      "loss": 3.8542,
+      "step": 531
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002982798165137614,
+      "loss": 4.008,
+      "step": 532
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002982110091743119,
+      "loss": 3.7908,
+      "step": 533
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029814220183486235,
+      "loss": 3.7293,
+      "step": 534
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029807339449541284,
+      "loss": 3.6423,
+      "step": 535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029800458715596327,
+      "loss": 3.8278,
+      "step": 536
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029793577981651376,
+      "loss": 3.634,
+      "step": 537
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002978669724770642,
+      "loss": 3.6174,
+      "step": 538
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002977981651376146,
+      "loss": 3.5131,
+      "step": 539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002977293577981651,
+      "loss": 3.6799,
+      "step": 540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029766055045871555,
+      "loss": 3.5297,
+      "step": 541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029759174311926604,
+      "loss": 3.3935,
+      "step": 542
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029752293577981647,
+      "loss": 3.3705,
+      "step": 543
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029745412844036696,
+      "loss": 3.1697,
+      "step": 544
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002973853211009174,
+      "loss": 3.2117,
+      "step": 545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002973165137614679,
+      "loss": 3.0101,
+      "step": 546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002972477064220183,
+      "loss": 3.0621,
+      "step": 547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029717889908256875,
+      "loss": 2.76,
+      "step": 548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029711009174311924,
+      "loss": 2.775,
+      "step": 549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029704128440366967,
+      "loss": 2.6342,
+      "step": 550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029697247706422016,
+      "loss": 5.6387,
+      "step": 551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002969036697247706,
+      "loss": 5.1001,
+      "step": 552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002968348623853211,
+      "loss": 4.7732,
+      "step": 553
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002967660550458715,
+      "loss": 4.6534,
+      "step": 554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000296697247706422,
+      "loss": 4.3346,
+      "step": 555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029662844036697244,
+      "loss": 4.3084,
+      "step": 556
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029655963302752293,
+      "loss": 4.3598,
+      "step": 557
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029649082568807336,
+      "loss": 4.4294,
+      "step": 558
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029642201834862385,
+      "loss": 4.2965,
+      "step": 559
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002963532110091743,
+      "loss": 4.2361,
+      "step": 560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002962844036697248,
+      "loss": 4.1292,
+      "step": 561
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002962155963302752,
+      "loss": 4.1621,
+      "step": 562
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002961467889908257,
+      "loss": 4.1383,
+      "step": 563
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029607798165137613,
+      "loss": 4.2404,
+      "step": 564
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029600917431192656,
+      "loss": 4.1117,
+      "step": 565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029594036697247705,
+      "loss": 4.0631,
+      "step": 566
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002958715596330275,
+      "loss": 4.1108,
+      "step": 567
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000295802752293578,
+      "loss": 3.9427,
+      "step": 568
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002957339449541284,
+      "loss": 4.0572,
+      "step": 569
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029566513761467884,
+      "loss": 3.9819,
+      "step": 570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029559633027522933,
+      "loss": 3.9515,
+      "step": 571
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029552752293577977,
+      "loss": 3.9857,
+      "step": 572
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029545871559633025,
+      "loss": 4.061,
+      "step": 573
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002953899082568807,
+      "loss": 4.0238,
+      "step": 574
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002953211009174312,
+      "loss": 3.9119,
+      "step": 575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002952522935779816,
+      "loss": 3.743,
+      "step": 576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002951834862385321,
+      "loss": 3.9652,
+      "step": 577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029511467889908253,
+      "loss": 3.8645,
+      "step": 578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000295045871559633,
+      "loss": 3.8511,
+      "step": 579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029497706422018346,
+      "loss": 3.8194,
+      "step": 580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002949082568807339,
+      "loss": 3.9449,
+      "step": 581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002948394495412844,
+      "loss": 3.8325,
+      "step": 582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002947706422018348,
+      "loss": 3.6292,
+      "step": 583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002947018348623853,
+      "loss": 3.6024,
+      "step": 584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029463302752293574,
+      "loss": 3.7113,
+      "step": 585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002945642201834862,
+      "loss": 3.6525,
+      "step": 586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029449541284403666,
+      "loss": 3.4445,
+      "step": 587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029442660550458715,
+      "loss": 3.432,
+      "step": 588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002943577981651376,
+      "loss": 3.5035,
+      "step": 589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029428899082568807,
+      "loss": 3.4721,
+      "step": 590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002942201834862385,
+      "loss": 3.4675,
+      "step": 591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000294151376146789,
+      "loss": 3.2225,
+      "step": 592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002940825688073394,
+      "loss": 3.1528,
+      "step": 593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002940137614678899,
+      "loss": 3.307,
+      "step": 594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029394495412844035,
+      "loss": 3.2358,
+      "step": 595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002938761467889908,
+      "loss": 3.0446,
+      "step": 596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029380733944954127,
+      "loss": 2.8091,
+      "step": 597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002937385321100917,
+      "loss": 2.8645,
+      "step": 598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002936697247706422,
+      "loss": 2.6645,
+      "step": 599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029360091743119263,
+      "loss": 2.5688,
+      "step": 600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029353211009174306,
+      "loss": 5.9051,
+      "step": 601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029346330275229355,
+      "loss": 5.1959,
+      "step": 602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000293394495412844,
+      "loss": 4.8426,
+      "step": 603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002933256880733945,
+      "loss": 4.7532,
+      "step": 604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002932568807339449,
+      "loss": 4.6745,
+      "step": 605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002931880733944954,
+      "loss": 4.4316,
+      "step": 606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029311926605504583,
+      "loss": 4.3115,
+      "step": 607
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002930504587155963,
+      "loss": 4.3938,
+      "step": 608
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029298165137614675,
+      "loss": 4.3118,
+      "step": 609
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029291284403669724,
+      "loss": 4.2531,
+      "step": 610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002928440366972477,
+      "loss": 4.2168,
+      "step": 611
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002927752293577981,
+      "loss": 4.1789,
+      "step": 612
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002927064220183486,
+      "loss": 4.2307,
+      "step": 613
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029263761467889903,
+      "loss": 3.9621,
+      "step": 614
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002925688073394495,
+      "loss": 4.07,
+      "step": 615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029249999999999995,
+      "loss": 4.1799,
+      "step": 616
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029243119266055044,
+      "loss": 4.1643,
+      "step": 617
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002923623853211009,
+      "loss": 4.1368,
+      "step": 618
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029229357798165136,
+      "loss": 4.0929,
+      "step": 619
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002922247706422018,
+      "loss": 4.104,
+      "step": 620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002921559633027523,
+      "loss": 4.3348,
+      "step": 621
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002920871559633027,
+      "loss": 4.1143,
+      "step": 622
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002920183486238532,
+      "loss": 4.1904,
+      "step": 623
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029194954128440364,
+      "loss": 3.9873,
+      "step": 624
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029188073394495413,
+      "loss": 4.0171,
+      "step": 625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029181192660550457,
+      "loss": 4.1695,
+      "step": 626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029174311926605506,
+      "loss": 4.1384,
+      "step": 627
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002916743119266055,
+      "loss": 3.822,
+      "step": 628
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002916055045871559,
+      "loss": 3.7661,
+      "step": 629
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002915366972477064,
+      "loss": 3.8224,
+      "step": 630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029146788990825685,
+      "loss": 3.7437,
+      "step": 631
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002913990825688073,
+      "loss": 3.8119,
+      "step": 632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029133027522935777,
+      "loss": 3.668,
+      "step": 633
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002912614678899082,
+      "loss": 3.6149,
+      "step": 634
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002911926605504587,
+      "loss": 3.6475,
+      "step": 635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002911238532110091,
+      "loss": 3.5243,
+      "step": 636
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002910550458715596,
+      "loss": 3.5894,
+      "step": 637
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029098623853211005,
+      "loss": 3.5434,
+      "step": 638
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029091743119266054,
+      "loss": 3.5424,
+      "step": 639
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029084862385321097,
+      "loss": 3.2515,
+      "step": 640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029077981651376146,
+      "loss": 3.3912,
+      "step": 641
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002907110091743119,
+      "loss": 3.2805,
+      "step": 642
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002906422018348624,
+      "loss": 3.3486,
+      "step": 643
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002905733944954128,
+      "loss": 3.1337,
+      "step": 644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029050458715596325,
+      "loss": 2.9084,
+      "step": 645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029043577981651374,
+      "loss": 2.7774,
+      "step": 646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029036697247706417,
+      "loss": 2.8928,
+      "step": 647
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029029816513761466,
+      "loss": 2.6588,
+      "step": 648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002902293577981651,
+      "loss": 2.5163,
+      "step": 649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002901605504587156,
+      "loss": 2.4516,
+      "step": 650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000290091743119266,
+      "loss": 5.7477,
+      "step": 651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002900229357798165,
+      "loss": 5.2403,
+      "step": 652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028995412844036694,
+      "loss": 4.5533,
+      "step": 653
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028988532110091743,
+      "loss": 4.4385,
+      "step": 654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028981651376146786,
+      "loss": 4.5647,
+      "step": 655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028974770642201835,
+      "loss": 4.3326,
+      "step": 656
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002896788990825688,
+      "loss": 4.3062,
+      "step": 657
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002896100917431193,
+      "loss": 4.3222,
+      "step": 658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002895412844036697,
+      "loss": 4.2311,
+      "step": 659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028947247706422014,
+      "loss": 4.3512,
+      "step": 660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028940366972477063,
+      "loss": 4.2595,
+      "step": 661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028933486238532106,
+      "loss": 4.3594,
+      "step": 662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028926605504587155,
+      "loss": 4.0522,
+      "step": 663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000289197247706422,
+      "loss": 4.1923,
+      "step": 664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002891284403669724,
+      "loss": 4.1089,
+      "step": 665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002890596330275229,
+      "loss": 4.0241,
+      "step": 666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028899082568807334,
+      "loss": 3.9912,
+      "step": 667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028892201834862383,
+      "loss": 4.0207,
+      "step": 668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028885321100917427,
+      "loss": 4.0843,
+      "step": 669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00028878440366972475,
+      "loss": 4.1729,
+      "step": 670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002887155963302752,
+      "loss": 4.0149,
+      "step": 671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002886467889908257,
+      "loss": 3.9101,
+      "step": 672
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002885779816513761,
+      "loss": 3.8613,
+      "step": 673
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002885091743119266,
+      "loss": 3.9378,
+      "step": 674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00028844036697247703,
+      "loss": 3.7407,
+      "step": 675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00028837155963302747,
+      "loss": 3.8231,
+      "step": 676
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00028830275229357796,
+      "loss": 3.8906,
+      "step": 677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002882339449541284,
+      "loss": 3.7453,
+      "step": 678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002881651376146789,
+      "loss": 3.7523,
+      "step": 679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002880963302752293,
+      "loss": 3.7658,
+      "step": 680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002880275229357798,
+      "loss": 3.6904,
+      "step": 681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00028795871559633024,
+      "loss": 3.6741,
+      "step": 682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002878899082568807,
+      "loss": 3.7134,
+      "step": 683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00028782110091743116,
+      "loss": 3.6732,
+      "step": 684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00028775229357798165,
+      "loss": 3.424,
+      "step": 685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002876834862385321,
+      "loss": 3.4771,
+      "step": 686
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00028761467889908257,
+      "loss": 3.5035,
+      "step": 687
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000287545871559633,
+      "loss": 3.3055,
+      "step": 688
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002874770642201835,
+      "loss": 3.4968,
+      "step": 689
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002874082568807339,
+      "loss": 3.2907,
+      "step": 690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002873394495412844,
+      "loss": 3.0678,
+      "step": 691
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00028727064220183485,
+      "loss": 3.2938,
+      "step": 692
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002872018348623853,
+      "loss": 3.3574,
+      "step": 693
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00028713302752293577,
+      "loss": 3.0665,
+      "step": 694
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002870642201834862,
+      "loss": 3.056,
+      "step": 695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00028699541284403664,
+      "loss": 2.8224,
+      "step": 696
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00028692660550458713,
+      "loss": 2.6976,
+      "step": 697
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00028685779816513756,
+      "loss": 2.6324,
+      "step": 698
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00028678899082568805,
+      "loss": 2.5969,
+      "step": 699
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002867201834862385,
+      "loss": 2.3751,
+      "step": 700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00028665137614678897,
+      "loss": 5.6773,
+      "step": 701
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002865825688073394,
+      "loss": 4.9661,
+      "step": 702
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002865137614678899,
+      "loss": 4.5686,
+      "step": 703
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00028644495412844033,
+      "loss": 4.6135,
+      "step": 704
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002863761467889908,
+      "loss": 4.2898,
+      "step": 705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028630733944954125,
+      "loss": 4.2391,
+      "step": 706
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028623853211009174,
+      "loss": 4.2093,
+      "step": 707
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002861697247706422,
+      "loss": 4.1607,
+      "step": 708
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002861009174311926,
+      "loss": 3.9372,
+      "step": 709
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002860321100917431,
+      "loss": 4.1236,
+      "step": 710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028596330275229353,
+      "loss": 4.0751,
+      "step": 711
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000285894495412844,
+      "loss": 4.1068,
+      "step": 712
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028582568807339445,
+      "loss": 3.8714,
+      "step": 713
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028575688073394494,
+      "loss": 3.9801,
+      "step": 714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002856880733944954,
+      "loss": 4.0199,
+      "step": 715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00028561926605504586,
+      "loss": 3.8776,
+      "step": 716
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002855504587155963,
+      "loss": 3.9058,
+      "step": 717
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002854816513761468,
+      "loss": 3.8545,
+      "step": 718
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002854128440366972,
+      "loss": 3.9446,
+      "step": 719
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002853440366972477,
+      "loss": 4.0342,
+      "step": 720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028527522935779814,
+      "loss": 3.8863,
+      "step": 721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028520642201834863,
+      "loss": 3.8662,
+      "step": 722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028513761467889907,
+      "loss": 3.8483,
+      "step": 723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002850688073394495,
+      "loss": 3.8917,
+      "step": 724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000285,
+      "loss": 3.7865,
+      "step": 725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002849311926605504,
+      "loss": 3.8303,
+      "step": 726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028486238532110086,
+      "loss": 3.6543,
+      "step": 727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028479357798165135,
+      "loss": 3.7781,
+      "step": 728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002847247706422018,
+      "loss": 3.536,
+      "step": 729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028465596330275227,
+      "loss": 3.6209,
+      "step": 730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002845871559633027,
+      "loss": 3.6361,
+      "step": 731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002845183486238532,
+      "loss": 3.6567,
+      "step": 732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002844495412844036,
+      "loss": 3.6093,
+      "step": 733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002843807339449541,
+      "loss": 3.6645,
+      "step": 734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028431192660550455,
+      "loss": 3.559,
+      "step": 735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028424311926605504,
+      "loss": 3.4329,
+      "step": 736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028417431192660547,
+      "loss": 3.5064,
+      "step": 737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028410550458715596,
+      "loss": 3.1375,
+      "step": 738
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002840366972477064,
+      "loss": 3.3257,
+      "step": 739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002839678899082568,
+      "loss": 3.2393,
+      "step": 740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002838990825688073,
+      "loss": 3.0156,
+      "step": 741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028383027522935775,
+      "loss": 2.9855,
+      "step": 742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028376146788990824,
+      "loss": 2.9816,
+      "step": 743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028369266055045867,
+      "loss": 2.9994,
+      "step": 744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028362385321100916,
+      "loss": 2.8371,
+      "step": 745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002835550458715596,
+      "loss": 2.6648,
+      "step": 746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002834862385321101,
+      "loss": 2.4887,
+      "step": 747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002834174311926605,
+      "loss": 2.4057,
+      "step": 748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000283348623853211,
+      "loss": 2.4427,
+      "step": 749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028327981651376144,
+      "loss": 2.3271,
+      "step": 750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028321100917431193,
+      "loss": 5.6453,
+      "step": 751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028314220183486236,
+      "loss": 5.1783,
+      "step": 752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028307339449541285,
+      "loss": 4.7046,
+      "step": 753
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002830045871559633,
+      "loss": 4.6663,
+      "step": 754
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00028293577981651377,
+      "loss": 4.3295,
+      "step": 755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002828669724770642,
+      "loss": 4.2772,
+      "step": 756
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00028279816513761464,
+      "loss": 4.1738,
+      "step": 757
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002827293577981651,
+      "loss": 4.2217,
+      "step": 758
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00028266055045871556,
+      "loss": 3.9638,
+      "step": 759
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000282591743119266,
+      "loss": 3.9321,
+      "step": 760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002825229357798165,
+      "loss": 3.9311,
+      "step": 761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002824541284403669,
+      "loss": 4.0939,
+      "step": 762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002823853211009174,
+      "loss": 3.7752,
+      "step": 763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00028231651376146784,
+      "loss": 3.9306,
+      "step": 764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00028224770642201833,
+      "loss": 4.0087,
+      "step": 765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00028217889908256877,
+      "loss": 3.9266,
+      "step": 766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00028211009174311925,
+      "loss": 3.8301,
+      "step": 767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002820412844036697,
+      "loss": 3.7035,
+      "step": 768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002819724770642202,
+      "loss": 3.8257,
+      "step": 769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002819036697247706,
+      "loss": 3.786,
+      "step": 770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002818348623853211,
+      "loss": 3.6439,
+      "step": 771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00028176605504587153,
+      "loss": 3.5426,
+      "step": 772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00028169724770642197,
+      "loss": 3.6834,
+      "step": 773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00028162844036697246,
+      "loss": 3.617,
+      "step": 774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002815596330275229,
+      "loss": 3.5549,
+      "step": 775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002814908256880734,
+      "loss": 3.6517,
+      "step": 776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002814220183486238,
+      "loss": 3.6253,
+      "step": 777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002813532110091743,
+      "loss": 3.5861,
+      "step": 778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00028128440366972473,
+      "loss": 3.5194,
+      "step": 779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002812155963302752,
+      "loss": 3.5125,
+      "step": 780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00028114678899082566,
+      "loss": 3.4713,
+      "step": 781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00028107798165137615,
+      "loss": 3.5415,
+      "step": 782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002810091743119266,
+      "loss": 3.2948,
+      "step": 783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00028094036697247707,
+      "loss": 3.7558,
+      "step": 784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002808715596330275,
+      "loss": 3.4264,
+      "step": 785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000280802752293578,
+      "loss": 3.5141,
+      "step": 786
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002807339449541284,
+      "loss": 3.2201,
+      "step": 787
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00028066513761467886,
+      "loss": 3.3493,
+      "step": 788
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002805963302752293,
+      "loss": 3.3462,
+      "step": 789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002805275229357798,
+      "loss": 3.088,
+      "step": 790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002804587155963302,
+      "loss": 3.1689,
+      "step": 791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002803899082568807,
+      "loss": 3.0794,
+      "step": 792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00028032110091743114,
+      "loss": 3.018,
+      "step": 793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002802522935779816,
+      "loss": 2.9667,
+      "step": 794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00028018348623853206,
+      "loss": 3.0386,
+      "step": 795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00028011467889908255,
+      "loss": 2.8074,
+      "step": 796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000280045871559633,
+      "loss": 2.5167,
+      "step": 797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00027997706422018347,
+      "loss": 2.6814,
+      "step": 798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002799082568807339,
+      "loss": 2.2415,
+      "step": 799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002798394495412844,
+      "loss": 2.3352,
+      "step": 800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00027977064220183483,
+      "loss": 5.8354,
+      "step": 801
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002797018348623853,
+      "loss": 5.0938,
+      "step": 802
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00027963302752293575,
+      "loss": 4.5163,
+      "step": 803
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002795642201834862,
+      "loss": 4.4829,
+      "step": 804
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002794954128440367,
+      "loss": 4.3322,
+      "step": 805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002794266055045871,
+      "loss": 4.1584,
+      "step": 806
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002793577981651376,
+      "loss": 3.9791,
+      "step": 807
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00027928899082568803,
+      "loss": 3.9365,
+      "step": 808
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002792201834862385,
+      "loss": 3.9113,
+      "step": 809
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00027915137614678895,
+      "loss": 3.8803,
+      "step": 810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00027908256880733944,
+      "loss": 3.8092,
+      "step": 811
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002790137614678899,
+      "loss": 3.8322,
+      "step": 812
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00027894495412844036,
+      "loss": 3.7589,
+      "step": 813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002788761467889908,
+      "loss": 3.8974,
+      "step": 814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002788073394495413,
+      "loss": 3.6453,
+      "step": 815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002787385321100917,
+      "loss": 3.7593,
+      "step": 816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002786697247706422,
+      "loss": 3.6785,
+      "step": 817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00027860091743119264,
+      "loss": 3.6274,
+      "step": 818
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027853211009174313,
+      "loss": 3.6767,
+      "step": 819
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002784633027522935,
+      "loss": 3.7273,
+      "step": 820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000278394495412844,
+      "loss": 3.6245,
+      "step": 821
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027832568807339443,
+      "loss": 3.5374,
+      "step": 822
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002782568807339449,
+      "loss": 3.7519,
+      "step": 823
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027818807339449536,
+      "loss": 3.6401,
+      "step": 824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027811926605504584,
+      "loss": 3.6166,
+      "step": 825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002780504587155963,
+      "loss": 3.5436,
+      "step": 826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027798165137614677,
+      "loss": 3.5718,
+      "step": 827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002779128440366972,
+      "loss": 3.7381,
+      "step": 828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002778440366972477,
+      "loss": 3.4856,
+      "step": 829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002777752293577981,
+      "loss": 3.5257,
+      "step": 830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002777064220183486,
+      "loss": 3.5684,
+      "step": 831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027763761467889905,
+      "loss": 3.5243,
+      "step": 832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027756880733944953,
+      "loss": 3.2728,
+      "step": 833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027749999999999997,
+      "loss": 3.2295,
+      "step": 834
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027743119266055046,
+      "loss": 3.2984,
+      "step": 835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002773623853211009,
+      "loss": 3.212,
+      "step": 836
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002772935779816513,
+      "loss": 3.0495,
+      "step": 837
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002772247706422018,
+      "loss": 3.044,
+      "step": 838
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027715596330275225,
+      "loss": 3.1158,
+      "step": 839
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027708715596330274,
+      "loss": 3.0624,
+      "step": 840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027701834862385317,
+      "loss": 2.9915,
+      "step": 841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027694954128440366,
+      "loss": 3.0146,
+      "step": 842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002768807339449541,
+      "loss": 2.9135,
+      "step": 843
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002768119266055046,
+      "loss": 2.8511,
+      "step": 844
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000276743119266055,
+      "loss": 2.7068,
+      "step": 845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002766743119266055,
+      "loss": 2.611,
+      "step": 846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027660550458715594,
+      "loss": 2.5175,
+      "step": 847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002765366972477064,
+      "loss": 2.453,
+      "step": 848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027646788990825686,
+      "loss": 2.2522,
+      "step": 849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027639908256880735,
+      "loss": 2.2374,
+      "step": 850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002763302752293578,
+      "loss": 5.4623,
+      "step": 851
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002762614678899082,
+      "loss": 5.1737,
+      "step": 852
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027619266055045865,
+      "loss": 4.7505,
+      "step": 853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027612385321100914,
+      "loss": 4.5885,
+      "step": 854
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002760550458715596,
+      "loss": 4.3554,
+      "step": 855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027598623853211006,
+      "loss": 4.3703,
+      "step": 856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002759174311926605,
+      "loss": 4.1621,
+      "step": 857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000275848623853211,
+      "loss": 3.9429,
+      "step": 858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002757798165137614,
+      "loss": 3.7094,
+      "step": 859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002757110091743119,
+      "loss": 3.8591,
+      "step": 860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027564220183486234,
+      "loss": 3.8113,
+      "step": 861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027557339449541283,
+      "loss": 3.6976,
+      "step": 862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027550458715596326,
+      "loss": 3.6116,
+      "step": 863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027543577981651375,
+      "loss": 3.5698,
+      "step": 864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002753669724770642,
+      "loss": 3.5438,
+      "step": 865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002752981651376147,
+      "loss": 3.6478,
+      "step": 866
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002752293577981651,
+      "loss": 3.6013,
+      "step": 867
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027516055045871554,
+      "loss": 3.6408,
+      "step": 868
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027509174311926603,
+      "loss": 3.5644,
+      "step": 869
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027502293577981647,
+      "loss": 3.4828,
+      "step": 870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027495412844036695,
+      "loss": 3.5369,
+      "step": 871
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002748853211009174,
+      "loss": 3.5172,
+      "step": 872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002748165137614679,
+      "loss": 3.302,
+      "step": 873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002747477064220183,
+      "loss": 3.4129,
+      "step": 874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002746788990825688,
+      "loss": 3.2188,
+      "step": 875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027461009174311923,
+      "loss": 3.37,
+      "step": 876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002745412844036697,
+      "loss": 3.2573,
+      "step": 877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027447247706422016,
+      "loss": 3.3593,
+      "step": 878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027440366972477064,
+      "loss": 3.3993,
+      "step": 879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002743348623853211,
+      "loss": 3.3669,
+      "step": 880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027426605504587157,
+      "loss": 3.3771,
+      "step": 881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000274197247706422,
+      "loss": 3.1646,
+      "step": 882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002741284403669725,
+      "loss": 3.2597,
+      "step": 883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00027405963302752287,
+      "loss": 3.0931,
+      "step": 884
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00027399082568807336,
+      "loss": 3.0211,
+      "step": 885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002739220183486238,
+      "loss": 3.0644,
+      "step": 886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002738532110091743,
+      "loss": 2.9487,
+      "step": 887
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002737844036697247,
+      "loss": 2.8871,
+      "step": 888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002737155963302752,
+      "loss": 2.9525,
+      "step": 889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00027364678899082564,
+      "loss": 2.8661,
+      "step": 890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002735779816513761,
+      "loss": 2.4642,
+      "step": 891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00027350917431192656,
+      "loss": 2.7808,
+      "step": 892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00027344036697247705,
+      "loss": 2.8034,
+      "step": 893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002733715596330275,
+      "loss": 2.6838,
+      "step": 894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00027330275229357797,
+      "loss": 2.4445,
+      "step": 895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002732339449541284,
+      "loss": 2.375,
+      "step": 896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002731651376146789,
+      "loss": 2.4147,
+      "step": 897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00027309633027522933,
+      "loss": 2.159,
+      "step": 898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002730275229357798,
+      "loss": 2.087,
+      "step": 899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00027295871559633025,
+      "loss": 1.8963,
+      "step": 900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002728899082568807,
+      "loss": 5.6814,
+      "step": 901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00027282110091743117,
+      "loss": 4.8834,
+      "step": 902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002727522935779816,
+      "loss": 4.5176,
+      "step": 903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002726834862385321,
+      "loss": 4.1832,
+      "step": 904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00027261467889908253,
+      "loss": 4.2073,
+      "step": 905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000272545871559633,
+      "loss": 3.726,
+      "step": 906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00027247706422018345,
+      "loss": 3.7203,
+      "step": 907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00027240825688073394,
+      "loss": 3.5696,
+      "step": 908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002723394495412844,
+      "loss": 3.5516,
+      "step": 909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00027227064220183486,
+      "loss": 3.5932,
+      "step": 910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002722018348623853,
+      "loss": 3.621,
+      "step": 911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002721330275229358,
+      "loss": 3.6297,
+      "step": 912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002720642201834862,
+      "loss": 3.4574,
+      "step": 913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002719954128440367,
+      "loss": 3.429,
+      "step": 914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00027192660550458714,
+      "loss": 3.3761,
+      "step": 915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002718577981651376,
+      "loss": 3.2979,
+      "step": 916
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000271788990825688,
+      "loss": 3.4376,
+      "step": 917
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002717201834862385,
+      "loss": 3.4223,
+      "step": 918
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00027165137614678893,
+      "loss": 3.4307,
+      "step": 919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002715825688073394,
+      "loss": 3.2741,
+      "step": 920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00027151376146788986,
+      "loss": 3.229,
+      "step": 921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00027144495412844034,
+      "loss": 3.387,
+      "step": 922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002713761467889908,
+      "loss": 3.3286,
+      "step": 923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00027130733944954127,
+      "loss": 3.2406,
+      "step": 924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002712385321100917,
+      "loss": 3.0068,
+      "step": 925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002711697247706422,
+      "loss": 3.2257,
+      "step": 926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002711009174311926,
+      "loss": 3.4286,
+      "step": 927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002710321100917431,
+      "loss": 3.1836,
+      "step": 928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00027096330275229355,
+      "loss": 3.1256,
+      "step": 929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00027089449541284403,
+      "loss": 2.9933,
+      "step": 930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00027082568807339447,
+      "loss": 3.0308,
+      "step": 931
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002707568807339449,
+      "loss": 2.9226,
+      "step": 932
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002706880733944954,
+      "loss": 2.9666,
+      "step": 933
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002706192660550458,
+      "loss": 2.8008,
+      "step": 934
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002705504587155963,
+      "loss": 2.8944,
+      "step": 935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00027048165137614675,
+      "loss": 2.9266,
+      "step": 936
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00027041284403669724,
+      "loss": 2.8044,
+      "step": 937
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00027034403669724767,
+      "loss": 2.9476,
+      "step": 938
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00027027522935779816,
+      "loss": 2.7552,
+      "step": 939
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002702064220183486,
+      "loss": 2.7592,
+      "step": 940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002701376146788991,
+      "loss": 2.7915,
+      "step": 941
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002700688073394495,
+      "loss": 2.8133,
+      "step": 942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00027,
+      "loss": 2.5633,
+      "step": 943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00026993119266055044,
+      "loss": 2.4147,
+      "step": 944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002698623853211009,
+      "loss": 2.5834,
+      "step": 945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00026979357798165136,
+      "loss": 2.3341,
+      "step": 946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00026972477064220185,
+      "loss": 2.5304,
+      "step": 947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00026965596330275223,
+      "loss": 2.3964,
+      "step": 948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002695871559633027,
+      "loss": 3.3065,
+      "step": 949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00026951834862385315,
+      "loss": 2.194,
+      "step": 950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00026944954128440364,
+      "loss": 5.7219,
+      "step": 951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002693807339449541,
+      "loss": 5.0634,
+      "step": 952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00026931192660550456,
+      "loss": 4.3777,
+      "step": 953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000269243119266055,
+      "loss": 4.4132,
+      "step": 954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002691743119266055,
+      "loss": 4.1163,
+      "step": 955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002691055045871559,
+      "loss": 3.8913,
+      "step": 956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002690366972477064,
+      "loss": 3.8223,
+      "step": 957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00026896788990825684,
+      "loss": 3.7676,
+      "step": 958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00026889908256880733,
+      "loss": 3.7502,
+      "step": 959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00026883027522935776,
+      "loss": 3.5551,
+      "step": 960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00026876146788990825,
+      "loss": 3.7311,
+      "step": 961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002686926605504587,
+      "loss": 3.7123,
+      "step": 962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002686238532110092,
+      "loss": 3.4324,
+      "step": 963
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002685550458715596,
+      "loss": 3.4605,
+      "step": 964
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026848623853211004,
+      "loss": 3.3101,
+      "step": 965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026841743119266053,
+      "loss": 3.4289,
+      "step": 966
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026834862385321097,
+      "loss": 3.4608,
+      "step": 967
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026827981651376145,
+      "loss": 3.2693,
+      "step": 968
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002682110091743119,
+      "loss": 3.1958,
+      "step": 969
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002681422018348624,
+      "loss": 3.2383,
+      "step": 970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002680733944954128,
+      "loss": 3.149,
+      "step": 971
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002680045871559633,
+      "loss": 3.1712,
+      "step": 972
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026793577981651373,
+      "loss": 3.2276,
+      "step": 973
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002678669724770642,
+      "loss": 3.1085,
+      "step": 974
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026779816513761466,
+      "loss": 3.3517,
+      "step": 975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026772935779816514,
+      "loss": 3.0752,
+      "step": 976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002676605504587156,
+      "loss": 3.1543,
+      "step": 977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026759174311926607,
+      "loss": 3.4293,
+      "step": 978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002675229357798165,
+      "loss": 2.9209,
+      "step": 979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00026745412844036693,
+      "loss": 3.3701,
+      "step": 980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026738532110091737,
+      "loss": 3.0162,
+      "step": 981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026731651376146786,
+      "loss": 3.1484,
+      "step": 982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002672477064220183,
+      "loss": 2.8463,
+      "step": 983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002671788990825688,
+      "loss": 2.8031,
+      "step": 984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002671100917431192,
+      "loss": 2.8604,
+      "step": 985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002670412844036697,
+      "loss": 2.7669,
+      "step": 986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026697247706422014,
+      "loss": 2.739,
+      "step": 987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002669036697247706,
+      "loss": 2.846,
+      "step": 988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026683486238532106,
+      "loss": 2.7752,
+      "step": 989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026676605504587155,
+      "loss": 2.6357,
+      "step": 990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000266697247706422,
+      "loss": 2.5999,
+      "step": 991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026662844036697247,
+      "loss": 2.4139,
+      "step": 992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002665596330275229,
+      "loss": 2.448,
+      "step": 993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002664908256880734,
+      "loss": 2.3161,
+      "step": 994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002664220183486238,
+      "loss": 2.1951,
+      "step": 995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00026635321100917426,
+      "loss": 2.3582,
+      "step": 996
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00026628440366972475,
+      "loss": 2.4091,
+      "step": 997
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002662155963302752,
+      "loss": 2.3661,
+      "step": 998
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00026614678899082567,
+      "loss": 2.1907,
+      "step": 999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002660779816513761,
+      "loss": 2.2033,
+      "step": 1000
+    },
+    {
+      "epoch": 0.62,
+      "eval_bleu": 1.471304930449711e-13,
+      "eval_loss": 4.178153991699219,
+      "eval_runtime": 2571.1504,
+      "eval_samples_per_second": 5.741,
+      "eval_steps_per_second": 0.718,
+      "step": 1000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002660091743119266,
+      "loss": 5.5762,
+      "step": 1001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00026594036697247703,
+      "loss": 4.7959,
+      "step": 1002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002658715596330275,
+      "loss": 4.3913,
+      "step": 1003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00026580275229357795,
+      "loss": 3.9288,
+      "step": 1004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00026573394495412844,
+      "loss": 3.9872,
+      "step": 1005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002656651376146789,
+      "loss": 3.6041,
+      "step": 1006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00026559633027522936,
+      "loss": 3.6125,
+      "step": 1007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002655275229357798,
+      "loss": 3.7386,
+      "step": 1008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002654587155963303,
+      "loss": 3.4695,
+      "step": 1009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002653899082568807,
+      "loss": 3.5328,
+      "step": 1010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002653211009174312,
+      "loss": 3.3167,
+      "step": 1011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002652522935779816,
+      "loss": 3.3963,
+      "step": 1012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002651834862385321,
+      "loss": 3.205,
+      "step": 1013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002651146788990825,
+      "loss": 3.3196,
+      "step": 1014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000265045871559633,
+      "loss": 3.4569,
+      "step": 1015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00026497706422018343,
+      "loss": 3.2782,
+      "step": 1016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002649082568807339,
+      "loss": 3.3713,
+      "step": 1017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00026483944954128435,
+      "loss": 3.3488,
+      "step": 1018
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00026477064220183484,
+      "loss": 3.0157,
+      "step": 1019
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002647018348623853,
+      "loss": 3.1181,
+      "step": 1020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00026463302752293577,
+      "loss": 3.3601,
+      "step": 1021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002645642201834862,
+      "loss": 3.1228,
+      "step": 1022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002644954128440367,
+      "loss": 3.1386,
+      "step": 1023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002644266055045871,
+      "loss": 3.0566,
+      "step": 1024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002643577981651376,
+      "loss": 3.0509,
+      "step": 1025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00026428899082568804,
+      "loss": 2.9676,
+      "step": 1026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00026422018348623853,
+      "loss": 2.8082,
+      "step": 1027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00026415137614678897,
+      "loss": 2.8932,
+      "step": 1028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002640825688073394,
+      "loss": 3.0243,
+      "step": 1029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002640137614678899,
+      "loss": 2.8623,
+      "step": 1030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002639449541284403,
+      "loss": 2.9837,
+      "step": 1031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002638761467889908,
+      "loss": 2.9066,
+      "step": 1032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00026380733944954125,
+      "loss": 2.7663,
+      "step": 1033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00026373853211009174,
+      "loss": 2.696,
+      "step": 1034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00026366972477064217,
+      "loss": 2.8101,
+      "step": 1035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00026360091743119266,
+      "loss": 2.6771,
+      "step": 1036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002635321100917431,
+      "loss": 2.707,
+      "step": 1037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002634633027522936,
+      "loss": 2.5267,
+      "step": 1038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000263394495412844,
+      "loss": 2.6367,
+      "step": 1039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002633256880733945,
+      "loss": 2.6826,
+      "step": 1040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00026325688073394494,
+      "loss": 2.4771,
+      "step": 1041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002631880733944954,
+      "loss": 2.4592,
+      "step": 1042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00026311926605504586,
+      "loss": 2.4441,
+      "step": 1043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002630504587155963,
+      "loss": 2.2774,
+      "step": 1044
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00026298165137614673,
+      "loss": 2.0834,
+      "step": 1045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002629128440366972,
+      "loss": 1.9322,
+      "step": 1046
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00026284403669724765,
+      "loss": 2.0929,
+      "step": 1047
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00026277522935779814,
+      "loss": 2.1183,
+      "step": 1048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00026270642201834857,
+      "loss": 1.8789,
+      "step": 1049
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00026263761467889906,
+      "loss": 1.7628,
+      "step": 1050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002625688073394495,
+      "loss": 4.9101,
+      "step": 1051
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002625,
+      "loss": 4.2879,
+      "step": 1052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002624311926605504,
+      "loss": 3.8733,
+      "step": 1053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002623623853211009,
+      "loss": 3.6801,
+      "step": 1054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00026229357798165134,
+      "loss": 3.679,
+      "step": 1055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00026222477064220183,
+      "loss": 3.5236,
+      "step": 1056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00026215596330275226,
+      "loss": 3.5384,
+      "step": 1057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00026208715596330275,
+      "loss": 3.3687,
+      "step": 1058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002620183486238532,
+      "loss": 3.4308,
+      "step": 1059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002619495412844037,
+      "loss": 3.2832,
+      "step": 1060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002618807339449541,
+      "loss": 3.0896,
+      "step": 1061
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026181192660550454,
+      "loss": 3.347,
+      "step": 1062
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026174311926605503,
+      "loss": 3.1642,
+      "step": 1063
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026167431192660546,
+      "loss": 3.1044,
+      "step": 1064
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026160550458715595,
+      "loss": 3.2826,
+      "step": 1065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002615366972477064,
+      "loss": 3.1107,
+      "step": 1066
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002614678899082569,
+      "loss": 3.148,
+      "step": 1067
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002613990825688073,
+      "loss": 3.0607,
+      "step": 1068
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002613302752293578,
+      "loss": 3.0048,
+      "step": 1069
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026126146788990823,
+      "loss": 2.9932,
+      "step": 1070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002611926605504587,
+      "loss": 3.0713,
+      "step": 1071
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026112385321100915,
+      "loss": 3.0098,
+      "step": 1072
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026105504587155964,
+      "loss": 2.9331,
+      "step": 1073
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002609862385321101,
+      "loss": 3.0307,
+      "step": 1074
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026091743119266057,
+      "loss": 2.8047,
+      "step": 1075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000260848623853211,
+      "loss": 2.8941,
+      "step": 1076
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026077981651376143,
+      "loss": 2.8448,
+      "step": 1077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026071100917431187,
+      "loss": 2.7392,
+      "step": 1078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026064220183486236,
+      "loss": 2.8165,
+      "step": 1079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002605733944954128,
+      "loss": 2.721,
+      "step": 1080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002605045871559633,
+      "loss": 2.7159,
+      "step": 1081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002604357798165137,
+      "loss": 2.6376,
+      "step": 1082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002603669724770642,
+      "loss": 2.6103,
+      "step": 1083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026029816513761464,
+      "loss": 2.5928,
+      "step": 1084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002602293577981651,
+      "loss": 2.7037,
+      "step": 1085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026016055045871556,
+      "loss": 2.6811,
+      "step": 1086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026009174311926605,
+      "loss": 2.714,
+      "step": 1087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002600229357798165,
+      "loss": 2.6176,
+      "step": 1088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025995412844036697,
+      "loss": 2.5062,
+      "step": 1089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002598853211009174,
+      "loss": 2.5099,
+      "step": 1090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002598165137614679,
+      "loss": 2.4763,
+      "step": 1091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002597477064220183,
+      "loss": 2.3145,
+      "step": 1092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025967889908256876,
+      "loss": 2.194,
+      "step": 1093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025961009174311925,
+      "loss": 2.333,
+      "step": 1094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002595412844036697,
+      "loss": 2.0515,
+      "step": 1095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025947247706422017,
+      "loss": 1.9804,
+      "step": 1096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002594036697247706,
+      "loss": 1.9431,
+      "step": 1097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002593348623853211,
+      "loss": 2.0674,
+      "step": 1098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025926605504587153,
+      "loss": 1.8007,
+      "step": 1099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000259197247706422,
+      "loss": 1.7359,
+      "step": 1100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025912844036697245,
+      "loss": 4.9063,
+      "step": 1101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025905963302752294,
+      "loss": 4.5578,
+      "step": 1102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002589908256880734,
+      "loss": 3.9415,
+      "step": 1103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025892201834862386,
+      "loss": 3.9376,
+      "step": 1104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002588532110091743,
+      "loss": 3.6935,
+      "step": 1105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002587844036697248,
+      "loss": 3.2228,
+      "step": 1106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002587155963302752,
+      "loss": 3.1785,
+      "step": 1107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002586467889908257,
+      "loss": 3.4129,
+      "step": 1108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002585779816513761,
+      "loss": 3.1752,
+      "step": 1109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002585091743119266,
+      "loss": 3.1613,
+      "step": 1110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000258440366972477,
+      "loss": 3.1191,
+      "step": 1111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002583715596330275,
+      "loss": 3.0338,
+      "step": 1112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025830275229357793,
+      "loss": 3.106,
+      "step": 1113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002582339449541284,
+      "loss": 3.1221,
+      "step": 1114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025816513761467885,
+      "loss": 2.9862,
+      "step": 1115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025809633027522934,
+      "loss": 2.9024,
+      "step": 1116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002580275229357798,
+      "loss": 3.0392,
+      "step": 1117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025795871559633027,
+      "loss": 2.8382,
+      "step": 1118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002578899082568807,
+      "loss": 2.7565,
+      "step": 1119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002578211009174312,
+      "loss": 2.9379,
+      "step": 1120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002577522935779816,
+      "loss": 2.8578,
+      "step": 1121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002576834862385321,
+      "loss": 2.8066,
+      "step": 1122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025761467889908254,
+      "loss": 2.8392,
+      "step": 1123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025754587155963303,
+      "loss": 2.6414,
+      "step": 1124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00025747706422018347,
+      "loss": 2.8674,
+      "step": 1125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002574082568807339,
+      "loss": 2.9095,
+      "step": 1126
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002573394495412844,
+      "loss": 2.7973,
+      "step": 1127
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002572706422018348,
+      "loss": 2.7233,
+      "step": 1128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002572018348623853,
+      "loss": 2.8347,
+      "step": 1129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00025713302752293575,
+      "loss": 2.692,
+      "step": 1130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00025706422018348623,
+      "loss": 2.6711,
+      "step": 1131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00025699541284403667,
+      "loss": 2.7745,
+      "step": 1132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00025692660550458716,
+      "loss": 2.5442,
+      "step": 1133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002568577981651376,
+      "loss": 2.5528,
+      "step": 1134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002567889908256881,
+      "loss": 2.6461,
+      "step": 1135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002567201834862385,
+      "loss": 2.5496,
+      "step": 1136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000256651376146789,
+      "loss": 2.5886,
+      "step": 1137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00025658256880733944,
+      "loss": 2.2997,
+      "step": 1138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002565137614678899,
+      "loss": 2.2357,
+      "step": 1139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00025644495412844036,
+      "loss": 2.333,
+      "step": 1140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002563761467889908,
+      "loss": 2.3503,
+      "step": 1141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002563073394495412,
+      "loss": 2.2457,
+      "step": 1142
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002562385321100917,
+      "loss": 2.2465,
+      "step": 1143
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00025616972477064215,
+      "loss": 2.3449,
+      "step": 1144
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00025610091743119264,
+      "loss": 1.9791,
+      "step": 1145
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00025603211009174307,
+      "loss": 2.0701,
+      "step": 1146
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00025596330275229356,
+      "loss": 2.0319,
+      "step": 1147
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000255894495412844,
+      "loss": 1.8884,
+      "step": 1148
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002558256880733945,
+      "loss": 1.7236,
+      "step": 1149
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002557568807339449,
+      "loss": 1.7023,
+      "step": 1150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002556880733944954,
+      "loss": 5.0322,
+      "step": 1151
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00025561926605504584,
+      "loss": 4.4311,
+      "step": 1152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00025555045871559633,
+      "loss": 3.7635,
+      "step": 1153
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00025548165137614676,
+      "loss": 3.4902,
+      "step": 1154
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00025541284403669725,
+      "loss": 3.5092,
+      "step": 1155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002553440366972477,
+      "loss": 3.1856,
+      "step": 1156
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002552752293577981,
+      "loss": 3.2636,
+      "step": 1157
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002552064220183486,
+      "loss": 3.2892,
+      "step": 1158
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00025513761467889904,
+      "loss": 3.0195,
+      "step": 1159
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00025506880733944953,
+      "loss": 3.2674,
+      "step": 1160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00025499999999999996,
+      "loss": 2.9148,
+      "step": 1161
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00025493119266055045,
+      "loss": 2.8722,
+      "step": 1162
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002548623853211009,
+      "loss": 2.8195,
+      "step": 1163
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002547935779816514,
+      "loss": 2.9372,
+      "step": 1164
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002547247706422018,
+      "loss": 2.9755,
+      "step": 1165
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002546559633027523,
+      "loss": 2.9018,
+      "step": 1166
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00025458715596330273,
+      "loss": 2.8212,
+      "step": 1167
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002545183486238532,
+      "loss": 2.9238,
+      "step": 1168
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00025444954128440365,
+      "loss": 2.7164,
+      "step": 1169
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00025438073394495414,
+      "loss": 2.8057,
+      "step": 1170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002543119266055046,
+      "loss": 2.7161,
+      "step": 1171
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00025424311926605507,
+      "loss": 2.9002,
+      "step": 1172
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00025417431192660545,
+      "loss": 2.6253,
+      "step": 1173
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00025410550458715593,
+      "loss": 2.6776,
+      "step": 1174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00025403669724770637,
+      "loss": 2.6572,
+      "step": 1175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00025396788990825686,
+      "loss": 2.7412,
+      "step": 1176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002538990825688073,
+      "loss": 2.5223,
+      "step": 1177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002538302752293578,
+      "loss": 2.6715,
+      "step": 1178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002537614678899082,
+      "loss": 2.549,
+      "step": 1179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002536926605504587,
+      "loss": 2.656,
+      "step": 1180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00025362385321100914,
+      "loss": 2.4706,
+      "step": 1181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002535550458715596,
+      "loss": 2.5941,
+      "step": 1182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00025348623853211006,
+      "loss": 2.5448,
+      "step": 1183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00025341743119266055,
+      "loss": 2.5535,
+      "step": 1184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000253348623853211,
+      "loss": 2.4567,
+      "step": 1185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00025327981651376147,
+      "loss": 2.43,
+      "step": 1186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002532110091743119,
+      "loss": 2.4834,
+      "step": 1187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002531422018348624,
+      "loss": 2.5156,
+      "step": 1188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002530733944954128,
+      "loss": 2.3824,
+      "step": 1189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00025300458715596326,
+      "loss": 2.3285,
+      "step": 1190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00025293577981651375,
+      "loss": 2.2779,
+      "step": 1191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002528669724770642,
+      "loss": 2.1884,
+      "step": 1192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00025279816513761467,
+      "loss": 2.3475,
+      "step": 1193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002527293577981651,
+      "loss": 2.2281,
+      "step": 1194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002526605504587156,
+      "loss": 1.9123,
+      "step": 1195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00025259174311926603,
+      "loss": 1.9846,
+      "step": 1196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002525229357798165,
+      "loss": 1.8164,
+      "step": 1197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00025245412844036695,
+      "loss": 1.9029,
+      "step": 1198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00025238532110091744,
+      "loss": 1.8056,
+      "step": 1199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00025231651376146787,
+      "loss": 1.703,
+      "step": 1200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00025224770642201836,
+      "loss": 4.4119,
+      "step": 1201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002521788990825688,
+      "loss": 3.8685,
+      "step": 1202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002521100917431193,
+      "loss": 3.3704,
+      "step": 1203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002520412844036697,
+      "loss": 3.2853,
+      "step": 1204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00025197247706422015,
+      "loss": 3.2298,
+      "step": 1205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002519036697247706,
+      "loss": 3.1027,
+      "step": 1206
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002518348623853211,
+      "loss": 2.976,
+      "step": 1207
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002517660550458715,
+      "loss": 2.9351,
+      "step": 1208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000251697247706422,
+      "loss": 2.8098,
+      "step": 1209
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00025162844036697243,
+      "loss": 2.8428,
+      "step": 1210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002515596330275229,
+      "loss": 3.0106,
+      "step": 1211
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00025149082568807335,
+      "loss": 2.9271,
+      "step": 1212
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00025142201834862384,
+      "loss": 2.7298,
+      "step": 1213
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002513532110091743,
+      "loss": 2.9279,
+      "step": 1214
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00025128440366972476,
+      "loss": 2.7588,
+      "step": 1215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002512155963302752,
+      "loss": 2.8592,
+      "step": 1216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002511467889908257,
+      "loss": 2.8537,
+      "step": 1217
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002510779816513761,
+      "loss": 2.7539,
+      "step": 1218
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002510091743119266,
+      "loss": 2.6405,
+      "step": 1219
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00025094036697247704,
+      "loss": 2.5818,
+      "step": 1220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002508715596330275,
+      "loss": 2.6985,
+      "step": 1221
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00025080275229357797,
+      "loss": 2.5826,
+      "step": 1222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002507339449541284,
+      "loss": 2.7408,
+      "step": 1223
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002506651376146789,
+      "loss": 2.7314,
+      "step": 1224
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002505963302752293,
+      "loss": 2.507,
+      "step": 1225
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002505275229357798,
+      "loss": 2.4916,
+      "step": 1226
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00025045871559633025,
+      "loss": 2.6332,
+      "step": 1227
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00025038990825688073,
+      "loss": 2.6837,
+      "step": 1228
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00025032110091743117,
+      "loss": 2.5342,
+      "step": 1229
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00025025229357798166,
+      "loss": 2.5196,
+      "step": 1230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002501834862385321,
+      "loss": 2.5154,
+      "step": 1231
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002501146788990826,
+      "loss": 2.4692,
+      "step": 1232
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000250045871559633,
+      "loss": 2.4471,
+      "step": 1233
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002499770642201835,
+      "loss": 2.3725,
+      "step": 1234
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00024990825688073394,
+      "loss": 2.4658,
+      "step": 1235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002498394495412844,
+      "loss": 2.4992,
+      "step": 1236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002497706422018348,
+      "loss": 2.3282,
+      "step": 1237
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002497018348623853,
+      "loss": 2.1878,
+      "step": 1238
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002496330275229357,
+      "loss": 2.2067,
+      "step": 1239
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002495642201834862,
+      "loss": 2.2267,
+      "step": 1240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024949541284403665,
+      "loss": 2.2462,
+      "step": 1241
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024942660550458714,
+      "loss": 2.1119,
+      "step": 1242
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024935779816513757,
+      "loss": 2.1393,
+      "step": 1243
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024928899082568806,
+      "loss": 2.0206,
+      "step": 1244
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002492201834862385,
+      "loss": 2.1669,
+      "step": 1245
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000249151376146789,
+      "loss": 1.8983,
+      "step": 1246
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002490825688073394,
+      "loss": 1.821,
+      "step": 1247
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002490137614678899,
+      "loss": 1.6637,
+      "step": 1248
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024894495412844034,
+      "loss": 1.6363,
+      "step": 1249
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024887614678899083,
+      "loss": 1.5859,
+      "step": 1250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024880733944954126,
+      "loss": 4.6515,
+      "step": 1251
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00024873853211009175,
+      "loss": 4.2094,
+      "step": 1252
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002486697247706422,
+      "loss": 3.7558,
+      "step": 1253
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002486009174311926,
+      "loss": 3.486,
+      "step": 1254
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002485321100917431,
+      "loss": 3.3385,
+      "step": 1255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024846330275229354,
+      "loss": 3.1675,
+      "step": 1256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024839449541284403,
+      "loss": 3.1744,
+      "step": 1257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024832568807339446,
+      "loss": 3.0137,
+      "step": 1258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024825688073394495,
+      "loss": 3.0504,
+      "step": 1259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002481880733944954,
+      "loss": 3.0661,
+      "step": 1260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002481192660550459,
+      "loss": 2.7988,
+      "step": 1261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002480504587155963,
+      "loss": 2.8546,
+      "step": 1262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002479816513761468,
+      "loss": 2.654,
+      "step": 1263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024791284403669723,
+      "loss": 2.6323,
+      "step": 1264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002478440366972477,
+      "loss": 2.7632,
+      "step": 1265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024777522935779815,
+      "loss": 2.8722,
+      "step": 1266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024770642201834864,
+      "loss": 2.7517,
+      "step": 1267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002476376146788991,
+      "loss": 2.8213,
+      "step": 1268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002475688073394495,
+      "loss": 2.6378,
+      "step": 1269
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024749999999999994,
+      "loss": 2.6336,
+      "step": 1270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024743119266055043,
+      "loss": 2.6516,
+      "step": 1271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024736238532110087,
+      "loss": 2.6232,
+      "step": 1272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024729357798165136,
+      "loss": 2.6596,
+      "step": 1273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002472247706422018,
+      "loss": 2.6417,
+      "step": 1274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002471559633027523,
+      "loss": 2.7287,
+      "step": 1275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002470871559633027,
+      "loss": 2.4056,
+      "step": 1276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002470183486238532,
+      "loss": 2.6234,
+      "step": 1277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024694954128440363,
+      "loss": 2.5816,
+      "step": 1278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002468807339449541,
+      "loss": 2.415,
+      "step": 1279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024681192660550456,
+      "loss": 2.5448,
+      "step": 1280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024674311926605505,
+      "loss": 2.5468,
+      "step": 1281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002466743119266055,
+      "loss": 2.5486,
+      "step": 1282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024660550458715597,
+      "loss": 2.5176,
+      "step": 1283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002465366972477064,
+      "loss": 2.1955,
+      "step": 1284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024646788990825684,
+      "loss": 2.3487,
+      "step": 1285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002463990825688073,
+      "loss": 2.502,
+      "step": 1286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00024633027522935776,
+      "loss": 2.1346,
+      "step": 1287
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024626146788990825,
+      "loss": 2.2725,
+      "step": 1288
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002461926605504587,
+      "loss": 2.3531,
+      "step": 1289
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024612385321100917,
+      "loss": 2.1908,
+      "step": 1290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002460550458715596,
+      "loss": 2.0789,
+      "step": 1291
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002459862385321101,
+      "loss": 2.1947,
+      "step": 1292
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002459174311926605,
+      "loss": 2.0086,
+      "step": 1293
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000245848623853211,
+      "loss": 1.9971,
+      "step": 1294
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024577981651376145,
+      "loss": 1.8451,
+      "step": 1295
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024571100917431194,
+      "loss": 1.756,
+      "step": 1296
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024564220183486237,
+      "loss": 1.6671,
+      "step": 1297
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024557339449541286,
+      "loss": 1.7778,
+      "step": 1298
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002455045871559633,
+      "loss": 1.3684,
+      "step": 1299
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002454357798165138,
+      "loss": 1.4714,
+      "step": 1300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024536697247706416,
+      "loss": 4.441,
+      "step": 1301
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00024529816513761465,
+      "loss": 3.8664,
+      "step": 1302
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002452293577981651,
+      "loss": 3.3995,
+      "step": 1303
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002451605504587156,
+      "loss": 3.3105,
+      "step": 1304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000245091743119266,
+      "loss": 3.1228,
+      "step": 1305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002450229357798165,
+      "loss": 2.9794,
+      "step": 1306
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024495412844036693,
+      "loss": 2.9494,
+      "step": 1307
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002448853211009174,
+      "loss": 3.0211,
+      "step": 1308
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024481651376146785,
+      "loss": 2.9939,
+      "step": 1309
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024474770642201834,
+      "loss": 3.0454,
+      "step": 1310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002446788990825688,
+      "loss": 2.7784,
+      "step": 1311
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024461009174311926,
+      "loss": 2.8133,
+      "step": 1312
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002445412844036697,
+      "loss": 2.7374,
+      "step": 1313
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002444724770642202,
+      "loss": 2.6958,
+      "step": 1314
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002444036697247706,
+      "loss": 2.6172,
+      "step": 1315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002443348623853211,
+      "loss": 2.6109,
+      "step": 1316
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024426605504587154,
+      "loss": 2.7339,
+      "step": 1317
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000244197247706422,
+      "loss": 2.5767,
+      "step": 1318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00024412844036697244,
+      "loss": 2.557,
+      "step": 1319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002440596330275229,
+      "loss": 2.586,
+      "step": 1320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024399082568807336,
+      "loss": 2.4573,
+      "step": 1321
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024392201834862382,
+      "loss": 2.5958,
+      "step": 1322
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024385321100917428,
+      "loss": 2.6669,
+      "step": 1323
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024378440366972474,
+      "loss": 2.3446,
+      "step": 1324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002437155963302752,
+      "loss": 2.4018,
+      "step": 1325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024364678899082567,
+      "loss": 2.6164,
+      "step": 1326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024357798165137613,
+      "loss": 2.5206,
+      "step": 1327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002435091743119266,
+      "loss": 2.4659,
+      "step": 1328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024344036697247705,
+      "loss": 2.1809,
+      "step": 1329
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002433715596330275,
+      "loss": 2.2729,
+      "step": 1330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024330275229357797,
+      "loss": 2.4214,
+      "step": 1331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024323394495412843,
+      "loss": 2.4258,
+      "step": 1332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024316513761467887,
+      "loss": 2.5117,
+      "step": 1333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024309633027522933,
+      "loss": 2.1816,
+      "step": 1334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002430275229357798,
+      "loss": 2.2469,
+      "step": 1335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00024295871559633025,
+      "loss": 2.3788,
+      "step": 1336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00024288990825688071,
+      "loss": 2.0899,
+      "step": 1337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00024282110091743118,
+      "loss": 2.263,
+      "step": 1338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00024275229357798164,
+      "loss": 2.202,
+      "step": 1339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002426834862385321,
+      "loss": 2.1724,
+      "step": 1340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00024261467889908256,
+      "loss": 2.0293,
+      "step": 1341
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00024254587155963302,
+      "loss": 2.1315,
+      "step": 1342
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00024247706422018348,
+      "loss": 2.0213,
+      "step": 1343
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00024240825688073394,
+      "loss": 1.9341,
+      "step": 1344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002423394495412844,
+      "loss": 1.7983,
+      "step": 1345
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00024227064220183487,
+      "loss": 1.77,
+      "step": 1346
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00024220183486238533,
+      "loss": 1.8767,
+      "step": 1347
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00024213302752293576,
+      "loss": 1.6549,
+      "step": 1348
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002420642201834862,
+      "loss": 1.5778,
+      "step": 1349
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00024199541284403666,
+      "loss": 1.5833,
+      "step": 1350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00024192660550458712,
+      "loss": 4.6938,
+      "step": 1351
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00024185779816513758,
+      "loss": 4.0161,
+      "step": 1352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00024178899082568804,
+      "loss": 3.5704,
+      "step": 1353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002417201834862385,
+      "loss": 3.1779,
+      "step": 1354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00024165137614678896,
+      "loss": 3.0573,
+      "step": 1355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00024158256880733942,
+      "loss": 3.1404,
+      "step": 1356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00024151376146788989,
+      "loss": 2.7964,
+      "step": 1357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00024144495412844035,
+      "loss": 2.9402,
+      "step": 1358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002413761467889908,
+      "loss": 2.7826,
+      "step": 1359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00024130733944954127,
+      "loss": 2.7306,
+      "step": 1360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00024123853211009173,
+      "loss": 2.8451,
+      "step": 1361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002411697247706422,
+      "loss": 2.6798,
+      "step": 1362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00024110091743119265,
+      "loss": 2.7597,
+      "step": 1363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00024103211009174311,
+      "loss": 2.8166,
+      "step": 1364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00024096330275229355,
+      "loss": 2.745,
+      "step": 1365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000240894495412844,
+      "loss": 2.5577,
+      "step": 1366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00024082568807339447,
+      "loss": 2.6861,
+      "step": 1367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00024075688073394493,
+      "loss": 2.6284,
+      "step": 1368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002406880733944954,
+      "loss": 2.4811,
+      "step": 1369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00024061926605504585,
+      "loss": 2.5559,
+      "step": 1370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00024055045871559632,
+      "loss": 2.5791,
+      "step": 1371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00024048165137614678,
+      "loss": 2.529,
+      "step": 1372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00024041284403669724,
+      "loss": 2.4775,
+      "step": 1373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002403440366972477,
+      "loss": 2.5257,
+      "step": 1374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00024027522935779816,
+      "loss": 2.5076,
+      "step": 1375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00024020642201834862,
+      "loss": 2.4796,
+      "step": 1376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00024013761467889908,
+      "loss": 2.3824,
+      "step": 1377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00024006880733944954,
+      "loss": 2.2693,
+      "step": 1378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 2.3754,
+      "step": 1379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023993119266055044,
+      "loss": 2.3962,
+      "step": 1380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023986238532110087,
+      "loss": 2.4415,
+      "step": 1381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023979357798165134,
+      "loss": 2.3266,
+      "step": 1382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002397247706422018,
+      "loss": 2.2076,
+      "step": 1383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023965596330275226,
+      "loss": 2.1958,
+      "step": 1384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00023958715596330272,
+      "loss": 2.3991,
+      "step": 1385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023951834862385318,
+      "loss": 2.0174,
+      "step": 1386
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023944954128440364,
+      "loss": 2.1168,
+      "step": 1387
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002393807339449541,
+      "loss": 2.0234,
+      "step": 1388
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023931192660550456,
+      "loss": 2.2305,
+      "step": 1389
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023924311926605503,
+      "loss": 2.2301,
+      "step": 1390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002391743119266055,
+      "loss": 2.1488,
+      "step": 1391
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023910550458715595,
+      "loss": 1.9698,
+      "step": 1392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002390366972477064,
+      "loss": 1.8753,
+      "step": 1393
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023896788990825687,
+      "loss": 1.9798,
+      "step": 1394
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023889908256880733,
+      "loss": 1.822,
+      "step": 1395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002388302752293578,
+      "loss": 1.6806,
+      "step": 1396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023876146788990823,
+      "loss": 1.7254,
+      "step": 1397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002386926605504587,
+      "loss": 1.5683,
+      "step": 1398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023862385321100915,
+      "loss": 1.5895,
+      "step": 1399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002385550458715596,
+      "loss": 1.4108,
+      "step": 1400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00023848623853211007,
+      "loss": 4.3234,
+      "step": 1401
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023841743119266053,
+      "loss": 3.7324,
+      "step": 1402
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000238348623853211,
+      "loss": 3.3879,
+      "step": 1403
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023827981651376146,
+      "loss": 3.0386,
+      "step": 1404
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023821100917431192,
+      "loss": 3.051,
+      "step": 1405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023814220183486238,
+      "loss": 2.7574,
+      "step": 1406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023807339449541284,
+      "loss": 2.788,
+      "step": 1407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002380045871559633,
+      "loss": 2.7725,
+      "step": 1408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023793577981651376,
+      "loss": 2.8618,
+      "step": 1409
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002378669724770642,
+      "loss": 2.4982,
+      "step": 1410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023779816513761466,
+      "loss": 2.6049,
+      "step": 1411
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023772935779816512,
+      "loss": 2.6398,
+      "step": 1412
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023766055045871555,
+      "loss": 2.6199,
+      "step": 1413
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023759174311926602,
+      "loss": 2.4028,
+      "step": 1414
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023752293577981648,
+      "loss": 2.6199,
+      "step": 1415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00023745412844036694,
+      "loss": 2.6152,
+      "step": 1416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002373853211009174,
+      "loss": 2.6174,
+      "step": 1417
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023731651376146786,
+      "loss": 2.6818,
+      "step": 1418
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023724770642201832,
+      "loss": 2.4989,
+      "step": 1419
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023717889908256878,
+      "loss": 2.5147,
+      "step": 1420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023711009174311924,
+      "loss": 2.2742,
+      "step": 1421
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002370412844036697,
+      "loss": 2.4624,
+      "step": 1422
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023697247706422017,
+      "loss": 2.511,
+      "step": 1423
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023690366972477063,
+      "loss": 2.3468,
+      "step": 1424
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002368348623853211,
+      "loss": 2.5092,
+      "step": 1425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023676605504587155,
+      "loss": 2.4016,
+      "step": 1426
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000236697247706422,
+      "loss": 2.3635,
+      "step": 1427
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023662844036697247,
+      "loss": 2.4423,
+      "step": 1428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002365596330275229,
+      "loss": 2.4538,
+      "step": 1429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023649082568807337,
+      "loss": 2.3316,
+      "step": 1430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023642201834862383,
+      "loss": 2.2705,
+      "step": 1431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002363532110091743,
+      "loss": 2.219,
+      "step": 1432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00023628440366972475,
+      "loss": 2.2657,
+      "step": 1433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002362155963302752,
+      "loss": 2.4393,
+      "step": 1434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023614678899082567,
+      "loss": 2.377,
+      "step": 1435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023607798165137614,
+      "loss": 2.2584,
+      "step": 1436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002360091743119266,
+      "loss": 2.1022,
+      "step": 1437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023594036697247706,
+      "loss": 2.1148,
+      "step": 1438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023587155963302752,
+      "loss": 1.9968,
+      "step": 1439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023580275229357798,
+      "loss": 1.9591,
+      "step": 1440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023573394495412842,
+      "loss": 1.9437,
+      "step": 1441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023566513761467888,
+      "loss": 1.9426,
+      "step": 1442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023559633027522934,
+      "loss": 2.0203,
+      "step": 1443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002355275229357798,
+      "loss": 1.8824,
+      "step": 1444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023545871559633023,
+      "loss": 1.9237,
+      "step": 1445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002353899082568807,
+      "loss": 1.7076,
+      "step": 1446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023532110091743116,
+      "loss": 1.5834,
+      "step": 1447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023525229357798162,
+      "loss": 1.5915,
+      "step": 1448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023518348623853208,
+      "loss": 1.4358,
+      "step": 1449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00023511467889908254,
+      "loss": 1.4267,
+      "step": 1450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000235045871559633,
+      "loss": 4.0403,
+      "step": 1451
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023497706422018346,
+      "loss": 3.6418,
+      "step": 1452
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023490825688073392,
+      "loss": 3.1654,
+      "step": 1453
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023483944954128438,
+      "loss": 2.7494,
+      "step": 1454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023477064220183485,
+      "loss": 2.7254,
+      "step": 1455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002347018348623853,
+      "loss": 2.7735,
+      "step": 1456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023463302752293577,
+      "loss": 2.7853,
+      "step": 1457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023456422018348623,
+      "loss": 2.7738,
+      "step": 1458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002344954128440367,
+      "loss": 2.6205,
+      "step": 1459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023442660550458715,
+      "loss": 2.657,
+      "step": 1460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023435779816513759,
+      "loss": 2.5686,
+      "step": 1461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023428899082568805,
+      "loss": 2.6935,
+      "step": 1462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002342201834862385,
+      "loss": 2.4596,
+      "step": 1463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023415137614678897,
+      "loss": 2.3145,
+      "step": 1464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00023408256880733943,
+      "loss": 2.5441,
+      "step": 1465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002340137614678899,
+      "loss": 2.4676,
+      "step": 1466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023394495412844035,
+      "loss": 2.3969,
+      "step": 1467
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023387614678899082,
+      "loss": 2.3363,
+      "step": 1468
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023380733944954128,
+      "loss": 2.6365,
+      "step": 1469
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023373853211009174,
+      "loss": 2.3824,
+      "step": 1470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002336697247706422,
+      "loss": 2.3885,
+      "step": 1471
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023360091743119266,
+      "loss": 2.4727,
+      "step": 1472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002335321100917431,
+      "loss": 2.3845,
+      "step": 1473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023346330275229356,
+      "loss": 2.5182,
+      "step": 1474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023339449541284402,
+      "loss": 2.4093,
+      "step": 1475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023332568807339448,
+      "loss": 2.2637,
+      "step": 1476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002332568807339449,
+      "loss": 2.3533,
+      "step": 1477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023318807339449537,
+      "loss": 2.3644,
+      "step": 1478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023311926605504583,
+      "loss": 2.394,
+      "step": 1479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002330504587155963,
+      "loss": 2.2024,
+      "step": 1480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023298165137614676,
+      "loss": 2.0975,
+      "step": 1481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00023291284403669722,
+      "loss": 2.2884,
+      "step": 1482
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00023284403669724768,
+      "loss": 2.148,
+      "step": 1483
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00023277522935779814,
+      "loss": 2.1001,
+      "step": 1484
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002327064220183486,
+      "loss": 2.0864,
+      "step": 1485
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00023263761467889906,
+      "loss": 2.061,
+      "step": 1486
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00023256880733944953,
+      "loss": 2.0502,
+      "step": 1487
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00023249999999999999,
+      "loss": 2.1436,
+      "step": 1488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00023243119266055045,
+      "loss": 1.9573,
+      "step": 1489
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002323623853211009,
+      "loss": 1.8843,
+      "step": 1490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00023229357798165137,
+      "loss": 1.9595,
+      "step": 1491
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00023222477064220183,
+      "loss": 1.7707,
+      "step": 1492
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00023215596330275227,
+      "loss": 1.8516,
+      "step": 1493
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00023208715596330273,
+      "loss": 1.8104,
+      "step": 1494
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002320183486238532,
+      "loss": 1.7862,
+      "step": 1495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00023194954128440365,
+      "loss": 1.7488,
+      "step": 1496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002318807339449541,
+      "loss": 1.4077,
+      "step": 1497
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00023181192660550457,
+      "loss": 1.4606,
+      "step": 1498
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00023174311926605503,
+      "loss": 1.498,
+      "step": 1499
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002316743119266055,
+      "loss": 1.4703,
+      "step": 1500
+    },
+    {
+      "epoch": 0.93,
+      "eval_bleu": 1.1928315591476622e-15,
+      "eval_loss": 2.897862195968628,
+      "eval_runtime": 2471.822,
+      "eval_samples_per_second": 5.971,
+      "eval_steps_per_second": 0.746,
+      "step": 1500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00023160550458715596,
+      "loss": 3.8798,
+      "step": 1501
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00023153669724770642,
+      "loss": 3.6521,
+      "step": 1502
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00023146788990825688,
+      "loss": 3.1413,
+      "step": 1503
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002313990825688073,
+      "loss": 2.9996,
+      "step": 1504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00023133027522935777,
+      "loss": 2.8918,
+      "step": 1505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00023126146788990824,
+      "loss": 2.615,
+      "step": 1506
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002311926605504587,
+      "loss": 2.7964,
+      "step": 1507
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00023112385321100916,
+      "loss": 2.7088,
+      "step": 1508
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002310550458715596,
+      "loss": 2.6699,
+      "step": 1509
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00023098623853211005,
+      "loss": 2.6601,
+      "step": 1510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00023091743119266051,
+      "loss": 2.5294,
+      "step": 1511
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00023084862385321098,
+      "loss": 2.5886,
+      "step": 1512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00023077981651376144,
+      "loss": 2.4782,
+      "step": 1513
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002307110091743119,
+      "loss": 2.444,
+      "step": 1514
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00023064220183486236,
+      "loss": 2.2496,
+      "step": 1515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00023057339449541282,
+      "loss": 2.3786,
+      "step": 1516
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00023050458715596328,
+      "loss": 2.3432,
+      "step": 1517
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00023043577981651374,
+      "loss": 2.6102,
+      "step": 1518
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002303669724770642,
+      "loss": 2.42,
+      "step": 1519
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00023029816513761467,
+      "loss": 2.4603,
+      "step": 1520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00023022935779816513,
+      "loss": 2.3453,
+      "step": 1521
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002301605504587156,
+      "loss": 2.4801,
+      "step": 1522
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00023009174311926605,
+      "loss": 2.3845,
+      "step": 1523
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002300229357798165,
+      "loss": 2.3283,
+      "step": 1524
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00022995412844036694,
+      "loss": 2.2907,
+      "step": 1525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002298853211009174,
+      "loss": 2.3028,
+      "step": 1526
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00022981651376146787,
+      "loss": 2.3779,
+      "step": 1527
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00022974770642201833,
+      "loss": 2.2853,
+      "step": 1528
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002296788990825688,
+      "loss": 2.2455,
+      "step": 1529
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00022961009174311925,
+      "loss": 2.3359,
+      "step": 1530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002295412844036697,
+      "loss": 2.2759,
+      "step": 1531
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022947247706422017,
+      "loss": 2.1927,
+      "step": 1532
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022940366972477064,
+      "loss": 2.0256,
+      "step": 1533
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002293348623853211,
+      "loss": 2.0893,
+      "step": 1534
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022926605504587153,
+      "loss": 1.9414,
+      "step": 1535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000229197247706422,
+      "loss": 2.0403,
+      "step": 1536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022912844036697245,
+      "loss": 2.151,
+      "step": 1537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022905963302752291,
+      "loss": 2.0342,
+      "step": 1538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022899082568807338,
+      "loss": 1.7482,
+      "step": 1539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022892201834862384,
+      "loss": 1.8345,
+      "step": 1540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022885321100917427,
+      "loss": 1.8883,
+      "step": 1541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022878440366972473,
+      "loss": 1.7642,
+      "step": 1542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002287155963302752,
+      "loss": 1.7402,
+      "step": 1543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022864678899082565,
+      "loss": 1.7481,
+      "step": 1544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022857798165137612,
+      "loss": 1.7225,
+      "step": 1545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022850917431192658,
+      "loss": 1.6718,
+      "step": 1546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00022844036697247704,
+      "loss": 1.434,
+      "step": 1547
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002283715596330275,
+      "loss": 1.3589,
+      "step": 1548
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022830275229357796,
+      "loss": 1.3906,
+      "step": 1549
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022823394495412842,
+      "loss": 1.3383,
+      "step": 1550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022816513761467888,
+      "loss": 3.8618,
+      "step": 1551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022809633027522935,
+      "loss": 3.47,
+      "step": 1552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002280275229357798,
+      "loss": 2.9152,
+      "step": 1553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022795871559633027,
+      "loss": 2.7829,
+      "step": 1554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022788990825688073,
+      "loss": 2.7614,
+      "step": 1555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002278211009174312,
+      "loss": 2.794,
+      "step": 1556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022775229357798162,
+      "loss": 2.6156,
+      "step": 1557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022768348623853209,
+      "loss": 2.6174,
+      "step": 1558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022761467889908255,
+      "loss": 2.6436,
+      "step": 1559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000227545871559633,
+      "loss": 2.4479,
+      "step": 1560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022747706422018347,
+      "loss": 2.5513,
+      "step": 1561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00022740825688073393,
+      "loss": 2.488,
+      "step": 1562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002273394495412844,
+      "loss": 2.613,
+      "step": 1563
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022727064220183485,
+      "loss": 2.4519,
+      "step": 1564
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022720183486238531,
+      "loss": 2.4788,
+      "step": 1565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022713302752293575,
+      "loss": 2.2182,
+      "step": 1566
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002270642201834862,
+      "loss": 2.4339,
+      "step": 1567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022699541284403667,
+      "loss": 2.4258,
+      "step": 1568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022692660550458713,
+      "loss": 2.3051,
+      "step": 1569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002268577981651376,
+      "loss": 2.4419,
+      "step": 1570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022678899082568806,
+      "loss": 2.4325,
+      "step": 1571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022672018348623852,
+      "loss": 2.325,
+      "step": 1572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022665137614678895,
+      "loss": 2.3338,
+      "step": 1573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002265825688073394,
+      "loss": 2.223,
+      "step": 1574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022651376146788987,
+      "loss": 2.1918,
+      "step": 1575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022644495412844033,
+      "loss": 2.3632,
+      "step": 1576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002263761467889908,
+      "loss": 2.1487,
+      "step": 1577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022630733944954126,
+      "loss": 2.3696,
+      "step": 1578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00022623853211009172,
+      "loss": 2.0953,
+      "step": 1579
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022616972477064218,
+      "loss": 2.1683,
+      "step": 1580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022610091743119264,
+      "loss": 2.0308,
+      "step": 1581
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002260321100917431,
+      "loss": 2.0683,
+      "step": 1582
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022596330275229356,
+      "loss": 2.0815,
+      "step": 1583
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022589449541284402,
+      "loss": 2.0315,
+      "step": 1584
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022582568807339449,
+      "loss": 1.9406,
+      "step": 1585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022575688073394495,
+      "loss": 2.0098,
+      "step": 1586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002256880733944954,
+      "loss": 1.9886,
+      "step": 1587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022561926605504587,
+      "loss": 1.9092,
+      "step": 1588
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002255504587155963,
+      "loss": 1.8882,
+      "step": 1589
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022548165137614676,
+      "loss": 1.8068,
+      "step": 1590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022541284403669723,
+      "loss": 1.7487,
+      "step": 1591
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002253440366972477,
+      "loss": 1.8935,
+      "step": 1592
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022527522935779815,
+      "loss": 1.8995,
+      "step": 1593
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002252064220183486,
+      "loss": 1.7237,
+      "step": 1594
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00022513761467889907,
+      "loss": 1.5711,
+      "step": 1595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022506880733944953,
+      "loss": 1.5474,
+      "step": 1596
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000225,
+      "loss": 1.4504,
+      "step": 1597
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022493119266055043,
+      "loss": 1.4231,
+      "step": 1598
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002248623853211009,
+      "loss": 1.2094,
+      "step": 1599
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022479357798165135,
+      "loss": 1.3736,
+      "step": 1600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002247247706422018,
+      "loss": 3.8124,
+      "step": 1601
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022465596330275227,
+      "loss": 3.1417,
+      "step": 1602
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022458715596330273,
+      "loss": 2.9373,
+      "step": 1603
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002245183486238532,
+      "loss": 2.7386,
+      "step": 1604
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022444954128440366,
+      "loss": 2.5797,
+      "step": 1605
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002243807339449541,
+      "loss": 2.62,
+      "step": 1606
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022431192660550455,
+      "loss": 2.4809,
+      "step": 1607
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022424311926605501,
+      "loss": 2.527,
+      "step": 1608
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022417431192660547,
+      "loss": 2.4485,
+      "step": 1609
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022410550458715594,
+      "loss": 2.4327,
+      "step": 1610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002240366972477064,
+      "loss": 2.3558,
+      "step": 1611
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00022396788990825686,
+      "loss": 2.2636,
+      "step": 1612
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022389908256880732,
+      "loss": 2.158,
+      "step": 1613
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022383027522935778,
+      "loss": 2.2312,
+      "step": 1614
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022376146788990824,
+      "loss": 1.8881,
+      "step": 1615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002236926605504587,
+      "loss": 1.9816,
+      "step": 1616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022362385321100917,
+      "loss": 1.8794,
+      "step": 1617
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022355504587155963,
+      "loss": 1.7875,
+      "step": 1618
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002234862385321101,
+      "loss": 1.4404,
+      "step": 1619
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022341743119266055,
+      "loss": 1.3255,
+      "step": 1620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000223348623853211,
+      "loss": 3.5384,
+      "step": 1621
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022327981651376144,
+      "loss": 3.0625,
+      "step": 1622
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002232110091743119,
+      "loss": 2.5833,
+      "step": 1623
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022314220183486237,
+      "loss": 2.7377,
+      "step": 1624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022307339449541283,
+      "loss": 2.5123,
+      "step": 1625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002230045871559633,
+      "loss": 2.0895,
+      "step": 1626
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00022293577981651375,
+      "loss": 2.4198,
+      "step": 1627
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002228669724770642,
+      "loss": 2.3877,
+      "step": 1628
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00022279816513761465,
+      "loss": 2.297,
+      "step": 1629
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002227293577981651,
+      "loss": 2.1656,
+      "step": 1630
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00022266055045871557,
+      "loss": 2.0874,
+      "step": 1631
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00022259174311926603,
+      "loss": 2.0864,
+      "step": 1632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002225229357798165,
+      "loss": 2.1314,
+      "step": 1633
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00022245412844036695,
+      "loss": 2.1392,
+      "step": 1634
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00022238532110091741,
+      "loss": 2.0376,
+      "step": 1635
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00022231651376146787,
+      "loss": 1.9865,
+      "step": 1636
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00022224770642201834,
+      "loss": 2.0534,
+      "step": 1637
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00022217889908256877,
+      "loss": 1.9698,
+      "step": 1638
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00022211009174311923,
+      "loss": 1.954,
+      "step": 1639
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002220412844036697,
+      "loss": 1.9595,
+      "step": 1640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00022197247706422015,
+      "loss": 1.9459,
+      "step": 1641
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00022190366972477062,
+      "loss": 1.9686,
+      "step": 1642
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00022183486238532108,
+      "loss": 1.9225,
+      "step": 1643
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00022176605504587154,
+      "loss": 2.0311,
+      "step": 1644
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000221697247706422,
+      "loss": 2.0121,
+      "step": 1645
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00022162844036697246,
+      "loss": 1.8775,
+      "step": 1646
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00022155963302752292,
+      "loss": 1.933,
+      "step": 1647
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00022149082568807338,
+      "loss": 1.823,
+      "step": 1648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00022142201834862384,
+      "loss": 1.8175,
+      "step": 1649
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002213532110091743,
+      "loss": 1.9192,
+      "step": 1650
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00022128440366972477,
+      "loss": 1.8093,
+      "step": 1651
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00022121559633027523,
+      "loss": 1.8638,
+      "step": 1652
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002211467889908257,
+      "loss": 1.8102,
+      "step": 1653
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00022107798165137612,
+      "loss": 1.645,
+      "step": 1654
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00022100917431192658,
+      "loss": 1.7941,
+      "step": 1655
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00022094036697247705,
+      "loss": 1.7335,
+      "step": 1656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002208715596330275,
+      "loss": 1.7053,
+      "step": 1657
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00022080275229357797,
+      "loss": 1.7755,
+      "step": 1658
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00022073394495412843,
+      "loss": 1.6338,
+      "step": 1659
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00022066513761467886,
+      "loss": 1.6653,
+      "step": 1660
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00022059633027522933,
+      "loss": 1.7996,
+      "step": 1661
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002205275229357798,
+      "loss": 1.4886,
+      "step": 1662
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00022045871559633025,
+      "loss": 1.4918,
+      "step": 1663
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002203899082568807,
+      "loss": 1.5135,
+      "step": 1664
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00022032110091743117,
+      "loss": 1.3269,
+      "step": 1665
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00022025229357798163,
+      "loss": 1.3267,
+      "step": 1666
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002201834862385321,
+      "loss": 1.2449,
+      "step": 1667
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00022011467889908255,
+      "loss": 1.2959,
+      "step": 1668
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00022004587155963302,
+      "loss": 1.1309,
+      "step": 1669
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00021997706422018345,
+      "loss": 1.244,
+      "step": 1670
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002199082568807339,
+      "loss": 3.6016,
+      "step": 1671
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00021983944954128437,
+      "loss": 3.0305,
+      "step": 1672
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00021977064220183483,
+      "loss": 2.8526,
+      "step": 1673
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002197018348623853,
+      "loss": 2.6346,
+      "step": 1674
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00021963302752293576,
+      "loss": 2.5328,
+      "step": 1675
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00021956422018348622,
+      "loss": 2.3991,
+      "step": 1676
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021949541284403668,
+      "loss": 2.2768,
+      "step": 1677
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021942660550458714,
+      "loss": 2.3904,
+      "step": 1678
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002193577981651376,
+      "loss": 2.3287,
+      "step": 1679
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021928899082568806,
+      "loss": 2.1604,
+      "step": 1680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021922018348623852,
+      "loss": 2.0885,
+      "step": 1681
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021915137614678898,
+      "loss": 2.2015,
+      "step": 1682
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021908256880733945,
+      "loss": 2.1696,
+      "step": 1683
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002190137614678899,
+      "loss": 2.2066,
+      "step": 1684
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021894495412844037,
+      "loss": 2.1416,
+      "step": 1685
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002188761467889908,
+      "loss": 2.0317,
+      "step": 1686
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021880733944954126,
+      "loss": 2.0359,
+      "step": 1687
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021873853211009173,
+      "loss": 2.1068,
+      "step": 1688
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002186697247706422,
+      "loss": 2.0652,
+      "step": 1689
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021860091743119265,
+      "loss": 2.0542,
+      "step": 1690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021853211009174308,
+      "loss": 2.0819,
+      "step": 1691
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00021846330275229354,
+      "loss": 1.9453,
+      "step": 1692
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000218394495412844,
+      "loss": 2.0384,
+      "step": 1693
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021832568807339447,
+      "loss": 2.0665,
+      "step": 1694
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021825688073394493,
+      "loss": 1.8885,
+      "step": 1695
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002181880733944954,
+      "loss": 1.9032,
+      "step": 1696
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021811926605504585,
+      "loss": 1.8915,
+      "step": 1697
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002180504587155963,
+      "loss": 1.8936,
+      "step": 1698
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021798165137614677,
+      "loss": 1.9411,
+      "step": 1699
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021791284403669723,
+      "loss": 1.9526,
+      "step": 1700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002178440366972477,
+      "loss": 1.7593,
+      "step": 1701
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021777522935779813,
+      "loss": 1.7568,
+      "step": 1702
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002177064220183486,
+      "loss": 1.7143,
+      "step": 1703
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021763761467889905,
+      "loss": 1.5789,
+      "step": 1704
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002175688073394495,
+      "loss": 1.8545,
+      "step": 1705
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021749999999999997,
+      "loss": 1.58,
+      "step": 1706
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021743119266055044,
+      "loss": 1.6144,
+      "step": 1707
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002173623853211009,
+      "loss": 1.6012,
+      "step": 1708
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00021729357798165136,
+      "loss": 1.615,
+      "step": 1709
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021722477064220182,
+      "loss": 1.6088,
+      "step": 1710
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021715596330275228,
+      "loss": 1.4713,
+      "step": 1711
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021708715596330274,
+      "loss": 1.3805,
+      "step": 1712
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002170183486238532,
+      "loss": 1.4281,
+      "step": 1713
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021694954128440366,
+      "loss": 1.4774,
+      "step": 1714
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021688073394495413,
+      "loss": 1.2971,
+      "step": 1715
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002168119266055046,
+      "loss": 1.1743,
+      "step": 1716
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021674311926605505,
+      "loss": 1.184,
+      "step": 1717
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021667431192660548,
+      "loss": 1.3324,
+      "step": 1718
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021660550458715594,
+      "loss": 1.16,
+      "step": 1719
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002165366972477064,
+      "loss": 1.2817,
+      "step": 1720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021646788990825687,
+      "loss": 3.5558,
+      "step": 1721
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002163990825688073,
+      "loss": 3.0307,
+      "step": 1722
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021633027522935776,
+      "loss": 2.6262,
+      "step": 1723
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021626146788990822,
+      "loss": 2.4631,
+      "step": 1724
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00021619266055045868,
+      "loss": 2.5842,
+      "step": 1725
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021612385321100915,
+      "loss": 2.4398,
+      "step": 1726
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002160550458715596,
+      "loss": 2.4056,
+      "step": 1727
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021598623853211007,
+      "loss": 2.4079,
+      "step": 1728
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021591743119266053,
+      "loss": 2.1906,
+      "step": 1729
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000215848623853211,
+      "loss": 2.2679,
+      "step": 1730
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021577981651376145,
+      "loss": 1.9226,
+      "step": 1731
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002157110091743119,
+      "loss": 2.0811,
+      "step": 1732
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021564220183486237,
+      "loss": 2.2899,
+      "step": 1733
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002155733944954128,
+      "loss": 2.0975,
+      "step": 1734
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021550458715596327,
+      "loss": 2.0941,
+      "step": 1735
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021543577981651373,
+      "loss": 1.97,
+      "step": 1736
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002153669724770642,
+      "loss": 2.0479,
+      "step": 1737
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021529816513761465,
+      "loss": 2.0814,
+      "step": 1738
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021522935779816511,
+      "loss": 1.9018,
+      "step": 1739
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021516055045871558,
+      "loss": 2.0003,
+      "step": 1740
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00021509174311926604,
+      "loss": 1.8782,
+      "step": 1741
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002150229357798165,
+      "loss": 1.8627,
+      "step": 1742
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021495412844036696,
+      "loss": 1.8832,
+      "step": 1743
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021488532110091742,
+      "loss": 1.8778,
+      "step": 1744
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021481651376146788,
+      "loss": 1.809,
+      "step": 1745
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021474770642201834,
+      "loss": 1.705,
+      "step": 1746
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002146788990825688,
+      "loss": 2.0336,
+      "step": 1747
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021461009174311927,
+      "loss": 1.9158,
+      "step": 1748
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021454128440366973,
+      "loss": 1.9045,
+      "step": 1749
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021447247706422016,
+      "loss": 1.9342,
+      "step": 1750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021440366972477062,
+      "loss": 1.9411,
+      "step": 1751
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021433486238532108,
+      "loss": 1.9156,
+      "step": 1752
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021426605504587155,
+      "loss": 1.8337,
+      "step": 1753
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021419724770642198,
+      "loss": 1.5281,
+      "step": 1754
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021412844036697244,
+      "loss": 1.6437,
+      "step": 1755
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002140596330275229,
+      "loss": 1.6482,
+      "step": 1756
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00021399082568807336,
+      "loss": 1.6374,
+      "step": 1757
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021392201834862382,
+      "loss": 1.6254,
+      "step": 1758
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021385321100917429,
+      "loss": 1.59,
+      "step": 1759
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021378440366972475,
+      "loss": 1.5096,
+      "step": 1760
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002137155963302752,
+      "loss": 1.5776,
+      "step": 1761
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021364678899082567,
+      "loss": 1.5149,
+      "step": 1762
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021357798165137613,
+      "loss": 1.5024,
+      "step": 1763
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002135091743119266,
+      "loss": 1.4055,
+      "step": 1764
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021344036697247705,
+      "loss": 1.2917,
+      "step": 1765
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002133715596330275,
+      "loss": 1.4198,
+      "step": 1766
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021330275229357795,
+      "loss": 1.4071,
+      "step": 1767
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002132339449541284,
+      "loss": 1.2486,
+      "step": 1768
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021316513761467887,
+      "loss": 1.1579,
+      "step": 1769
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021309633027522933,
+      "loss": 1.1932,
+      "step": 1770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002130275229357798,
+      "loss": 3.7279,
+      "step": 1771
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021295871559633026,
+      "loss": 3.1071,
+      "step": 1772
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00021288990825688072,
+      "loss": 2.5656,
+      "step": 1773
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00021282110091743118,
+      "loss": 2.5555,
+      "step": 1774
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00021275229357798164,
+      "loss": 2.4096,
+      "step": 1775
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002126834862385321,
+      "loss": 2.443,
+      "step": 1776
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00021261467889908256,
+      "loss": 2.247,
+      "step": 1777
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00021254587155963302,
+      "loss": 2.4117,
+      "step": 1778
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00021247706422018348,
+      "loss": 2.2131,
+      "step": 1779
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00021240825688073395,
+      "loss": 2.2154,
+      "step": 1780
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002123394495412844,
+      "loss": 2.1035,
+      "step": 1781
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00021227064220183484,
+      "loss": 2.099,
+      "step": 1782
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002122018348623853,
+      "loss": 1.9619,
+      "step": 1783
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00021213302752293576,
+      "loss": 2.0422,
+      "step": 1784
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002120642201834862,
+      "loss": 1.9492,
+      "step": 1785
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00021199541284403666,
+      "loss": 2.013,
+      "step": 1786
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00021192660550458712,
+      "loss": 2.0643,
+      "step": 1787
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00021185779816513758,
+      "loss": 2.1128,
+      "step": 1788
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00021178899082568804,
+      "loss": 2.069,
+      "step": 1789
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002117201834862385,
+      "loss": 2.1215,
+      "step": 1790
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00021165137614678897,
+      "loss": 1.9409,
+      "step": 1791
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00021158256880733943,
+      "loss": 1.8557,
+      "step": 1792
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002115137614678899,
+      "loss": 1.7989,
+      "step": 1793
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00021144495412844035,
+      "loss": 1.9805,
+      "step": 1794
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002113761467889908,
+      "loss": 1.8681,
+      "step": 1795
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00021130733944954127,
+      "loss": 1.8311,
+      "step": 1796
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00021123853211009173,
+      "loss": 1.8996,
+      "step": 1797
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00021116972477064217,
+      "loss": 1.9109,
+      "step": 1798
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00021110091743119263,
+      "loss": 1.8419,
+      "step": 1799
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002110321100917431,
+      "loss": 1.7898,
+      "step": 1800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00021096330275229355,
+      "loss": 1.6416,
+      "step": 1801
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000210894495412844,
+      "loss": 1.8661,
+      "step": 1802
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00021082568807339447,
+      "loss": 1.7898,
+      "step": 1803
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00021075688073394493,
+      "loss": 1.5966,
+      "step": 1804
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002106880733944954,
+      "loss": 1.6136,
+      "step": 1805
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00021061926605504586,
+      "loss": 1.7709,
+      "step": 1806
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00021055045871559632,
+      "loss": 1.6369,
+      "step": 1807
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00021048165137614678,
+      "loss": 1.6701,
+      "step": 1808
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00021041284403669724,
+      "loss": 1.5804,
+      "step": 1809
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002103440366972477,
+      "loss": 1.5023,
+      "step": 1810
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00021027522935779816,
+      "loss": 1.6007,
+      "step": 1811
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00021020642201834862,
+      "loss": 1.5211,
+      "step": 1812
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00021013761467889909,
+      "loss": 1.3998,
+      "step": 1813
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00021006880733944952,
+      "loss": 1.3186,
+      "step": 1814
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 1.3886,
+      "step": 1815
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020993119266055042,
+      "loss": 1.2077,
+      "step": 1816
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020986238532110088,
+      "loss": 1.1515,
+      "step": 1817
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020979357798165134,
+      "loss": 1.1892,
+      "step": 1818
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002097247706422018,
+      "loss": 1.1482,
+      "step": 1819
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020965596330275226,
+      "loss": 1.0993,
+      "step": 1820
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020958715596330272,
+      "loss": 3.5628,
+      "step": 1821
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00020951834862385318,
+      "loss": 3.0245,
+      "step": 1822
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020944954128440364,
+      "loss": 2.3985,
+      "step": 1823
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002093807339449541,
+      "loss": 2.2958,
+      "step": 1824
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020931192660550457,
+      "loss": 2.4073,
+      "step": 1825
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020924311926605503,
+      "loss": 2.2435,
+      "step": 1826
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002091743119266055,
+      "loss": 2.4422,
+      "step": 1827
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020910550458715595,
+      "loss": 2.2574,
+      "step": 1828
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002090366972477064,
+      "loss": 2.271,
+      "step": 1829
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020896788990825685,
+      "loss": 2.1299,
+      "step": 1830
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002088990825688073,
+      "loss": 2.2288,
+      "step": 1831
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020883027522935777,
+      "loss": 1.9979,
+      "step": 1832
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020876146788990823,
+      "loss": 2.3091,
+      "step": 1833
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002086926605504587,
+      "loss": 1.9847,
+      "step": 1834
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020862385321100915,
+      "loss": 2.1384,
+      "step": 1835
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020855504587155961,
+      "loss": 2.0983,
+      "step": 1836
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020848623853211008,
+      "loss": 2.1569,
+      "step": 1837
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00020841743119266054,
+      "loss": 2.1108,
+      "step": 1838
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000208348623853211,
+      "loss": 1.9124,
+      "step": 1839
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020827981651376146,
+      "loss": 1.9459,
+      "step": 1840
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020821100917431192,
+      "loss": 1.995,
+      "step": 1841
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020814220183486238,
+      "loss": 1.8326,
+      "step": 1842
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020807339449541284,
+      "loss": 1.9383,
+      "step": 1843
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002080045871559633,
+      "loss": 1.8695,
+      "step": 1844
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020793577981651377,
+      "loss": 1.8381,
+      "step": 1845
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002078669724770642,
+      "loss": 1.7778,
+      "step": 1846
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020779816513761463,
+      "loss": 1.9175,
+      "step": 1847
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002077293577981651,
+      "loss": 1.8504,
+      "step": 1848
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020766055045871556,
+      "loss": 1.7752,
+      "step": 1849
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020759174311926602,
+      "loss": 1.5618,
+      "step": 1850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020752293577981648,
+      "loss": 1.7225,
+      "step": 1851
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020745412844036694,
+      "loss": 1.8193,
+      "step": 1852
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002073853211009174,
+      "loss": 1.9269,
+      "step": 1853
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00020731651376146786,
+      "loss": 1.8738,
+      "step": 1854
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020724770642201832,
+      "loss": 1.7606,
+      "step": 1855
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020717889908256879,
+      "loss": 1.7587,
+      "step": 1856
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020711009174311925,
+      "loss": 1.5626,
+      "step": 1857
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002070412844036697,
+      "loss": 1.521,
+      "step": 1858
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020697247706422017,
+      "loss": 1.4554,
+      "step": 1859
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020690366972477063,
+      "loss": 1.5262,
+      "step": 1860
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002068348623853211,
+      "loss": 1.4124,
+      "step": 1861
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020676605504587153,
+      "loss": 1.5763,
+      "step": 1862
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000206697247706422,
+      "loss": 1.3704,
+      "step": 1863
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020662844036697245,
+      "loss": 1.3973,
+      "step": 1864
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002065596330275229,
+      "loss": 1.4078,
+      "step": 1865
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020649082568807337,
+      "loss": 1.2323,
+      "step": 1866
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020642201834862383,
+      "loss": 1.3553,
+      "step": 1867
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002063532110091743,
+      "loss": 1.159,
+      "step": 1868
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020628440366972475,
+      "loss": 1.1942,
+      "step": 1869
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020621559633027522,
+      "loss": 1.0791,
+      "step": 1870
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00020614678899082568,
+      "loss": 3.2909,
+      "step": 1871
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020607798165137614,
+      "loss": 2.9312,
+      "step": 1872
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002060091743119266,
+      "loss": 2.5817,
+      "step": 1873
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020594036697247706,
+      "loss": 2.2504,
+      "step": 1874
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020587155963302752,
+      "loss": 2.3404,
+      "step": 1875
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020580275229357798,
+      "loss": 2.4065,
+      "step": 1876
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020573394495412844,
+      "loss": 2.178,
+      "step": 1877
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020566513761467885,
+      "loss": 2.1733,
+      "step": 1878
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002055963302752293,
+      "loss": 2.3174,
+      "step": 1879
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020552752293577977,
+      "loss": 2.1098,
+      "step": 1880
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020545871559633024,
+      "loss": 2.2311,
+      "step": 1881
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002053899082568807,
+      "loss": 2.0851,
+      "step": 1882
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020532110091743116,
+      "loss": 2.0471,
+      "step": 1883
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020525229357798162,
+      "loss": 2.0604,
+      "step": 1884
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020518348623853208,
+      "loss": 2.0673,
+      "step": 1885
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00020511467889908254,
+      "loss": 1.9443,
+      "step": 1886
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000205045871559633,
+      "loss": 1.9506,
+      "step": 1887
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020497706422018346,
+      "loss": 2.0262,
+      "step": 1888
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020490825688073393,
+      "loss": 1.9836,
+      "step": 1889
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002048394495412844,
+      "loss": 2.0778,
+      "step": 1890
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020477064220183485,
+      "loss": 1.9802,
+      "step": 1891
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002047018348623853,
+      "loss": 1.8413,
+      "step": 1892
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020463302752293577,
+      "loss": 1.9037,
+      "step": 1893
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002045642201834862,
+      "loss": 1.8697,
+      "step": 1894
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020449541284403667,
+      "loss": 1.8834,
+      "step": 1895
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020442660550458713,
+      "loss": 1.7303,
+      "step": 1896
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002043577981651376,
+      "loss": 1.861,
+      "step": 1897
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020428899082568805,
+      "loss": 1.8614,
+      "step": 1898
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002042201834862385,
+      "loss": 1.772,
+      "step": 1899
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020415137614678897,
+      "loss": 1.8243,
+      "step": 1900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020408256880733943,
+      "loss": 1.695,
+      "step": 1901
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002040137614678899,
+      "loss": 1.6338,
+      "step": 1902
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00020394495412844036,
+      "loss": 1.8594,
+      "step": 1903
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020387614678899082,
+      "loss": 1.6646,
+      "step": 1904
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020380733944954128,
+      "loss": 1.5384,
+      "step": 1905
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020373853211009174,
+      "loss": 1.6177,
+      "step": 1906
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002036697247706422,
+      "loss": 1.5614,
+      "step": 1907
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020360091743119266,
+      "loss": 1.5878,
+      "step": 1908
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020353211009174312,
+      "loss": 1.5746,
+      "step": 1909
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020346330275229353,
+      "loss": 1.5444,
+      "step": 1910
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000203394495412844,
+      "loss": 1.5805,
+      "step": 1911
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020332568807339445,
+      "loss": 1.5324,
+      "step": 1912
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020325688073394492,
+      "loss": 1.3703,
+      "step": 1913
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020318807339449538,
+      "loss": 1.2171,
+      "step": 1914
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020311926605504584,
+      "loss": 1.3561,
+      "step": 1915
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002030504587155963,
+      "loss": 1.2731,
+      "step": 1916
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020298165137614676,
+      "loss": 1.1354,
+      "step": 1917
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020291284403669722,
+      "loss": 1.2504,
+      "step": 1918
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00020284403669724768,
+      "loss": 1.1603,
+      "step": 1919
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00020277522935779814,
+      "loss": 1.0721,
+      "step": 1920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002027064220183486,
+      "loss": 3.2151,
+      "step": 1921
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00020263761467889907,
+      "loss": 2.8365,
+      "step": 1922
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00020256880733944953,
+      "loss": 2.3569,
+      "step": 1923
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002025,
+      "loss": 2.439,
+      "step": 1924
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00020243119266055045,
+      "loss": 2.2511,
+      "step": 1925
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00020236238532110088,
+      "loss": 2.1246,
+      "step": 1926
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00020229357798165135,
+      "loss": 2.1318,
+      "step": 1927
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002022247706422018,
+      "loss": 2.1898,
+      "step": 1928
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00020215596330275227,
+      "loss": 2.0258,
+      "step": 1929
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00020208715596330273,
+      "loss": 2.0232,
+      "step": 1930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002020183486238532,
+      "loss": 2.1099,
+      "step": 1931
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00020194954128440365,
+      "loss": 2.1024,
+      "step": 1932
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002018807339449541,
+      "loss": 1.9652,
+      "step": 1933
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00020181192660550457,
+      "loss": 1.7831,
+      "step": 1934
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00020174311926605504,
+      "loss": 1.9607,
+      "step": 1935
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002016743119266055,
+      "loss": 1.9349,
+      "step": 1936
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00020160550458715596,
+      "loss": 1.9872,
+      "step": 1937
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00020153669724770642,
+      "loss": 1.8682,
+      "step": 1938
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00020146788990825688,
+      "loss": 1.8741,
+      "step": 1939
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00020139908256880734,
+      "loss": 1.9381,
+      "step": 1940
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002013302752293578,
+      "loss": 1.8689,
+      "step": 1941
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002012614678899082,
+      "loss": 1.8468,
+      "step": 1942
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00020119266055045867,
+      "loss": 1.8137,
+      "step": 1943
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00020112385321100913,
+      "loss": 1.6493,
+      "step": 1944
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002010550458715596,
+      "loss": 1.7694,
+      "step": 1945
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00020098623853211006,
+      "loss": 1.7595,
+      "step": 1946
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00020091743119266052,
+      "loss": 1.7865,
+      "step": 1947
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00020084862385321098,
+      "loss": 1.9081,
+      "step": 1948
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00020077981651376144,
+      "loss": 1.7594,
+      "step": 1949
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002007110091743119,
+      "loss": 1.6857,
+      "step": 1950
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00020064220183486236,
+      "loss": 1.7442,
+      "step": 1951
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00020057339449541282,
+      "loss": 1.6748,
+      "step": 1952
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00020050458715596328,
+      "loss": 1.6105,
+      "step": 1953
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00020043577981651375,
+      "loss": 1.6602,
+      "step": 1954
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002003669724770642,
+      "loss": 1.6126,
+      "step": 1955
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00020029816513761467,
+      "loss": 1.639,
+      "step": 1956
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00020022935779816513,
+      "loss": 1.509,
+      "step": 1957
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00020016055045871556,
+      "loss": 1.6552,
+      "step": 1958
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00020009174311926603,
+      "loss": 1.6348,
+      "step": 1959
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00020002293577981649,
+      "loss": 1.4605,
+      "step": 1960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019995412844036695,
+      "loss": 1.3822,
+      "step": 1961
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001998853211009174,
+      "loss": 1.3392,
+      "step": 1962
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019981651376146787,
+      "loss": 1.4034,
+      "step": 1963
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019974770642201833,
+      "loss": 1.2861,
+      "step": 1964
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001996788990825688,
+      "loss": 1.2665,
+      "step": 1965
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019961009174311925,
+      "loss": 1.2643,
+      "step": 1966
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019954128440366972,
+      "loss": 1.1004,
+      "step": 1967
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00019947247706422018,
+      "loss": 1.0879,
+      "step": 1968
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019940366972477064,
+      "loss": 1.0452,
+      "step": 1969
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001993348623853211,
+      "loss": 1.0075,
+      "step": 1970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019926605504587156,
+      "loss": 3.1939,
+      "step": 1971
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019919724770642202,
+      "loss": 2.6152,
+      "step": 1972
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019912844036697248,
+      "loss": 2.4699,
+      "step": 1973
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001990596330275229,
+      "loss": 2.4007,
+      "step": 1974
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019899082568807335,
+      "loss": 2.1458,
+      "step": 1975
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001989220183486238,
+      "loss": 2.3048,
+      "step": 1976
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019885321100917427,
+      "loss": 2.1876,
+      "step": 1977
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019878440366972474,
+      "loss": 2.0,
+      "step": 1978
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001987155963302752,
+      "loss": 1.9698,
+      "step": 1979
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019864678899082566,
+      "loss": 2.104,
+      "step": 1980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019857798165137612,
+      "loss": 2.1339,
+      "step": 1981
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019850917431192658,
+      "loss": 2.0213,
+      "step": 1982
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00019844036697247704,
+      "loss": 1.992,
+      "step": 1983
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001983715596330275,
+      "loss": 2.2249,
+      "step": 1984
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019830275229357796,
+      "loss": 2.0143,
+      "step": 1985
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019823394495412843,
+      "loss": 1.853,
+      "step": 1986
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019816513761467889,
+      "loss": 2.0188,
+      "step": 1987
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019809633027522935,
+      "loss": 1.884,
+      "step": 1988
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001980275229357798,
+      "loss": 1.9637,
+      "step": 1989
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019795871559633024,
+      "loss": 1.77,
+      "step": 1990
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001978899082568807,
+      "loss": 1.8534,
+      "step": 1991
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019782110091743117,
+      "loss": 1.9073,
+      "step": 1992
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019775229357798163,
+      "loss": 1.8242,
+      "step": 1993
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001976834862385321,
+      "loss": 1.9404,
+      "step": 1994
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019761467889908255,
+      "loss": 1.6926,
+      "step": 1995
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000197545871559633,
+      "loss": 1.8429,
+      "step": 1996
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019747706422018347,
+      "loss": 1.8468,
+      "step": 1997
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019740825688073393,
+      "loss": 1.7018,
+      "step": 1998
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001973394495412844,
+      "loss": 1.7915,
+      "step": 1999
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00019727064220183486,
+      "loss": 1.6507,
+      "step": 2000
+    },
+    {
+      "epoch": 1.23,
+      "eval_bleu": 1.7162192922202172e-16,
+      "eval_loss": 2.2250263690948486,
+      "eval_runtime": 2534.2548,
+      "eval_samples_per_second": 5.824,
+      "eval_steps_per_second": 0.728,
+      "step": 2000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019720183486238532,
+      "loss": 1.709,
+      "step": 2001
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019713302752293578,
+      "loss": 1.6782,
+      "step": 2002
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019706422018348624,
+      "loss": 1.5424,
+      "step": 2003
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001969954128440367,
+      "loss": 1.6469,
+      "step": 2004
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019692660550458716,
+      "loss": 1.8118,
+      "step": 2005
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019685779816513757,
+      "loss": 1.5385,
+      "step": 2006
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019678899082568803,
+      "loss": 1.4475,
+      "step": 2007
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001967201834862385,
+      "loss": 1.6445,
+      "step": 2008
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019665137614678895,
+      "loss": 1.4671,
+      "step": 2009
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019658256880733941,
+      "loss": 1.5471,
+      "step": 2010
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019651376146788988,
+      "loss": 1.532,
+      "step": 2011
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019644495412844034,
+      "loss": 1.4689,
+      "step": 2012
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001963761467889908,
+      "loss": 1.4248,
+      "step": 2013
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019630733944954126,
+      "loss": 1.362,
+      "step": 2014
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019623853211009172,
+      "loss": 1.3883,
+      "step": 2015
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00019616972477064218,
+      "loss": 1.2565,
+      "step": 2016
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019610091743119264,
+      "loss": 1.2136,
+      "step": 2017
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001960321100917431,
+      "loss": 1.1219,
+      "step": 2018
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019596330275229357,
+      "loss": 1.0602,
+      "step": 2019
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019589449541284403,
+      "loss": 0.9342,
+      "step": 2020
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001958256880733945,
+      "loss": 3.1753,
+      "step": 2021
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019575688073394492,
+      "loss": 2.5807,
+      "step": 2022
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019568807339449538,
+      "loss": 2.4483,
+      "step": 2023
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019561926605504585,
+      "loss": 2.1783,
+      "step": 2024
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001955504587155963,
+      "loss": 2.2004,
+      "step": 2025
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019548165137614677,
+      "loss": 2.2108,
+      "step": 2026
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019541284403669723,
+      "loss": 2.2816,
+      "step": 2027
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001953440366972477,
+      "loss": 2.0895,
+      "step": 2028
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019527522935779815,
+      "loss": 2.143,
+      "step": 2029
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001952064220183486,
+      "loss": 2.005,
+      "step": 2030
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019513761467889907,
+      "loss": 1.9381,
+      "step": 2031
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00019506880733944954,
+      "loss": 1.8762,
+      "step": 2032
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000195,
+      "loss": 2.1054,
+      "step": 2033
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019493119266055046,
+      "loss": 1.8601,
+      "step": 2034
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019486238532110092,
+      "loss": 1.9703,
+      "step": 2035
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019479357798165138,
+      "loss": 2.0773,
+      "step": 2036
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019472477064220184,
+      "loss": 1.8942,
+      "step": 2037
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019465596330275225,
+      "loss": 1.9202,
+      "step": 2038
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001945871559633027,
+      "loss": 1.7541,
+      "step": 2039
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019451834862385317,
+      "loss": 1.7267,
+      "step": 2040
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019444954128440363,
+      "loss": 1.8747,
+      "step": 2041
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001943807339449541,
+      "loss": 2.0165,
+      "step": 2042
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019431192660550455,
+      "loss": 1.8279,
+      "step": 2043
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019424311926605502,
+      "loss": 1.7813,
+      "step": 2044
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019417431192660548,
+      "loss": 1.7252,
+      "step": 2045
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019410550458715594,
+      "loss": 1.6401,
+      "step": 2046
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001940366972477064,
+      "loss": 1.6962,
+      "step": 2047
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019396788990825686,
+      "loss": 1.7297,
+      "step": 2048
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00019389908256880732,
+      "loss": 1.738,
+      "step": 2049
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019383027522935778,
+      "loss": 1.7066,
+      "step": 2050
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019376146788990825,
+      "loss": 1.6057,
+      "step": 2051
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001936926605504587,
+      "loss": 1.7023,
+      "step": 2052
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019362385321100917,
+      "loss": 1.528,
+      "step": 2053
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001935550458715596,
+      "loss": 1.5115,
+      "step": 2054
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019348623853211006,
+      "loss": 1.4674,
+      "step": 2055
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019341743119266052,
+      "loss": 1.3539,
+      "step": 2056
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019334862385321099,
+      "loss": 1.6316,
+      "step": 2057
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019327981651376145,
+      "loss": 1.7101,
+      "step": 2058
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001932110091743119,
+      "loss": 1.4022,
+      "step": 2059
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019314220183486237,
+      "loss": 1.5265,
+      "step": 2060
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019307339449541283,
+      "loss": 1.3397,
+      "step": 2061
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001930045871559633,
+      "loss": 1.4825,
+      "step": 2062
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019293577981651375,
+      "loss": 1.3488,
+      "step": 2063
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019286697247706421,
+      "loss": 1.3008,
+      "step": 2064
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00019279816513761468,
+      "loss": 1.227,
+      "step": 2065
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00019272935779816514,
+      "loss": 1.1107,
+      "step": 2066
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001926605504587156,
+      "loss": 1.061,
+      "step": 2067
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00019259174311926606,
+      "loss": 1.1078,
+      "step": 2068
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00019252293577981652,
+      "loss": 0.998,
+      "step": 2069
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00019245412844036693,
+      "loss": 0.9788,
+      "step": 2070
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001923853211009174,
+      "loss": 3.155,
+      "step": 2071
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00019231651376146785,
+      "loss": 2.6992,
+      "step": 2072
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001922477064220183,
+      "loss": 2.3502,
+      "step": 2073
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00019217889908256877,
+      "loss": 2.2841,
+      "step": 2074
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00019211009174311923,
+      "loss": 2.0852,
+      "step": 2075
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001920412844036697,
+      "loss": 2.0464,
+      "step": 2076
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00019197247706422016,
+      "loss": 2.1334,
+      "step": 2077
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00019190366972477062,
+      "loss": 2.1189,
+      "step": 2078
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00019183486238532108,
+      "loss": 2.1277,
+      "step": 2079
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00019176605504587154,
+      "loss": 2.1875,
+      "step": 2080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000191697247706422,
+      "loss": 1.8635,
+      "step": 2081
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00019162844036697246,
+      "loss": 2.1077,
+      "step": 2082
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00019155963302752292,
+      "loss": 1.978,
+      "step": 2083
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00019149082568807339,
+      "loss": 1.9189,
+      "step": 2084
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00019142201834862385,
+      "loss": 2.0011,
+      "step": 2085
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00019135321100917428,
+      "loss": 1.7829,
+      "step": 2086
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00019128440366972474,
+      "loss": 1.8061,
+      "step": 2087
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001912155963302752,
+      "loss": 1.9791,
+      "step": 2088
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00019114678899082566,
+      "loss": 1.9986,
+      "step": 2089
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00019107798165137613,
+      "loss": 1.7943,
+      "step": 2090
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001910091743119266,
+      "loss": 1.8172,
+      "step": 2091
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00019094036697247705,
+      "loss": 1.8964,
+      "step": 2092
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001908715596330275,
+      "loss": 1.7612,
+      "step": 2093
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00019080275229357797,
+      "loss": 1.6614,
+      "step": 2094
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00019073394495412843,
+      "loss": 1.794,
+      "step": 2095
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001906651376146789,
+      "loss": 1.7443,
+      "step": 2096
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00019059633027522936,
+      "loss": 1.8229,
+      "step": 2097
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00019052752293577982,
+      "loss": 1.7122,
+      "step": 2098
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00019045871559633028,
+      "loss": 1.5759,
+      "step": 2099
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00019038990825688074,
+      "loss": 1.7618,
+      "step": 2100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001903211009174312,
+      "loss": 1.603,
+      "step": 2101
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001902522935779816,
+      "loss": 1.5539,
+      "step": 2102
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00019018348623853207,
+      "loss": 1.5407,
+      "step": 2103
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00019011467889908253,
+      "loss": 1.6749,
+      "step": 2104
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000190045871559633,
+      "loss": 1.5617,
+      "step": 2105
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018997706422018345,
+      "loss": 1.6097,
+      "step": 2106
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018990825688073391,
+      "loss": 1.6182,
+      "step": 2107
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018983944954128437,
+      "loss": 1.5297,
+      "step": 2108
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018977064220183484,
+      "loss": 1.6091,
+      "step": 2109
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001897018348623853,
+      "loss": 1.4417,
+      "step": 2110
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018963302752293576,
+      "loss": 1.2602,
+      "step": 2111
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018956422018348622,
+      "loss": 1.4811,
+      "step": 2112
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018949541284403668,
+      "loss": 1.372,
+      "step": 2113
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018942660550458714,
+      "loss": 1.2413,
+      "step": 2114
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001893577981651376,
+      "loss": 1.1568,
+      "step": 2115
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018928899082568807,
+      "loss": 1.2316,
+      "step": 2116
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018922018348623853,
+      "loss": 1.1584,
+      "step": 2117
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018915137614678896,
+      "loss": 1.1388,
+      "step": 2118
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018908256880733942,
+      "loss": 1.0518,
+      "step": 2119
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018901376146788988,
+      "loss": 1.0085,
+      "step": 2120
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018894495412844034,
+      "loss": 3.0042,
+      "step": 2121
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001888761467889908,
+      "loss": 2.5236,
+      "step": 2122
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018880733944954127,
+      "loss": 2.2717,
+      "step": 2123
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018873853211009173,
+      "loss": 2.1288,
+      "step": 2124
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001886697247706422,
+      "loss": 2.0028,
+      "step": 2125
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018860091743119265,
+      "loss": 2.0841,
+      "step": 2126
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001885321100917431,
+      "loss": 2.1967,
+      "step": 2127
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018846330275229357,
+      "loss": 1.833,
+      "step": 2128
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018839449541284403,
+      "loss": 2.0022,
+      "step": 2129
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001883256880733945,
+      "loss": 2.0113,
+      "step": 2130
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018825688073394496,
+      "loss": 1.9534,
+      "step": 2131
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018818807339449542,
+      "loss": 1.8513,
+      "step": 2132
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018811926605504588,
+      "loss": 1.8559,
+      "step": 2133
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001880504587155963,
+      "loss": 1.9258,
+      "step": 2134
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018798165137614675,
+      "loss": 1.8945,
+      "step": 2135
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001879128440366972,
+      "loss": 1.8376,
+      "step": 2136
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018784403669724767,
+      "loss": 1.8323,
+      "step": 2137
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018777522935779813,
+      "loss": 1.8517,
+      "step": 2138
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001877064220183486,
+      "loss": 1.8802,
+      "step": 2139
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018763761467889905,
+      "loss": 1.8161,
+      "step": 2140
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018756880733944952,
+      "loss": 1.7722,
+      "step": 2141
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018749999999999998,
+      "loss": 1.7697,
+      "step": 2142
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018743119266055044,
+      "loss": 1.9684,
+      "step": 2143
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001873623853211009,
+      "loss": 1.8625,
+      "step": 2144
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018729357798165136,
+      "loss": 1.68,
+      "step": 2145
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018722477064220182,
+      "loss": 1.7741,
+      "step": 2146
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018715596330275228,
+      "loss": 1.8442,
+      "step": 2147
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018708715596330274,
+      "loss": 1.7399,
+      "step": 2148
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001870183486238532,
+      "loss": 1.7896,
+      "step": 2149
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018694954128440367,
+      "loss": 1.7657,
+      "step": 2150
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001868807339449541,
+      "loss": 1.7501,
+      "step": 2151
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018681192660550456,
+      "loss": 1.6063,
+      "step": 2152
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018674311926605502,
+      "loss": 1.5323,
+      "step": 2153
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018667431192660548,
+      "loss": 1.5954,
+      "step": 2154
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018660550458715595,
+      "loss": 1.4607,
+      "step": 2155
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001865366972477064,
+      "loss": 1.6201,
+      "step": 2156
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018646788990825687,
+      "loss": 1.3629,
+      "step": 2157
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018639908256880733,
+      "loss": 1.3295,
+      "step": 2158
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001863302752293578,
+      "loss": 1.2844,
+      "step": 2159
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018626146788990825,
+      "loss": 1.3591,
+      "step": 2160
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018619266055045871,
+      "loss": 1.3799,
+      "step": 2161
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018612385321100918,
+      "loss": 1.4718,
+      "step": 2162
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018605504587155964,
+      "loss": 1.3193,
+      "step": 2163
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001859862385321101,
+      "loss": 1.2736,
+      "step": 2164
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018591743119266056,
+      "loss": 1.2972,
+      "step": 2165
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018584862385321102,
+      "loss": 1.2635,
+      "step": 2166
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018577981651376143,
+      "loss": 1.1203,
+      "step": 2167
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001857110091743119,
+      "loss": 1.0905,
+      "step": 2168
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018564220183486235,
+      "loss": 0.9921,
+      "step": 2169
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001855733944954128,
+      "loss": 1.0113,
+      "step": 2170
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018550458715596327,
+      "loss": 3.0903,
+      "step": 2171
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018543577981651373,
+      "loss": 2.6619,
+      "step": 2172
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001853669724770642,
+      "loss": 2.2993,
+      "step": 2173
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018529816513761466,
+      "loss": 2.1252,
+      "step": 2174
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018522935779816512,
+      "loss": 2.1888,
+      "step": 2175
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018516055045871558,
+      "loss": 2.1205,
+      "step": 2176
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018509174311926604,
+      "loss": 1.9359,
+      "step": 2177
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001850229357798165,
+      "loss": 2.0144,
+      "step": 2178
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018495412844036696,
+      "loss": 2.1088,
+      "step": 2179
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018488532110091742,
+      "loss": 2.1593,
+      "step": 2180
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018481651376146789,
+      "loss": 2.0112,
+      "step": 2181
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018474770642201835,
+      "loss": 1.8821,
+      "step": 2182
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018467889908256878,
+      "loss": 1.8886,
+      "step": 2183
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018461009174311924,
+      "loss": 2.0067,
+      "step": 2184
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001845412844036697,
+      "loss": 1.814,
+      "step": 2185
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018447247706422016,
+      "loss": 1.9341,
+      "step": 2186
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018440366972477063,
+      "loss": 1.8397,
+      "step": 2187
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001843348623853211,
+      "loss": 1.7073,
+      "step": 2188
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018426605504587155,
+      "loss": 1.8368,
+      "step": 2189
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000184197247706422,
+      "loss": 1.8199,
+      "step": 2190
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018412844036697247,
+      "loss": 1.8054,
+      "step": 2191
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018405963302752293,
+      "loss": 1.7279,
+      "step": 2192
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001839908256880734,
+      "loss": 1.8178,
+      "step": 2193
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018392201834862385,
+      "loss": 1.778,
+      "step": 2194
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018385321100917432,
+      "loss": 1.6164,
+      "step": 2195
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018378440366972478,
+      "loss": 1.653,
+      "step": 2196
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018371559633027524,
+      "loss": 1.6927,
+      "step": 2197
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001836467889908257,
+      "loss": 1.6417,
+      "step": 2198
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001835779816513761,
+      "loss": 1.5684,
+      "step": 2199
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018350917431192657,
+      "loss": 1.5989,
+      "step": 2200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018344036697247703,
+      "loss": 1.577,
+      "step": 2201
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001833715596330275,
+      "loss": 1.5353,
+      "step": 2202
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018330275229357795,
+      "loss": 1.5282,
+      "step": 2203
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001832339449541284,
+      "loss": 1.4547,
+      "step": 2204
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018316513761467887,
+      "loss": 1.3983,
+      "step": 2205
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018309633027522934,
+      "loss": 1.4681,
+      "step": 2206
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001830275229357798,
+      "loss": 1.4729,
+      "step": 2207
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018295871559633026,
+      "loss": 1.5094,
+      "step": 2208
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018288990825688072,
+      "loss": 1.4348,
+      "step": 2209
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018282110091743118,
+      "loss": 1.3821,
+      "step": 2210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018275229357798164,
+      "loss": 1.4742,
+      "step": 2211
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001826834862385321,
+      "loss": 1.4456,
+      "step": 2212
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018261467889908256,
+      "loss": 1.3444,
+      "step": 2213
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018254587155963303,
+      "loss": 1.195,
+      "step": 2214
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018247706422018346,
+      "loss": 1.4143,
+      "step": 2215
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018240825688073392,
+      "loss": 1.2932,
+      "step": 2216
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018233944954128438,
+      "loss": 1.1407,
+      "step": 2217
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018227064220183484,
+      "loss": 1.0317,
+      "step": 2218
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001822018348623853,
+      "loss": 0.9022,
+      "step": 2219
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018213302752293577,
+      "loss": 1.1126,
+      "step": 2220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018206422018348623,
+      "loss": 3.0819,
+      "step": 2221
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001819954128440367,
+      "loss": 2.5293,
+      "step": 2222
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018192660550458715,
+      "loss": 2.3659,
+      "step": 2223
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001818577981651376,
+      "loss": 2.2317,
+      "step": 2224
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018178899082568807,
+      "loss": 2.1227,
+      "step": 2225
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018172018348623853,
+      "loss": 2.0041,
+      "step": 2226
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000181651376146789,
+      "loss": 2.0501,
+      "step": 2227
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00018158256880733946,
+      "loss": 1.9934,
+      "step": 2228
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00018151376146788992,
+      "loss": 1.9805,
+      "step": 2229
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00018144495412844038,
+      "loss": 1.9281,
+      "step": 2230
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00018137614678899079,
+      "loss": 2.092,
+      "step": 2231
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00018130733944954125,
+      "loss": 1.8745,
+      "step": 2232
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001812385321100917,
+      "loss": 1.8711,
+      "step": 2233
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00018116972477064217,
+      "loss": 1.9596,
+      "step": 2234
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00018110091743119263,
+      "loss": 1.9111,
+      "step": 2235
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001810321100917431,
+      "loss": 1.8684,
+      "step": 2236
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00018096330275229355,
+      "loss": 1.8264,
+      "step": 2237
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00018089449541284401,
+      "loss": 1.8391,
+      "step": 2238
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00018082568807339448,
+      "loss": 1.8479,
+      "step": 2239
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00018075688073394494,
+      "loss": 1.9619,
+      "step": 2240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001806880733944954,
+      "loss": 1.8425,
+      "step": 2241
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00018061926605504586,
+      "loss": 1.8801,
+      "step": 2242
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00018055045871559632,
+      "loss": 1.8278,
+      "step": 2243
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00018048165137614678,
+      "loss": 1.7898,
+      "step": 2244
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00018041284403669724,
+      "loss": 1.7335,
+      "step": 2245
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001803440366972477,
+      "loss": 1.6493,
+      "step": 2246
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00018027522935779814,
+      "loss": 1.6948,
+      "step": 2247
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001802064220183486,
+      "loss": 1.6788,
+      "step": 2248
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00018013761467889906,
+      "loss": 1.7021,
+      "step": 2249
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00018006880733944952,
+      "loss": 1.5245,
+      "step": 2250
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 1.4886,
+      "step": 2251
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017993119266055045,
+      "loss": 1.5184,
+      "step": 2252
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001798623853211009,
+      "loss": 1.3526,
+      "step": 2253
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017979357798165137,
+      "loss": 1.4944,
+      "step": 2254
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017972477064220183,
+      "loss": 1.4891,
+      "step": 2255
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001796559633027523,
+      "loss": 1.5575,
+      "step": 2256
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017958715596330275,
+      "loss": 1.4377,
+      "step": 2257
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001795183486238532,
+      "loss": 1.4172,
+      "step": 2258
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017944954128440367,
+      "loss": 1.3875,
+      "step": 2259
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017938073394495414,
+      "loss": 1.4868,
+      "step": 2260
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001793119266055046,
+      "loss": 1.335,
+      "step": 2261
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017924311926605506,
+      "loss": 1.3073,
+      "step": 2262
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017917431192660547,
+      "loss": 1.238,
+      "step": 2263
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017910550458715593,
+      "loss": 1.2034,
+      "step": 2264
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001790366972477064,
+      "loss": 1.2099,
+      "step": 2265
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017896788990825685,
+      "loss": 1.2746,
+      "step": 2266
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001788990825688073,
+      "loss": 1.1214,
+      "step": 2267
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017883027522935777,
+      "loss": 1.141,
+      "step": 2268
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017876146788990823,
+      "loss": 1.1525,
+      "step": 2269
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001786926605504587,
+      "loss": 0.9149,
+      "step": 2270
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017862385321100916,
+      "loss": 2.7913,
+      "step": 2271
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017855504587155962,
+      "loss": 2.5904,
+      "step": 2272
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017848623853211008,
+      "loss": 2.4769,
+      "step": 2273
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017841743119266054,
+      "loss": 2.2475,
+      "step": 2274
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000178348623853211,
+      "loss": 1.9642,
+      "step": 2275
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017827981651376146,
+      "loss": 1.957,
+      "step": 2276
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017821100917431192,
+      "loss": 2.0239,
+      "step": 2277
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017814220183486238,
+      "loss": 2.0304,
+      "step": 2278
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017807339449541282,
+      "loss": 2.0484,
+      "step": 2279
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017800458715596328,
+      "loss": 1.8238,
+      "step": 2280
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017793577981651374,
+      "loss": 1.9024,
+      "step": 2281
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001778669724770642,
+      "loss": 1.7931,
+      "step": 2282
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017779816513761466,
+      "loss": 1.7485,
+      "step": 2283
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017772935779816512,
+      "loss": 1.8179,
+      "step": 2284
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017766055045871559,
+      "loss": 1.9105,
+      "step": 2285
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017759174311926605,
+      "loss": 1.7327,
+      "step": 2286
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001775229357798165,
+      "loss": 1.8671,
+      "step": 2287
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017745412844036697,
+      "loss": 1.8943,
+      "step": 2288
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017738532110091743,
+      "loss": 1.765,
+      "step": 2289
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001773165137614679,
+      "loss": 1.704,
+      "step": 2290
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017724770642201835,
+      "loss": 1.7748,
+      "step": 2291
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017717889908256882,
+      "loss": 1.74,
+      "step": 2292
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017711009174311928,
+      "loss": 1.8968,
+      "step": 2293
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017704128440366974,
+      "loss": 1.7546,
+      "step": 2294
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017697247706422014,
+      "loss": 1.7636,
+      "step": 2295
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001769036697247706,
+      "loss": 1.7031,
+      "step": 2296
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017683486238532107,
+      "loss": 1.6796,
+      "step": 2297
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017676605504587153,
+      "loss": 1.5201,
+      "step": 2298
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000176697247706422,
+      "loss": 1.7066,
+      "step": 2299
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017662844036697245,
+      "loss": 1.7178,
+      "step": 2300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001765596330275229,
+      "loss": 1.5365,
+      "step": 2301
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017649082568807337,
+      "loss": 1.6511,
+      "step": 2302
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017642201834862383,
+      "loss": 1.4195,
+      "step": 2303
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001763532110091743,
+      "loss": 1.5129,
+      "step": 2304
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017628440366972476,
+      "loss": 1.5296,
+      "step": 2305
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017621559633027522,
+      "loss": 1.5313,
+      "step": 2306
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017614678899082568,
+      "loss": 1.5138,
+      "step": 2307
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017607798165137614,
+      "loss": 1.3979,
+      "step": 2308
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001760091743119266,
+      "loss": 1.4173,
+      "step": 2309
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017594036697247706,
+      "loss": 1.3764,
+      "step": 2310
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001758715596330275,
+      "loss": 1.2286,
+      "step": 2311
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017580275229357796,
+      "loss": 1.2268,
+      "step": 2312
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017573394495412842,
+      "loss": 1.29,
+      "step": 2313
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017566513761467888,
+      "loss": 1.2289,
+      "step": 2314
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017559633027522934,
+      "loss": 1.2193,
+      "step": 2315
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001755275229357798,
+      "loss": 1.1629,
+      "step": 2316
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017545871559633027,
+      "loss": 1.0802,
+      "step": 2317
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017538990825688073,
+      "loss": 1.0788,
+      "step": 2318
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001753211009174312,
+      "loss": 1.044,
+      "step": 2319
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017525229357798165,
+      "loss": 0.9904,
+      "step": 2320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001751834862385321,
+      "loss": 2.7566,
+      "step": 2321
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017511467889908257,
+      "loss": 2.4943,
+      "step": 2322
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017504587155963303,
+      "loss": 2.2526,
+      "step": 2323
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001749770642201835,
+      "loss": 2.0902,
+      "step": 2324
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017490825688073396,
+      "loss": 2.0279,
+      "step": 2325
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017483944954128442,
+      "loss": 2.1295,
+      "step": 2326
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017477064220183482,
+      "loss": 2.0603,
+      "step": 2327
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017470183486238529,
+      "loss": 2.1415,
+      "step": 2328
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017463302752293575,
+      "loss": 1.8661,
+      "step": 2329
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001745642201834862,
+      "loss": 1.8926,
+      "step": 2330
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017449541284403667,
+      "loss": 1.8352,
+      "step": 2331
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017442660550458713,
+      "loss": 1.8973,
+      "step": 2332
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001743577981651376,
+      "loss": 1.7612,
+      "step": 2333
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017428899082568805,
+      "loss": 1.8255,
+      "step": 2334
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017422018348623851,
+      "loss": 1.7843,
+      "step": 2335
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017415137614678898,
+      "loss": 1.7908,
+      "step": 2336
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017408256880733944,
+      "loss": 1.7631,
+      "step": 2337
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001740137614678899,
+      "loss": 1.8242,
+      "step": 2338
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017394495412844036,
+      "loss": 1.8142,
+      "step": 2339
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017387614678899082,
+      "loss": 1.743,
+      "step": 2340
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017380733944954128,
+      "loss": 1.8156,
+      "step": 2341
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017373853211009174,
+      "loss": 1.8141,
+      "step": 2342
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017366972477064218,
+      "loss": 1.653,
+      "step": 2343
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017360091743119264,
+      "loss": 1.7386,
+      "step": 2344
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001735321100917431,
+      "loss": 1.6271,
+      "step": 2345
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017346330275229356,
+      "loss": 1.5777,
+      "step": 2346
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017339449541284402,
+      "loss": 1.5498,
+      "step": 2347
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017332568807339448,
+      "loss": 1.6421,
+      "step": 2348
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017325688073394494,
+      "loss": 1.667,
+      "step": 2349
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001731880733944954,
+      "loss": 1.6007,
+      "step": 2350
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017311926605504587,
+      "loss": 1.5331,
+      "step": 2351
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017305045871559633,
+      "loss": 1.4682,
+      "step": 2352
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001729816513761468,
+      "loss": 1.6398,
+      "step": 2353
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017291284403669725,
+      "loss": 1.4617,
+      "step": 2354
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001728440366972477,
+      "loss": 1.3861,
+      "step": 2355
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017277522935779817,
+      "loss": 1.3548,
+      "step": 2356
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017270642201834863,
+      "loss": 1.4008,
+      "step": 2357
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001726376146788991,
+      "loss": 1.2727,
+      "step": 2358
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001725688073394495,
+      "loss": 1.2984,
+      "step": 2359
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017249999999999996,
+      "loss": 1.2728,
+      "step": 2360
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017243119266055043,
+      "loss": 1.255,
+      "step": 2361
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001723623853211009,
+      "loss": 1.2775,
+      "step": 2362
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017229357798165135,
+      "loss": 1.3776,
+      "step": 2363
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001722247706422018,
+      "loss": 1.2018,
+      "step": 2364
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017215596330275227,
+      "loss": 1.0652,
+      "step": 2365
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017208715596330273,
+      "loss": 1.1631,
+      "step": 2366
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001720183486238532,
+      "loss": 1.1331,
+      "step": 2367
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017194954128440365,
+      "loss": 0.9859,
+      "step": 2368
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017188073394495412,
+      "loss": 1.1434,
+      "step": 2369
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017181192660550458,
+      "loss": 0.8219,
+      "step": 2370
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017174311926605504,
+      "loss": 2.6552,
+      "step": 2371
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001716743119266055,
+      "loss": 2.6183,
+      "step": 2372
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017160550458715596,
+      "loss": 2.2357,
+      "step": 2373
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017153669724770642,
+      "loss": 2.1045,
+      "step": 2374
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017146788990825686,
+      "loss": 2.2228,
+      "step": 2375
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017139908256880732,
+      "loss": 1.99,
+      "step": 2376
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017133027522935778,
+      "loss": 1.8998,
+      "step": 2377
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017126146788990824,
+      "loss": 2.0302,
+      "step": 2378
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001711926605504587,
+      "loss": 1.7413,
+      "step": 2379
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017112385321100916,
+      "loss": 1.6514,
+      "step": 2380
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017105504587155962,
+      "loss": 1.9371,
+      "step": 2381
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017098623853211009,
+      "loss": 1.7878,
+      "step": 2382
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017091743119266055,
+      "loss": 1.8151,
+      "step": 2383
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000170848623853211,
+      "loss": 1.8037,
+      "step": 2384
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017077981651376147,
+      "loss": 1.8958,
+      "step": 2385
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017071100917431193,
+      "loss": 1.8613,
+      "step": 2386
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001706422018348624,
+      "loss": 1.8085,
+      "step": 2387
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017057339449541285,
+      "loss": 1.6826,
+      "step": 2388
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017050458715596331,
+      "loss": 1.7469,
+      "step": 2389
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017043577981651378,
+      "loss": 1.8051,
+      "step": 2390
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017036697247706418,
+      "loss": 1.9276,
+      "step": 2391
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017029816513761464,
+      "loss": 1.7462,
+      "step": 2392
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001702293577981651,
+      "loss": 1.6551,
+      "step": 2393
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017016055045871557,
+      "loss": 1.6347,
+      "step": 2394
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017009174311926603,
+      "loss": 1.6097,
+      "step": 2395
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001700229357798165,
+      "loss": 1.6692,
+      "step": 2396
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016995412844036695,
+      "loss": 1.5917,
+      "step": 2397
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001698853211009174,
+      "loss": 1.6417,
+      "step": 2398
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016981651376146787,
+      "loss": 1.6644,
+      "step": 2399
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016974770642201833,
+      "loss": 1.4968,
+      "step": 2400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001696788990825688,
+      "loss": 1.5209,
+      "step": 2401
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016961009174311926,
+      "loss": 1.5565,
+      "step": 2402
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016954128440366972,
+      "loss": 1.456,
+      "step": 2403
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016947247706422018,
+      "loss": 1.5134,
+      "step": 2404
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016940366972477064,
+      "loss": 1.4441,
+      "step": 2405
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001693348623853211,
+      "loss": 1.3615,
+      "step": 2406
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016926605504587154,
+      "loss": 1.3585,
+      "step": 2407
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000169197247706422,
+      "loss": 1.4528,
+      "step": 2408
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016912844036697246,
+      "loss": 1.3891,
+      "step": 2409
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016905963302752292,
+      "loss": 1.3184,
+      "step": 2410
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016899082568807338,
+      "loss": 1.35,
+      "step": 2411
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016892201834862384,
+      "loss": 1.2573,
+      "step": 2412
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001688532110091743,
+      "loss": 1.279,
+      "step": 2413
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016878440366972476,
+      "loss": 1.1925,
+      "step": 2414
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016871559633027523,
+      "loss": 1.0992,
+      "step": 2415
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001686467889908257,
+      "loss": 1.1728,
+      "step": 2416
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016857798165137615,
+      "loss": 1.1968,
+      "step": 2417
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001685091743119266,
+      "loss": 1.0496,
+      "step": 2418
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016844036697247707,
+      "loss": 0.9928,
+      "step": 2419
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016837155963302753,
+      "loss": 0.9409,
+      "step": 2420
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000168302752293578,
+      "loss": 2.8201,
+      "step": 2421
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016823394495412843,
+      "loss": 2.5297,
+      "step": 2422
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016816513761467886,
+      "loss": 2.2047,
+      "step": 2423
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016809633027522932,
+      "loss": 2.2145,
+      "step": 2424
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016802752293577978,
+      "loss": 2.1811,
+      "step": 2425
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016795871559633025,
+      "loss": 2.0704,
+      "step": 2426
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001678899082568807,
+      "loss": 1.9387,
+      "step": 2427
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016782110091743117,
+      "loss": 1.8659,
+      "step": 2428
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016775229357798163,
+      "loss": 2.0039,
+      "step": 2429
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001676834862385321,
+      "loss": 1.8378,
+      "step": 2430
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016761467889908255,
+      "loss": 1.9685,
+      "step": 2431
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000167545871559633,
+      "loss": 1.8136,
+      "step": 2432
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016747706422018347,
+      "loss": 1.7035,
+      "step": 2433
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016740825688073394,
+      "loss": 1.7615,
+      "step": 2434
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001673394495412844,
+      "loss": 1.768,
+      "step": 2435
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016727064220183486,
+      "loss": 1.8968,
+      "step": 2436
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016720183486238532,
+      "loss": 1.8514,
+      "step": 2437
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016713302752293578,
+      "loss": 1.5916,
+      "step": 2438
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016706422018348622,
+      "loss": 1.5929,
+      "step": 2439
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016699541284403668,
+      "loss": 1.6183,
+      "step": 2440
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016692660550458714,
+      "loss": 1.6629,
+      "step": 2441
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001668577981651376,
+      "loss": 1.8363,
+      "step": 2442
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016678899082568806,
+      "loss": 1.636,
+      "step": 2443
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016672018348623852,
+      "loss": 1.7425,
+      "step": 2444
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016665137614678898,
+      "loss": 1.6772,
+      "step": 2445
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016658256880733944,
+      "loss": 1.675,
+      "step": 2446
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001665137614678899,
+      "loss": 1.6034,
+      "step": 2447
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016644495412844037,
+      "loss": 1.4182,
+      "step": 2448
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016637614678899083,
+      "loss": 1.6199,
+      "step": 2449
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001663073394495413,
+      "loss": 1.4876,
+      "step": 2450
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016623853211009175,
+      "loss": 1.5222,
+      "step": 2451
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001661697247706422,
+      "loss": 1.4074,
+      "step": 2452
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016610091743119265,
+      "loss": 1.3419,
+      "step": 2453
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001660321100917431,
+      "loss": 1.372,
+      "step": 2454
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016596330275229354,
+      "loss": 1.6406,
+      "step": 2455
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.000165894495412844,
+      "loss": 1.4601,
+      "step": 2456
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016582568807339446,
+      "loss": 1.5373,
+      "step": 2457
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016575688073394493,
+      "loss": 1.3078,
+      "step": 2458
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016568807339449539,
+      "loss": 1.3279,
+      "step": 2459
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016561926605504585,
+      "loss": 1.1597,
+      "step": 2460
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001655504587155963,
+      "loss": 1.1937,
+      "step": 2461
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016548165137614677,
+      "loss": 1.2922,
+      "step": 2462
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016541284403669723,
+      "loss": 1.0909,
+      "step": 2463
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001653440366972477,
+      "loss": 1.086,
+      "step": 2464
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016527522935779815,
+      "loss": 1.0827,
+      "step": 2465
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016520642201834862,
+      "loss": 1.0222,
+      "step": 2466
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016513761467889908,
+      "loss": 1.0251,
+      "step": 2467
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016506880733944954,
+      "loss": 0.8553,
+      "step": 2468
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.000165,
+      "loss": 1.0218,
+      "step": 2469
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016493119266055046,
+      "loss": 0.9624,
+      "step": 2470
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001648623853211009,
+      "loss": 2.7151,
+      "step": 2471
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016479357798165136,
+      "loss": 2.4307,
+      "step": 2472
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016472477064220182,
+      "loss": 2.171,
+      "step": 2473
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016465596330275228,
+      "loss": 1.968,
+      "step": 2474
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016458715596330274,
+      "loss": 1.9974,
+      "step": 2475
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001645183486238532,
+      "loss": 2.0983,
+      "step": 2476
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016444954128440366,
+      "loss": 1.88,
+      "step": 2477
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016438073394495412,
+      "loss": 1.8526,
+      "step": 2478
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016431192660550458,
+      "loss": 1.9102,
+      "step": 2479
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016424311926605505,
+      "loss": 1.8006,
+      "step": 2480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001641743119266055,
+      "loss": 1.7544,
+      "step": 2481
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016410550458715597,
+      "loss": 2.0858,
+      "step": 2482
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016403669724770643,
+      "loss": 1.7207,
+      "step": 2483
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001639678899082569,
+      "loss": 1.8056,
+      "step": 2484
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016389908256880733,
+      "loss": 1.6819,
+      "step": 2485
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001638302752293578,
+      "loss": 1.7391,
+      "step": 2486
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016376146788990822,
+      "loss": 1.6962,
+      "step": 2487
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016369266055045868,
+      "loss": 1.7615,
+      "step": 2488
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016362385321100914,
+      "loss": 1.7741,
+      "step": 2489
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001635550458715596,
+      "loss": 1.792,
+      "step": 2490
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016348623853211007,
+      "loss": 1.6932,
+      "step": 2491
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016341743119266053,
+      "loss": 1.6707,
+      "step": 2492
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.000163348623853211,
+      "loss": 1.7434,
+      "step": 2493
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016327981651376145,
+      "loss": 1.713,
+      "step": 2494
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001632110091743119,
+      "loss": 1.6532,
+      "step": 2495
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016314220183486237,
+      "loss": 1.5183,
+      "step": 2496
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016307339449541283,
+      "loss": 1.5773,
+      "step": 2497
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001630045871559633,
+      "loss": 1.5828,
+      "step": 2498
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016293577981651376,
+      "loss": 1.539,
+      "step": 2499
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016286697247706422,
+      "loss": 1.6791,
+      "step": 2500
+    },
+    {
+      "epoch": 1.54,
+      "eval_bleu": 9.104417999122743e-14,
+      "eval_loss": 2.0529630184173584,
+      "eval_runtime": 2550.0568,
+      "eval_samples_per_second": 5.788,
+      "eval_steps_per_second": 0.724,
+      "step": 2500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016279816513761468,
+      "loss": 1.5201,
+      "step": 2501
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016272935779816514,
+      "loss": 1.4272,
+      "step": 2502
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016266055045871557,
+      "loss": 1.4311,
+      "step": 2503
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016259174311926604,
+      "loss": 1.4258,
+      "step": 2504
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001625229357798165,
+      "loss": 1.4489,
+      "step": 2505
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016245412844036696,
+      "loss": 1.3071,
+      "step": 2506
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016238532110091742,
+      "loss": 1.289,
+      "step": 2507
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016231651376146788,
+      "loss": 1.4723,
+      "step": 2508
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016224770642201834,
+      "loss": 1.325,
+      "step": 2509
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001621788990825688,
+      "loss": 1.2669,
+      "step": 2510
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016211009174311926,
+      "loss": 1.3747,
+      "step": 2511
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016204128440366973,
+      "loss": 1.2665,
+      "step": 2512
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001619724770642202,
+      "loss": 1.2726,
+      "step": 2513
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016190366972477065,
+      "loss": 0.9968,
+      "step": 2514
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001618348623853211,
+      "loss": 1.1225,
+      "step": 2515
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016176605504587154,
+      "loss": 1.0416,
+      "step": 2516
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000161697247706422,
+      "loss": 1.1329,
+      "step": 2517
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016162844036697247,
+      "loss": 0.9321,
+      "step": 2518
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001615596330275229,
+      "loss": 0.9791,
+      "step": 2519
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016149082568807336,
+      "loss": 0.8882,
+      "step": 2520
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016142201834862382,
+      "loss": 2.6526,
+      "step": 2521
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016135321100917428,
+      "loss": 2.365,
+      "step": 2522
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016128440366972475,
+      "loss": 2.2288,
+      "step": 2523
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001612155963302752,
+      "loss": 2.0181,
+      "step": 2524
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016114678899082567,
+      "loss": 2.0531,
+      "step": 2525
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016107798165137613,
+      "loss": 1.8414,
+      "step": 2526
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001610091743119266,
+      "loss": 2.1092,
+      "step": 2527
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016094036697247705,
+      "loss": 1.8598,
+      "step": 2528
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001608715596330275,
+      "loss": 1.9902,
+      "step": 2529
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016080275229357797,
+      "loss": 1.6837,
+      "step": 2530
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016073394495412844,
+      "loss": 1.8995,
+      "step": 2531
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001606651376146789,
+      "loss": 1.7875,
+      "step": 2532
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016059633027522936,
+      "loss": 1.7717,
+      "step": 2533
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016052752293577982,
+      "loss": 1.8133,
+      "step": 2534
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016045871559633025,
+      "loss": 1.8402,
+      "step": 2535
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016038990825688071,
+      "loss": 1.7249,
+      "step": 2536
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016032110091743118,
+      "loss": 1.7668,
+      "step": 2537
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016025229357798164,
+      "loss": 1.7421,
+      "step": 2538
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001601834862385321,
+      "loss": 1.5446,
+      "step": 2539
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016011467889908256,
+      "loss": 1.7223,
+      "step": 2540
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016004587155963302,
+      "loss": 1.7396,
+      "step": 2541
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015997706422018348,
+      "loss": 1.8328,
+      "step": 2542
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015990825688073394,
+      "loss": 1.7118,
+      "step": 2543
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001598394495412844,
+      "loss": 1.6988,
+      "step": 2544
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015977064220183487,
+      "loss": 1.4266,
+      "step": 2545
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015970183486238533,
+      "loss": 1.5278,
+      "step": 2546
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015963302752293576,
+      "loss": 1.484,
+      "step": 2547
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015956422018348622,
+      "loss": 1.515,
+      "step": 2548
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015949541284403668,
+      "loss": 1.2707,
+      "step": 2549
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015942660550458715,
+      "loss": 1.4544,
+      "step": 2550
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015935779816513758,
+      "loss": 1.4689,
+      "step": 2551
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015928899082568804,
+      "loss": 1.412,
+      "step": 2552
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001592201834862385,
+      "loss": 1.4757,
+      "step": 2553
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015915137614678896,
+      "loss": 1.3614,
+      "step": 2554
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015908256880733942,
+      "loss": 1.4429,
+      "step": 2555
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015901376146788989,
+      "loss": 1.4448,
+      "step": 2556
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015894495412844035,
+      "loss": 1.2,
+      "step": 2557
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001588761467889908,
+      "loss": 1.2744,
+      "step": 2558
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015880733944954127,
+      "loss": 1.122,
+      "step": 2559
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015873853211009173,
+      "loss": 1.2659,
+      "step": 2560
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001586697247706422,
+      "loss": 1.1515,
+      "step": 2561
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015860091743119265,
+      "loss": 1.1739,
+      "step": 2562
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015853211009174311,
+      "loss": 1.1951,
+      "step": 2563
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015846330275229358,
+      "loss": 1.2862,
+      "step": 2564
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015839449541284404,
+      "loss": 1.1841,
+      "step": 2565
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001583256880733945,
+      "loss": 1.111,
+      "step": 2566
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015825688073394493,
+      "loss": 1.049,
+      "step": 2567
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001581880733944954,
+      "loss": 0.8997,
+      "step": 2568
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015811926605504586,
+      "loss": 0.8725,
+      "step": 2569
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015805045871559632,
+      "loss": 0.8984,
+      "step": 2570
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015798165137614678,
+      "loss": 2.6616,
+      "step": 2571
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015791284403669724,
+      "loss": 2.2496,
+      "step": 2572
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001578440366972477,
+      "loss": 2.1663,
+      "step": 2573
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015777522935779816,
+      "loss": 1.9776,
+      "step": 2574
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015770642201834862,
+      "loss": 1.9404,
+      "step": 2575
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015763761467889908,
+      "loss": 2.0084,
+      "step": 2576
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015756880733944955,
+      "loss": 1.9548,
+      "step": 2577
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015749999999999998,
+      "loss": 1.8605,
+      "step": 2578
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015743119266055044,
+      "loss": 1.9351,
+      "step": 2579
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001573623853211009,
+      "loss": 1.5703,
+      "step": 2580
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015729357798165136,
+      "loss": 1.7759,
+      "step": 2581
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015722477064220182,
+      "loss": 1.9455,
+      "step": 2582
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015715596330275226,
+      "loss": 1.7536,
+      "step": 2583
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015708715596330272,
+      "loss": 1.8178,
+      "step": 2584
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015701834862385318,
+      "loss": 1.812,
+      "step": 2585
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015694954128440364,
+      "loss": 1.7461,
+      "step": 2586
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001568807339449541,
+      "loss": 1.6504,
+      "step": 2587
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015681192660550457,
+      "loss": 1.6768,
+      "step": 2588
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015674311926605503,
+      "loss": 1.6623,
+      "step": 2589
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001566743119266055,
+      "loss": 1.746,
+      "step": 2590
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015660550458715595,
+      "loss": 1.5993,
+      "step": 2591
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001565366972477064,
+      "loss": 1.6449,
+      "step": 2592
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015646788990825687,
+      "loss": 1.6487,
+      "step": 2593
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015639908256880733,
+      "loss": 1.6896,
+      "step": 2594
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001563302752293578,
+      "loss": 1.6811,
+      "step": 2595
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015626146788990826,
+      "loss": 1.5278,
+      "step": 2596
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015619266055045872,
+      "loss": 1.5104,
+      "step": 2597
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015612385321100918,
+      "loss": 1.602,
+      "step": 2598
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001560550458715596,
+      "loss": 1.5557,
+      "step": 2599
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015598623853211007,
+      "loss": 1.4968,
+      "step": 2600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015591743119266053,
+      "loss": 1.3723,
+      "step": 2601
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000155848623853211,
+      "loss": 1.5659,
+      "step": 2602
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015577981651376146,
+      "loss": 1.344,
+      "step": 2603
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015571100917431192,
+      "loss": 1.4165,
+      "step": 2604
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015564220183486238,
+      "loss": 1.3802,
+      "step": 2605
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015557339449541284,
+      "loss": 1.3201,
+      "step": 2606
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001555045871559633,
+      "loss": 1.3926,
+      "step": 2607
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015543577981651376,
+      "loss": 1.2347,
+      "step": 2608
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001553669724770642,
+      "loss": 1.3002,
+      "step": 2609
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015529816513761466,
+      "loss": 1.2601,
+      "step": 2610
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015522935779816512,
+      "loss": 1.426,
+      "step": 2611
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015516055045871558,
+      "loss": 1.1477,
+      "step": 2612
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015509174311926604,
+      "loss": 0.9914,
+      "step": 2613
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001550229357798165,
+      "loss": 1.178,
+      "step": 2614
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015495412844036694,
+      "loss": 1.0769,
+      "step": 2615
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001548853211009174,
+      "loss": 1.0737,
+      "step": 2616
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015481651376146786,
+      "loss": 0.9871,
+      "step": 2617
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015474770642201832,
+      "loss": 0.8882,
+      "step": 2618
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015467889908256878,
+      "loss": 0.9982,
+      "step": 2619
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015461009174311924,
+      "loss": 0.8103,
+      "step": 2620
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001545412844036697,
+      "loss": 2.3961,
+      "step": 2621
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015447247706422017,
+      "loss": 2.2256,
+      "step": 2622
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015440366972477063,
+      "loss": 1.9595,
+      "step": 2623
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001543348623853211,
+      "loss": 2.0277,
+      "step": 2624
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015426605504587155,
+      "loss": 2.0765,
+      "step": 2625
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.000154197247706422,
+      "loss": 1.9673,
+      "step": 2626
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015412844036697247,
+      "loss": 1.7778,
+      "step": 2627
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015405963302752293,
+      "loss": 1.9129,
+      "step": 2628
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001539908256880734,
+      "loss": 1.8891,
+      "step": 2629
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015392201834862386,
+      "loss": 1.7402,
+      "step": 2630
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001538532110091743,
+      "loss": 1.7574,
+      "step": 2631
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015378440366972475,
+      "loss": 1.7611,
+      "step": 2632
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015371559633027521,
+      "loss": 1.8226,
+      "step": 2633
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015364678899082568,
+      "loss": 1.5753,
+      "step": 2634
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015357798165137614,
+      "loss": 1.7198,
+      "step": 2635
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001535091743119266,
+      "loss": 1.6867,
+      "step": 2636
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015344036697247706,
+      "loss": 1.722,
+      "step": 2637
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015337155963302752,
+      "loss": 1.6694,
+      "step": 2638
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015330275229357798,
+      "loss": 1.655,
+      "step": 2639
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015323394495412844,
+      "loss": 1.5777,
+      "step": 2640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015316513761467888,
+      "loss": 1.6608,
+      "step": 2641
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015309633027522934,
+      "loss": 1.6519,
+      "step": 2642
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001530275229357798,
+      "loss": 1.623,
+      "step": 2643
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015295871559633026,
+      "loss": 1.4663,
+      "step": 2644
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015288990825688072,
+      "loss": 1.6839,
+      "step": 2645
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015282110091743118,
+      "loss": 1.4738,
+      "step": 2646
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015275229357798162,
+      "loss": 1.5288,
+      "step": 2647
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015268348623853208,
+      "loss": 1.5614,
+      "step": 2648
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015261467889908254,
+      "loss": 1.3896,
+      "step": 2649
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000152545871559633,
+      "loss": 1.4337,
+      "step": 2650
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015247706422018346,
+      "loss": 1.5167,
+      "step": 2651
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015240825688073392,
+      "loss": 1.308,
+      "step": 2652
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015233944954128438,
+      "loss": 1.4404,
+      "step": 2653
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015227064220183485,
+      "loss": 1.4403,
+      "step": 2654
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001522018348623853,
+      "loss": 1.433,
+      "step": 2655
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015213302752293577,
+      "loss": 1.3051,
+      "step": 2656
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015206422018348623,
+      "loss": 1.2807,
+      "step": 2657
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001519954128440367,
+      "loss": 1.3028,
+      "step": 2658
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015192660550458715,
+      "loss": 1.1943,
+      "step": 2659
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015185779816513761,
+      "loss": 1.2522,
+      "step": 2660
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015178899082568808,
+      "loss": 1.1313,
+      "step": 2661
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015172018348623854,
+      "loss": 1.1258,
+      "step": 2662
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015165137614678897,
+      "loss": 1.0589,
+      "step": 2663
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015158256880733943,
+      "loss": 1.1104,
+      "step": 2664
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001515137614678899,
+      "loss": 1.011,
+      "step": 2665
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015144495412844035,
+      "loss": 0.9751,
+      "step": 2666
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015137614678899082,
+      "loss": 0.9338,
+      "step": 2667
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015130733944954128,
+      "loss": 0.9432,
+      "step": 2668
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015123853211009174,
+      "loss": 0.8753,
+      "step": 2669
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001511697247706422,
+      "loss": 0.8526,
+      "step": 2670
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015110091743119266,
+      "loss": 2.5274,
+      "step": 2671
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001510321100917431,
+      "loss": 2.506,
+      "step": 2672
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015096330275229356,
+      "loss": 2.1575,
+      "step": 2673
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015089449541284402,
+      "loss": 2.0803,
+      "step": 2674
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015082568807339448,
+      "loss": 1.9175,
+      "step": 2675
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015075688073394494,
+      "loss": 1.9355,
+      "step": 2676
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001506880733944954,
+      "loss": 2.0204,
+      "step": 2677
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015061926605504586,
+      "loss": 1.889,
+      "step": 2678
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001505504587155963,
+      "loss": 2.0395,
+      "step": 2679
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015048165137614676,
+      "loss": 1.8121,
+      "step": 2680
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015041284403669722,
+      "loss": 1.6833,
+      "step": 2681
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015034403669724768,
+      "loss": 1.7684,
+      "step": 2682
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015027522935779814,
+      "loss": 1.5983,
+      "step": 2683
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001502064220183486,
+      "loss": 1.7694,
+      "step": 2684
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015013761467889906,
+      "loss": 1.7431,
+      "step": 2685
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015006880733944953,
+      "loss": 1.7292,
+      "step": 2686
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015,
+      "loss": 1.565,
+      "step": 2687
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014993119266055045,
+      "loss": 1.6728,
+      "step": 2688
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001498623853211009,
+      "loss": 1.6175,
+      "step": 2689
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014979357798165137,
+      "loss": 1.627,
+      "step": 2690
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014972477064220183,
+      "loss": 1.4694,
+      "step": 2691
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014965596330275227,
+      "loss": 1.6295,
+      "step": 2692
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014958715596330273,
+      "loss": 1.5165,
+      "step": 2693
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001495183486238532,
+      "loss": 1.6623,
+      "step": 2694
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014944954128440365,
+      "loss": 1.6055,
+      "step": 2695
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001493807339449541,
+      "loss": 1.6107,
+      "step": 2696
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00014931192660550457,
+      "loss": 1.4241,
+      "step": 2697
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014924311926605503,
+      "loss": 1.5645,
+      "step": 2698
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001491743119266055,
+      "loss": 1.5751,
+      "step": 2699
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014910550458715596,
+      "loss": 1.4571,
+      "step": 2700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014903669724770642,
+      "loss": 1.4935,
+      "step": 2701
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014896788990825688,
+      "loss": 1.366,
+      "step": 2702
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001488990825688073,
+      "loss": 1.4008,
+      "step": 2703
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014883027522935777,
+      "loss": 1.3579,
+      "step": 2704
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014876146788990824,
+      "loss": 1.2239,
+      "step": 2705
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001486926605504587,
+      "loss": 1.4696,
+      "step": 2706
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014862385321100916,
+      "loss": 1.3272,
+      "step": 2707
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014855504587155962,
+      "loss": 1.1287,
+      "step": 2708
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014848623853211008,
+      "loss": 1.2084,
+      "step": 2709
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014841743119266054,
+      "loss": 1.0919,
+      "step": 2710
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.000148348623853211,
+      "loss": 1.1432,
+      "step": 2711
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014827981651376146,
+      "loss": 1.1812,
+      "step": 2712
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00014821100917431193,
+      "loss": 1.2734,
+      "step": 2713
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001481422018348624,
+      "loss": 1.1117,
+      "step": 2714
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014807339449541285,
+      "loss": 1.1729,
+      "step": 2715
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014800458715596328,
+      "loss": 0.9826,
+      "step": 2716
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014793577981651374,
+      "loss": 0.8985,
+      "step": 2717
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001478669724770642,
+      "loss": 0.8843,
+      "step": 2718
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014779816513761467,
+      "loss": 0.9639,
+      "step": 2719
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014772935779816513,
+      "loss": 0.9416,
+      "step": 2720
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001476605504587156,
+      "loss": 2.504,
+      "step": 2721
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014759174311926605,
+      "loss": 2.3732,
+      "step": 2722
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001475229357798165,
+      "loss": 1.9512,
+      "step": 2723
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014745412844036695,
+      "loss": 1.9638,
+      "step": 2724
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001473853211009174,
+      "loss": 1.9552,
+      "step": 2725
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014731651376146787,
+      "loss": 1.9798,
+      "step": 2726
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014724770642201833,
+      "loss": 1.8179,
+      "step": 2727
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001471788990825688,
+      "loss": 1.8957,
+      "step": 2728
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00014711009174311925,
+      "loss": 1.9607,
+      "step": 2729
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001470412844036697,
+      "loss": 1.832,
+      "step": 2730
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014697247706422017,
+      "loss": 1.7741,
+      "step": 2731
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014690366972477064,
+      "loss": 1.8495,
+      "step": 2732
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001468348623853211,
+      "loss": 1.8138,
+      "step": 2733
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014676605504587153,
+      "loss": 1.7175,
+      "step": 2734
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.000146697247706422,
+      "loss": 1.8109,
+      "step": 2735
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014662844036697245,
+      "loss": 1.6513,
+      "step": 2736
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014655963302752291,
+      "loss": 1.7149,
+      "step": 2737
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014649082568807338,
+      "loss": 1.6808,
+      "step": 2738
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014642201834862384,
+      "loss": 1.6835,
+      "step": 2739
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001463532110091743,
+      "loss": 1.7405,
+      "step": 2740
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014628440366972476,
+      "loss": 1.5068,
+      "step": 2741
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014621559633027522,
+      "loss": 1.6685,
+      "step": 2742
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014614678899082568,
+      "loss": 1.5865,
+      "step": 2743
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00014607798165137614,
+      "loss": 1.6137,
+      "step": 2744
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001460091743119266,
+      "loss": 1.4727,
+      "step": 2745
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014594036697247707,
+      "loss": 1.5242,
+      "step": 2746
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014587155963302753,
+      "loss": 1.4384,
+      "step": 2747
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014580275229357796,
+      "loss": 1.5791,
+      "step": 2748
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014573394495412842,
+      "loss": 1.3213,
+      "step": 2749
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014566513761467888,
+      "loss": 1.5016,
+      "step": 2750
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014559633027522935,
+      "loss": 1.3458,
+      "step": 2751
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001455275229357798,
+      "loss": 1.4696,
+      "step": 2752
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014545871559633027,
+      "loss": 1.4191,
+      "step": 2753
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014538990825688073,
+      "loss": 1.1216,
+      "step": 2754
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001453211009174312,
+      "loss": 1.331,
+      "step": 2755
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014525229357798162,
+      "loss": 1.4856,
+      "step": 2756
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014518348623853209,
+      "loss": 1.3681,
+      "step": 2757
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014511467889908255,
+      "loss": 1.1923,
+      "step": 2758
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000145045871559633,
+      "loss": 1.1456,
+      "step": 2759
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014497706422018347,
+      "loss": 1.2025,
+      "step": 2760
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00014490825688073393,
+      "loss": 1.1414,
+      "step": 2761
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001448394495412844,
+      "loss": 1.1824,
+      "step": 2762
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014477064220183485,
+      "loss": 1.1105,
+      "step": 2763
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014470183486238531,
+      "loss": 1.1726,
+      "step": 2764
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014463302752293578,
+      "loss": 0.9236,
+      "step": 2765
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001445642201834862,
+      "loss": 0.8958,
+      "step": 2766
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014449541284403667,
+      "loss": 1.0156,
+      "step": 2767
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014442660550458713,
+      "loss": 1.0257,
+      "step": 2768
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001443577981651376,
+      "loss": 0.8883,
+      "step": 2769
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014428899082568806,
+      "loss": 0.7913,
+      "step": 2770
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014422018348623852,
+      "loss": 2.459,
+      "step": 2771
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014415137614678898,
+      "loss": 2.1902,
+      "step": 2772
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014408256880733944,
+      "loss": 1.8838,
+      "step": 2773
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001440137614678899,
+      "loss": 1.9688,
+      "step": 2774
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014394495412844036,
+      "loss": 1.9483,
+      "step": 2775
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014387614678899082,
+      "loss": 1.7471,
+      "step": 2776
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014380733944954128,
+      "loss": 1.7646,
+      "step": 2777
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00014373853211009175,
+      "loss": 1.7671,
+      "step": 2778
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001436697247706422,
+      "loss": 1.7177,
+      "step": 2779
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014360091743119264,
+      "loss": 1.6666,
+      "step": 2780
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001435321100917431,
+      "loss": 1.8098,
+      "step": 2781
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014346330275229356,
+      "loss": 1.6711,
+      "step": 2782
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014339449541284402,
+      "loss": 1.6581,
+      "step": 2783
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014332568807339449,
+      "loss": 1.8288,
+      "step": 2784
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014325688073394495,
+      "loss": 1.6395,
+      "step": 2785
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001431880733944954,
+      "loss": 1.5874,
+      "step": 2786
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014311926605504587,
+      "loss": 1.6246,
+      "step": 2787
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001430504587155963,
+      "loss": 1.794,
+      "step": 2788
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014298165137614677,
+      "loss": 1.5566,
+      "step": 2789
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014291284403669723,
+      "loss": 1.5721,
+      "step": 2790
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001428440366972477,
+      "loss": 1.5802,
+      "step": 2791
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014277522935779815,
+      "loss": 1.5288,
+      "step": 2792
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001427064220183486,
+      "loss": 1.5479,
+      "step": 2793
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00014263761467889907,
+      "loss": 1.486,
+      "step": 2794
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014256880733944953,
+      "loss": 1.5892,
+      "step": 2795
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001425,
+      "loss": 1.4275,
+      "step": 2796
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014243119266055043,
+      "loss": 1.6013,
+      "step": 2797
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001423623853211009,
+      "loss": 1.6975,
+      "step": 2798
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014229357798165135,
+      "loss": 1.4974,
+      "step": 2799
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001422247706422018,
+      "loss": 1.4928,
+      "step": 2800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014215596330275227,
+      "loss": 1.2671,
+      "step": 2801
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014208715596330273,
+      "loss": 1.3971,
+      "step": 2802
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001420183486238532,
+      "loss": 1.3389,
+      "step": 2803
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014194954128440366,
+      "loss": 1.3498,
+      "step": 2804
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014188073394495412,
+      "loss": 1.3314,
+      "step": 2805
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014181192660550458,
+      "loss": 1.2777,
+      "step": 2806
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014174311926605504,
+      "loss": 1.2437,
+      "step": 2807
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001416743119266055,
+      "loss": 1.2039,
+      "step": 2808
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014160550458715596,
+      "loss": 1.2735,
+      "step": 2809
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00014153669724770642,
+      "loss": 1.1883,
+      "step": 2810
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00014146788990825689,
+      "loss": 1.1214,
+      "step": 2811
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00014139908256880732,
+      "loss": 1.012,
+      "step": 2812
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00014133027522935778,
+      "loss": 1.1341,
+      "step": 2813
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00014126146788990824,
+      "loss": 1.1077,
+      "step": 2814
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001411926605504587,
+      "loss": 1.0662,
+      "step": 2815
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00014112385321100917,
+      "loss": 0.9541,
+      "step": 2816
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00014105504587155963,
+      "loss": 0.906,
+      "step": 2817
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001409862385321101,
+      "loss": 0.8681,
+      "step": 2818
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00014091743119266055,
+      "loss": 0.806,
+      "step": 2819
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00014084862385321098,
+      "loss": 1.0313,
+      "step": 2820
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00014077981651376144,
+      "loss": 2.4214,
+      "step": 2821
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001407110091743119,
+      "loss": 2.2117,
+      "step": 2822
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00014064220183486237,
+      "loss": 1.9729,
+      "step": 2823
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00014057339449541283,
+      "loss": 1.9015,
+      "step": 2824
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001405045871559633,
+      "loss": 1.8548,
+      "step": 2825
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00014043577981651375,
+      "loss": 1.8444,
+      "step": 2826
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001403669724770642,
+      "loss": 1.8318,
+      "step": 2827
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00014029816513761465,
+      "loss": 1.7103,
+      "step": 2828
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001402293577981651,
+      "loss": 1.869,
+      "step": 2829
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00014016055045871557,
+      "loss": 1.6276,
+      "step": 2830
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00014009174311926603,
+      "loss": 1.6946,
+      "step": 2831
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001400229357798165,
+      "loss": 1.7185,
+      "step": 2832
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013995412844036695,
+      "loss": 1.5489,
+      "step": 2833
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013988532110091741,
+      "loss": 1.4284,
+      "step": 2834
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013981651376146788,
+      "loss": 1.6394,
+      "step": 2835
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013974770642201834,
+      "loss": 1.6487,
+      "step": 2836
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001396788990825688,
+      "loss": 1.7531,
+      "step": 2837
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013961009174311926,
+      "loss": 1.5455,
+      "step": 2838
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013954128440366972,
+      "loss": 1.5839,
+      "step": 2839
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013947247706422018,
+      "loss": 1.4558,
+      "step": 2840
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013940366972477064,
+      "loss": 1.4974,
+      "step": 2841
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001393348623853211,
+      "loss": 1.5057,
+      "step": 2842
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013926605504587157,
+      "loss": 1.5581,
+      "step": 2843
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.000139197247706422,
+      "loss": 1.6725,
+      "step": 2844
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013912844036697246,
+      "loss": 1.4783,
+      "step": 2845
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013905963302752292,
+      "loss": 1.5513,
+      "step": 2846
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013899082568807338,
+      "loss": 1.569,
+      "step": 2847
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013892201834862384,
+      "loss": 1.3348,
+      "step": 2848
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001388532110091743,
+      "loss": 1.4112,
+      "step": 2849
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013878440366972477,
+      "loss": 1.4632,
+      "step": 2850
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013871559633027523,
+      "loss": 1.411,
+      "step": 2851
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013864678899082566,
+      "loss": 1.3533,
+      "step": 2852
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013857798165137612,
+      "loss": 1.3169,
+      "step": 2853
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013850917431192659,
+      "loss": 1.2661,
+      "step": 2854
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013844036697247705,
+      "loss": 1.3256,
+      "step": 2855
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001383715596330275,
+      "loss": 1.2034,
+      "step": 2856
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013830275229357797,
+      "loss": 1.3393,
+      "step": 2857
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00013823394495412843,
+      "loss": 1.2033,
+      "step": 2858
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001381651376146789,
+      "loss": 1.0357,
+      "step": 2859
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013809633027522933,
+      "loss": 1.0992,
+      "step": 2860
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001380275229357798,
+      "loss": 1.1742,
+      "step": 2861
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013795871559633025,
+      "loss": 1.0391,
+      "step": 2862
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001378899082568807,
+      "loss": 0.967,
+      "step": 2863
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013782110091743117,
+      "loss": 1.2282,
+      "step": 2864
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013775229357798163,
+      "loss": 0.9983,
+      "step": 2865
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001376834862385321,
+      "loss": 1.0236,
+      "step": 2866
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013761467889908255,
+      "loss": 0.9401,
+      "step": 2867
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013754587155963302,
+      "loss": 0.7745,
+      "step": 2868
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013747706422018348,
+      "loss": 0.8361,
+      "step": 2869
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013740825688073394,
+      "loss": 0.9759,
+      "step": 2870
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001373394495412844,
+      "loss": 2.5971,
+      "step": 2871
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013727064220183486,
+      "loss": 2.1958,
+      "step": 2872
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013720183486238532,
+      "loss": 1.958,
+      "step": 2873
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013713302752293578,
+      "loss": 1.8039,
+      "step": 2874
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00013706422018348624,
+      "loss": 1.859,
+      "step": 2875
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013699541284403668,
+      "loss": 1.7978,
+      "step": 2876
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013692660550458714,
+      "loss": 1.8082,
+      "step": 2877
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001368577981651376,
+      "loss": 1.6707,
+      "step": 2878
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013678899082568806,
+      "loss": 1.671,
+      "step": 2879
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013672018348623852,
+      "loss": 1.6763,
+      "step": 2880
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013665137614678899,
+      "loss": 1.6665,
+      "step": 2881
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013658256880733945,
+      "loss": 1.6413,
+      "step": 2882
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001365137614678899,
+      "loss": 1.5468,
+      "step": 2883
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013644495412844034,
+      "loss": 1.5592,
+      "step": 2884
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001363761467889908,
+      "loss": 1.6336,
+      "step": 2885
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013630733944954126,
+      "loss": 1.6274,
+      "step": 2886
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013623853211009173,
+      "loss": 1.6902,
+      "step": 2887
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001361697247706422,
+      "loss": 1.6418,
+      "step": 2888
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013610091743119265,
+      "loss": 1.5073,
+      "step": 2889
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001360321100917431,
+      "loss": 1.4728,
+      "step": 2890
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00013596330275229357,
+      "loss": 1.6102,
+      "step": 2891
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.000135894495412844,
+      "loss": 1.6547,
+      "step": 2892
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013582568807339447,
+      "loss": 1.6617,
+      "step": 2893
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013575688073394493,
+      "loss": 1.5925,
+      "step": 2894
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001356880733944954,
+      "loss": 1.3673,
+      "step": 2895
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013561926605504585,
+      "loss": 1.4199,
+      "step": 2896
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001355504587155963,
+      "loss": 1.4639,
+      "step": 2897
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013548165137614677,
+      "loss": 1.4093,
+      "step": 2898
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013541284403669723,
+      "loss": 1.3339,
+      "step": 2899
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001353440366972477,
+      "loss": 1.3778,
+      "step": 2900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013527522935779816,
+      "loss": 1.3368,
+      "step": 2901
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013520642201834862,
+      "loss": 1.4735,
+      "step": 2902
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013513761467889908,
+      "loss": 1.2763,
+      "step": 2903
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013506880733944954,
+      "loss": 1.2386,
+      "step": 2904
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.000135,
+      "loss": 1.3115,
+      "step": 2905
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013493119266055046,
+      "loss": 1.2399,
+      "step": 2906
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00013486238532110092,
+      "loss": 1.2276,
+      "step": 2907
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013479357798165136,
+      "loss": 1.1559,
+      "step": 2908
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013472477064220182,
+      "loss": 1.1065,
+      "step": 2909
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013465596330275228,
+      "loss": 1.213,
+      "step": 2910
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013458715596330274,
+      "loss": 1.0436,
+      "step": 2911
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001345183486238532,
+      "loss": 1.1155,
+      "step": 2912
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013444954128440366,
+      "loss": 1.0094,
+      "step": 2913
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013438073394495413,
+      "loss": 0.9752,
+      "step": 2914
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001343119266055046,
+      "loss": 0.9204,
+      "step": 2915
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013424311926605502,
+      "loss": 0.9557,
+      "step": 2916
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013417431192660548,
+      "loss": 0.8189,
+      "step": 2917
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013410550458715594,
+      "loss": 0.9168,
+      "step": 2918
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001340366972477064,
+      "loss": 0.7607,
+      "step": 2919
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013396788990825687,
+      "loss": 0.9393,
+      "step": 2920
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013389908256880733,
+      "loss": 2.4956,
+      "step": 2921
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001338302752293578,
+      "loss": 2.2171,
+      "step": 2922
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013376146788990825,
+      "loss": 1.8456,
+      "step": 2923
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00013369266055045868,
+      "loss": 1.7762,
+      "step": 2924
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013362385321100915,
+      "loss": 2.0005,
+      "step": 2925
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001335550458715596,
+      "loss": 1.8162,
+      "step": 2926
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013348623853211007,
+      "loss": 1.8116,
+      "step": 2927
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013341743119266053,
+      "loss": 1.7126,
+      "step": 2928
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.000133348623853211,
+      "loss": 1.672,
+      "step": 2929
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013327981651376145,
+      "loss": 1.8054,
+      "step": 2930
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001332110091743119,
+      "loss": 1.6762,
+      "step": 2931
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013314220183486237,
+      "loss": 1.7362,
+      "step": 2932
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013307339449541284,
+      "loss": 1.7796,
+      "step": 2933
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001330045871559633,
+      "loss": 1.6385,
+      "step": 2934
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013293577981651376,
+      "loss": 1.5444,
+      "step": 2935
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013286697247706422,
+      "loss": 1.6065,
+      "step": 2936
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013279816513761468,
+      "loss": 1.6448,
+      "step": 2937
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013272935779816514,
+      "loss": 1.5449,
+      "step": 2938
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001326605504587156,
+      "loss": 1.5301,
+      "step": 2939
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00013259174311926604,
+      "loss": 1.3936,
+      "step": 2940
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001325229357798165,
+      "loss": 1.6669,
+      "step": 2941
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013245412844036696,
+      "loss": 1.4067,
+      "step": 2942
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013238532110091742,
+      "loss": 1.559,
+      "step": 2943
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013231651376146788,
+      "loss": 1.6738,
+      "step": 2944
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013224770642201834,
+      "loss": 1.5091,
+      "step": 2945
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001321788990825688,
+      "loss": 1.5064,
+      "step": 2946
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013211009174311927,
+      "loss": 1.3895,
+      "step": 2947
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001320412844036697,
+      "loss": 1.4321,
+      "step": 2948
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013197247706422016,
+      "loss": 1.4752,
+      "step": 2949
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013190366972477062,
+      "loss": 1.3213,
+      "step": 2950
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013183486238532108,
+      "loss": 1.3146,
+      "step": 2951
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013176605504587155,
+      "loss": 1.3893,
+      "step": 2952
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000131697247706422,
+      "loss": 1.5642,
+      "step": 2953
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013162844036697247,
+      "loss": 1.3631,
+      "step": 2954
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013155963302752293,
+      "loss": 1.3061,
+      "step": 2955
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00013149082568807336,
+      "loss": 1.1956,
+      "step": 2956
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013142201834862383,
+      "loss": 1.1796,
+      "step": 2957
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013135321100917429,
+      "loss": 1.1048,
+      "step": 2958
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013128440366972475,
+      "loss": 1.0927,
+      "step": 2959
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001312155963302752,
+      "loss": 1.0363,
+      "step": 2960
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013114678899082567,
+      "loss": 1.127,
+      "step": 2961
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013107798165137613,
+      "loss": 1.1476,
+      "step": 2962
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001310091743119266,
+      "loss": 0.9952,
+      "step": 2963
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013094036697247705,
+      "loss": 0.9473,
+      "step": 2964
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013087155963302752,
+      "loss": 1.0435,
+      "step": 2965
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013080275229357798,
+      "loss": 0.8027,
+      "step": 2966
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013073394495412844,
+      "loss": 0.8835,
+      "step": 2967
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001306651376146789,
+      "loss": 0.8866,
+      "step": 2968
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013059633027522936,
+      "loss": 0.8132,
+      "step": 2969
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013052752293577982,
+      "loss": 0.8458,
+      "step": 2970
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013045871559633028,
+      "loss": 2.4443,
+      "step": 2971
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00013038990825688072,
+      "loss": 2.0481,
+      "step": 2972
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013032110091743118,
+      "loss": 2.0024,
+      "step": 2973
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013025229357798164,
+      "loss": 1.7469,
+      "step": 2974
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001301834862385321,
+      "loss": 1.7955,
+      "step": 2975
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013011467889908256,
+      "loss": 1.9425,
+      "step": 2976
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00013004587155963302,
+      "loss": 1.8198,
+      "step": 2977
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012997706422018348,
+      "loss": 1.9084,
+      "step": 2978
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012990825688073395,
+      "loss": 1.7141,
+      "step": 2979
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012983944954128438,
+      "loss": 1.6393,
+      "step": 2980
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012977064220183484,
+      "loss": 1.7444,
+      "step": 2981
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001297018348623853,
+      "loss": 1.6021,
+      "step": 2982
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012963302752293576,
+      "loss": 1.5906,
+      "step": 2983
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012956422018348623,
+      "loss": 1.7553,
+      "step": 2984
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001294954128440367,
+      "loss": 1.673,
+      "step": 2985
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012942660550458715,
+      "loss": 1.5965,
+      "step": 2986
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001293577981651376,
+      "loss": 1.7012,
+      "step": 2987
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012928899082568804,
+      "loss": 1.6297,
+      "step": 2988
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001292201834862385,
+      "loss": 1.5626,
+      "step": 2989
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012915137614678897,
+      "loss": 1.6843,
+      "step": 2990
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012908256880733943,
+      "loss": 1.5795,
+      "step": 2991
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001290137614678899,
+      "loss": 1.6206,
+      "step": 2992
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012894495412844035,
+      "loss": 1.5836,
+      "step": 2993
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001288761467889908,
+      "loss": 1.4308,
+      "step": 2994
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012880733944954127,
+      "loss": 1.675,
+      "step": 2995
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012873853211009173,
+      "loss": 1.4567,
+      "step": 2996
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001286697247706422,
+      "loss": 1.4889,
+      "step": 2997
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012860091743119266,
+      "loss": 1.5249,
+      "step": 2998
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012853211009174312,
+      "loss": 1.4043,
+      "step": 2999
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012846330275229358,
+      "loss": 1.4587,
+      "step": 3000
+    },
+    {
+      "epoch": 1.85,
+      "eval_bleu": 2.0001405343797235e-14,
+      "eval_loss": 1.9121204614639282,
+      "eval_runtime": 2596.1075,
+      "eval_samples_per_second": 5.685,
+      "eval_steps_per_second": 0.711,
+      "step": 3000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012839449541284404,
+      "loss": 1.39,
+      "step": 3001
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001283256880733945,
+      "loss": 1.3469,
+      "step": 3002
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012825688073394496,
+      "loss": 1.1199,
+      "step": 3003
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001281880733944954,
+      "loss": 1.1479,
+      "step": 3004
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012811926605504586,
+      "loss": 1.4059,
+      "step": 3005
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012805045871559632,
+      "loss": 1.1349,
+      "step": 3006
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012798165137614678,
+      "loss": 1.3401,
+      "step": 3007
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012791284403669724,
+      "loss": 1.1072,
+      "step": 3008
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001278440366972477,
+      "loss": 1.2175,
+      "step": 3009
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012777522935779816,
+      "loss": 1.2285,
+      "step": 3010
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012770642201834863,
+      "loss": 1.0367,
+      "step": 3011
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012763761467889906,
+      "loss": 1.0945,
+      "step": 3012
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012756880733944952,
+      "loss": 1.0363,
+      "step": 3013
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012749999999999998,
+      "loss": 1.0025,
+      "step": 3014
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012743119266055044,
+      "loss": 0.8898,
+      "step": 3015
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001273623853211009,
+      "loss": 0.8455,
+      "step": 3016
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012729357798165137,
+      "loss": 0.9232,
+      "step": 3017
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012722477064220183,
+      "loss": 0.9279,
+      "step": 3018
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001271559633027523,
+      "loss": 0.7919,
+      "step": 3019
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012708715596330272,
+      "loss": 0.8098,
+      "step": 3020
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012701834862385318,
+      "loss": 2.407,
+      "step": 3021
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012694954128440365,
+      "loss": 2.0285,
+      "step": 3022
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001268807339449541,
+      "loss": 1.9198,
+      "step": 3023
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012681192660550457,
+      "loss": 1.8226,
+      "step": 3024
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012674311926605503,
+      "loss": 1.5921,
+      "step": 3025
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001266743119266055,
+      "loss": 1.6613,
+      "step": 3026
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012660550458715595,
+      "loss": 1.8662,
+      "step": 3027
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001265366972477064,
+      "loss": 1.6606,
+      "step": 3028
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012646788990825687,
+      "loss": 1.6165,
+      "step": 3029
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012639908256880734,
+      "loss": 1.7079,
+      "step": 3030
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001263302752293578,
+      "loss": 1.5306,
+      "step": 3031
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012626146788990826,
+      "loss": 1.6343,
+      "step": 3032
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012619266055045872,
+      "loss": 1.5417,
+      "step": 3033
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012612385321100918,
+      "loss": 1.4634,
+      "step": 3034
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012605504587155964,
+      "loss": 1.5174,
+      "step": 3035
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012598623853211008,
+      "loss": 1.5647,
+      "step": 3036
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012591743119266054,
+      "loss": 1.6638,
+      "step": 3037
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000125848623853211,
+      "loss": 1.589,
+      "step": 3038
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012577981651376146,
+      "loss": 1.7321,
+      "step": 3039
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012571100917431192,
+      "loss": 1.4931,
+      "step": 3040
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012564220183486238,
+      "loss": 1.4464,
+      "step": 3041
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012557339449541284,
+      "loss": 1.4127,
+      "step": 3042
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001255045871559633,
+      "loss": 1.5282,
+      "step": 3043
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012543577981651374,
+      "loss": 1.5018,
+      "step": 3044
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001253669724770642,
+      "loss": 1.5265,
+      "step": 3045
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012529816513761466,
+      "loss": 1.4623,
+      "step": 3046
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012522935779816512,
+      "loss": 1.4353,
+      "step": 3047
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012516055045871558,
+      "loss": 1.2729,
+      "step": 3048
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012509174311926605,
+      "loss": 1.4302,
+      "step": 3049
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001250229357798165,
+      "loss": 1.3645,
+      "step": 3050
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012495412844036697,
+      "loss": 1.4171,
+      "step": 3051
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001248853211009174,
+      "loss": 1.3164,
+      "step": 3052
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012481651376146786,
+      "loss": 1.3799,
+      "step": 3053
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012474770642201832,
+      "loss": 1.3677,
+      "step": 3054
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012467889908256879,
+      "loss": 1.166,
+      "step": 3055
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012461009174311925,
+      "loss": 1.2092,
+      "step": 3056
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001245412844036697,
+      "loss": 1.1978,
+      "step": 3057
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012447247706422017,
+      "loss": 1.2847,
+      "step": 3058
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012440366972477063,
+      "loss": 1.0555,
+      "step": 3059
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001243348623853211,
+      "loss": 0.9679,
+      "step": 3060
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012426605504587155,
+      "loss": 1.1345,
+      "step": 3061
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012419724770642201,
+      "loss": 1.111,
+      "step": 3062
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012412844036697248,
+      "loss": 1.0165,
+      "step": 3063
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012405963302752294,
+      "loss": 1.1669,
+      "step": 3064
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001239908256880734,
+      "loss": 1.0485,
+      "step": 3065
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012392201834862386,
+      "loss": 0.885,
+      "step": 3066
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012385321100917432,
+      "loss": 0.9282,
+      "step": 3067
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012378440366972476,
+      "loss": 0.8279,
+      "step": 3068
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012371559633027522,
+      "loss": 0.8235,
+      "step": 3069
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012364678899082568,
+      "loss": 0.7094,
+      "step": 3070
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012357798165137614,
+      "loss": 2.3064,
+      "step": 3071
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001235091743119266,
+      "loss": 2.1128,
+      "step": 3072
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012344036697247706,
+      "loss": 1.8536,
+      "step": 3073
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012337155963302752,
+      "loss": 1.8792,
+      "step": 3074
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012330275229357798,
+      "loss": 1.8256,
+      "step": 3075
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012323394495412842,
+      "loss": 1.6448,
+      "step": 3076
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012316513761467888,
+      "loss": 1.5534,
+      "step": 3077
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012309633027522934,
+      "loss": 1.8025,
+      "step": 3078
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001230275229357798,
+      "loss": 1.7608,
+      "step": 3079
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012295871559633026,
+      "loss": 1.6532,
+      "step": 3080
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012288990825688072,
+      "loss": 1.6359,
+      "step": 3081
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012282110091743119,
+      "loss": 1.5604,
+      "step": 3082
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012275229357798165,
+      "loss": 1.5728,
+      "step": 3083
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012268348623853208,
+      "loss": 1.4838,
+      "step": 3084
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012261467889908254,
+      "loss": 1.6424,
+      "step": 3085
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.000122545871559633,
+      "loss": 1.6252,
+      "step": 3086
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012247706422018347,
+      "loss": 1.5319,
+      "step": 3087
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012240825688073393,
+      "loss": 1.5595,
+      "step": 3088
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001223394495412844,
+      "loss": 1.5667,
+      "step": 3089
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012227064220183485,
+      "loss": 1.4773,
+      "step": 3090
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001222018348623853,
+      "loss": 1.4734,
+      "step": 3091
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012213302752293577,
+      "loss": 1.4411,
+      "step": 3092
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012206422018348622,
+      "loss": 1.4964,
+      "step": 3093
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012199541284403668,
+      "loss": 1.4735,
+      "step": 3094
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012192660550458714,
+      "loss": 1.5404,
+      "step": 3095
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001218577981651376,
+      "loss": 1.3542,
+      "step": 3096
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012178899082568806,
+      "loss": 1.4582,
+      "step": 3097
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012172018348623853,
+      "loss": 1.3486,
+      "step": 3098
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012165137614678899,
+      "loss": 1.2559,
+      "step": 3099
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012158256880733943,
+      "loss": 1.2626,
+      "step": 3100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001215137614678899,
+      "loss": 1.332,
+      "step": 3101
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012144495412844036,
+      "loss": 1.319,
+      "step": 3102
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012137614678899082,
+      "loss": 1.2508,
+      "step": 3103
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012130733944954128,
+      "loss": 1.1567,
+      "step": 3104
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012123853211009174,
+      "loss": 1.2438,
+      "step": 3105
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001211697247706422,
+      "loss": 1.1211,
+      "step": 3106
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012110091743119266,
+      "loss": 1.2314,
+      "step": 3107
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001210321100917431,
+      "loss": 1.1626,
+      "step": 3108
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012096330275229356,
+      "loss": 1.0391,
+      "step": 3109
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012089449541284402,
+      "loss": 1.0671,
+      "step": 3110
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012082568807339448,
+      "loss": 0.9932,
+      "step": 3111
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012075688073394494,
+      "loss": 1.0327,
+      "step": 3112
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001206880733944954,
+      "loss": 1.0487,
+      "step": 3113
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012061926605504587,
+      "loss": 1.0331,
+      "step": 3114
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012055045871559633,
+      "loss": 0.8263,
+      "step": 3115
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012048165137614677,
+      "loss": 0.9046,
+      "step": 3116
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012041284403669724,
+      "loss": 0.864,
+      "step": 3117
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001203440366972477,
+      "loss": 0.8924,
+      "step": 3118
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00012027522935779816,
+      "loss": 0.7518,
+      "step": 3119
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00012020642201834862,
+      "loss": 0.6808,
+      "step": 3120
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00012013761467889908,
+      "loss": 2.3114,
+      "step": 3121
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00012006880733944954,
+      "loss": 2.2515,
+      "step": 3122
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 1.7827,
+      "step": 3123
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011993119266055044,
+      "loss": 1.7588,
+      "step": 3124
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001198623853211009,
+      "loss": 1.742,
+      "step": 3125
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011979357798165136,
+      "loss": 1.6805,
+      "step": 3126
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011972477064220182,
+      "loss": 1.7548,
+      "step": 3127
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011965596330275228,
+      "loss": 1.6756,
+      "step": 3128
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011958715596330274,
+      "loss": 1.6954,
+      "step": 3129
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001195183486238532,
+      "loss": 1.573,
+      "step": 3130
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011944954128440367,
+      "loss": 1.6479,
+      "step": 3131
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011938073394495411,
+      "loss": 1.6855,
+      "step": 3132
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011931192660550458,
+      "loss": 1.5933,
+      "step": 3133
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00011924311926605504,
+      "loss": 1.5387,
+      "step": 3134
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001191743119266055,
+      "loss": 1.5856,
+      "step": 3135
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011910550458715596,
+      "loss": 1.5024,
+      "step": 3136
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011903669724770642,
+      "loss": 1.4833,
+      "step": 3137
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011896788990825688,
+      "loss": 1.4254,
+      "step": 3138
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011889908256880733,
+      "loss": 1.4986,
+      "step": 3139
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011883027522935778,
+      "loss": 1.5669,
+      "step": 3140
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011876146788990824,
+      "loss": 1.4592,
+      "step": 3141
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001186926605504587,
+      "loss": 1.3417,
+      "step": 3142
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011862385321100916,
+      "loss": 1.4769,
+      "step": 3143
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011855504587155962,
+      "loss": 1.5297,
+      "step": 3144
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011848623853211008,
+      "loss": 1.5434,
+      "step": 3145
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011841743119266054,
+      "loss": 1.4386,
+      "step": 3146
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.000118348623853211,
+      "loss": 1.4406,
+      "step": 3147
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011827981651376145,
+      "loss": 1.4261,
+      "step": 3148
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011821100917431191,
+      "loss": 1.3689,
+      "step": 3149
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011814220183486238,
+      "loss": 1.2491,
+      "step": 3150
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00011807339449541284,
+      "loss": 1.3364,
+      "step": 3151
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001180045871559633,
+      "loss": 1.256,
+      "step": 3152
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011793577981651376,
+      "loss": 1.3198,
+      "step": 3153
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011786697247706421,
+      "loss": 1.1744,
+      "step": 3154
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011779816513761467,
+      "loss": 1.2297,
+      "step": 3155
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011772935779816512,
+      "loss": 1.0893,
+      "step": 3156
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011766055045871558,
+      "loss": 1.2357,
+      "step": 3157
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011759174311926604,
+      "loss": 1.0637,
+      "step": 3158
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001175229357798165,
+      "loss": 0.9954,
+      "step": 3159
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011745412844036696,
+      "loss": 1.1149,
+      "step": 3160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011738532110091742,
+      "loss": 1.0899,
+      "step": 3161
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011731651376146788,
+      "loss": 0.9103,
+      "step": 3162
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011724770642201835,
+      "loss": 1.0167,
+      "step": 3163
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011717889908256879,
+      "loss": 0.9564,
+      "step": 3164
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011711009174311925,
+      "loss": 0.9899,
+      "step": 3165
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011704128440366972,
+      "loss": 0.89,
+      "step": 3166
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00011697247706422018,
+      "loss": 0.7885,
+      "step": 3167
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011690366972477064,
+      "loss": 0.8358,
+      "step": 3168
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001168348623853211,
+      "loss": 0.7267,
+      "step": 3169
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011676605504587155,
+      "loss": 0.8229,
+      "step": 3170
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011669724770642201,
+      "loss": 2.4191,
+      "step": 3171
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011662844036697246,
+      "loss": 2.0197,
+      "step": 3172
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011655963302752292,
+      "loss": 1.9992,
+      "step": 3173
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011649082568807338,
+      "loss": 1.7866,
+      "step": 3174
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011642201834862384,
+      "loss": 1.6166,
+      "step": 3175
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001163532110091743,
+      "loss": 1.642,
+      "step": 3176
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011628440366972476,
+      "loss": 1.7037,
+      "step": 3177
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011621559633027522,
+      "loss": 1.6492,
+      "step": 3178
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011614678899082569,
+      "loss": 1.6288,
+      "step": 3179
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011607798165137613,
+      "loss": 1.5632,
+      "step": 3180
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001160091743119266,
+      "loss": 1.5532,
+      "step": 3181
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011594036697247706,
+      "loss": 1.7433,
+      "step": 3182
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00011587155963302752,
+      "loss": 1.5728,
+      "step": 3183
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011580275229357798,
+      "loss": 1.7057,
+      "step": 3184
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011573394495412844,
+      "loss": 1.5485,
+      "step": 3185
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011566513761467889,
+      "loss": 1.528,
+      "step": 3186
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011559633027522935,
+      "loss": 1.4964,
+      "step": 3187
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001155275229357798,
+      "loss": 1.516,
+      "step": 3188
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011545871559633026,
+      "loss": 1.6188,
+      "step": 3189
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011538990825688072,
+      "loss": 1.417,
+      "step": 3190
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011532110091743118,
+      "loss": 1.437,
+      "step": 3191
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011525229357798164,
+      "loss": 1.446,
+      "step": 3192
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001151834862385321,
+      "loss": 1.4484,
+      "step": 3193
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011511467889908256,
+      "loss": 1.3121,
+      "step": 3194
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011504587155963302,
+      "loss": 1.4484,
+      "step": 3195
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011497706422018347,
+      "loss": 1.5076,
+      "step": 3196
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011490825688073393,
+      "loss": 1.3903,
+      "step": 3197
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001148394495412844,
+      "loss": 1.3124,
+      "step": 3198
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00011477064220183486,
+      "loss": 1.1794,
+      "step": 3199
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011470183486238532,
+      "loss": 1.2289,
+      "step": 3200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011463302752293577,
+      "loss": 1.3136,
+      "step": 3201
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011456422018348623,
+      "loss": 1.2753,
+      "step": 3202
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011449541284403669,
+      "loss": 1.2726,
+      "step": 3203
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011442660550458714,
+      "loss": 1.2318,
+      "step": 3204
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001143577981651376,
+      "loss": 1.2314,
+      "step": 3205
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011428899082568806,
+      "loss": 1.2216,
+      "step": 3206
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011422018348623852,
+      "loss": 1.0592,
+      "step": 3207
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011415137614678898,
+      "loss": 1.1086,
+      "step": 3208
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011408256880733944,
+      "loss": 1.01,
+      "step": 3209
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001140137614678899,
+      "loss": 0.9835,
+      "step": 3210
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011394495412844036,
+      "loss": 1.0515,
+      "step": 3211
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011387614678899081,
+      "loss": 1.0544,
+      "step": 3212
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011380733944954127,
+      "loss": 1.0031,
+      "step": 3213
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00011373853211009173,
+      "loss": 0.8895,
+      "step": 3214
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001136697247706422,
+      "loss": 0.869,
+      "step": 3215
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011360091743119266,
+      "loss": 0.8822,
+      "step": 3216
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001135321100917431,
+      "loss": 0.8238,
+      "step": 3217
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011346330275229357,
+      "loss": 0.9088,
+      "step": 3218
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011339449541284403,
+      "loss": 0.8089,
+      "step": 3219
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011332568807339448,
+      "loss": 0.7532,
+      "step": 3220
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011325688073394494,
+      "loss": 2.1718,
+      "step": 3221
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001131880733944954,
+      "loss": 1.8052,
+      "step": 3222
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011311926605504586,
+      "loss": 1.6693,
+      "step": 3223
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011305045871559632,
+      "loss": 1.6312,
+      "step": 3224
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011298165137614678,
+      "loss": 1.5112,
+      "step": 3225
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011291284403669724,
+      "loss": 1.5097,
+      "step": 3226
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001128440366972477,
+      "loss": 1.5781,
+      "step": 3227
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011277522935779815,
+      "loss": 1.5083,
+      "step": 3228
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011270642201834861,
+      "loss": 1.5071,
+      "step": 3229
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011263761467889907,
+      "loss": 1.3595,
+      "step": 3230
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00011256880733944954,
+      "loss": 1.3151,
+      "step": 3231
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001125,
+      "loss": 1.4075,
+      "step": 3232
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011243119266055044,
+      "loss": 1.438,
+      "step": 3233
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001123623853211009,
+      "loss": 1.3186,
+      "step": 3234
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011229357798165137,
+      "loss": 1.1342,
+      "step": 3235
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011222477064220183,
+      "loss": 1.0946,
+      "step": 3236
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011215596330275228,
+      "loss": 1.0398,
+      "step": 3237
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011208715596330274,
+      "loss": 0.9152,
+      "step": 3238
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001120183486238532,
+      "loss": 0.8311,
+      "step": 3239
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011194954128440366,
+      "loss": 0.8015,
+      "step": 3240
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011188073394495412,
+      "loss": 1.8998,
+      "step": 3241
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011181192660550458,
+      "loss": 1.6994,
+      "step": 3242
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011174311926605504,
+      "loss": 1.449,
+      "step": 3243
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001116743119266055,
+      "loss": 1.48,
+      "step": 3244
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011160550458715595,
+      "loss": 1.45,
+      "step": 3245
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011153669724770641,
+      "loss": 1.3864,
+      "step": 3246
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011146788990825688,
+      "loss": 1.2641,
+      "step": 3247
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011139908256880732,
+      "loss": 1.3071,
+      "step": 3248
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011133027522935778,
+      "loss": 1.2711,
+      "step": 3249
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011126146788990825,
+      "loss": 1.1952,
+      "step": 3250
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011119266055045871,
+      "loss": 1.1409,
+      "step": 3251
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011112385321100917,
+      "loss": 1.2432,
+      "step": 3252
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011105504587155962,
+      "loss": 1.3421,
+      "step": 3253
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011098623853211008,
+      "loss": 1.1626,
+      "step": 3254
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011091743119266054,
+      "loss": 1.2161,
+      "step": 3255
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.000110848623853211,
+      "loss": 1.27,
+      "step": 3256
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011077981651376146,
+      "loss": 1.2663,
+      "step": 3257
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011071100917431192,
+      "loss": 1.1851,
+      "step": 3258
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011064220183486238,
+      "loss": 1.1674,
+      "step": 3259
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011057339449541284,
+      "loss": 1.1489,
+      "step": 3260
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011050458715596329,
+      "loss": 1.1105,
+      "step": 3261
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011043577981651375,
+      "loss": 1.2066,
+      "step": 3262
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011036697247706421,
+      "loss": 1.1664,
+      "step": 3263
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011029816513761466,
+      "loss": 1.0803,
+      "step": 3264
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00011022935779816512,
+      "loss": 0.9479,
+      "step": 3265
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00011016055045871559,
+      "loss": 1.0562,
+      "step": 3266
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00011009174311926605,
+      "loss": 0.9889,
+      "step": 3267
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00011002293577981651,
+      "loss": 0.9113,
+      "step": 3268
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010995412844036696,
+      "loss": 0.9999,
+      "step": 3269
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010988532110091742,
+      "loss": 1.0455,
+      "step": 3270
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010981651376146788,
+      "loss": 0.8844,
+      "step": 3271
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010974770642201834,
+      "loss": 0.9276,
+      "step": 3272
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001096788990825688,
+      "loss": 0.9944,
+      "step": 3273
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010961009174311926,
+      "loss": 0.8686,
+      "step": 3274
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010954128440366972,
+      "loss": 0.8649,
+      "step": 3275
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010947247706422018,
+      "loss": 0.9012,
+      "step": 3276
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010940366972477063,
+      "loss": 0.8939,
+      "step": 3277
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001093348623853211,
+      "loss": 0.6883,
+      "step": 3278
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00010926605504587154,
+      "loss": 0.8001,
+      "step": 3279
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.000109197247706422,
+      "loss": 0.8376,
+      "step": 3280
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010912844036697246,
+      "loss": 0.7515,
+      "step": 3281
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010905963302752292,
+      "loss": 0.7799,
+      "step": 3282
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010899082568807339,
+      "loss": 0.815,
+      "step": 3283
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010892201834862385,
+      "loss": 0.6693,
+      "step": 3284
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001088532110091743,
+      "loss": 0.697,
+      "step": 3285
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010878440366972476,
+      "loss": 0.7176,
+      "step": 3286
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010871559633027522,
+      "loss": 0.5935,
+      "step": 3287
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010864678899082568,
+      "loss": 0.6069,
+      "step": 3288
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010857798165137614,
+      "loss": 0.5867,
+      "step": 3289
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001085091743119266,
+      "loss": 0.6292,
+      "step": 3290
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010844036697247706,
+      "loss": 1.9927,
+      "step": 3291
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010837155963302752,
+      "loss": 1.8382,
+      "step": 3292
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010830275229357797,
+      "loss": 1.6172,
+      "step": 3293
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010823394495412843,
+      "loss": 1.3558,
+      "step": 3294
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010816513761467888,
+      "loss": 1.3915,
+      "step": 3295
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00010809633027522934,
+      "loss": 1.3112,
+      "step": 3296
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001080275229357798,
+      "loss": 1.351,
+      "step": 3297
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010795871559633026,
+      "loss": 1.5167,
+      "step": 3298
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010788990825688073,
+      "loss": 1.4182,
+      "step": 3299
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010782110091743119,
+      "loss": 1.2447,
+      "step": 3300
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010775229357798163,
+      "loss": 1.2202,
+      "step": 3301
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001076834862385321,
+      "loss": 1.2386,
+      "step": 3302
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010761467889908256,
+      "loss": 1.2798,
+      "step": 3303
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010754587155963302,
+      "loss": 1.2992,
+      "step": 3304
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010747706422018348,
+      "loss": 1.2656,
+      "step": 3305
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010740825688073394,
+      "loss": 1.187,
+      "step": 3306
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001073394495412844,
+      "loss": 1.1373,
+      "step": 3307
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010727064220183486,
+      "loss": 1.1892,
+      "step": 3308
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010720183486238531,
+      "loss": 1.196,
+      "step": 3309
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010713302752293577,
+      "loss": 1.113,
+      "step": 3310
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010706422018348622,
+      "loss": 1.0907,
+      "step": 3311
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00010699541284403668,
+      "loss": 1.0878,
+      "step": 3312
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010692660550458714,
+      "loss": 1.0999,
+      "step": 3313
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001068577981651376,
+      "loss": 1.0508,
+      "step": 3314
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010678899082568807,
+      "loss": 1.1594,
+      "step": 3315
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010672018348623853,
+      "loss": 1.0458,
+      "step": 3316
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010665137614678897,
+      "loss": 0.9389,
+      "step": 3317
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010658256880733944,
+      "loss": 1.0557,
+      "step": 3318
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001065137614678899,
+      "loss": 0.924,
+      "step": 3319
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010644495412844036,
+      "loss": 1.0157,
+      "step": 3320
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010637614678899082,
+      "loss": 0.9298,
+      "step": 3321
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010630733944954128,
+      "loss": 0.9338,
+      "step": 3322
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010623853211009174,
+      "loss": 0.9585,
+      "step": 3323
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001061697247706422,
+      "loss": 0.8398,
+      "step": 3324
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010610091743119265,
+      "loss": 0.8221,
+      "step": 3325
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001060321100917431,
+      "loss": 0.8746,
+      "step": 3326
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010596330275229356,
+      "loss": 0.8275,
+      "step": 3327
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010589449541284402,
+      "loss": 0.8398,
+      "step": 3328
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00010582568807339448,
+      "loss": 0.8013,
+      "step": 3329
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010575688073394494,
+      "loss": 0.738,
+      "step": 3330
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001056880733944954,
+      "loss": 0.7928,
+      "step": 3331
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010561926605504587,
+      "loss": 0.7466,
+      "step": 3332
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010555045871559631,
+      "loss": 0.7771,
+      "step": 3333
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010548165137614678,
+      "loss": 0.7779,
+      "step": 3334
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010541284403669724,
+      "loss": 0.6664,
+      "step": 3335
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001053440366972477,
+      "loss": 0.5874,
+      "step": 3336
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010527522935779816,
+      "loss": 0.6239,
+      "step": 3337
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010520642201834862,
+      "loss": 0.6225,
+      "step": 3338
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010513761467889908,
+      "loss": 0.5751,
+      "step": 3339
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010506880733944954,
+      "loss": 0.6024,
+      "step": 3340
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010499999999999999,
+      "loss": 1.9908,
+      "step": 3341
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010493119266055044,
+      "loss": 1.8556,
+      "step": 3342
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001048623853211009,
+      "loss": 1.4763,
+      "step": 3343
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010479357798165136,
+      "loss": 1.3352,
+      "step": 3344
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00010472477064220182,
+      "loss": 1.4443,
+      "step": 3345
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010465596330275228,
+      "loss": 1.3799,
+      "step": 3346
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010458715596330274,
+      "loss": 1.3652,
+      "step": 3347
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001045183486238532,
+      "loss": 1.4642,
+      "step": 3348
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010444954128440365,
+      "loss": 1.3314,
+      "step": 3349
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010438073394495412,
+      "loss": 1.2554,
+      "step": 3350
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010431192660550458,
+      "loss": 1.3257,
+      "step": 3351
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010424311926605504,
+      "loss": 1.1991,
+      "step": 3352
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001041743119266055,
+      "loss": 1.1797,
+      "step": 3353
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010410550458715596,
+      "loss": 1.3033,
+      "step": 3354
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010403669724770642,
+      "loss": 1.35,
+      "step": 3355
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010396788990825688,
+      "loss": 1.2393,
+      "step": 3356
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010389908256880732,
+      "loss": 1.2471,
+      "step": 3357
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010383027522935778,
+      "loss": 1.2599,
+      "step": 3358
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010376146788990824,
+      "loss": 1.2458,
+      "step": 3359
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001036926605504587,
+      "loss": 1.0753,
+      "step": 3360
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00010362385321100916,
+      "loss": 1.1329,
+      "step": 3361
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010355504587155962,
+      "loss": 1.134,
+      "step": 3362
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010348623853211008,
+      "loss": 1.112,
+      "step": 3363
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010341743119266055,
+      "loss": 1.1027,
+      "step": 3364
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000103348623853211,
+      "loss": 0.9633,
+      "step": 3365
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010327981651376145,
+      "loss": 1.0392,
+      "step": 3366
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010321100917431192,
+      "loss": 1.1153,
+      "step": 3367
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010314220183486238,
+      "loss": 1.0044,
+      "step": 3368
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010307339449541284,
+      "loss": 0.9651,
+      "step": 3369
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001030045871559633,
+      "loss": 0.9368,
+      "step": 3370
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010293577981651376,
+      "loss": 0.8179,
+      "step": 3371
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010286697247706422,
+      "loss": 1.0389,
+      "step": 3372
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010279816513761466,
+      "loss": 0.8948,
+      "step": 3373
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010272935779816512,
+      "loss": 0.8474,
+      "step": 3374
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010266055045871558,
+      "loss": 0.9365,
+      "step": 3375
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00010259174311926604,
+      "loss": 0.9399,
+      "step": 3376
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001025229357798165,
+      "loss": 0.818,
+      "step": 3377
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010245412844036696,
+      "loss": 0.8144,
+      "step": 3378
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010238532110091742,
+      "loss": 0.8105,
+      "step": 3379
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010231651376146789,
+      "loss": 0.6731,
+      "step": 3380
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010224770642201833,
+      "loss": 0.8096,
+      "step": 3381
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001021788990825688,
+      "loss": 0.7844,
+      "step": 3382
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010211009174311926,
+      "loss": 0.7096,
+      "step": 3383
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010204128440366972,
+      "loss": 0.5881,
+      "step": 3384
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010197247706422018,
+      "loss": 0.7115,
+      "step": 3385
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010190366972477064,
+      "loss": 0.6648,
+      "step": 3386
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001018348623853211,
+      "loss": 0.6421,
+      "step": 3387
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010176605504587156,
+      "loss": 0.5482,
+      "step": 3388
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.000101697247706422,
+      "loss": 0.597,
+      "step": 3389
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010162844036697246,
+      "loss": 0.7062,
+      "step": 3390
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010155963302752292,
+      "loss": 1.9705,
+      "step": 3391
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010149082568807338,
+      "loss": 1.6736,
+      "step": 3392
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00010142201834862384,
+      "loss": 1.641,
+      "step": 3393
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001013532110091743,
+      "loss": 1.5655,
+      "step": 3394
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00010128440366972476,
+      "loss": 1.4845,
+      "step": 3395
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00010121559633027523,
+      "loss": 1.3657,
+      "step": 3396
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00010114678899082567,
+      "loss": 1.4253,
+      "step": 3397
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00010107798165137613,
+      "loss": 1.4804,
+      "step": 3398
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001010091743119266,
+      "loss": 1.4445,
+      "step": 3399
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00010094036697247706,
+      "loss": 1.2404,
+      "step": 3400
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00010087155963302752,
+      "loss": 1.2933,
+      "step": 3401
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00010080275229357798,
+      "loss": 1.25,
+      "step": 3402
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00010073394495412844,
+      "loss": 1.1904,
+      "step": 3403
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001006651376146789,
+      "loss": 1.3232,
+      "step": 3404
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00010059633027522934,
+      "loss": 1.2416,
+      "step": 3405
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001005275229357798,
+      "loss": 1.2251,
+      "step": 3406
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00010045871559633026,
+      "loss": 1.1865,
+      "step": 3407
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00010038990825688072,
+      "loss": 1.2328,
+      "step": 3408
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00010032110091743118,
+      "loss": 1.1768,
+      "step": 3409
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00010025229357798164,
+      "loss": 1.1402,
+      "step": 3410
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001001834862385321,
+      "loss": 1.2011,
+      "step": 3411
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00010011467889908256,
+      "loss": 1.1794,
+      "step": 3412
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00010004587155963301,
+      "loss": 1.0074,
+      "step": 3413
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.997706422018347e-05,
+      "loss": 1.0444,
+      "step": 3414
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.990825688073394e-05,
+      "loss": 1.1296,
+      "step": 3415
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.98394495412844e-05,
+      "loss": 1.0497,
+      "step": 3416
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.977064220183486e-05,
+      "loss": 1.0577,
+      "step": 3417
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.970183486238532e-05,
+      "loss": 0.9681,
+      "step": 3418
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.963302752293578e-05,
+      "loss": 1.0792,
+      "step": 3419
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.956422018348624e-05,
+      "loss": 1.0182,
+      "step": 3420
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.949541284403668e-05,
+      "loss": 0.9079,
+      "step": 3421
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.942660550458714e-05,
+      "loss": 0.878,
+      "step": 3422
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.93577981651376e-05,
+      "loss": 0.9249,
+      "step": 3423
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.928899082568806e-05,
+      "loss": 0.9437,
+      "step": 3424
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.922018348623852e-05,
+      "loss": 0.9596,
+      "step": 3425
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.915137614678898e-05,
+      "loss": 1.0187,
+      "step": 3426
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.908256880733944e-05,
+      "loss": 0.8151,
+      "step": 3427
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.90137614678899e-05,
+      "loss": 0.8303,
+      "step": 3428
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.894495412844035e-05,
+      "loss": 0.8061,
+      "step": 3429
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.887614678899081e-05,
+      "loss": 0.7894,
+      "step": 3430
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.880733944954127e-05,
+      "loss": 0.7404,
+      "step": 3431
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.873853211009174e-05,
+      "loss": 0.7282,
+      "step": 3432
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.86697247706422e-05,
+      "loss": 0.7364,
+      "step": 3433
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.860091743119266e-05,
+      "loss": 0.7194,
+      "step": 3434
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.853211009174312e-05,
+      "loss": 0.6544,
+      "step": 3435
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.846330275229358e-05,
+      "loss": 0.7254,
+      "step": 3436
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.839449541284402e-05,
+      "loss": 0.6503,
+      "step": 3437
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.832568807339448e-05,
+      "loss": 0.6116,
+      "step": 3438
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.825688073394494e-05,
+      "loss": 0.6194,
+      "step": 3439
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.81880733944954e-05,
+      "loss": 0.6522,
+      "step": 3440
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.811926605504586e-05,
+      "loss": 1.935,
+      "step": 3441
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.805045871559632e-05,
+      "loss": 1.8507,
+      "step": 3442
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.798165137614678e-05,
+      "loss": 1.5835,
+      "step": 3443
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.791284403669724e-05,
+      "loss": 1.3481,
+      "step": 3444
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.784403669724769e-05,
+      "loss": 1.4314,
+      "step": 3445
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.777522935779815e-05,
+      "loss": 1.302,
+      "step": 3446
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.770642201834861e-05,
+      "loss": 1.4149,
+      "step": 3447
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.763761467889908e-05,
+      "loss": 1.2766,
+      "step": 3448
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.756880733944954e-05,
+      "loss": 1.3728,
+      "step": 3449
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.75e-05,
+      "loss": 1.1997,
+      "step": 3450
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.743119266055046e-05,
+      "loss": 1.33,
+      "step": 3451
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.736238532110092e-05,
+      "loss": 1.3506,
+      "step": 3452
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.729357798165135e-05,
+      "loss": 1.2247,
+      "step": 3453
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.722477064220182e-05,
+      "loss": 1.1914,
+      "step": 3454
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.715596330275228e-05,
+      "loss": 1.1132,
+      "step": 3455
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.708715596330274e-05,
+      "loss": 1.139,
+      "step": 3456
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.70183486238532e-05,
+      "loss": 1.1673,
+      "step": 3457
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.694954128440366e-05,
+      "loss": 1.1573,
+      "step": 3458
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.688073394495412e-05,
+      "loss": 1.198,
+      "step": 3459
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.681192660550458e-05,
+      "loss": 1.138,
+      "step": 3460
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.674311926605503e-05,
+      "loss": 1.115,
+      "step": 3461
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.667431192660549e-05,
+      "loss": 1.1485,
+      "step": 3462
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.660550458715595e-05,
+      "loss": 1.0184,
+      "step": 3463
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.653669724770642e-05,
+      "loss": 1.0353,
+      "step": 3464
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.646788990825688e-05,
+      "loss": 1.1045,
+      "step": 3465
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.639908256880734e-05,
+      "loss": 1.0095,
+      "step": 3466
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.63302752293578e-05,
+      "loss": 1.0109,
+      "step": 3467
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.626146788990826e-05,
+      "loss": 1.0251,
+      "step": 3468
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.61926605504587e-05,
+      "loss": 1.0543,
+      "step": 3469
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.612385321100916e-05,
+      "loss": 0.9435,
+      "step": 3470
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.605504587155962e-05,
+      "loss": 0.9223,
+      "step": 3471
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.598623853211008e-05,
+      "loss": 0.9981,
+      "step": 3472
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.591743119266054e-05,
+      "loss": 0.9542,
+      "step": 3473
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.5848623853211e-05,
+      "loss": 0.8352,
+      "step": 3474
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.577981651376146e-05,
+      "loss": 0.8954,
+      "step": 3475
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.571100917431192e-05,
+      "loss": 0.7515,
+      "step": 3476
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.564220183486237e-05,
+      "loss": 0.8236,
+      "step": 3477
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.557339449541283e-05,
+      "loss": 0.902,
+      "step": 3478
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.55045871559633e-05,
+      "loss": 0.6544,
+      "step": 3479
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.543577981651376e-05,
+      "loss": 0.9736,
+      "step": 3480
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.536697247706422e-05,
+      "loss": 0.8371,
+      "step": 3481
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.529816513761468e-05,
+      "loss": 0.7488,
+      "step": 3482
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.522935779816514e-05,
+      "loss": 0.7905,
+      "step": 3483
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.51605504587156e-05,
+      "loss": 0.6648,
+      "step": 3484
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.509174311926603e-05,
+      "loss": 0.7115,
+      "step": 3485
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.50229357798165e-05,
+      "loss": 0.6977,
+      "step": 3486
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.495412844036696e-05,
+      "loss": 0.7411,
+      "step": 3487
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.488532110091742e-05,
+      "loss": 0.6174,
+      "step": 3488
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.481651376146788e-05,
+      "loss": 0.6065,
+      "step": 3489
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.474770642201834e-05,
+      "loss": 0.6839,
+      "step": 3490
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.46788990825688e-05,
+      "loss": 1.9471,
+      "step": 3491
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.461009174311926e-05,
+      "loss": 1.8161,
+      "step": 3492
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.454128440366971e-05,
+      "loss": 1.5657,
+      "step": 3493
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.447247706422017e-05,
+      "loss": 1.3685,
+      "step": 3494
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.440366972477063e-05,
+      "loss": 1.4222,
+      "step": 3495
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.43348623853211e-05,
+      "loss": 1.289,
+      "step": 3496
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.426605504587156e-05,
+      "loss": 1.4978,
+      "step": 3497
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.419724770642202e-05,
+      "loss": 1.2522,
+      "step": 3498
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.412844036697248e-05,
+      "loss": 1.1857,
+      "step": 3499
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.405963302752294e-05,
+      "loss": 1.288,
+      "step": 3500
+    },
+    {
+      "epoch": 2.16,
+      "eval_bleu": 4.138692653029469e-15,
+      "eval_loss": 1.8705445528030396,
+      "eval_runtime": 2629.971,
+      "eval_samples_per_second": 5.612,
+      "eval_steps_per_second": 0.702,
+      "step": 3500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.399082568807337e-05,
+      "loss": 1.2475,
+      "step": 3501
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.392201834862384e-05,
+      "loss": 1.3109,
+      "step": 3502
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.38532110091743e-05,
+      "loss": 1.3751,
+      "step": 3503
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.378440366972476e-05,
+      "loss": 1.1444,
+      "step": 3504
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.371559633027522e-05,
+      "loss": 1.185,
+      "step": 3505
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.364678899082568e-05,
+      "loss": 1.2135,
+      "step": 3506
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.357798165137614e-05,
+      "loss": 1.1444,
+      "step": 3507
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.35091743119266e-05,
+      "loss": 1.0526,
+      "step": 3508
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.344036697247705e-05,
+      "loss": 1.1024,
+      "step": 3509
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.337155963302751e-05,
+      "loss": 1.0709,
+      "step": 3510
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.330275229357797e-05,
+      "loss": 0.9865,
+      "step": 3511
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.323394495412843e-05,
+      "loss": 1.1247,
+      "step": 3512
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.31651376146789e-05,
+      "loss": 1.1548,
+      "step": 3513
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.309633027522936e-05,
+      "loss": 1.207,
+      "step": 3514
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.302752293577982e-05,
+      "loss": 1.0402,
+      "step": 3515
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.295871559633028e-05,
+      "loss": 1.0626,
+      "step": 3516
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.288990825688071e-05,
+      "loss": 1.0654,
+      "step": 3517
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.282110091743117e-05,
+      "loss": 0.9486,
+      "step": 3518
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.275229357798164e-05,
+      "loss": 0.8802,
+      "step": 3519
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.26834862385321e-05,
+      "loss": 1.0842,
+      "step": 3520
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.261467889908256e-05,
+      "loss": 0.8901,
+      "step": 3521
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.254587155963302e-05,
+      "loss": 1.0226,
+      "step": 3522
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.247706422018348e-05,
+      "loss": 0.7663,
+      "step": 3523
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.240825688073394e-05,
+      "loss": 0.8909,
+      "step": 3524
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.233944954128439e-05,
+      "loss": 0.9203,
+      "step": 3525
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.227064220183485e-05,
+      "loss": 0.8367,
+      "step": 3526
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.220183486238531e-05,
+      "loss": 0.6632,
+      "step": 3527
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.213302752293577e-05,
+      "loss": 0.7502,
+      "step": 3528
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.206422018348624e-05,
+      "loss": 0.7843,
+      "step": 3529
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.19954128440367e-05,
+      "loss": 0.7437,
+      "step": 3530
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.192660550458716e-05,
+      "loss": 0.762,
+      "step": 3531
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.185779816513762e-05,
+      "loss": 0.7725,
+      "step": 3532
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.178899082568805e-05,
+      "loss": 0.6965,
+      "step": 3533
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.172018348623851e-05,
+      "loss": 0.7457,
+      "step": 3534
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.165137614678898e-05,
+      "loss": 0.6144,
+      "step": 3535
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.158256880733944e-05,
+      "loss": 0.6706,
+      "step": 3536
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.15137614678899e-05,
+      "loss": 0.6769,
+      "step": 3537
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.144495412844036e-05,
+      "loss": 0.5808,
+      "step": 3538
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.137614678899082e-05,
+      "loss": 0.6593,
+      "step": 3539
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.130733944954128e-05,
+      "loss": 0.7093,
+      "step": 3540
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.123853211009173e-05,
+      "loss": 1.8793,
+      "step": 3541
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.116972477064219e-05,
+      "loss": 1.7687,
+      "step": 3542
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.110091743119265e-05,
+      "loss": 1.4532,
+      "step": 3543
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.103211009174311e-05,
+      "loss": 1.5811,
+      "step": 3544
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.096330275229357e-05,
+      "loss": 1.5387,
+      "step": 3545
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.089449541284404e-05,
+      "loss": 1.35,
+      "step": 3546
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.08256880733945e-05,
+      "loss": 1.355,
+      "step": 3547
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.075688073394496e-05,
+      "loss": 1.2898,
+      "step": 3548
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.068807339449539e-05,
+      "loss": 1.2948,
+      "step": 3549
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.061926605504585e-05,
+      "loss": 1.3558,
+      "step": 3550
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.055045871559632e-05,
+      "loss": 1.2416,
+      "step": 3551
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.048165137614678e-05,
+      "loss": 1.2496,
+      "step": 3552
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.041284403669724e-05,
+      "loss": 1.3016,
+      "step": 3553
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.03440366972477e-05,
+      "loss": 1.0363,
+      "step": 3554
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.027522935779816e-05,
+      "loss": 1.1286,
+      "step": 3555
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.020642201834862e-05,
+      "loss": 1.1571,
+      "step": 3556
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.013761467889907e-05,
+      "loss": 1.1658,
+      "step": 3557
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.006880733944953e-05,
+      "loss": 1.1334,
+      "step": 3558
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 1.0564,
+      "step": 3559
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.993119266055045e-05,
+      "loss": 1.227,
+      "step": 3560
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.986238532110091e-05,
+      "loss": 1.2322,
+      "step": 3561
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.979357798165138e-05,
+      "loss": 1.1075,
+      "step": 3562
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.972477064220184e-05,
+      "loss": 1.0504,
+      "step": 3563
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.96559633027523e-05,
+      "loss": 1.1546,
+      "step": 3564
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.958715596330273e-05,
+      "loss": 1.0451,
+      "step": 3565
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.95183486238532e-05,
+      "loss": 1.0315,
+      "step": 3566
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.944954128440366e-05,
+      "loss": 1.054,
+      "step": 3567
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.938073394495412e-05,
+      "loss": 1.0141,
+      "step": 3568
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.931192660550458e-05,
+      "loss": 0.941,
+      "step": 3569
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.924311926605504e-05,
+      "loss": 0.865,
+      "step": 3570
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.91743119266055e-05,
+      "loss": 0.8931,
+      "step": 3571
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.910550458715596e-05,
+      "loss": 0.8717,
+      "step": 3572
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.903669724770641e-05,
+      "loss": 0.977,
+      "step": 3573
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.896788990825687e-05,
+      "loss": 0.8469,
+      "step": 3574
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.889908256880733e-05,
+      "loss": 0.8662,
+      "step": 3575
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.883027522935779e-05,
+      "loss": 0.9046,
+      "step": 3576
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.876146788990825e-05,
+      "loss": 0.7823,
+      "step": 3577
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.869266055045872e-05,
+      "loss": 0.8266,
+      "step": 3578
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.862385321100918e-05,
+      "loss": 0.8557,
+      "step": 3579
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.855504587155964e-05,
+      "loss": 0.7704,
+      "step": 3580
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.848623853211007e-05,
+      "loss": 0.7383,
+      "step": 3581
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.841743119266053e-05,
+      "loss": 0.7076,
+      "step": 3582
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.8348623853211e-05,
+      "loss": 0.667,
+      "step": 3583
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.827981651376146e-05,
+      "loss": 0.7423,
+      "step": 3584
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.821100917431192e-05,
+      "loss": 0.6872,
+      "step": 3585
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.814220183486238e-05,
+      "loss": 0.8133,
+      "step": 3586
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.807339449541284e-05,
+      "loss": 0.5794,
+      "step": 3587
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.80045871559633e-05,
+      "loss": 0.6116,
+      "step": 3588
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.793577981651375e-05,
+      "loss": 0.5995,
+      "step": 3589
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.786697247706421e-05,
+      "loss": 0.6264,
+      "step": 3590
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.779816513761467e-05,
+      "loss": 1.9586,
+      "step": 3591
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.772935779816513e-05,
+      "loss": 1.6907,
+      "step": 3592
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.76605504587156e-05,
+      "loss": 1.4812,
+      "step": 3593
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.759174311926606e-05,
+      "loss": 1.4334,
+      "step": 3594
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.752293577981652e-05,
+      "loss": 1.208,
+      "step": 3595
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.745412844036698e-05,
+      "loss": 1.2686,
+      "step": 3596
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.738532110091741e-05,
+      "loss": 1.3475,
+      "step": 3597
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.731651376146787e-05,
+      "loss": 1.2428,
+      "step": 3598
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.724770642201833e-05,
+      "loss": 1.1171,
+      "step": 3599
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.71788990825688e-05,
+      "loss": 1.2145,
+      "step": 3600
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.711009174311926e-05,
+      "loss": 1.2259,
+      "step": 3601
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.704128440366972e-05,
+      "loss": 1.2641,
+      "step": 3602
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.697247706422018e-05,
+      "loss": 1.1741,
+      "step": 3603
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.690366972477064e-05,
+      "loss": 1.2045,
+      "step": 3604
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.683486238532109e-05,
+      "loss": 1.2254,
+      "step": 3605
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.676605504587155e-05,
+      "loss": 1.2583,
+      "step": 3606
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.669724770642201e-05,
+      "loss": 1.1906,
+      "step": 3607
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.662844036697247e-05,
+      "loss": 1.0979,
+      "step": 3608
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.655963302752293e-05,
+      "loss": 1.1706,
+      "step": 3609
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.64908256880734e-05,
+      "loss": 1.0694,
+      "step": 3610
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.642201834862386e-05,
+      "loss": 1.0504,
+      "step": 3611
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.635321100917432e-05,
+      "loss": 1.0533,
+      "step": 3612
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.628440366972475e-05,
+      "loss": 1.1704,
+      "step": 3613
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.621559633027521e-05,
+      "loss": 1.0594,
+      "step": 3614
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.614678899082567e-05,
+      "loss": 0.9619,
+      "step": 3615
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.607798165137614e-05,
+      "loss": 0.9403,
+      "step": 3616
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.60091743119266e-05,
+      "loss": 0.9973,
+      "step": 3617
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.594036697247706e-05,
+      "loss": 1.0126,
+      "step": 3618
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.587155963302752e-05,
+      "loss": 0.8992,
+      "step": 3619
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.580275229357798e-05,
+      "loss": 0.9369,
+      "step": 3620
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.573394495412843e-05,
+      "loss": 0.9439,
+      "step": 3621
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.566513761467889e-05,
+      "loss": 0.8866,
+      "step": 3622
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.559633027522935e-05,
+      "loss": 0.9114,
+      "step": 3623
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.552752293577981e-05,
+      "loss": 0.8652,
+      "step": 3624
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.545871559633027e-05,
+      "loss": 0.9351,
+      "step": 3625
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.538990825688073e-05,
+      "loss": 0.8621,
+      "step": 3626
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.53211009174312e-05,
+      "loss": 0.8522,
+      "step": 3627
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.525229357798166e-05,
+      "loss": 0.8697,
+      "step": 3628
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.518348623853209e-05,
+      "loss": 0.7722,
+      "step": 3629
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.511467889908255e-05,
+      "loss": 0.7667,
+      "step": 3630
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.504587155963301e-05,
+      "loss": 0.7215,
+      "step": 3631
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.497706422018348e-05,
+      "loss": 0.67,
+      "step": 3632
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.490825688073394e-05,
+      "loss": 0.7736,
+      "step": 3633
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.48394495412844e-05,
+      "loss": 0.6512,
+      "step": 3634
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.477064220183486e-05,
+      "loss": 0.5156,
+      "step": 3635
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.470183486238532e-05,
+      "loss": 0.6425,
+      "step": 3636
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.463302752293577e-05,
+      "loss": 0.5601,
+      "step": 3637
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.456422018348623e-05,
+      "loss": 0.5592,
+      "step": 3638
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.449541284403669e-05,
+      "loss": 0.733,
+      "step": 3639
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.442660550458715e-05,
+      "loss": 0.6353,
+      "step": 3640
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.435779816513761e-05,
+      "loss": 1.9434,
+      "step": 3641
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.428899082568807e-05,
+      "loss": 1.5979,
+      "step": 3642
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.422018348623854e-05,
+      "loss": 1.5136,
+      "step": 3643
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.4151376146789e-05,
+      "loss": 1.4489,
+      "step": 3644
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.408256880733943e-05,
+      "loss": 1.3091,
+      "step": 3645
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.401376146788989e-05,
+      "loss": 1.3172,
+      "step": 3646
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.394495412844035e-05,
+      "loss": 1.3353,
+      "step": 3647
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.387614678899081e-05,
+      "loss": 1.4128,
+      "step": 3648
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.380733944954128e-05,
+      "loss": 1.1792,
+      "step": 3649
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.373853211009174e-05,
+      "loss": 1.353,
+      "step": 3650
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.36697247706422e-05,
+      "loss": 1.2037,
+      "step": 3651
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.360091743119266e-05,
+      "loss": 1.1121,
+      "step": 3652
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.353211009174311e-05,
+      "loss": 1.2247,
+      "step": 3653
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.346330275229357e-05,
+      "loss": 1.1077,
+      "step": 3654
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.339449541284403e-05,
+      "loss": 1.2596,
+      "step": 3655
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.332568807339449e-05,
+      "loss": 1.0809,
+      "step": 3656
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.325688073394495e-05,
+      "loss": 1.2514,
+      "step": 3657
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.318807339449541e-05,
+      "loss": 1.2214,
+      "step": 3658
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.311926605504588e-05,
+      "loss": 1.05,
+      "step": 3659
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.305045871559632e-05,
+      "loss": 1.0321,
+      "step": 3660
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.298165137614677e-05,
+      "loss": 1.1047,
+      "step": 3661
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.291284403669723e-05,
+      "loss": 1.143,
+      "step": 3662
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.284403669724769e-05,
+      "loss": 1.1161,
+      "step": 3663
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.277522935779815e-05,
+      "loss": 1.0904,
+      "step": 3664
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.270642201834862e-05,
+      "loss": 0.9625,
+      "step": 3665
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.263761467889908e-05,
+      "loss": 1.0028,
+      "step": 3666
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.256880733944954e-05,
+      "loss": 0.9684,
+      "step": 3667
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.25e-05,
+      "loss": 1.0138,
+      "step": 3668
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.243119266055045e-05,
+      "loss": 0.8684,
+      "step": 3669
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.236238532110091e-05,
+      "loss": 0.9249,
+      "step": 3670
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.229357798165137e-05,
+      "loss": 0.9015,
+      "step": 3671
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.222477064220183e-05,
+      "loss": 1.019,
+      "step": 3672
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.215596330275229e-05,
+      "loss": 0.8567,
+      "step": 3673
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.208715596330275e-05,
+      "loss": 0.8534,
+      "step": 3674
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.201834862385321e-05,
+      "loss": 0.8346,
+      "step": 3675
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.194954128440366e-05,
+      "loss": 0.7746,
+      "step": 3676
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.188073394495411e-05,
+      "loss": 0.8471,
+      "step": 3677
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.181192660550457e-05,
+      "loss": 0.7325,
+      "step": 3678
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.174311926605503e-05,
+      "loss": 0.7493,
+      "step": 3679
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.16743119266055e-05,
+      "loss": 0.8608,
+      "step": 3680
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.160550458715596e-05,
+      "loss": 0.753,
+      "step": 3681
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.153669724770642e-05,
+      "loss": 0.63,
+      "step": 3682
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.146788990825688e-05,
+      "loss": 0.7141,
+      "step": 3683
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.139908256880734e-05,
+      "loss": 0.6127,
+      "step": 3684
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.133027522935779e-05,
+      "loss": 0.7448,
+      "step": 3685
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.126146788990825e-05,
+      "loss": 0.5936,
+      "step": 3686
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.119266055045871e-05,
+      "loss": 0.6129,
+      "step": 3687
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.112385321100917e-05,
+      "loss": 0.6168,
+      "step": 3688
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.105504587155963e-05,
+      "loss": 0.5678,
+      "step": 3689
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.09862385321101e-05,
+      "loss": 0.6078,
+      "step": 3690
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.091743119266055e-05,
+      "loss": 2.0308,
+      "step": 3691
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.0848623853211e-05,
+      "loss": 1.6064,
+      "step": 3692
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.077981651376145e-05,
+      "loss": 1.5351,
+      "step": 3693
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.071100917431191e-05,
+      "loss": 1.4937,
+      "step": 3694
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.064220183486237e-05,
+      "loss": 1.4196,
+      "step": 3695
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.057339449541283e-05,
+      "loss": 1.382,
+      "step": 3696
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.05045871559633e-05,
+      "loss": 1.2651,
+      "step": 3697
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.043577981651376e-05,
+      "loss": 1.3193,
+      "step": 3698
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.036697247706422e-05,
+      "loss": 1.2006,
+      "step": 3699
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.029816513761468e-05,
+      "loss": 1.2163,
+      "step": 3700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.022935779816513e-05,
+      "loss": 1.1444,
+      "step": 3701
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.016055045871559e-05,
+      "loss": 1.2515,
+      "step": 3702
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.009174311926605e-05,
+      "loss": 1.1149,
+      "step": 3703
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.002293577981651e-05,
+      "loss": 1.1447,
+      "step": 3704
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.995412844036697e-05,
+      "loss": 1.2943,
+      "step": 3705
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.988532110091743e-05,
+      "loss": 1.28,
+      "step": 3706
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.981651376146788e-05,
+      "loss": 1.0291,
+      "step": 3707
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.974770642201834e-05,
+      "loss": 1.1908,
+      "step": 3708
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.967889908256879e-05,
+      "loss": 1.087,
+      "step": 3709
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.961009174311925e-05,
+      "loss": 1.0814,
+      "step": 3710
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.954128440366971e-05,
+      "loss": 1.0838,
+      "step": 3711
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.947247706422017e-05,
+      "loss": 1.1268,
+      "step": 3712
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.940366972477063e-05,
+      "loss": 1.1201,
+      "step": 3713
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.93348623853211e-05,
+      "loss": 0.9538,
+      "step": 3714
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.926605504587156e-05,
+      "loss": 1.0952,
+      "step": 3715
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.919724770642202e-05,
+      "loss": 0.9965,
+      "step": 3716
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.912844036697247e-05,
+      "loss": 1.0483,
+      "step": 3717
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.905963302752293e-05,
+      "loss": 1.0054,
+      "step": 3718
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.899082568807339e-05,
+      "loss": 0.9281,
+      "step": 3719
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.892201834862385e-05,
+      "loss": 0.9692,
+      "step": 3720
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.885321100917431e-05,
+      "loss": 0.9048,
+      "step": 3721
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.878440366972477e-05,
+      "loss": 0.9481,
+      "step": 3722
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.871559633027522e-05,
+      "loss": 0.8283,
+      "step": 3723
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.864678899082568e-05,
+      "loss": 0.8775,
+      "step": 3724
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.857798165137613e-05,
+      "loss": 0.9424,
+      "step": 3725
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.850917431192659e-05,
+      "loss": 0.7042,
+      "step": 3726
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.844036697247705e-05,
+      "loss": 0.7994,
+      "step": 3727
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.837155963302751e-05,
+      "loss": 0.7188,
+      "step": 3728
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.830275229357797e-05,
+      "loss": 0.7178,
+      "step": 3729
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.823394495412844e-05,
+      "loss": 0.7984,
+      "step": 3730
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.81651376146789e-05,
+      "loss": 0.8928,
+      "step": 3731
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.809633027522936e-05,
+      "loss": 0.7836,
+      "step": 3732
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.80275229357798e-05,
+      "loss": 0.7521,
+      "step": 3733
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.795871559633027e-05,
+      "loss": 0.7392,
+      "step": 3734
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.788990825688073e-05,
+      "loss": 0.6345,
+      "step": 3735
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.782110091743119e-05,
+      "loss": 0.5659,
+      "step": 3736
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.775229357798165e-05,
+      "loss": 0.6035,
+      "step": 3737
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.76834862385321e-05,
+      "loss": 0.5845,
+      "step": 3738
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.761467889908256e-05,
+      "loss": 0.5404,
+      "step": 3739
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.754587155963302e-05,
+      "loss": 0.5461,
+      "step": 3740
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.747706422018347e-05,
+      "loss": 2.0229,
+      "step": 3741
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.740825688073393e-05,
+      "loss": 1.6285,
+      "step": 3742
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.733944954128439e-05,
+      "loss": 1.5521,
+      "step": 3743
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.727064220183485e-05,
+      "loss": 1.4716,
+      "step": 3744
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.720183486238531e-05,
+      "loss": 1.3636,
+      "step": 3745
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.713302752293578e-05,
+      "loss": 1.3513,
+      "step": 3746
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.706422018348624e-05,
+      "loss": 1.2689,
+      "step": 3747
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.69954128440367e-05,
+      "loss": 1.1857,
+      "step": 3748
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.692660550458715e-05,
+      "loss": 1.2468,
+      "step": 3749
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.685779816513761e-05,
+      "loss": 1.0738,
+      "step": 3750
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.678899082568807e-05,
+      "loss": 1.2023,
+      "step": 3751
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.672018348623853e-05,
+      "loss": 1.4627,
+      "step": 3752
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.665137614678899e-05,
+      "loss": 1.1457,
+      "step": 3753
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.658256880733944e-05,
+      "loss": 1.2178,
+      "step": 3754
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.65137614678899e-05,
+      "loss": 1.1336,
+      "step": 3755
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.644495412844036e-05,
+      "loss": 1.1109,
+      "step": 3756
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.637614678899081e-05,
+      "loss": 1.1721,
+      "step": 3757
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.630733944954127e-05,
+      "loss": 1.2285,
+      "step": 3758
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.623853211009173e-05,
+      "loss": 1.2159,
+      "step": 3759
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.616972477064219e-05,
+      "loss": 1.1024,
+      "step": 3760
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.610091743119265e-05,
+      "loss": 1.194,
+      "step": 3761
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.603211009174312e-05,
+      "loss": 1.027,
+      "step": 3762
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.596330275229358e-05,
+      "loss": 1.1311,
+      "step": 3763
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.589449541284404e-05,
+      "loss": 1.0079,
+      "step": 3764
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.582568807339449e-05,
+      "loss": 1.048,
+      "step": 3765
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.575688073394495e-05,
+      "loss": 0.8928,
+      "step": 3766
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.568807339449541e-05,
+      "loss": 0.9437,
+      "step": 3767
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.561926605504587e-05,
+      "loss": 0.9593,
+      "step": 3768
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.555045871559633e-05,
+      "loss": 0.885,
+      "step": 3769
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.548165137614678e-05,
+      "loss": 0.9193,
+      "step": 3770
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.541284403669724e-05,
+      "loss": 0.9879,
+      "step": 3771
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.53440366972477e-05,
+      "loss": 0.9309,
+      "step": 3772
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.527522935779815e-05,
+      "loss": 0.9125,
+      "step": 3773
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.520642201834861e-05,
+      "loss": 0.8467,
+      "step": 3774
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.513761467889907e-05,
+      "loss": 0.829,
+      "step": 3775
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.506880733944953e-05,
+      "loss": 0.8285,
+      "step": 3776
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.5e-05,
+      "loss": 0.818,
+      "step": 3777
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.493119266055045e-05,
+      "loss": 0.8252,
+      "step": 3778
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.486238532110092e-05,
+      "loss": 0.7033,
+      "step": 3779
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.479357798165136e-05,
+      "loss": 0.7164,
+      "step": 3780
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.472477064220182e-05,
+      "loss": 0.6598,
+      "step": 3781
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.465596330275229e-05,
+      "loss": 0.774,
+      "step": 3782
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.458715596330275e-05,
+      "loss": 0.7169,
+      "step": 3783
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.451834862385321e-05,
+      "loss": 0.6549,
+      "step": 3784
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.444954128440366e-05,
+      "loss": 0.6953,
+      "step": 3785
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.438073394495412e-05,
+      "loss": 0.5774,
+      "step": 3786
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.431192660550458e-05,
+      "loss": 0.5607,
+      "step": 3787
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.424311926605504e-05,
+      "loss": 0.5631,
+      "step": 3788
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.41743119266055e-05,
+      "loss": 0.6497,
+      "step": 3789
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.410550458715596e-05,
+      "loss": 0.7251,
+      "step": 3790
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.403669724770642e-05,
+      "loss": 2.0469,
+      "step": 3791
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.396788990825687e-05,
+      "loss": 1.5945,
+      "step": 3792
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.389908256880733e-05,
+      "loss": 1.5784,
+      "step": 3793
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.38302752293578e-05,
+      "loss": 1.2069,
+      "step": 3794
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.376146788990826e-05,
+      "loss": 1.3418,
+      "step": 3795
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.36926605504587e-05,
+      "loss": 1.3097,
+      "step": 3796
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.362385321100916e-05,
+      "loss": 1.3082,
+      "step": 3797
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.355504587155963e-05,
+      "loss": 1.302,
+      "step": 3798
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.348623853211009e-05,
+      "loss": 1.2243,
+      "step": 3799
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.341743119266055e-05,
+      "loss": 1.2325,
+      "step": 3800
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.3348623853211e-05,
+      "loss": 1.1625,
+      "step": 3801
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.327981651376146e-05,
+      "loss": 1.1826,
+      "step": 3802
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.321100917431192e-05,
+      "loss": 1.2711,
+      "step": 3803
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.314220183486238e-05,
+      "loss": 1.1629,
+      "step": 3804
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.307339449541284e-05,
+      "loss": 1.1474,
+      "step": 3805
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.30045871559633e-05,
+      "loss": 1.1057,
+      "step": 3806
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.293577981651376e-05,
+      "loss": 1.1821,
+      "step": 3807
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.286697247706421e-05,
+      "loss": 1.1355,
+      "step": 3808
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.279816513761467e-05,
+      "loss": 1.0072,
+      "step": 3809
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.272935779816513e-05,
+      "loss": 1.0652,
+      "step": 3810
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.26605504587156e-05,
+      "loss": 1.0744,
+      "step": 3811
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.259174311926604e-05,
+      "loss": 1.0275,
+      "step": 3812
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.25229357798165e-05,
+      "loss": 1.1359,
+      "step": 3813
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.245412844036697e-05,
+      "loss": 1.0698,
+      "step": 3814
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.238532110091743e-05,
+      "loss": 0.9387,
+      "step": 3815
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.231651376146789e-05,
+      "loss": 1.0375,
+      "step": 3816
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.224770642201834e-05,
+      "loss": 0.9125,
+      "step": 3817
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.21788990825688e-05,
+      "loss": 0.9368,
+      "step": 3818
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.211009174311926e-05,
+      "loss": 0.9507,
+      "step": 3819
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.204128440366972e-05,
+      "loss": 0.8843,
+      "step": 3820
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.197247706422018e-05,
+      "loss": 0.9185,
+      "step": 3821
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.190366972477064e-05,
+      "loss": 0.8446,
+      "step": 3822
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.18348623853211e-05,
+      "loss": 0.9292,
+      "step": 3823
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.176605504587155e-05,
+      "loss": 0.7549,
+      "step": 3824
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.169724770642201e-05,
+      "loss": 0.7597,
+      "step": 3825
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.162844036697247e-05,
+      "loss": 0.8596,
+      "step": 3826
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.155963302752293e-05,
+      "loss": 0.7265,
+      "step": 3827
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.149082568807338e-05,
+      "loss": 0.6695,
+      "step": 3828
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.142201834862384e-05,
+      "loss": 0.7114,
+      "step": 3829
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.13532110091743e-05,
+      "loss": 0.7298,
+      "step": 3830
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.128440366972477e-05,
+      "loss": 0.6657,
+      "step": 3831
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.121559633027521e-05,
+      "loss": 0.6855,
+      "step": 3832
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.114678899082568e-05,
+      "loss": 0.6794,
+      "step": 3833
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.107798165137614e-05,
+      "loss": 0.6444,
+      "step": 3834
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.10091743119266e-05,
+      "loss": 0.6803,
+      "step": 3835
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.094036697247706e-05,
+      "loss": 0.6631,
+      "step": 3836
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.087155963302752e-05,
+      "loss": 0.6324,
+      "step": 3837
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.080275229357798e-05,
+      "loss": 0.6049,
+      "step": 3838
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.073394495412844e-05,
+      "loss": 0.5095,
+      "step": 3839
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.066513761467889e-05,
+      "loss": 0.6426,
+      "step": 3840
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.059633027522935e-05,
+      "loss": 1.5919,
+      "step": 3841
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.052752293577981e-05,
+      "loss": 1.5759,
+      "step": 3842
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.045871559633027e-05,
+      "loss": 1.5045,
+      "step": 3843
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.038990825688072e-05,
+      "loss": 1.4559,
+      "step": 3844
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.032110091743118e-05,
+      "loss": 1.2657,
+      "step": 3845
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.025229357798164e-05,
+      "loss": 1.318,
+      "step": 3846
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.01834862385321e-05,
+      "loss": 1.2932,
+      "step": 3847
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.011467889908255e-05,
+      "loss": 1.2322,
+      "step": 3848
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.004587155963302e-05,
+      "loss": 1.3297,
+      "step": 3849
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.997706422018348e-05,
+      "loss": 1.1616,
+      "step": 3850
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.990825688073394e-05,
+      "loss": 1.1907,
+      "step": 3851
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.98394495412844e-05,
+      "loss": 1.2661,
+      "step": 3852
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.977064220183486e-05,
+      "loss": 1.1496,
+      "step": 3853
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.970183486238532e-05,
+      "loss": 1.1133,
+      "step": 3854
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.963302752293578e-05,
+      "loss": 1.1594,
+      "step": 3855
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.956422018348623e-05,
+      "loss": 1.14,
+      "step": 3856
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.949541284403669e-05,
+      "loss": 1.1135,
+      "step": 3857
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.942660550458715e-05,
+      "loss": 1.0633,
+      "step": 3858
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.935779816513761e-05,
+      "loss": 1.2905,
+      "step": 3859
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.928899082568806e-05,
+      "loss": 1.0853,
+      "step": 3860
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.922018348623852e-05,
+      "loss": 0.9948,
+      "step": 3861
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.915137614678898e-05,
+      "loss": 0.927,
+      "step": 3862
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.908256880733945e-05,
+      "loss": 0.9915,
+      "step": 3863
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.90137614678899e-05,
+      "loss": 0.927,
+      "step": 3864
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.894495412844035e-05,
+      "loss": 1.1382,
+      "step": 3865
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.887614678899082e-05,
+      "loss": 1.1153,
+      "step": 3866
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.880733944954128e-05,
+      "loss": 1.0171,
+      "step": 3867
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.873853211009174e-05,
+      "loss": 0.9905,
+      "step": 3868
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.86697247706422e-05,
+      "loss": 0.9552,
+      "step": 3869
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.860091743119266e-05,
+      "loss": 0.9987,
+      "step": 3870
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.853211009174312e-05,
+      "loss": 0.8147,
+      "step": 3871
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.846330275229357e-05,
+      "loss": 0.8816,
+      "step": 3872
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.839449541284403e-05,
+      "loss": 0.9317,
+      "step": 3873
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.832568807339449e-05,
+      "loss": 0.9196,
+      "step": 3874
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.825688073394495e-05,
+      "loss": 0.7313,
+      "step": 3875
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.81880733944954e-05,
+      "loss": 0.8846,
+      "step": 3876
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.811926605504586e-05,
+      "loss": 0.8747,
+      "step": 3877
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.805045871559632e-05,
+      "loss": 0.7789,
+      "step": 3878
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.798165137614679e-05,
+      "loss": 0.7371,
+      "step": 3879
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.791284403669723e-05,
+      "loss": 0.8152,
+      "step": 3880
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.78440366972477e-05,
+      "loss": 0.7001,
+      "step": 3881
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.777522935779816e-05,
+      "loss": 0.7366,
+      "step": 3882
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.770642201834862e-05,
+      "loss": 0.7287,
+      "step": 3883
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.763761467889908e-05,
+      "loss": 0.6601,
+      "step": 3884
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.756880733944954e-05,
+      "loss": 0.7124,
+      "step": 3885
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.75e-05,
+      "loss": 0.659,
+      "step": 3886
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.743119266055046e-05,
+      "loss": 0.567,
+      "step": 3887
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.736238532110091e-05,
+      "loss": 0.5553,
+      "step": 3888
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.729357798165137e-05,
+      "loss": 0.522,
+      "step": 3889
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.722477064220183e-05,
+      "loss": 0.5788,
+      "step": 3890
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.71559633027523e-05,
+      "loss": 1.8061,
+      "step": 3891
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.708715596330274e-05,
+      "loss": 1.462,
+      "step": 3892
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.70183486238532e-05,
+      "loss": 1.3906,
+      "step": 3893
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.694954128440366e-05,
+      "loss": 1.3512,
+      "step": 3894
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.688073394495413e-05,
+      "loss": 1.3055,
+      "step": 3895
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.681192660550457e-05,
+      "loss": 1.1685,
+      "step": 3896
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.674311926605503e-05,
+      "loss": 1.2855,
+      "step": 3897
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.66743119266055e-05,
+      "loss": 1.3616,
+      "step": 3898
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.660550458715596e-05,
+      "loss": 1.3278,
+      "step": 3899
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.653669724770642e-05,
+      "loss": 1.2772,
+      "step": 3900
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.646788990825688e-05,
+      "loss": 1.1643,
+      "step": 3901
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.639908256880734e-05,
+      "loss": 1.1848,
+      "step": 3902
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.63302752293578e-05,
+      "loss": 1.179,
+      "step": 3903
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.626146788990825e-05,
+      "loss": 1.0882,
+      "step": 3904
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.619266055045871e-05,
+      "loss": 1.197,
+      "step": 3905
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.612385321100917e-05,
+      "loss": 1.0708,
+      "step": 3906
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.605504587155963e-05,
+      "loss": 1.1216,
+      "step": 3907
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.598623853211008e-05,
+      "loss": 1.1411,
+      "step": 3908
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.591743119266054e-05,
+      "loss": 1.1111,
+      "step": 3909
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.5848623853211e-05,
+      "loss": 1.1304,
+      "step": 3910
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.577981651376146e-05,
+      "loss": 1.0557,
+      "step": 3911
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.571100917431191e-05,
+      "loss": 1.0124,
+      "step": 3912
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.564220183486237e-05,
+      "loss": 1.0451,
+      "step": 3913
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.557339449541284e-05,
+      "loss": 0.9642,
+      "step": 3914
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.55045871559633e-05,
+      "loss": 1.0452,
+      "step": 3915
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.543577981651376e-05,
+      "loss": 0.8989,
+      "step": 3916
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.536697247706422e-05,
+      "loss": 1.001,
+      "step": 3917
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.529816513761468e-05,
+      "loss": 1.017,
+      "step": 3918
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.522935779816514e-05,
+      "loss": 0.86,
+      "step": 3919
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.516055045871559e-05,
+      "loss": 0.9105,
+      "step": 3920
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.509174311926605e-05,
+      "loss": 0.9058,
+      "step": 3921
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.502293577981651e-05,
+      "loss": 0.7566,
+      "step": 3922
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.495412844036697e-05,
+      "loss": 0.8233,
+      "step": 3923
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.488532110091742e-05,
+      "loss": 0.8129,
+      "step": 3924
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.481651376146788e-05,
+      "loss": 0.6511,
+      "step": 3925
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.474770642201834e-05,
+      "loss": 0.7709,
+      "step": 3926
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.46788990825688e-05,
+      "loss": 0.7738,
+      "step": 3927
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.461009174311925e-05,
+      "loss": 0.7771,
+      "step": 3928
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.454128440366971e-05,
+      "loss": 0.7731,
+      "step": 3929
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.447247706422017e-05,
+      "loss": 0.6962,
+      "step": 3930
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.440366972477064e-05,
+      "loss": 0.6278,
+      "step": 3931
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.43348623853211e-05,
+      "loss": 0.619,
+      "step": 3932
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.426605504587156e-05,
+      "loss": 0.6599,
+      "step": 3933
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.419724770642202e-05,
+      "loss": 0.6381,
+      "step": 3934
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.412844036697248e-05,
+      "loss": 0.6831,
+      "step": 3935
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.405963302752293e-05,
+      "loss": 0.6431,
+      "step": 3936
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.399082568807339e-05,
+      "loss": 0.4903,
+      "step": 3937
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.392201834862385e-05,
+      "loss": 0.5067,
+      "step": 3938
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.385321100917431e-05,
+      "loss": 0.5121,
+      "step": 3939
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.378440366972476e-05,
+      "loss": 0.5709,
+      "step": 3940
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.371559633027522e-05,
+      "loss": 1.9488,
+      "step": 3941
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.364678899082568e-05,
+      "loss": 1.7386,
+      "step": 3942
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.357798165137614e-05,
+      "loss": 1.5199,
+      "step": 3943
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.350917431192659e-05,
+      "loss": 1.442,
+      "step": 3944
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.344036697247705e-05,
+      "loss": 1.3376,
+      "step": 3945
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.337155963302751e-05,
+      "loss": 1.376,
+      "step": 3946
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.330275229357798e-05,
+      "loss": 1.2876,
+      "step": 3947
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.323394495412844e-05,
+      "loss": 1.1852,
+      "step": 3948
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.31651376146789e-05,
+      "loss": 1.4057,
+      "step": 3949
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.309633027522936e-05,
+      "loss": 1.2245,
+      "step": 3950
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.302752293577982e-05,
+      "loss": 1.0874,
+      "step": 3951
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.295871559633027e-05,
+      "loss": 1.185,
+      "step": 3952
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.288990825688073e-05,
+      "loss": 1.1653,
+      "step": 3953
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.282110091743119e-05,
+      "loss": 1.1425,
+      "step": 3954
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.275229357798165e-05,
+      "loss": 1.1903,
+      "step": 3955
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.26834862385321e-05,
+      "loss": 1.094,
+      "step": 3956
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.261467889908256e-05,
+      "loss": 1.1175,
+      "step": 3957
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.254587155963302e-05,
+      "loss": 0.9874,
+      "step": 3958
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.247706422018348e-05,
+      "loss": 1.056,
+      "step": 3959
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.240825688073393e-05,
+      "loss": 1.0672,
+      "step": 3960
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.233944954128439e-05,
+      "loss": 1.1297,
+      "step": 3961
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.227064220183485e-05,
+      "loss": 1.0381,
+      "step": 3962
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.220183486238532e-05,
+      "loss": 0.9999,
+      "step": 3963
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.213302752293578e-05,
+      "loss": 0.9969,
+      "step": 3964
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.206422018348624e-05,
+      "loss": 1.1175,
+      "step": 3965
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.19954128440367e-05,
+      "loss": 1.0107,
+      "step": 3966
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.192660550458716e-05,
+      "loss": 1.0708,
+      "step": 3967
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.185779816513761e-05,
+      "loss": 1.0417,
+      "step": 3968
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.178899082568807e-05,
+      "loss": 0.9759,
+      "step": 3969
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.172018348623853e-05,
+      "loss": 0.9711,
+      "step": 3970
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.165137614678899e-05,
+      "loss": 0.9798,
+      "step": 3971
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.158256880733944e-05,
+      "loss": 0.9775,
+      "step": 3972
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.15137614678899e-05,
+      "loss": 0.8265,
+      "step": 3973
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.144495412844036e-05,
+      "loss": 0.7628,
+      "step": 3974
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.137614678899082e-05,
+      "loss": 0.8038,
+      "step": 3975
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.130733944954127e-05,
+      "loss": 0.7388,
+      "step": 3976
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.123853211009173e-05,
+      "loss": 0.8612,
+      "step": 3977
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.11697247706422e-05,
+      "loss": 0.7079,
+      "step": 3978
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.110091743119266e-05,
+      "loss": 0.7777,
+      "step": 3979
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.103211009174311e-05,
+      "loss": 0.7212,
+      "step": 3980
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.096330275229357e-05,
+      "loss": 0.6824,
+      "step": 3981
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.089449541284403e-05,
+      "loss": 0.6714,
+      "step": 3982
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.082568807339449e-05,
+      "loss": 0.6988,
+      "step": 3983
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.075688073394495e-05,
+      "loss": 0.611,
+      "step": 3984
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.068807339449541e-05,
+      "loss": 0.5702,
+      "step": 3985
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.061926605504587e-05,
+      "loss": 0.5123,
+      "step": 3986
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.055045871559633e-05,
+      "loss": 0.584,
+      "step": 3987
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.048165137614678e-05,
+      "loss": 0.5251,
+      "step": 3988
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.041284403669724e-05,
+      "loss": 0.5232,
+      "step": 3989
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.03440366972477e-05,
+      "loss": 0.6507,
+      "step": 3990
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.027522935779816e-05,
+      "loss": 1.8332,
+      "step": 3991
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.020642201834862e-05,
+      "loss": 1.6924,
+      "step": 3992
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.013761467889908e-05,
+      "loss": 1.4632,
+      "step": 3993
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.006880733944954e-05,
+      "loss": 1.3165,
+      "step": 3994
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 1.4082,
+      "step": 3995
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.993119266055045e-05,
+      "loss": 1.3622,
+      "step": 3996
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.986238532110091e-05,
+      "loss": 1.2599,
+      "step": 3997
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.979357798165137e-05,
+      "loss": 1.2599,
+      "step": 3998
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.972477064220183e-05,
+      "loss": 1.3481,
+      "step": 3999
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.965596330275229e-05,
+      "loss": 1.2244,
+      "step": 4000
+    },
+    {
+      "epoch": 2.47,
+      "eval_bleu": 3.550386321342364e-14,
+      "eval_loss": 1.794046401977539,
+      "eval_runtime": 2535.6975,
+      "eval_samples_per_second": 5.821,
+      "eval_steps_per_second": 0.728,
+      "step": 4000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.958715596330275e-05,
+      "loss": 1.2282,
+      "step": 4001
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.951834862385321e-05,
+      "loss": 1.2254,
+      "step": 4002
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.9449541284403665e-05,
+      "loss": 1.2155,
+      "step": 4003
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.938073394495412e-05,
+      "loss": 1.2088,
+      "step": 4004
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.931192660550458e-05,
+      "loss": 1.1898,
+      "step": 4005
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.924311926605504e-05,
+      "loss": 1.1753,
+      "step": 4006
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.91743119266055e-05,
+      "loss": 1.1453,
+      "step": 4007
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.910550458715596e-05,
+      "loss": 1.1096,
+      "step": 4008
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.903669724770642e-05,
+      "loss": 1.0013,
+      "step": 4009
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.896788990825688e-05,
+      "loss": 1.1856,
+      "step": 4010
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.8899082568807334e-05,
+      "loss": 1.0922,
+      "step": 4011
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.883027522935779e-05,
+      "loss": 0.9892,
+      "step": 4012
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.876146788990825e-05,
+      "loss": 1.0373,
+      "step": 4013
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.869266055045871e-05,
+      "loss": 1.0258,
+      "step": 4014
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.862385321100917e-05,
+      "loss": 1.0715,
+      "step": 4015
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.855504587155963e-05,
+      "loss": 0.9778,
+      "step": 4016
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.848623853211009e-05,
+      "loss": 0.8886,
+      "step": 4017
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.841743119266055e-05,
+      "loss": 1.0301,
+      "step": 4018
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.8348623853211004e-05,
+      "loss": 0.9745,
+      "step": 4019
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.827981651376146e-05,
+      "loss": 0.8738,
+      "step": 4020
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.821100917431192e-05,
+      "loss": 0.8567,
+      "step": 4021
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.814220183486238e-05,
+      "loss": 0.8669,
+      "step": 4022
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.807339449541284e-05,
+      "loss": 0.8761,
+      "step": 4023
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.80045871559633e-05,
+      "loss": 0.817,
+      "step": 4024
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.793577981651376e-05,
+      "loss": 0.7875,
+      "step": 4025
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.786697247706422e-05,
+      "loss": 0.8503,
+      "step": 4026
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.7798165137614674e-05,
+      "loss": 0.8447,
+      "step": 4027
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.772935779816513e-05,
+      "loss": 0.7419,
+      "step": 4028
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.766055045871559e-05,
+      "loss": 0.6784,
+      "step": 4029
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.759174311926605e-05,
+      "loss": 0.6791,
+      "step": 4030
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.752293577981651e-05,
+      "loss": 0.6456,
+      "step": 4031
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.745412844036697e-05,
+      "loss": 0.7289,
+      "step": 4032
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.738532110091743e-05,
+      "loss": 0.7594,
+      "step": 4033
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.731651376146788e-05,
+      "loss": 0.5625,
+      "step": 4034
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.7247706422018344e-05,
+      "loss": 0.6974,
+      "step": 4035
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.71788990825688e-05,
+      "loss": 0.5923,
+      "step": 4036
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.711009174311926e-05,
+      "loss": 0.6235,
+      "step": 4037
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.704128440366972e-05,
+      "loss": 0.5517,
+      "step": 4038
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.697247706422018e-05,
+      "loss": 0.5743,
+      "step": 4039
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.690366972477064e-05,
+      "loss": 0.5629,
+      "step": 4040
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.68348623853211e-05,
+      "loss": 1.781,
+      "step": 4041
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.676605504587155e-05,
+      "loss": 1.5586,
+      "step": 4042
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.6697247706422014e-05,
+      "loss": 1.1529,
+      "step": 4043
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.662844036697247e-05,
+      "loss": 1.323,
+      "step": 4044
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.655963302752293e-05,
+      "loss": 1.2483,
+      "step": 4045
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.649082568807339e-05,
+      "loss": 1.3055,
+      "step": 4046
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.642201834862385e-05,
+      "loss": 1.2784,
+      "step": 4047
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.6353211009174307e-05,
+      "loss": 1.189,
+      "step": 4048
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.628440366972477e-05,
+      "loss": 1.231,
+      "step": 4049
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.621559633027522e-05,
+      "loss": 1.2197,
+      "step": 4050
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.6146788990825684e-05,
+      "loss": 1.1427,
+      "step": 4051
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.607798165137614e-05,
+      "loss": 1.0593,
+      "step": 4052
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.60091743119266e-05,
+      "loss": 1.0903,
+      "step": 4053
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.594036697247706e-05,
+      "loss": 1.0469,
+      "step": 4054
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.587155963302752e-05,
+      "loss": 1.1623,
+      "step": 4055
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.5802752293577976e-05,
+      "loss": 1.0757,
+      "step": 4056
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.573394495412844e-05,
+      "loss": 1.1588,
+      "step": 4057
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.566513761467889e-05,
+      "loss": 1.127,
+      "step": 4058
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.5596330275229353e-05,
+      "loss": 0.9677,
+      "step": 4059
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.552752293577981e-05,
+      "loss": 1.0443,
+      "step": 4060
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.545871559633027e-05,
+      "loss": 1.0866,
+      "step": 4061
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.538990825688073e-05,
+      "loss": 1.1641,
+      "step": 4062
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.532110091743119e-05,
+      "loss": 1.075,
+      "step": 4063
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.5252293577981646e-05,
+      "loss": 1.0053,
+      "step": 4064
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.518348623853211e-05,
+      "loss": 0.8862,
+      "step": 4065
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.511467889908256e-05,
+      "loss": 0.9135,
+      "step": 4066
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.504587155963302e-05,
+      "loss": 0.922,
+      "step": 4067
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.497706422018348e-05,
+      "loss": 0.8063,
+      "step": 4068
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.490825688073394e-05,
+      "loss": 0.9106,
+      "step": 4069
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.48394495412844e-05,
+      "loss": 0.829,
+      "step": 4070
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.477064220183486e-05,
+      "loss": 1.0013,
+      "step": 4071
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.4701834862385316e-05,
+      "loss": 0.8004,
+      "step": 4072
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.463302752293577e-05,
+      "loss": 0.8024,
+      "step": 4073
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.456422018348623e-05,
+      "loss": 0.7838,
+      "step": 4074
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.449541284403669e-05,
+      "loss": 0.8898,
+      "step": 4075
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.442660550458715e-05,
+      "loss": 0.7599,
+      "step": 4076
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.435779816513761e-05,
+      "loss": 0.7727,
+      "step": 4077
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.428899082568807e-05,
+      "loss": 0.7514,
+      "step": 4078
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.422018348623853e-05,
+      "loss": 0.7568,
+      "step": 4079
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.4151376146788986e-05,
+      "loss": 0.6024,
+      "step": 4080
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.408256880733944e-05,
+      "loss": 0.5901,
+      "step": 4081
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.40137614678899e-05,
+      "loss": 0.6672,
+      "step": 4082
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.394495412844036e-05,
+      "loss": 0.7881,
+      "step": 4083
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.387614678899082e-05,
+      "loss": 0.5734,
+      "step": 4084
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.380733944954128e-05,
+      "loss": 0.6074,
+      "step": 4085
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.373853211009174e-05,
+      "loss": 0.5452,
+      "step": 4086
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.36697247706422e-05,
+      "loss": 0.5534,
+      "step": 4087
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.3600917431192656e-05,
+      "loss": 0.571,
+      "step": 4088
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.353211009174311e-05,
+      "loss": 0.4237,
+      "step": 4089
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.346330275229357e-05,
+      "loss": 0.5453,
+      "step": 4090
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.339449541284403e-05,
+      "loss": 1.7752,
+      "step": 4091
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.332568807339449e-05,
+      "loss": 1.5751,
+      "step": 4092
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.325688073394495e-05,
+      "loss": 1.3436,
+      "step": 4093
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.318807339449541e-05,
+      "loss": 1.3601,
+      "step": 4094
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.311926605504587e-05,
+      "loss": 1.4734,
+      "step": 4095
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.3050458715596326e-05,
+      "loss": 1.3058,
+      "step": 4096
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.298165137614678e-05,
+      "loss": 1.2182,
+      "step": 4097
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.291284403669724e-05,
+      "loss": 1.2523,
+      "step": 4098
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.28440366972477e-05,
+      "loss": 1.3629,
+      "step": 4099
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.277522935779816e-05,
+      "loss": 1.3259,
+      "step": 4100
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.270642201834862e-05,
+      "loss": 1.3156,
+      "step": 4101
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.263761467889908e-05,
+      "loss": 1.1659,
+      "step": 4102
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.256880733944954e-05,
+      "loss": 1.17,
+      "step": 4103
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.2499999999999995e-05,
+      "loss": 0.9798,
+      "step": 4104
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.243119266055045e-05,
+      "loss": 1.2055,
+      "step": 4105
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.236238532110091e-05,
+      "loss": 1.1754,
+      "step": 4106
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.229357798165137e-05,
+      "loss": 1.0331,
+      "step": 4107
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.222477064220183e-05,
+      "loss": 1.0595,
+      "step": 4108
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.215596330275229e-05,
+      "loss": 1.0183,
+      "step": 4109
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.208715596330275e-05,
+      "loss": 1.0931,
+      "step": 4110
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.201834862385321e-05,
+      "loss": 1.0373,
+      "step": 4111
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.194954128440366e-05,
+      "loss": 0.924,
+      "step": 4112
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.188073394495412e-05,
+      "loss": 0.9767,
+      "step": 4113
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.181192660550458e-05,
+      "loss": 0.9825,
+      "step": 4114
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.174311926605504e-05,
+      "loss": 1.0085,
+      "step": 4115
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.16743119266055e-05,
+      "loss": 0.8032,
+      "step": 4116
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.160550458715596e-05,
+      "loss": 0.9641,
+      "step": 4117
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.153669724770642e-05,
+      "loss": 0.918,
+      "step": 4118
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.146788990825688e-05,
+      "loss": 0.8796,
+      "step": 4119
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.139908256880733e-05,
+      "loss": 0.8951,
+      "step": 4120
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.133027522935779e-05,
+      "loss": 0.8549,
+      "step": 4121
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.126146788990825e-05,
+      "loss": 0.8465,
+      "step": 4122
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.119266055045871e-05,
+      "loss": 0.914,
+      "step": 4123
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.1123853211009167e-05,
+      "loss": 0.7472,
+      "step": 4124
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.105504587155963e-05,
+      "loss": 0.7595,
+      "step": 4125
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.098623853211009e-05,
+      "loss": 0.8311,
+      "step": 4126
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.091743119266055e-05,
+      "loss": 0.7757,
+      "step": 4127
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.0848623853211e-05,
+      "loss": 0.8456,
+      "step": 4128
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.077981651376146e-05,
+      "loss": 0.7343,
+      "step": 4129
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.071100917431192e-05,
+      "loss": 0.7085,
+      "step": 4130
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.064220183486238e-05,
+      "loss": 0.7026,
+      "step": 4131
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.0573394495412836e-05,
+      "loss": 0.6747,
+      "step": 4132
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.05045871559633e-05,
+      "loss": 0.6524,
+      "step": 4133
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.043577981651376e-05,
+      "loss": 0.6328,
+      "step": 4134
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.036697247706422e-05,
+      "loss": 0.5689,
+      "step": 4135
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.029816513761467e-05,
+      "loss": 0.5405,
+      "step": 4136
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.022935779816513e-05,
+      "loss": 0.5766,
+      "step": 4137
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.016055045871559e-05,
+      "loss": 0.6067,
+      "step": 4138
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.009174311926605e-05,
+      "loss": 0.4424,
+      "step": 4139
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.0022935779816506e-05,
+      "loss": 0.5854,
+      "step": 4140
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.995412844036697e-05,
+      "loss": 1.7243,
+      "step": 4141
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.988532110091743e-05,
+      "loss": 1.5123,
+      "step": 4142
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.981651376146789e-05,
+      "loss": 1.4056,
+      "step": 4143
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.974770642201834e-05,
+      "loss": 1.3029,
+      "step": 4144
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.96788990825688e-05,
+      "loss": 1.2879,
+      "step": 4145
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.961009174311926e-05,
+      "loss": 1.1951,
+      "step": 4146
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.954128440366972e-05,
+      "loss": 1.2032,
+      "step": 4147
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.9472477064220176e-05,
+      "loss": 1.1617,
+      "step": 4148
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.940366972477064e-05,
+      "loss": 1.2706,
+      "step": 4149
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.93348623853211e-05,
+      "loss": 1.1968,
+      "step": 4150
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.926605504587156e-05,
+      "loss": 1.1593,
+      "step": 4151
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.919724770642201e-05,
+      "loss": 1.1046,
+      "step": 4152
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.912844036697247e-05,
+      "loss": 1.0335,
+      "step": 4153
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.905963302752293e-05,
+      "loss": 0.9551,
+      "step": 4154
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.899082568807339e-05,
+      "loss": 1.0888,
+      "step": 4155
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.8922018348623846e-05,
+      "loss": 1.1191,
+      "step": 4156
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.885321100917431e-05,
+      "loss": 1.0991,
+      "step": 4157
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.878440366972477e-05,
+      "loss": 1.206,
+      "step": 4158
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.871559633027523e-05,
+      "loss": 1.1113,
+      "step": 4159
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.864678899082568e-05,
+      "loss": 1.0501,
+      "step": 4160
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.857798165137614e-05,
+      "loss": 0.9874,
+      "step": 4161
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.85091743119266e-05,
+      "loss": 1.1126,
+      "step": 4162
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.844036697247706e-05,
+      "loss": 1.0347,
+      "step": 4163
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.8371559633027516e-05,
+      "loss": 0.9851,
+      "step": 4164
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.830275229357798e-05,
+      "loss": 0.9094,
+      "step": 4165
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.823394495412844e-05,
+      "loss": 0.9265,
+      "step": 4166
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.81651376146789e-05,
+      "loss": 0.9256,
+      "step": 4167
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.809633027522935e-05,
+      "loss": 1.0236,
+      "step": 4168
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.802752293577981e-05,
+      "loss": 0.8734,
+      "step": 4169
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.795871559633027e-05,
+      "loss": 0.8474,
+      "step": 4170
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.788990825688073e-05,
+      "loss": 0.91,
+      "step": 4171
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.7821100917431186e-05,
+      "loss": 0.869,
+      "step": 4172
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.775229357798165e-05,
+      "loss": 0.8564,
+      "step": 4173
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.768348623853211e-05,
+      "loss": 0.8149,
+      "step": 4174
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.761467889908257e-05,
+      "loss": 0.7767,
+      "step": 4175
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.754587155963302e-05,
+      "loss": 0.8687,
+      "step": 4176
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.747706422018348e-05,
+      "loss": 0.6592,
+      "step": 4177
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.740825688073394e-05,
+      "loss": 0.7391,
+      "step": 4178
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.73394495412844e-05,
+      "loss": 0.7144,
+      "step": 4179
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.7270642201834855e-05,
+      "loss": 0.6974,
+      "step": 4180
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.720183486238532e-05,
+      "loss": 0.7498,
+      "step": 4181
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.713302752293578e-05,
+      "loss": 0.7182,
+      "step": 4182
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.706422018348624e-05,
+      "loss": 0.6398,
+      "step": 4183
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.699541284403669e-05,
+      "loss": 0.618,
+      "step": 4184
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.692660550458715e-05,
+      "loss": 0.4744,
+      "step": 4185
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.685779816513761e-05,
+      "loss": 0.5971,
+      "step": 4186
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.678899082568807e-05,
+      "loss": 0.5964,
+      "step": 4187
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.6720183486238525e-05,
+      "loss": 0.5752,
+      "step": 4188
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.6651376146788987e-05,
+      "loss": 0.5247,
+      "step": 4189
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.658256880733945e-05,
+      "loss": 0.6126,
+      "step": 4190
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.651376146788991e-05,
+      "loss": 1.6758,
+      "step": 4191
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.644495412844036e-05,
+      "loss": 1.4452,
+      "step": 4192
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.637614678899082e-05,
+      "loss": 1.4002,
+      "step": 4193
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.630733944954128e-05,
+      "loss": 1.3457,
+      "step": 4194
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.623853211009174e-05,
+      "loss": 1.3434,
+      "step": 4195
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.6169724770642195e-05,
+      "loss": 1.2629,
+      "step": 4196
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.6100917431192656e-05,
+      "loss": 1.3153,
+      "step": 4197
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.603211009174312e-05,
+      "loss": 1.2419,
+      "step": 4198
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.596330275229358e-05,
+      "loss": 1.2093,
+      "step": 4199
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.589449541284403e-05,
+      "loss": 1.1671,
+      "step": 4200
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.582568807339449e-05,
+      "loss": 1.1328,
+      "step": 4201
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.575688073394495e-05,
+      "loss": 1.0832,
+      "step": 4202
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.568807339449541e-05,
+      "loss": 1.0791,
+      "step": 4203
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5619266055045865e-05,
+      "loss": 0.9577,
+      "step": 4204
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5550458715596326e-05,
+      "loss": 1.0808,
+      "step": 4205
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.548165137614679e-05,
+      "loss": 1.1195,
+      "step": 4206
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.541284403669725e-05,
+      "loss": 0.9716,
+      "step": 4207
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5344036697247696e-05,
+      "loss": 1.0299,
+      "step": 4208
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.527522935779816e-05,
+      "loss": 0.9451,
+      "step": 4209
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.520642201834862e-05,
+      "loss": 0.8694,
+      "step": 4210
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.513761467889908e-05,
+      "loss": 1.0465,
+      "step": 4211
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5068807339449535e-05,
+      "loss": 1.0414,
+      "step": 4212
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 0.9537,
+      "step": 4213
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.493119266055046e-05,
+      "loss": 0.9595,
+      "step": 4214
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.486238532110092e-05,
+      "loss": 0.8366,
+      "step": 4215
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.4793577981651366e-05,
+      "loss": 0.8978,
+      "step": 4216
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.472477064220183e-05,
+      "loss": 0.9412,
+      "step": 4217
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.465596330275229e-05,
+      "loss": 1.0137,
+      "step": 4218
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.458715596330275e-05,
+      "loss": 0.8054,
+      "step": 4219
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.4518348623853205e-05,
+      "loss": 0.9263,
+      "step": 4220
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.4449541284403666e-05,
+      "loss": 0.7642,
+      "step": 4221
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.438073394495413e-05,
+      "loss": 0.7591,
+      "step": 4222
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.431192660550459e-05,
+      "loss": 0.8161,
+      "step": 4223
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.4243119266055036e-05,
+      "loss": 0.7657,
+      "step": 4224
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.41743119266055e-05,
+      "loss": 0.9252,
+      "step": 4225
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.410550458715596e-05,
+      "loss": 0.725,
+      "step": 4226
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.403669724770642e-05,
+      "loss": 0.6823,
+      "step": 4227
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.3967889908256874e-05,
+      "loss": 0.7195,
+      "step": 4228
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.3899082568807336e-05,
+      "loss": 0.6832,
+      "step": 4229
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.38302752293578e-05,
+      "loss": 0.7753,
+      "step": 4230
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.376146788990826e-05,
+      "loss": 0.6637,
+      "step": 4231
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.3692660550458706e-05,
+      "loss": 0.6432,
+      "step": 4232
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.362385321100917e-05,
+      "loss": 0.6546,
+      "step": 4233
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.355504587155963e-05,
+      "loss": 0.5591,
+      "step": 4234
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.348623853211009e-05,
+      "loss": 0.477,
+      "step": 4235
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.3417431192660544e-05,
+      "loss": 0.5643,
+      "step": 4236
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.3348623853211006e-05,
+      "loss": 0.5023,
+      "step": 4237
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.327981651376147e-05,
+      "loss": 0.4696,
+      "step": 4238
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.321100917431193e-05,
+      "loss": 0.4686,
+      "step": 4239
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.3142201834862376e-05,
+      "loss": 0.5683,
+      "step": 4240
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.307339449541284e-05,
+      "loss": 1.6947,
+      "step": 4241
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.30045871559633e-05,
+      "loss": 1.579,
+      "step": 4242
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.293577981651376e-05,
+      "loss": 1.3698,
+      "step": 4243
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.2866972477064214e-05,
+      "loss": 1.2538,
+      "step": 4244
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.2798165137614675e-05,
+      "loss": 1.2957,
+      "step": 4245
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.272935779816514e-05,
+      "loss": 1.3994,
+      "step": 4246
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.26605504587156e-05,
+      "loss": 1.1448,
+      "step": 4247
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.2591743119266046e-05,
+      "loss": 1.1225,
+      "step": 4248
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.252293577981651e-05,
+      "loss": 1.1653,
+      "step": 4249
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.245412844036697e-05,
+      "loss": 1.2382,
+      "step": 4250
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.238532110091743e-05,
+      "loss": 1.1891,
+      "step": 4251
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.2316513761467884e-05,
+      "loss": 1.1272,
+      "step": 4252
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.2247706422018345e-05,
+      "loss": 1.0963,
+      "step": 4253
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.2178899082568807e-05,
+      "loss": 1.1827,
+      "step": 4254
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.211009174311927e-05,
+      "loss": 1.0944,
+      "step": 4255
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.2041284403669715e-05,
+      "loss": 0.9591,
+      "step": 4256
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.197247706422018e-05,
+      "loss": 1.0556,
+      "step": 4257
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.190366972477064e-05,
+      "loss": 1.154,
+      "step": 4258
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.18348623853211e-05,
+      "loss": 0.9204,
+      "step": 4259
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.1766055045871554e-05,
+      "loss": 1.0697,
+      "step": 4260
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.1697247706422015e-05,
+      "loss": 0.9823,
+      "step": 4261
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.1628440366972476e-05,
+      "loss": 1.0943,
+      "step": 4262
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.155963302752294e-05,
+      "loss": 1.0037,
+      "step": 4263
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.1490825688073385e-05,
+      "loss": 1.0757,
+      "step": 4264
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.1422018348623847e-05,
+      "loss": 1.0198,
+      "step": 4265
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.135321100917431e-05,
+      "loss": 0.8978,
+      "step": 4266
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.128440366972477e-05,
+      "loss": 0.9337,
+      "step": 4267
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.1215596330275224e-05,
+      "loss": 0.8985,
+      "step": 4268
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.1146788990825685e-05,
+      "loss": 0.9095,
+      "step": 4269
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.1077981651376146e-05,
+      "loss": 0.8912,
+      "step": 4270
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.100917431192661e-05,
+      "loss": 0.7621,
+      "step": 4271
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.0940366972477055e-05,
+      "loss": 0.7871,
+      "step": 4272
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.0871559633027516e-05,
+      "loss": 0.7856,
+      "step": 4273
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.080275229357798e-05,
+      "loss": 0.8142,
+      "step": 4274
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.073394495412844e-05,
+      "loss": 0.8532,
+      "step": 4275
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.0665137614678893e-05,
+      "loss": 0.7337,
+      "step": 4276
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.0596330275229355e-05,
+      "loss": 0.7877,
+      "step": 4277
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.0527522935779816e-05,
+      "loss": 0.751,
+      "step": 4278
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.045871559633028e-05,
+      "loss": 0.6837,
+      "step": 4279
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.0389908256880725e-05,
+      "loss": 0.736,
+      "step": 4280
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.0321100917431186e-05,
+      "loss": 0.6137,
+      "step": 4281
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.025229357798165e-05,
+      "loss": 0.6763,
+      "step": 4282
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.018348623853211e-05,
+      "loss": 0.6719,
+      "step": 4283
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.011467889908256e-05,
+      "loss": 0.6475,
+      "step": 4284
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.0045871559633025e-05,
+      "loss": 0.5891,
+      "step": 4285
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.9977064220183486e-05,
+      "loss": 0.5518,
+      "step": 4286
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.990825688073394e-05,
+      "loss": 0.5259,
+      "step": 4287
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.9839449541284395e-05,
+      "loss": 0.414,
+      "step": 4288
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.9770642201834856e-05,
+      "loss": 0.4134,
+      "step": 4289
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.970183486238532e-05,
+      "loss": 0.5864,
+      "step": 4290
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.963302752293578e-05,
+      "loss": 1.6473,
+      "step": 4291
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.956422018348623e-05,
+      "loss": 1.5565,
+      "step": 4292
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.9495412844036694e-05,
+      "loss": 1.336,
+      "step": 4293
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.9426605504587156e-05,
+      "loss": 1.2156,
+      "step": 4294
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.935779816513761e-05,
+      "loss": 1.2194,
+      "step": 4295
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.9288990825688065e-05,
+      "loss": 1.1811,
+      "step": 4296
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.9220183486238526e-05,
+      "loss": 1.2588,
+      "step": 4297
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.915137614678899e-05,
+      "loss": 1.1802,
+      "step": 4298
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.908256880733945e-05,
+      "loss": 1.1653,
+      "step": 4299
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.90137614678899e-05,
+      "loss": 1.1736,
+      "step": 4300
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.8944954128440364e-05,
+      "loss": 1.2001,
+      "step": 4301
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.8876146788990826e-05,
+      "loss": 1.1908,
+      "step": 4302
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.880733944954128e-05,
+      "loss": 1.0737,
+      "step": 4303
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.8738532110091735e-05,
+      "loss": 1.0277,
+      "step": 4304
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.8669724770642196e-05,
+      "loss": 1.1035,
+      "step": 4305
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.860091743119266e-05,
+      "loss": 1.1159,
+      "step": 4306
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.853211009174312e-05,
+      "loss": 1.0814,
+      "step": 4307
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.846330275229357e-05,
+      "loss": 1.1012,
+      "step": 4308
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.8394495412844034e-05,
+      "loss": 0.9931,
+      "step": 4309
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.8325688073394495e-05,
+      "loss": 0.981,
+      "step": 4310
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.825688073394495e-05,
+      "loss": 1.042,
+      "step": 4311
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.8188073394495404e-05,
+      "loss": 0.9217,
+      "step": 4312
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.8119266055045866e-05,
+      "loss": 1.0243,
+      "step": 4313
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.805045871559633e-05,
+      "loss": 0.9812,
+      "step": 4314
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.798165137614679e-05,
+      "loss": 0.9574,
+      "step": 4315
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.791284403669724e-05,
+      "loss": 0.9598,
+      "step": 4316
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7844036697247704e-05,
+      "loss": 0.9847,
+      "step": 4317
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7775229357798165e-05,
+      "loss": 0.9159,
+      "step": 4318
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.770642201834862e-05,
+      "loss": 0.8966,
+      "step": 4319
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7637614678899074e-05,
+      "loss": 0.9458,
+      "step": 4320
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7568807339449535e-05,
+      "loss": 0.7979,
+      "step": 4321
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.75e-05,
+      "loss": 0.8373,
+      "step": 4322
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.743119266055046e-05,
+      "loss": 0.7266,
+      "step": 4323
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.736238532110091e-05,
+      "loss": 0.7917,
+      "step": 4324
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7293577981651374e-05,
+      "loss": 0.8114,
+      "step": 4325
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.722477064220183e-05,
+      "loss": 0.7576,
+      "step": 4326
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.715596330275229e-05,
+      "loss": 0.7017,
+      "step": 4327
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.708715596330275e-05,
+      "loss": 0.6844,
+      "step": 4328
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.701834862385321e-05,
+      "loss": 0.6881,
+      "step": 4329
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6949541284403667e-05,
+      "loss": 0.6883,
+      "step": 4330
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.688073394495413e-05,
+      "loss": 0.6264,
+      "step": 4331
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.681192660550458e-05,
+      "loss": 0.6504,
+      "step": 4332
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6743119266055044e-05,
+      "loss": 0.6262,
+      "step": 4333
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.66743119266055e-05,
+      "loss": 0.6282,
+      "step": 4334
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.660550458715596e-05,
+      "loss": 0.5891,
+      "step": 4335
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.653669724770642e-05,
+      "loss": 0.4968,
+      "step": 4336
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.646788990825688e-05,
+      "loss": 0.5388,
+      "step": 4337
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6399082568807336e-05,
+      "loss": 0.5422,
+      "step": 4338
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.63302752293578e-05,
+      "loss": 0.5698,
+      "step": 4339
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.626146788990825e-05,
+      "loss": 0.5959,
+      "step": 4340
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6192660550458713e-05,
+      "loss": 1.5995,
+      "step": 4341
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.612385321100917e-05,
+      "loss": 1.4782,
+      "step": 4342
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.605504587155963e-05,
+      "loss": 1.4813,
+      "step": 4343
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.598623853211009e-05,
+      "loss": 1.4207,
+      "step": 4344
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.591743119266055e-05,
+      "loss": 1.1731,
+      "step": 4345
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.5848623853211006e-05,
+      "loss": 1.179,
+      "step": 4346
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.577981651376147e-05,
+      "loss": 1.0944,
+      "step": 4347
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.571100917431192e-05,
+      "loss": 1.1699,
+      "step": 4348
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.564220183486238e-05,
+      "loss": 1.2315,
+      "step": 4349
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.557339449541284e-05,
+      "loss": 1.1727,
+      "step": 4350
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.55045871559633e-05,
+      "loss": 1.0606,
+      "step": 4351
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.543577981651376e-05,
+      "loss": 1.1568,
+      "step": 4352
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.536697247706422e-05,
+      "loss": 1.1571,
+      "step": 4353
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.5298165137614676e-05,
+      "loss": 0.9995,
+      "step": 4354
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.522935779816514e-05,
+      "loss": 1.0495,
+      "step": 4355
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.516055045871559e-05,
+      "loss": 1.0534,
+      "step": 4356
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.509174311926605e-05,
+      "loss": 0.9739,
+      "step": 4357
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.502293577981651e-05,
+      "loss": 0.9951,
+      "step": 4358
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.495412844036697e-05,
+      "loss": 1.0041,
+      "step": 4359
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.488532110091743e-05,
+      "loss": 0.977,
+      "step": 4360
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.481651376146789e-05,
+      "loss": 0.9856,
+      "step": 4361
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4747706422018346e-05,
+      "loss": 1.0146,
+      "step": 4362
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.467889908256881e-05,
+      "loss": 0.9208,
+      "step": 4363
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.461009174311926e-05,
+      "loss": 0.9545,
+      "step": 4364
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.454128440366972e-05,
+      "loss": 0.9631,
+      "step": 4365
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.447247706422018e-05,
+      "loss": 0.9365,
+      "step": 4366
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.440366972477064e-05,
+      "loss": 1.0486,
+      "step": 4367
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.43348623853211e-05,
+      "loss": 0.788,
+      "step": 4368
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.426605504587156e-05,
+      "loss": 0.814,
+      "step": 4369
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.4197247706422016e-05,
+      "loss": 0.8769,
+      "step": 4370
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.412844036697248e-05,
+      "loss": 0.7888,
+      "step": 4371
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.405963302752293e-05,
+      "loss": 0.8163,
+      "step": 4372
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.399082568807339e-05,
+      "loss": 0.8102,
+      "step": 4373
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.392201834862385e-05,
+      "loss": 0.8255,
+      "step": 4374
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.385321100917431e-05,
+      "loss": 0.7503,
+      "step": 4375
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.378440366972477e-05,
+      "loss": 0.6332,
+      "step": 4376
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.371559633027523e-05,
+      "loss": 0.7194,
+      "step": 4377
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.3646788990825686e-05,
+      "loss": 0.6134,
+      "step": 4378
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.357798165137615e-05,
+      "loss": 0.5928,
+      "step": 4379
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.35091743119266e-05,
+      "loss": 0.691,
+      "step": 4380
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.344036697247706e-05,
+      "loss": 0.6328,
+      "step": 4381
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.337155963302752e-05,
+      "loss": 0.6643,
+      "step": 4382
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.330275229357798e-05,
+      "loss": 0.5593,
+      "step": 4383
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.323394495412844e-05,
+      "loss": 0.5403,
+      "step": 4384
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.31651376146789e-05,
+      "loss": 0.6108,
+      "step": 4385
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.3096330275229355e-05,
+      "loss": 0.5947,
+      "step": 4386
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.302752293577982e-05,
+      "loss": 0.5018,
+      "step": 4387
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.295871559633027e-05,
+      "loss": 0.4964,
+      "step": 4388
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.288990825688073e-05,
+      "loss": 0.5018,
+      "step": 4389
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.282110091743119e-05,
+      "loss": 0.6679,
+      "step": 4390
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.275229357798165e-05,
+      "loss": 1.8382,
+      "step": 4391
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.268348623853211e-05,
+      "loss": 1.597,
+      "step": 4392
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.261467889908257e-05,
+      "loss": 1.3316,
+      "step": 4393
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2545871559633025e-05,
+      "loss": 1.1517,
+      "step": 4394
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2477064220183487e-05,
+      "loss": 1.2313,
+      "step": 4395
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.240825688073394e-05,
+      "loss": 1.2299,
+      "step": 4396
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.23394495412844e-05,
+      "loss": 1.1572,
+      "step": 4397
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.227064220183486e-05,
+      "loss": 1.2136,
+      "step": 4398
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.220183486238532e-05,
+      "loss": 1.1076,
+      "step": 4399
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.213302752293578e-05,
+      "loss": 1.1897,
+      "step": 4400
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.206422018348624e-05,
+      "loss": 1.1483,
+      "step": 4401
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1995412844036695e-05,
+      "loss": 1.0785,
+      "step": 4402
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1926605504587156e-05,
+      "loss": 0.9209,
+      "step": 4403
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.185779816513761e-05,
+      "loss": 0.9945,
+      "step": 4404
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.178899082568807e-05,
+      "loss": 1.1352,
+      "step": 4405
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1720183486238527e-05,
+      "loss": 1.0646,
+      "step": 4406
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.165137614678899e-05,
+      "loss": 0.9019,
+      "step": 4407
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.158256880733945e-05,
+      "loss": 0.9739,
+      "step": 4408
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.151376146788991e-05,
+      "loss": 1.0171,
+      "step": 4409
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1444954128440365e-05,
+      "loss": 1.0374,
+      "step": 4410
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1376146788990826e-05,
+      "loss": 0.9943,
+      "step": 4411
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.130733944954128e-05,
+      "loss": 0.9283,
+      "step": 4412
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.123853211009174e-05,
+      "loss": 0.9359,
+      "step": 4413
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1169724770642196e-05,
+      "loss": 0.9158,
+      "step": 4414
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.110091743119266e-05,
+      "loss": 0.848,
+      "step": 4415
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.103211009174312e-05,
+      "loss": 0.9724,
+      "step": 4416
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.096330275229358e-05,
+      "loss": 0.7664,
+      "step": 4417
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0894495412844035e-05,
+      "loss": 0.9433,
+      "step": 4418
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0825688073394496e-05,
+      "loss": 0.7457,
+      "step": 4419
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.075688073394495e-05,
+      "loss": 0.8187,
+      "step": 4420
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.068807339449541e-05,
+      "loss": 0.8369,
+      "step": 4421
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0619266055045866e-05,
+      "loss": 0.9324,
+      "step": 4422
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.055045871559633e-05,
+      "loss": 0.8421,
+      "step": 4423
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0481651376146785e-05,
+      "loss": 0.8011,
+      "step": 4424
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0412844036697247e-05,
+      "loss": 0.8359,
+      "step": 4425
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0344036697247705e-05,
+      "loss": 0.8032,
+      "step": 4426
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0275229357798166e-05,
+      "loss": 0.6778,
+      "step": 4427
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.020642201834862e-05,
+      "loss": 0.6758,
+      "step": 4428
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.013761467889908e-05,
+      "loss": 0.7229,
+      "step": 4429
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.006880733944954e-05,
+      "loss": 0.6191,
+      "step": 4430
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 0.5859,
+      "step": 4431
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9931192660550455e-05,
+      "loss": 0.7384,
+      "step": 4432
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9862385321100917e-05,
+      "loss": 0.5568,
+      "step": 4433
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9793577981651374e-05,
+      "loss": 0.621,
+      "step": 4434
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9724770642201832e-05,
+      "loss": 0.4893,
+      "step": 4435
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.965596330275229e-05,
+      "loss": 0.5297,
+      "step": 4436
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.958715596330275e-05,
+      "loss": 0.4829,
+      "step": 4437
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.951834862385321e-05,
+      "loss": 0.4738,
+      "step": 4438
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9449541284403667e-05,
+      "loss": 0.4318,
+      "step": 4439
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9380733944954125e-05,
+      "loss": 0.5377,
+      "step": 4440
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9311926605504586e-05,
+      "loss": 1.7343,
+      "step": 4441
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9243119266055044e-05,
+      "loss": 1.5264,
+      "step": 4442
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9174311926605502e-05,
+      "loss": 1.3026,
+      "step": 4443
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.910550458715596e-05,
+      "loss": 1.1369,
+      "step": 4444
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.903669724770642e-05,
+      "loss": 1.3275,
+      "step": 4445
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.896788990825688e-05,
+      "loss": 1.1836,
+      "step": 4446
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8899082568807337e-05,
+      "loss": 1.1694,
+      "step": 4447
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8830275229357795e-05,
+      "loss": 1.179,
+      "step": 4448
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8761467889908256e-05,
+      "loss": 1.0569,
+      "step": 4449
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8692660550458714e-05,
+      "loss": 1.1042,
+      "step": 4450
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8623853211009172e-05,
+      "loss": 1.1164,
+      "step": 4451
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.855504587155963e-05,
+      "loss": 1.1172,
+      "step": 4452
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.848623853211009e-05,
+      "loss": 1.0317,
+      "step": 4453
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.841743119266055e-05,
+      "loss": 1.1035,
+      "step": 4454
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8348623853211007e-05,
+      "loss": 1.0552,
+      "step": 4455
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8279816513761465e-05,
+      "loss": 1.2677,
+      "step": 4456
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8211009174311926e-05,
+      "loss": 0.9247,
+      "step": 4457
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8142201834862384e-05,
+      "loss": 1.042,
+      "step": 4458
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8073394495412842e-05,
+      "loss": 1.0111,
+      "step": 4459
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.80045871559633e-05,
+      "loss": 0.9646,
+      "step": 4460
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.793577981651376e-05,
+      "loss": 1.032,
+      "step": 4461
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.786697247706422e-05,
+      "loss": 0.8874,
+      "step": 4462
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7798165137614677e-05,
+      "loss": 0.9166,
+      "step": 4463
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7729357798165135e-05,
+      "loss": 1.0155,
+      "step": 4464
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7660550458715596e-05,
+      "loss": 0.9373,
+      "step": 4465
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7591743119266054e-05,
+      "loss": 0.9618,
+      "step": 4466
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.752293577981651e-05,
+      "loss": 0.9627,
+      "step": 4467
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.745412844036697e-05,
+      "loss": 0.8467,
+      "step": 4468
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.738532110091743e-05,
+      "loss": 0.9347,
+      "step": 4469
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7316513761467885e-05,
+      "loss": 0.8594,
+      "step": 4470
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7247706422018347e-05,
+      "loss": 0.8523,
+      "step": 4471
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7178899082568804e-05,
+      "loss": 0.8178,
+      "step": 4472
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7110091743119266e-05,
+      "loss": 0.8881,
+      "step": 4473
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.704128440366972e-05,
+      "loss": 0.6914,
+      "step": 4474
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.697247706422018e-05,
+      "loss": 0.7926,
+      "step": 4475
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.690366972477064e-05,
+      "loss": 0.811,
+      "step": 4476
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.68348623853211e-05,
+      "loss": 0.725,
+      "step": 4477
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.6766055045871555e-05,
+      "loss": 0.7814,
+      "step": 4478
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.6697247706422016e-05,
+      "loss": 0.6713,
+      "step": 4479
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.6628440366972474e-05,
+      "loss": 0.6762,
+      "step": 4480
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.6559633027522936e-05,
+      "loss": 0.6274,
+      "step": 4481
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.649082568807339e-05,
+      "loss": 0.6817,
+      "step": 4482
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.642201834862385e-05,
+      "loss": 0.601,
+      "step": 4483
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.635321100917431e-05,
+      "loss": 0.5633,
+      "step": 4484
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.628440366972477e-05,
+      "loss": 0.4726,
+      "step": 4485
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.6215596330275225e-05,
+      "loss": 0.4709,
+      "step": 4486
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.6146788990825686e-05,
+      "loss": 0.5094,
+      "step": 4487
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.6077981651376144e-05,
+      "loss": 0.511,
+      "step": 4488
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.6009174311926605e-05,
+      "loss": 0.4216,
+      "step": 4489
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.594036697247706e-05,
+      "loss": 0.4931,
+      "step": 4490
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.587155963302752e-05,
+      "loss": 1.599,
+      "step": 4491
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.580275229357798e-05,
+      "loss": 1.4101,
+      "step": 4492
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.573394495412844e-05,
+      "loss": 1.3726,
+      "step": 4493
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.5665137614678895e-05,
+      "loss": 1.3044,
+      "step": 4494
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.5596330275229356e-05,
+      "loss": 1.3374,
+      "step": 4495
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.5527522935779814e-05,
+      "loss": 1.196,
+      "step": 4496
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.5458715596330275e-05,
+      "loss": 1.2508,
+      "step": 4497
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.538990825688073e-05,
+      "loss": 1.1229,
+      "step": 4498
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.532110091743119e-05,
+      "loss": 1.0638,
+      "step": 4499
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.525229357798165e-05,
+      "loss": 1.0364,
+      "step": 4500
+    },
+    {
+      "epoch": 2.78,
+      "eval_bleu": 1.3741745601573323e-19,
+      "eval_loss": 1.776824951171875,
+      "eval_runtime": 2494.6636,
+      "eval_samples_per_second": 5.917,
+      "eval_steps_per_second": 0.74,
+      "step": 4500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.518348623853211e-05,
+      "loss": 1.0992,
+      "step": 4501
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.5114678899082565e-05,
+      "loss": 1.1023,
+      "step": 4502
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.5045871559633026e-05,
+      "loss": 1.0038,
+      "step": 4503
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4977064220183484e-05,
+      "loss": 1.0275,
+      "step": 4504
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4908256880733945e-05,
+      "loss": 1.1966,
+      "step": 4505
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.48394495412844e-05,
+      "loss": 1.0445,
+      "step": 4506
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.477064220183486e-05,
+      "loss": 0.988,
+      "step": 4507
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.470183486238532e-05,
+      "loss": 0.9769,
+      "step": 4508
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.463302752293578e-05,
+      "loss": 0.9518,
+      "step": 4509
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4564220183486234e-05,
+      "loss": 1.1336,
+      "step": 4510
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4495412844036696e-05,
+      "loss": 0.9184,
+      "step": 4511
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4426605504587154e-05,
+      "loss": 0.846,
+      "step": 4512
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4357798165137615e-05,
+      "loss": 1.0451,
+      "step": 4513
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.428899082568807e-05,
+      "loss": 0.9108,
+      "step": 4514
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.422018348623853e-05,
+      "loss": 0.8252,
+      "step": 4515
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.415137614678899e-05,
+      "loss": 0.9341,
+      "step": 4516
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.408256880733945e-05,
+      "loss": 0.8154,
+      "step": 4517
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4013761467889904e-05,
+      "loss": 0.8433,
+      "step": 4518
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3944954128440366e-05,
+      "loss": 0.8591,
+      "step": 4519
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3876146788990823e-05,
+      "loss": 0.8198,
+      "step": 4520
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3807339449541285e-05,
+      "loss": 0.839,
+      "step": 4521
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.373853211009174e-05,
+      "loss": 0.7614,
+      "step": 4522
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.36697247706422e-05,
+      "loss": 0.8214,
+      "step": 4523
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.360091743119266e-05,
+      "loss": 0.7016,
+      "step": 4524
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.353211009174312e-05,
+      "loss": 0.668,
+      "step": 4525
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3463302752293574e-05,
+      "loss": 0.7033,
+      "step": 4526
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3394495412844035e-05,
+      "loss": 0.7296,
+      "step": 4527
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3325688073394493e-05,
+      "loss": 0.6356,
+      "step": 4528
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3256880733944955e-05,
+      "loss": 0.5682,
+      "step": 4529
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.318807339449541e-05,
+      "loss": 0.649,
+      "step": 4530
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.311926605504587e-05,
+      "loss": 0.6647,
+      "step": 4531
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3050458715596328e-05,
+      "loss": 0.5715,
+      "step": 4532
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.298165137614679e-05,
+      "loss": 0.5845,
+      "step": 4533
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2912844036697244e-05,
+      "loss": 0.5445,
+      "step": 4534
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2844036697247705e-05,
+      "loss": 0.524,
+      "step": 4535
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2775229357798163e-05,
+      "loss": 0.5271,
+      "step": 4536
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2706422018348624e-05,
+      "loss": 0.5474,
+      "step": 4537
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.263761467889908e-05,
+      "loss": 0.3952,
+      "step": 4538
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.256880733944954e-05,
+      "loss": 0.4909,
+      "step": 4539
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2499999999999998e-05,
+      "loss": 0.5566,
+      "step": 4540
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.243119266055046e-05,
+      "loss": 1.7067,
+      "step": 4541
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2362385321100914e-05,
+      "loss": 1.3893,
+      "step": 4542
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2293577981651375e-05,
+      "loss": 1.2802,
+      "step": 4543
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2224770642201833e-05,
+      "loss": 1.1312,
+      "step": 4544
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2155963302752294e-05,
+      "loss": 1.1352,
+      "step": 4545
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.208715596330275e-05,
+      "loss": 1.2833,
+      "step": 4546
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.201834862385321e-05,
+      "loss": 1.1461,
+      "step": 4547
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1949541284403668e-05,
+      "loss": 1.1973,
+      "step": 4548
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.188073394495413e-05,
+      "loss": 1.1169,
+      "step": 4549
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1811926605504584e-05,
+      "loss": 1.0031,
+      "step": 4550
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1743119266055045e-05,
+      "loss": 1.071,
+      "step": 4551
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1674311926605503e-05,
+      "loss": 1.2425,
+      "step": 4552
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1605504587155964e-05,
+      "loss": 0.9873,
+      "step": 4553
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.153669724770642e-05,
+      "loss": 1.0584,
+      "step": 4554
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.146788990825688e-05,
+      "loss": 1.208,
+      "step": 4555
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1399082568807338e-05,
+      "loss": 1.1719,
+      "step": 4556
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.13302752293578e-05,
+      "loss": 1.0729,
+      "step": 4557
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1261467889908253e-05,
+      "loss": 1.0485,
+      "step": 4558
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1192660550458715e-05,
+      "loss": 1.036,
+      "step": 4559
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1123853211009173e-05,
+      "loss": 0.8543,
+      "step": 4560
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1055045871559634e-05,
+      "loss": 1.0383,
+      "step": 4561
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.098623853211009e-05,
+      "loss": 0.9774,
+      "step": 4562
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.091743119266055e-05,
+      "loss": 0.9882,
+      "step": 4563
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0848623853211008e-05,
+      "loss": 0.9675,
+      "step": 4564
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.077981651376147e-05,
+      "loss": 0.9406,
+      "step": 4565
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0711009174311923e-05,
+      "loss": 0.9818,
+      "step": 4566
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0642201834862385e-05,
+      "loss": 0.9217,
+      "step": 4567
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0573394495412842e-05,
+      "loss": 0.8368,
+      "step": 4568
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0504587155963304e-05,
+      "loss": 0.9413,
+      "step": 4569
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0435779816513758e-05,
+      "loss": 0.788,
+      "step": 4570
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.036697247706422e-05,
+      "loss": 0.9156,
+      "step": 4571
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0298165137614677e-05,
+      "loss": 0.8273,
+      "step": 4572
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.022935779816514e-05,
+      "loss": 0.6147,
+      "step": 4573
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0160550458715593e-05,
+      "loss": 0.7959,
+      "step": 4574
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0091743119266054e-05,
+      "loss": 0.8803,
+      "step": 4575
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0022935779816512e-05,
+      "loss": 0.6465,
+      "step": 4576
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.995412844036697e-05,
+      "loss": 0.76,
+      "step": 4577
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9885321100917428e-05,
+      "loss": 0.7173,
+      "step": 4578
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.981651376146789e-05,
+      "loss": 0.5321,
+      "step": 4579
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9747706422018347e-05,
+      "loss": 0.6674,
+      "step": 4580
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9678899082568805e-05,
+      "loss": 0.6767,
+      "step": 4581
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9610091743119263e-05,
+      "loss": 0.6509,
+      "step": 4582
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9541284403669724e-05,
+      "loss": 0.6361,
+      "step": 4583
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9472477064220182e-05,
+      "loss": 0.5187,
+      "step": 4584
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.940366972477064e-05,
+      "loss": 0.5913,
+      "step": 4585
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9334862385321098e-05,
+      "loss": 0.5777,
+      "step": 4586
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.926605504587156e-05,
+      "loss": 0.5197,
+      "step": 4587
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9197247706422017e-05,
+      "loss": 0.4075,
+      "step": 4588
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9128440366972475e-05,
+      "loss": 0.4522,
+      "step": 4589
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9059633027522933e-05,
+      "loss": 0.4476,
+      "step": 4590
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8990825688073394e-05,
+      "loss": 1.7788,
+      "step": 4591
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8922018348623852e-05,
+      "loss": 1.4156,
+      "step": 4592
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.885321100917431e-05,
+      "loss": 1.3459,
+      "step": 4593
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8784403669724768e-05,
+      "loss": 1.1863,
+      "step": 4594
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.871559633027523e-05,
+      "loss": 1.2492,
+      "step": 4595
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8646788990825687e-05,
+      "loss": 1.1565,
+      "step": 4596
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8577981651376145e-05,
+      "loss": 1.3033,
+      "step": 4597
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8509174311926606e-05,
+      "loss": 1.1299,
+      "step": 4598
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8440366972477064e-05,
+      "loss": 1.3174,
+      "step": 4599
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8371559633027522e-05,
+      "loss": 1.005,
+      "step": 4600
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.830275229357798e-05,
+      "loss": 1.1638,
+      "step": 4601
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.823394495412844e-05,
+      "loss": 1.028,
+      "step": 4602
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.81651376146789e-05,
+      "loss": 1.081,
+      "step": 4603
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8096330275229357e-05,
+      "loss": 1.0723,
+      "step": 4604
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8027522935779815e-05,
+      "loss": 1.0672,
+      "step": 4605
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7958715596330276e-05,
+      "loss": 0.9622,
+      "step": 4606
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7889908256880734e-05,
+      "loss": 0.9519,
+      "step": 4607
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.782110091743119e-05,
+      "loss": 0.912,
+      "step": 4608
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.775229357798165e-05,
+      "loss": 0.8903,
+      "step": 4609
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.768348623853211e-05,
+      "loss": 1.0945,
+      "step": 4610
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.761467889908257e-05,
+      "loss": 0.9083,
+      "step": 4611
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7545871559633027e-05,
+      "loss": 0.8767,
+      "step": 4612
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7477064220183484e-05,
+      "loss": 0.9596,
+      "step": 4613
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7408256880733946e-05,
+      "loss": 1.0027,
+      "step": 4614
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7339449541284404e-05,
+      "loss": 0.8588,
+      "step": 4615
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.727064220183486e-05,
+      "loss": 0.9063,
+      "step": 4616
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.720183486238532e-05,
+      "loss": 0.7942,
+      "step": 4617
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.713302752293578e-05,
+      "loss": 0.82,
+      "step": 4618
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.706422018348624e-05,
+      "loss": 0.8289,
+      "step": 4619
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6995412844036696e-05,
+      "loss": 0.8573,
+      "step": 4620
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6926605504587154e-05,
+      "loss": 0.7957,
+      "step": 4621
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6857798165137616e-05,
+      "loss": 0.8278,
+      "step": 4622
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6788990825688073e-05,
+      "loss": 0.8664,
+      "step": 4623
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.672018348623853e-05,
+      "loss": 0.6976,
+      "step": 4624
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.665137614678899e-05,
+      "loss": 0.7771,
+      "step": 4625
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.658256880733945e-05,
+      "loss": 0.6409,
+      "step": 4626
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.651376146788991e-05,
+      "loss": 0.624,
+      "step": 4627
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6444954128440366e-05,
+      "loss": 0.5996,
+      "step": 4628
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6376146788990824e-05,
+      "loss": 0.7131,
+      "step": 4629
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6307339449541285e-05,
+      "loss": 0.6636,
+      "step": 4630
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6238532110091743e-05,
+      "loss": 0.6364,
+      "step": 4631
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.61697247706422e-05,
+      "loss": 0.5541,
+      "step": 4632
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.610091743119266e-05,
+      "loss": 0.6066,
+      "step": 4633
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.603211009174312e-05,
+      "loss": 0.5724,
+      "step": 4634
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5963302752293578e-05,
+      "loss": 0.5396,
+      "step": 4635
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5894495412844036e-05,
+      "loss": 0.5158,
+      "step": 4636
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5825688073394494e-05,
+      "loss": 0.4567,
+      "step": 4637
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5756880733944955e-05,
+      "loss": 0.4934,
+      "step": 4638
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5688073394495413e-05,
+      "loss": 0.5748,
+      "step": 4639
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.561926605504587e-05,
+      "loss": 0.5306,
+      "step": 4640
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.555045871559633e-05,
+      "loss": 1.515,
+      "step": 4641
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.548165137614679e-05,
+      "loss": 1.4593,
+      "step": 4642
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5412844036697248e-05,
+      "loss": 1.348,
+      "step": 4643
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5344036697247706e-05,
+      "loss": 1.2722,
+      "step": 4644
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5275229357798164e-05,
+      "loss": 1.1665,
+      "step": 4645
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5206422018348623e-05,
+      "loss": 1.2146,
+      "step": 4646
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5137614678899083e-05,
+      "loss": 1.2046,
+      "step": 4647
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.506880733944954e-05,
+      "loss": 1.1509,
+      "step": 4648
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 1.2011,
+      "step": 4649
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4931192660550458e-05,
+      "loss": 1.1445,
+      "step": 4650
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4862385321100916e-05,
+      "loss": 1.0681,
+      "step": 4651
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4793577981651376e-05,
+      "loss": 1.0935,
+      "step": 4652
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4724770642201834e-05,
+      "loss": 1.0531,
+      "step": 4653
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4655963302752293e-05,
+      "loss": 1.1714,
+      "step": 4654
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4587155963302751e-05,
+      "loss": 1.0124,
+      "step": 4655
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.451834862385321e-05,
+      "loss": 1.0096,
+      "step": 4656
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4449541284403669e-05,
+      "loss": 1.0099,
+      "step": 4657
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4380733944954128e-05,
+      "loss": 0.9615,
+      "step": 4658
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4311926605504586e-05,
+      "loss": 0.9102,
+      "step": 4659
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4243119266055046e-05,
+      "loss": 1.0185,
+      "step": 4660
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4174311926605503e-05,
+      "loss": 1.0161,
+      "step": 4661
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4105504587155963e-05,
+      "loss": 1.0585,
+      "step": 4662
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4036697247706421e-05,
+      "loss": 0.9831,
+      "step": 4663
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.396788990825688e-05,
+      "loss": 0.9231,
+      "step": 4664
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3899082568807338e-05,
+      "loss": 0.8305,
+      "step": 4665
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3830275229357798e-05,
+      "loss": 0.8973,
+      "step": 4666
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3761467889908256e-05,
+      "loss": 0.9529,
+      "step": 4667
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3692660550458715e-05,
+      "loss": 0.8387,
+      "step": 4668
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3623853211009173e-05,
+      "loss": 0.7796,
+      "step": 4669
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3555045871559633e-05,
+      "loss": 0.8061,
+      "step": 4670
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.348623853211009e-05,
+      "loss": 0.7489,
+      "step": 4671
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.341743119266055e-05,
+      "loss": 0.7882,
+      "step": 4672
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3348623853211008e-05,
+      "loss": 0.6364,
+      "step": 4673
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3279816513761468e-05,
+      "loss": 0.7082,
+      "step": 4674
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3211009174311926e-05,
+      "loss": 0.7344,
+      "step": 4675
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3142201834862385e-05,
+      "loss": 0.8378,
+      "step": 4676
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3073394495412843e-05,
+      "loss": 0.7075,
+      "step": 4677
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3004587155963303e-05,
+      "loss": 0.6722,
+      "step": 4678
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.293577981651376e-05,
+      "loss": 0.5465,
+      "step": 4679
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.286697247706422e-05,
+      "loss": 0.5903,
+      "step": 4680
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2798165137614678e-05,
+      "loss": 0.6214,
+      "step": 4681
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2729357798165138e-05,
+      "loss": 0.617,
+      "step": 4682
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2660550458715595e-05,
+      "loss": 0.5606,
+      "step": 4683
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2591743119266055e-05,
+      "loss": 0.4589,
+      "step": 4684
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2522935779816513e-05,
+      "loss": 0.5961,
+      "step": 4685
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2454128440366973e-05,
+      "loss": 0.6141,
+      "step": 4686
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.238532110091743e-05,
+      "loss": 0.4514,
+      "step": 4687
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.231651376146789e-05,
+      "loss": 0.5933,
+      "step": 4688
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2247706422018348e-05,
+      "loss": 0.4379,
+      "step": 4689
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2178899082568807e-05,
+      "loss": 0.5371,
+      "step": 4690
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2110091743119265e-05,
+      "loss": 1.647,
+      "step": 4691
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2041284403669725e-05,
+      "loss": 1.3897,
+      "step": 4692
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1972477064220183e-05,
+      "loss": 1.2873,
+      "step": 4693
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1903669724770642e-05,
+      "loss": 1.2577,
+      "step": 4694
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.18348623853211e-05,
+      "loss": 1.2482,
+      "step": 4695
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.176605504587156e-05,
+      "loss": 1.2372,
+      "step": 4696
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1697247706422018e-05,
+      "loss": 1.29,
+      "step": 4697
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1628440366972477e-05,
+      "loss": 1.0405,
+      "step": 4698
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1559633027522935e-05,
+      "loss": 1.1388,
+      "step": 4699
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1490825688073395e-05,
+      "loss": 1.0377,
+      "step": 4700
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1422018348623853e-05,
+      "loss": 1.1141,
+      "step": 4701
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1353211009174312e-05,
+      "loss": 1.0284,
+      "step": 4702
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.128440366972477e-05,
+      "loss": 1.0036,
+      "step": 4703
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.121559633027523e-05,
+      "loss": 1.0257,
+      "step": 4704
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1146788990825688e-05,
+      "loss": 0.9324,
+      "step": 4705
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1077981651376147e-05,
+      "loss": 1.076,
+      "step": 4706
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1009174311926605e-05,
+      "loss": 1.0192,
+      "step": 4707
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0940366972477065e-05,
+      "loss": 1.1122,
+      "step": 4708
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0871559633027522e-05,
+      "loss": 1.0317,
+      "step": 4709
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0802752293577982e-05,
+      "loss": 0.8847,
+      "step": 4710
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.073394495412844e-05,
+      "loss": 1.0597,
+      "step": 4711
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.06651376146789e-05,
+      "loss": 1.056,
+      "step": 4712
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0596330275229357e-05,
+      "loss": 0.9446,
+      "step": 4713
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0527522935779817e-05,
+      "loss": 0.9083,
+      "step": 4714
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0458715596330275e-05,
+      "loss": 0.8394,
+      "step": 4715
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0389908256880734e-05,
+      "loss": 0.8397,
+      "step": 4716
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0321100917431192e-05,
+      "loss": 0.8365,
+      "step": 4717
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0252293577981652e-05,
+      "loss": 0.8737,
+      "step": 4718
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.018348623853211e-05,
+      "loss": 0.8334,
+      "step": 4719
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.011467889908257e-05,
+      "loss": 0.9145,
+      "step": 4720
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0045871559633027e-05,
+      "loss": 0.9948,
+      "step": 4721
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.977064220183485e-06,
+      "loss": 0.7694,
+      "step": 4722
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.908256880733945e-06,
+      "loss": 0.7595,
+      "step": 4723
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.839449541284403e-06,
+      "loss": 0.6905,
+      "step": 4724
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.770642201834862e-06,
+      "loss": 0.6627,
+      "step": 4725
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.70183486238532e-06,
+      "loss": 0.7632,
+      "step": 4726
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.63302752293578e-06,
+      "loss": 0.7008,
+      "step": 4727
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.564220183486237e-06,
+      "loss": 0.6408,
+      "step": 4728
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.495412844036697e-06,
+      "loss": 0.6336,
+      "step": 4729
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.426605504587155e-06,
+      "loss": 0.7248,
+      "step": 4730
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.357798165137614e-06,
+      "loss": 0.6665,
+      "step": 4731
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.288990825688072e-06,
+      "loss": 0.5397,
+      "step": 4732
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.220183486238532e-06,
+      "loss": 0.5865,
+      "step": 4733
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.15137614678899e-06,
+      "loss": 0.6092,
+      "step": 4734
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.08256880733945e-06,
+      "loss": 0.5846,
+      "step": 4735
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.013761467889907e-06,
+      "loss": 0.5967,
+      "step": 4736
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.944954128440367e-06,
+      "loss": 0.5171,
+      "step": 4737
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.876146788990825e-06,
+      "loss": 0.4322,
+      "step": 4738
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.807339449541284e-06,
+      "loss": 0.437,
+      "step": 4739
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.738532110091742e-06,
+      "loss": 0.4653,
+      "step": 4740
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.669724770642202e-06,
+      "loss": 1.597,
+      "step": 4741
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.60091743119266e-06,
+      "loss": 1.6638,
+      "step": 4742
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.53211009174312e-06,
+      "loss": 1.2833,
+      "step": 4743
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.463302752293577e-06,
+      "loss": 1.2478,
+      "step": 4744
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.394495412844037e-06,
+      "loss": 1.1786,
+      "step": 4745
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.325688073394495e-06,
+      "loss": 1.1102,
+      "step": 4746
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.256880733944954e-06,
+      "loss": 1.2521,
+      "step": 4747
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.188073394495412e-06,
+      "loss": 1.193,
+      "step": 4748
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.119266055045872e-06,
+      "loss": 0.9963,
+      "step": 4749
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.05045871559633e-06,
+      "loss": 1.1012,
+      "step": 4750
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.981651376146789e-06,
+      "loss": 1.0408,
+      "step": 4751
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.912844036697247e-06,
+      "loss": 1.0984,
+      "step": 4752
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.844036697247707e-06,
+      "loss": 0.9055,
+      "step": 4753
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.775229357798164e-06,
+      "loss": 0.9933,
+      "step": 4754
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.706422018348624e-06,
+      "loss": 0.9191,
+      "step": 4755
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.637614678899082e-06,
+      "loss": 0.8848,
+      "step": 4756
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.5688073394495415e-06,
+      "loss": 0.9051,
+      "step": 4757
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.499999999999999e-06,
+      "loss": 0.9039,
+      "step": 4758
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.431192660550458e-06,
+      "loss": 1.0504,
+      "step": 4759
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.362385321100917e-06,
+      "loss": 0.9298,
+      "step": 4760
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.2935779816513755e-06,
+      "loss": 0.8676,
+      "step": 4761
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.224770642201834e-06,
+      "loss": 0.9159,
+      "step": 4762
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.155963302752293e-06,
+      "loss": 1.0064,
+      "step": 4763
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.087155963302752e-06,
+      "loss": 0.8841,
+      "step": 4764
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.0183486238532104e-06,
+      "loss": 0.9083,
+      "step": 4765
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.949541284403669e-06,
+      "loss": 0.8359,
+      "step": 4766
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.880733944954128e-06,
+      "loss": 0.9571,
+      "step": 4767
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.811926605504587e-06,
+      "loss": 0.8202,
+      "step": 4768
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.743119266055045e-06,
+      "loss": 0.7132,
+      "step": 4769
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.674311926605504e-06,
+      "loss": 0.8124,
+      "step": 4770
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.605504587155963e-06,
+      "loss": 0.6306,
+      "step": 4771
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.5366972477064216e-06,
+      "loss": 0.8336,
+      "step": 4772
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.46788990825688e-06,
+      "loss": 0.7276,
+      "step": 4773
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.399082568807339e-06,
+      "loss": 0.7068,
+      "step": 4774
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.330275229357798e-06,
+      "loss": 0.7737,
+      "step": 4775
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.2614678899082565e-06,
+      "loss": 0.7396,
+      "step": 4776
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.192660550458715e-06,
+      "loss": 0.6719,
+      "step": 4777
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.123853211009174e-06,
+      "loss": 0.678,
+      "step": 4778
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.055045871559633e-06,
+      "loss": 0.6683,
+      "step": 4779
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.986238532110091e-06,
+      "loss": 0.5581,
+      "step": 4780
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.91743119266055e-06,
+      "loss": 0.4987,
+      "step": 4781
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.848623853211009e-06,
+      "loss": 0.6311,
+      "step": 4782
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.779816513761468e-06,
+      "loss": 0.5812,
+      "step": 4783
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.711009174311926e-06,
+      "loss": 0.5674,
+      "step": 4784
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.642201834862385e-06,
+      "loss": 0.436,
+      "step": 4785
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.573394495412844e-06,
+      "loss": 0.5947,
+      "step": 4786
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.5045871559633025e-06,
+      "loss": 0.4523,
+      "step": 4787
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.435779816513761e-06,
+      "loss": 0.5239,
+      "step": 4788
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.36697247706422e-06,
+      "loss": 0.4447,
+      "step": 4789
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.298165137614679e-06,
+      "loss": 0.5378,
+      "step": 4790
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.229357798165137e-06,
+      "loss": 1.77,
+      "step": 4791
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.160550458715596e-06,
+      "loss": 1.4126,
+      "step": 4792
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.091743119266055e-06,
+      "loss": 1.2132,
+      "step": 4793
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.022935779816514e-06,
+      "loss": 1.263,
+      "step": 4794
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.954128440366972e-06,
+      "loss": 1.3289,
+      "step": 4795
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.885321100917431e-06,
+      "loss": 1.1501,
+      "step": 4796
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.81651376146789e-06,
+      "loss": 1.2185,
+      "step": 4797
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.7477064220183485e-06,
+      "loss": 1.11,
+      "step": 4798
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.678899082568807e-06,
+      "loss": 1.1127,
+      "step": 4799
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.610091743119266e-06,
+      "loss": 1.0948,
+      "step": 4800
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.541284403669725e-06,
+      "loss": 1.0986,
+      "step": 4801
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4724770642201834e-06,
+      "loss": 1.0068,
+      "step": 4802
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.403669724770642e-06,
+      "loss": 0.9905,
+      "step": 4803
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.334862385321101e-06,
+      "loss": 1.0702,
+      "step": 4804
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.26605504587156e-06,
+      "loss": 1.031,
+      "step": 4805
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.197247706422018e-06,
+      "loss": 1.066,
+      "step": 4806
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.128440366972477e-06,
+      "loss": 0.991,
+      "step": 4807
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.059633027522936e-06,
+      "loss": 0.9512,
+      "step": 4808
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.9908256880733945e-06,
+      "loss": 1.0002,
+      "step": 4809
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.922018348623853e-06,
+      "loss": 0.9506,
+      "step": 4810
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.853211009174312e-06,
+      "loss": 0.8638,
+      "step": 4811
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.7844036697247707e-06,
+      "loss": 0.9749,
+      "step": 4812
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.715596330275229e-06,
+      "loss": 0.9649,
+      "step": 4813
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6467889908256878e-06,
+      "loss": 0.8234,
+      "step": 4814
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5779816513761465e-06,
+      "loss": 0.931,
+      "step": 4815
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5091743119266052e-06,
+      "loss": 0.8197,
+      "step": 4816
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.440366972477064e-06,
+      "loss": 0.8216,
+      "step": 4817
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.3715596330275227e-06,
+      "loss": 0.8401,
+      "step": 4818
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.3027522935779814e-06,
+      "loss": 0.8022,
+      "step": 4819
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.23394495412844e-06,
+      "loss": 0.7684,
+      "step": 4820
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.165137614678899e-06,
+      "loss": 0.9382,
+      "step": 4821
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0963302752293576e-06,
+      "loss": 0.8536,
+      "step": 4822
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0275229357798163e-06,
+      "loss": 0.7438,
+      "step": 4823
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.958715596330275e-06,
+      "loss": 0.7,
+      "step": 4824
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.889908256880734e-06,
+      "loss": 0.7312,
+      "step": 4825
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8211009174311925e-06,
+      "loss": 0.6984,
+      "step": 4826
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7522935779816512e-06,
+      "loss": 0.7788,
+      "step": 4827
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.68348623853211e-06,
+      "loss": 0.6525,
+      "step": 4828
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6146788990825687e-06,
+      "loss": 0.7049,
+      "step": 4829
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5458715596330274e-06,
+      "loss": 0.6016,
+      "step": 4830
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.477064220183486e-06,
+      "loss": 0.6061,
+      "step": 4831
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.408256880733945e-06,
+      "loss": 0.6008,
+      "step": 4832
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3394495412844036e-06,
+      "loss": 0.506,
+      "step": 4833
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2706422018348624e-06,
+      "loss": 0.4979,
+      "step": 4834
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.201834862385321e-06,
+      "loss": 0.4849,
+      "step": 4835
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.13302752293578e-06,
+      "loss": 0.468,
+      "step": 4836
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0642201834862385e-06,
+      "loss": 0.5054,
+      "step": 4837
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9954128440366973e-06,
+      "loss": 0.5039,
+      "step": 4838
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.926605504587156e-06,
+      "loss": 0.4429,
+      "step": 4839
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8577981651376145e-06,
+      "loss": 0.4599,
+      "step": 4840
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7889908256880732e-06,
+      "loss": 1.5099,
+      "step": 4841
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.720183486238532e-06,
+      "loss": 1.2594,
+      "step": 4842
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6513761467889907e-06,
+      "loss": 1.1815,
+      "step": 4843
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5825688073394494e-06,
+      "loss": 1.0883,
+      "step": 4844
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5137614678899082e-06,
+      "loss": 1.1598,
+      "step": 4845
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.444954128440367e-06,
+      "loss": 0.9655,
+      "step": 4846
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3761467889908256e-06,
+      "loss": 0.8867,
+      "step": 4847
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3073394495412844e-06,
+      "loss": 0.9597,
+      "step": 4848
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.238532110091743e-06,
+      "loss": 1.0178,
+      "step": 4849
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1697247706422018e-06,
+      "loss": 0.8696,
+      "step": 4850
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1009174311926605e-06,
+      "loss": 0.906,
+      "step": 4851
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0321100917431193e-06,
+      "loss": 0.8132,
+      "step": 4852
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.63302752293578e-07,
+      "loss": 0.8195,
+      "step": 4853
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.944954128440366e-07,
+      "loss": 0.7609,
+      "step": 4854
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.256880733944954e-07,
+      "loss": 0.7066,
+      "step": 4855
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.568807339449541e-07,
+      "loss": 0.5607,
+      "step": 4856
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.880733944954128e-07,
+      "loss": 0.6017,
+      "step": 4857
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.192660550458715e-07,
+      "loss": 0.473,
+      "step": 4858
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.504587155963303e-07,
+      "loss": 0.4602,
+      "step": 4859
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.81651376146789e-07,
+      "loss": 0.4551,
+      "step": 4860
+    },
+    {
+      "epoch": 3.0,
+      "step": 4860,
+      "total_flos": 0.0,
+      "train_loss": 2.012981654225308,
+      "train_runtime": 68108.5497,
+      "train_samples_per_second": 9.134,
+      "train_steps_per_second": 0.071
+    }
+  ],
+  "max_steps": 4860,
+  "num_train_epochs": 3,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}